1 2017-12-28 Andreas Schwab <schwab@linux-m68k.org>
3 * config/m68k/m68k.md (ashrdi3_const1, lshrdi3_const1): Add
6 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
8 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
9 (aarch64_ld1x2<VDC:mode>): Likewise.
10 (aarch64_simd_ld1<mode>_x2): Likewise.
11 (aarch64_simd_ld1<mode>_x2): Likewise.
12 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
13 (vld1_s8_x2): Likewise.
14 (vld1_u16_x2): Likewise.
15 (vld1_s16_x2): Likewise.
16 (vld1_u32_x2): Likewise.
17 (vld1_s32_x2): Likewise.
18 (vld1_u64_x2): Likewise.
19 (vld1_s64_x2): Likewise.
20 (vld1_f16_x2): Likewise.
21 (vld1_f32_x2): Likewise.
22 (vld1_f64_x2): Likewise.
23 (vld1_p8_x2): Likewise.
24 (vld1_p16_x2): Likewise.
25 (vld1_p64_x2): Likewise.
26 (vld1q_u8_x2): Likewise.
27 (vld1q_s8_x2): Likewise.
28 (vld1q_u16_x2): Likewise.
29 (vld1q_s16_x2): Likewise.
30 (vld1q_u32_x2): Likewise.
31 (vld1q_s32_x2): Likewise.
32 (vld1q_u64_x2): Likewise.
33 (vld1q_s64_x2): Likewise.
34 (vld1q_f16_x2): Likewise.
35 (vld1q_f32_x2): Likewise.
36 (vld1q_f64_x2): Likewise.
37 (vld1q_p8_x2): Likewise.
38 (vld1q_p16_x2): Likewise.
39 (vld1q_p64_x2): Likewise.
41 2017-12-27 Martin Liska <mliska@suse.cz>
43 PR tree-optimization/83552
44 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
45 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
47 2017-12-27 Tom de Vries <tom@codesourcery.com>
49 * config/nvptx/nvptx.c (nvptx_option_override): Disable
50 -gstatement-frontiers.
52 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
54 PR rtl-optimization/83513
55 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
56 before priority comparison.
58 2017-12-25 Jakub Jelinek <jakub@redhat.com>
61 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
62 ix86_isa_flags2 to ix86_isa_flags.
63 * config/i386/i386-c.c (ix86_target_macros_internal): Test
64 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
65 isa_flags rather than isa_flags2.
66 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
67 and -mavx512bitalg from isa2_opts to isa_opts.
68 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
69 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
70 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
72 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
73 OPTION_MASK_ISA_AVX512F ored with another option similarly to
74 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
75 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
76 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
77 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
78 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
79 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
80 builtins from bdesc_args2 to bdesc_args section.
81 (__builtin_ia32_compressstoreuqi512_mask,
82 __builtin_ia32_compressstoreuhi512_mask,
83 __builtin_ia32_compressstoreuqi256_mask,
84 __builtin_ia32_expandloadqi512_mask,
85 __builtin_ia32_expandloadqi512_maskz,
86 __builtin_ia32_expandloadhi512_mask,
87 __builtin_ia32_expandloadhi512_maskz,
88 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
89 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
90 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
91 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
92 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
93 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
94 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
95 __builtin_ia32_vpshldv_v32hi_maskz,
96 __builtin_ia32_vpopcountb_v64qi_mask,
97 __builtin_ia32_vpopcountw_v32hi_mask,
98 __builtin_ia32_vpshufbitqmb512_mask,
99 __builtin_ia32_vpshufbitqmb256_mask): Add
100 " | OPTION_MASK_ISA_AVX512BW".
101 (__builtin_ia32_expandloadqi256_mask,
102 __builtin_ia32_expandloadqi256_maskz,
103 __builtin_ia32_vpopcountb_v32qi_mask): Add
104 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
105 (__builtin_ia32_expandloadhi256_mask,
106 __builtin_ia32_expandloadhi256_maskz,
107 __builtin_ia32_expandloadqi128_mask,
108 __builtin_ia32_expandloadqi128_maskz,
109 __builtin_ia32_expandloadhi128_mask,
110 __builtin_ia32_expandloadhi128_maskz,
111 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
112 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
113 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
114 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
115 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
116 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
117 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
118 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
119 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
120 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
121 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
122 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
123 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
124 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
125 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
126 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
127 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
128 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
129 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
130 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
131 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
132 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
133 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
134 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
135 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
136 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
137 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
138 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
139 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
140 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
141 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
142 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
143 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
144 __builtin_ia32_vpopcountw_v8hi_mask): Add
145 " | OPTION_MASK_ISA_AVX512VL".
146 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
147 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
148 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
149 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
150 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
151 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
152 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
153 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
154 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
155 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
156 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
157 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
158 _mm_mask_bitshuffle_epi64_mask): Likewise.
159 * common/config/i386/i386-common.c
160 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
161 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
162 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
163 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
164 OPTION_MASK_ISA_AVX512BITALG_UNSET.
165 (OPTION_MASK_ISA2_AVX512F_UNSET,
166 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
167 (ix86_handle_option): For -mno-general-regs-only, clear from
168 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
169 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
170 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
171 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
172 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
173 for moving from ix86_isa_flags2 to ix86_isa_flags.
175 2017-12-23 Jakub Jelinek <jakub@redhat.com>
178 * fold-const.c (struct contains_label_data): New type.
179 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
180 inside of a SWITCH_BODY seen during the walk.
181 (contains_label_p): Use walk_tree instead of
182 walk_tree_without_duplicates, prepare data for contains_label_1 and
185 2017-12-22 Jakub Jelinek <jakub@redhat.com>
188 * tree-iterator.c (alloc_stmt_list): Start with cleared
189 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
193 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
194 enable avx512f explicitly in #pragma GCC target.
195 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
196 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
197 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
198 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
199 __builtin_ia32_vpdpbusds_v8si_mask,
200 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
201 __builtin_ia32_vpdpbusds_v4si_mask,
202 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
203 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
204 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
205 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
206 __builtin_ia32_vpdpwssds_v8si_mask,
207 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
208 __builtin_ia32_vpdpwssds_v4si_mask,
209 __builtin_ia32_vpdpwssds_v4si_maskz): Use
210 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
211 just OPTION_MASK_ISA_AVX512VNNI.
213 2017-12-22 Martin Jambor <mjambor@suse.cz>
216 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
219 2017-12-22 Julia Koval <julia.koval@intel.com>
220 Sebastian Peryt <sebastian.peryt@intel.com>
222 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
223 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
224 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
225 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
226 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
227 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
228 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
229 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
230 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
231 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
232 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
233 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
234 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
235 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
236 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
237 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
238 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
239 _mm256_maskz_popcnt_epi64): New intrinsics.
240 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
241 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
242 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
243 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
244 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
245 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
246 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
247 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
248 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
249 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
250 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
251 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
252 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
253 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
254 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
255 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
256 __builtin_ia32_vpshufbitqmb256_mask,
257 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
258 * config/i386/i386-c.c (__AVX512BITALG__): New.
259 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
260 (ix86_valid_target_attribute_inner_p): Ditto.
261 (ix86_expand_args_builtin): Handle new types.
262 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
263 * config/i386/i386.opt: Add -mavx512bitalg.
264 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
265 avx512bitalgintrin.h.
266 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
267 (vpopcount<mode><mask_name>): Add more types.
268 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
269 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
271 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
273 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
274 Or in OPTION_MASK_ISA_AVX512F_SET.
275 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
276 OPTION_MASK_ISA_AVX512VNNI_UNSET.
277 (ix86_handle_option): Adjust for
278 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
279 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
281 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
282 isa_flag instead of isa_flag2.
283 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
284 isa_opts2 to isa_opts.
285 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
288 2017-12-22 Mike Stump <mikestump@comcast.net>
289 Eric Botcazou <ebotcazou@adacore.com>
291 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
293 2017-12-21 Jakub Jelinek <jakub@redhat.com>
296 * config/i386/i386.c (ix86_function_arg_boundary): Return
297 PARM_BOUNDARY for TYPE_EMPTY_P types.
300 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
301 if navail is >= dir.len.
303 2017-12-21 Steve Ellcey <sellcey@cavium.com>
305 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
308 2017-12-21 Jakub Jelinek <jakub@redhat.com>
310 PR rtl-optimization/80747
311 PR rtl-optimization/83512
312 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
313 succ edge from ENTRY, copy partition from e->dest to the newly
315 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
316 ENTRY, use BB_PARTITION of its successor block as current_partition.
317 Don't copy partition when splitting succ edge from ENTRY.
319 PR tree-optimization/83523
320 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
321 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
322 (convert_mult_to_fma): Likewise.
324 PR tree-optimization/83521
325 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
326 gimple_build_assign without code on result of
327 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
330 2017-12-21 Andrew Pinski <apinski@cavium.com>
331 Steve Ellcey <sellcey@cavium.com>
333 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
334 multi-arch for ilp32.
336 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
339 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
340 constraints to operand 2.
341 (*ashl<mode>3_mask_1): Ditto.
342 (*<shift_insn><mode>3_mask): Ditto.
343 (*<shift_insn><mode>3_mask_1): Ditto.
344 (*<rotate_insn><mode>3_mask): Ditto.
345 (*<rotate_insn><mode>3_mask_1): Ditto.
347 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
349 * reorg.c (make_return_insns): Reemit each insn with its own location.
351 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
353 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
354 generation for cases where splatting a value is not useful.
355 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
356 across a vec_duplicate and a paradoxical subreg forming a vector
357 mode to a vec_concat.
359 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
361 * combine.c (simplify_set): Do not transform subregs to zero_extends
362 if the destination is not a scalar int mode.
364 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
367 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
368 shared zero if the input has overflowed.
370 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
372 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
373 support for Cortex-A55 and Cortex-A75.
375 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
377 * common/config/arm/arm-common.c (compare_opt_names): Add function
378 comment. Use strcmp instead of manual loop.
380 2017-12-21 Martin Liska <mliska@suse.cz>
382 PR gcov-profile/83509
383 * gcov-dump.c (dump_gcov_file): Do not read info about
384 support_unexecuted_blocks for gcda files.
386 2017-12-21 Jakub Jelinek <jakub@redhat.com>
388 PR rtl-optimization/82973
389 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
390 (valid_for_const_vector_p): ... this.
391 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
392 (valid_for_const_vector_p): ... this. Adjust function comment.
393 (gen_vec_duplicate): Adjust caller.
394 * optabs.c (expand_vector_broadcast): Likewise.
395 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
396 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
398 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
399 instead of GET_CODE == CONST_FIXED.
400 (simplify_subreg): Use CONST_FIXED_P macro instead of
401 GET_CODE == CONST_FIXED.
404 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
405 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
406 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
407 (ix86_option_override_internal): Adjust for
408 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
409 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
410 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
411 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
412 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
413 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
414 def_builtin2 for CET builtins.
415 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
416 up formatting in IX86_BUILTIN_RDPID code.
417 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
418 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
419 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
420 ix86_isa_flags2 to ix86_isa_flags.
421 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
423 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
424 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
425 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
427 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
428 Or in OPTION_MASK_ISA_AVX512F_SET.
429 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
430 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
431 (ix86_handle_option): Adjust for
432 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
433 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
435 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
436 Alan Hayward <alan.hayward@arm.com>
437 David Sherwood <david.sherwood@arm.com>
439 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
440 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
441 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
442 Track polynomial offsets.
444 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
445 Alan Hayward <alan.hayward@arm.com>
446 David Sherwood <david.sherwood@arm.com>
448 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
449 Treat drb->init as a poly_int. Fail if its misalignment wrt
450 vector_alignment isn't known.
452 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
453 Alan Hayward <alan.hayward@arm.com>
454 David Sherwood <david.sherwood@arm.com>
456 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
458 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
459 (vect_loop_versioning): Take the loop versioning threshold as a
461 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
462 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
463 versioning_threshold.
464 (vect_analyze_loop_2): Compute the loop versioning threshold
465 whenever loop versioning is needed, and store it in the new
466 field rather than combining it with the cost model threshold.
467 (vect_transform_loop): Update call to vect_loop_versioning.
468 Try to combine the loop versioning and cost thresholds here.
470 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
471 Alan Hayward <alan.hayward@arm.com>
472 David Sherwood <david.sherwood@arm.com>
474 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
475 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
476 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
477 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
478 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
479 (group_compare_offset): Update accordingly.
480 (split_small_address_groups_p): Likewise.
481 (record_use): Take addr_offset as a poly_uint64 rather than
482 an unsigned HOST_WIDE_INT.
483 (strip_offset): Return the offset as a poly_uint64 rather than
484 an unsigned HOST_WIDE_INT.
485 (record_group_use, split_address_groups): Track polynomial offsets.
486 (add_iv_candidate_for_use): Likewise.
487 (addr_offset_valid_p): Take the offset as a poly_int64 rather
488 than a HOST_WIDE_INT.
489 (strip_offset_1): Return the offset as a poly_int64 rather than
492 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
493 Alan Hayward <alan.hayward@arm.com>
494 David Sherwood <david.sherwood@arm.com>
496 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
497 rather than a HOST_WIDE_INT.
498 * tree.c (get_binfo_at_offset): Likewise.
500 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
501 Alan Hayward <alan.hayward@arm.com>
502 David Sherwood <david.sherwood@arm.com>
504 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
505 rather than a HOST_WIDE_INT.
506 * tree-sra.c (build_ref_for_offset): Likewise.
508 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
509 Alan Hayward <alan.hayward@arm.com>
510 David Sherwood <david.sherwood@arm.com>
512 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
514 * tree.c (mem_ref_offset): Likewise.
515 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
516 * builtins.c (get_object_alignment_2): Likewise.
517 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
518 * gimple-fold.c (get_base_constructor): Likewise.
519 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
520 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
522 * ipa-polymorphic-call.c
523 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
524 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
525 (get_ancestor_addr_info): Likewise.
526 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
527 * match.pd: Likewise.
528 * tree-data-ref.c (dr_analyze_innermost): Likewise.
529 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
530 * tree-eh.c (tree_could_trap_p): Likewise.
531 * tree-object-size.c (addr_object_size): Likewise.
532 * tree-ssa-address.c (copy_ref_info): Likewise.
533 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
534 (indirect_refs_may_alias_p): Likewise.
535 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
536 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
537 (non_rewritable_mem_ref_base): Likewise.
538 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
539 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
540 * varasm.c (decode_addr_const): Likewise.
542 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
543 Alan Hayward <alan.hayward@arm.com>
544 David Sherwood <david.sherwood@arm.com>
546 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
547 offsets for MEM_REFs.
549 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
550 Alan Hayward <alan.hayward@arm.com>
551 David Sherwood <david.sherwood@arm.com>
553 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
554 as a poly_uint64 rather than an unsigned int.
555 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
557 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
558 Alan Hayward <alan.hayward@arm.com>
559 David Sherwood <david.sherwood@arm.com>
561 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
563 (decode_addr_const): Update accordingly.
565 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
566 Alan Hayward <alan.hayward@arm.com>
567 David Sherwood <david.sherwood@arm.com>
569 * tree.h (bit_field_size, bit_field_offset): New functions.
570 * hsa-gen.c (gen_hsa_addr): Use them.
571 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
572 (simplify_vector_constructor): Likewise.
573 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
574 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
575 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
576 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
579 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
580 Alan Hayward <alan.hayward@arm.com>
581 David Sherwood <david.sherwood@arm.com>
583 * expr.h (emit_group_load, emit_group_load_into_temps)
584 (emit_group_store): Take the size as a poly_int64 rather than an int.
585 * expr.c (emit_group_load_1, emit_group_load): Likewise.
586 (emit_group_load_into_temp, emit_group_store): Likewise.
588 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
589 Alan Hayward <alan.hayward@arm.com>
590 David Sherwood <david.sherwood@arm.com>
592 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
593 an unsigned int to a poly_uint64.
594 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
595 sizes as poly_uint64s rather than unsigned ints.
596 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
599 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
600 Alan Hayward <alan.hayward@arm.com>
601 David Sherwood <david.sherwood@arm.com>
603 * lra-constraints.c (emit_inc): Change inc_amount from an int
606 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
607 Alan Hayward <alan.hayward@arm.com>
608 David Sherwood <david.sherwood@arm.com>
610 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
612 (add_stack_var, stack_var_cmp, partition_stack_vars)
613 (dump_stack_var_partition): Update accordingly.
614 (alloc_stack_frame_space): Take the size as a poly_int64 rather
615 than a HOST_WIDE_INT.
616 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
617 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
618 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
619 rather than a HOST_WIDE_INT.
621 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
622 Alan Hayward <alan.hayward@arm.com>
623 David Sherwood <david.sherwood@arm.com>
625 * target.def (return_pops_args): Treat both the input and output
626 sizes as poly_int64s rather than HOST_WIDE_INTS.
627 * targhooks.h (default_return_pops_args): Update accordingly.
628 * targhooks.c (default_return_pops_args): Likewise.
629 * doc/tm.texi: Regenerate.
630 * emit-rtl.h (incoming_args): Change pops_args, size and
631 outgoing_args_size from int to poly_int64_pod.
632 * function.h (expr_status): Change x_pending_stack_adjust and
633 x_stack_pointer_delta from int to poly_int64.
634 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
635 (ARGS_SIZE_RTX): Update accordingly.
636 * calls.c (highest_outgoing_arg_in_use): Change from int to
638 (stack_usage_watermark, stored_args_watermark): New variables.
639 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
640 (emit_call_1): Change the stack_size and rounded_stack_size
641 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
643 (save_fixed_argument_area): Check stack_usage_watermark.
644 (initialize_argument_information): Change old_pending_adj from
645 a HOST_WIDE_INT * to a poly_int64_pod *.
646 (compute_argument_block_size): Return the size as a poly_int64
648 (finalize_must_preallocate): Track polynomial argument sizes.
649 (compute_argument_addresses): Likewise.
650 (internal_arg_pointer_based_exp): Track polynomial offsets.
651 (mem_overlaps_already_clobbered_arg_p): Rename to...
652 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
653 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
654 Check stored_args_used_watermark.
655 (load_register_parameters): Update accordingly.
656 (check_sibcall_argument_overlap_1): Likewise.
657 (combine_pending_stack_adjustment_and_call): Take the unadjusted
658 args size as a poly_int64 rather than an int. Return a bool
659 indicating whether the optimization was possible and return
660 the new adjustment by reference.
661 (check_sibcall_argument_overlap): Track polynomail argument sizes.
662 Update stored_args_watermark.
663 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
664 (expand_call): Likewise. Maintain stack_usage_watermark and
665 stored_args_watermark. Update calls to
666 combine_pending_stack_adjustment_and_call.
667 (emit_library_call_value_1): Handle polynomial argument sizes.
668 Call stack_region_maybe_used_p and mark_stack_region_used.
669 Maintain stack_usage_watermark.
670 (store_one_arg): Likewise. Update call to
671 mem_overlaps_already_clobbered_arg_p.
672 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
674 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
675 * config/microblaze/microblaze.c (microblaze_function_prologue):
677 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
678 TARGET_RETURN_POPS_ARGS interface.
679 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
681 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
682 * config/i386/i386.c (ix86_return_pops_args): Update for new
683 TARGET_RETURN_POPS_ARGS interface.
684 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
685 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
686 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
687 TARGET_RETURN_POPS_ARGS interface.
688 * config/vax/vax.c (vax_return_pops_args): Likewise.
689 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
690 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
691 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
692 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
693 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
694 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
695 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
696 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
697 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
698 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
699 and x_stack_pointer_delta from int to poly_int64.
700 * dojump.c (do_pending_stack_adjust): Update accordingly.
701 * explow.c (allocate_dynamic_stack_space): Handle polynomial
702 stack_pointer_deltas.
703 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
704 (pad_to_arg_alignment): Track polynomial offsets.
705 (assign_parm_find_stack_rtl): Likewise.
706 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
707 * toplev.c (output_stack_usage): Update reference to
708 current_function_pushed_stack_size.
710 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
711 Alan Hayward <alan.hayward@arm.com>
712 David Sherwood <david.sherwood@arm.com>
714 * function.c (in_arg_offset, var_offset, dynamic_offset)
715 (out_arg_offset, cfa_offset): Change from int to poly_int64.
716 (instantiate_new_reg): Return the new offset as a poly_int64_pod
717 rather than a HOST_WIDE_INT.
718 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
719 (instantiate_virtual_regs_in_insn): Likewise.
721 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
722 Alan Hayward <alan.hayward@arm.com>
723 David Sherwood <david.sherwood@arm.com>
725 * rtl.h (get_args_size, add_args_size_note): New functions.
726 (find_args_size_adjust): Return a poly_int64 rather than a
728 (fixup_args_size_notes): Likewise. Make the same change to the
729 end_args_size parameter.
730 * rtlanal.c (get_args_size, add_args_size_note): New functions.
731 * builtins.c (expand_builtin_trap): Use add_args_size_note.
732 * calls.c (emit_call_1): Likewise.
733 * explow.c (adjust_stack_1): Likewise.
734 * cfgcleanup.c (old_insns_match_p): Update use of
735 find_args_size_adjust.
736 * combine.c (distribute_notes): Track polynomial arg sizes.
737 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
738 end_true_args_size, beg_delay_args_size and end_delay_args_size
739 from HOST_WIDE_INT to poly_int64.
740 (add_cfi_args_size): Take the args_size as a poly_int64 rather
741 than a HOST_WIDE_INT.
742 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
743 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
744 polynomial arg sizes.
745 * emit-rtl.c (try_split): Use get_args_size.
746 * recog.c (peep2_attempt): Likewise.
747 * reload1.c (reload_as_needed): Likewise.
748 * expr.c (find_args_size_adjust): Return the adjustment as a
749 poly_int64 rather than a HOST_WIDE_INT.
750 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
751 to a poly_int64 and change the return type in the same way.
752 (emit_single_push_insn): Track polynomial arg sizes.
754 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
755 Alan Hayward <alan.hayward@arm.com>
756 David Sherwood <david.sherwood@arm.com>
758 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
759 from HOST_WIDE_INT to poly_int64.
760 * expr.c (push_block, emit_push_insn): Likewise.
762 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
763 Alan Hayward <alan.hayward@arm.com>
764 David Sherwood <david.sherwood@arm.com>
766 * function.h (frame_space): Change start and length from HOST_WIDE_INT
768 (get_frame_size): Return the size as a poly_int64 rather than a
770 (frame_offset_overflow): Take the offset as a poly_int64 rather
771 than a HOST_WIDE_INT.
772 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
773 (assign_stack_temp): Likewise for the size.
774 * function.c (get_frame_size): Return a poly_int64 rather than
776 (frame_offset_overflow): Take the offset as a poly_int64 rather
777 than a HOST_WIDE_INT.
778 (try_fit_stack_local): Take the start, length and size as poly_int64s
779 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
780 rather than a HOST_WIDE_INT.
781 (add_frame_space): Take the start and end as poly_int64s rather than
783 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
784 (assign_stack_temp): Likewise for the size.
785 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
787 (find_temp_slot_from_address): Handle polynomial offsets.
788 (combine_temp_slots): Likewise.
789 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
791 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
792 poly_int64 rather than a HOST_WIDE_INT.
793 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
794 than a HOST_WIDE_INT.
795 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
796 polynomial frame offsets.
797 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
798 as a poly_int64 rather than an int.
799 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
800 * config/v850/v850-protos.h (compute_frame_size): Likewise.
801 * config/v850/v850.c (compute_frame_size): Likewise.
802 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
803 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
804 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
805 * config/pa/pa.c (pa_compute_frame_size): Likewise.
806 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
807 rather than a HOST_WIDE_INT.
808 * explow.c (get_dynamic_stack_base): Likewise.
809 * final.c (final_start_function): Use the constant lower bound
810 of the frame size for -Wframe-larger-than.
811 * ira.c (do_reload): Adjust for new get_frame_size return type.
812 * lra.c (lra): Likewise.
813 * reload1.c (reload): Likewise.
814 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
815 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
816 * rtlanal.c (get_initial_register_offset): Return the offset as
817 a poly_int64 rather than a HOST_WIDE_INT.
819 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
820 Alan Hayward <alan.hayward@arm.com>
821 David Sherwood <david.sherwood@arm.com>
823 * reload1.c (elim_table): Change initial_offset, offset and
824 previous_offset from HOST_WIDE_INT to poly_int64_pod.
825 (offsets_at): Change the target array's element type from
826 HOST_WIDE_INT to poly_int64_pod.
827 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
828 (elimination_costs_in_insn, update_eliminable_offsets)
829 (verify_initial_elim_offsets, set_offsets_for_label)
830 (init_eliminable_invariants): Update after above changes.
832 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
833 Alan Hayward <alan.hayward@arm.com>
834 David Sherwood <david.sherwood@arm.com>
836 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
837 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
838 (decomposition): Change start and end from HOST_WIDE_INT
840 (decompose, immune_p): Update accordingly.
841 (find_inc_amount): Return a poly_int64 rather than an int.
842 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
845 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
846 Alan Hayward <alan.hayward@arm.com>
847 David Sherwood <david.sherwood@arm.com>
849 * tree.h (get_inner_reference): Return the bitsize and bitpos
850 as poly_int64_pods rather than HOST_WIDE_INT.
851 * fold-const.h (ptr_difference_const): Return the pointer difference
852 as a poly_int64_pod rather than a HOST_WIDE_INT.
853 * expr.c (get_inner_reference): Return the bitsize and bitpos
854 as poly_int64_pods rather than HOST_WIDE_INT.
855 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
857 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
858 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
859 (optimize_bit_field_compare): Update call to get_inner_reference.
860 (decode_field_reference): Likewise.
861 (fold_unary_loc): Track polynomial offsets and sizes.
862 (split_address_to_core_and_offset): Return the bitpos as a
863 poly_int64_pod rather than a HOST_WIDE_INT.
864 (ptr_difference_const): Likewise for the pointer difference.
865 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
866 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
867 * dbxout.c (dbxout_expand_expr): Likewise.
868 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
869 (loc_list_from_tree_1, fortran_common): Likewise.
870 * gimple-laddress.c (pass_laddress::execute): Likewise.
871 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
872 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
873 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
874 * tree-affine.c (tree_to_aff_combination): Likewise.
875 (get_inner_reference_aff): Likewise.
876 * tree-data-ref.c (split_constant_offset_1): Likewise.
877 (dr_analyze_innermost): Likewise.
878 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
879 * tree-sra.c (ipa_sra_check_caller): Likewise.
880 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
881 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
882 (instrument_bool_enum_load, instrument_object_size): Likewise.
883 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
884 to get_inner_reference.
885 * hsa-gen.c (gen_hsa_addr): Likewise.
886 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
887 * tsan.c (instrument_expr): Likewise.
888 * match.pd: Update call to ptr_difference_const.
890 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
891 Alan Hayward <alan.hayward@arm.com>
892 David Sherwood <david.sherwood@arm.com>
894 * fold-const.c (fold_comparison): Track sizes and offsets as
895 poly_int64s rather than HOST_WIDE_INTs when folding address
898 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
899 Alan Hayward <alan.hayward@arm.com>
900 David Sherwood <david.sherwood@arm.com>
902 * expr.h (get_bit_range): Return the bitstart and bitend as
903 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
904 as a poly_int64 rather than a HOST_WIDE_INT.
905 * expr.c (get_bit_range): Likewise.
906 (expand_assignment): Update call accordingly.
907 * fold-const.c (optimize_bit_field_compare): Likewise.
909 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
910 Alan Hayward <alan.hayward@arm.com>
911 David Sherwood <david.sherwood@arm.com>
913 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
914 (aarch64_legitimate_address_p): Use it instead of an rtx code,
915 as an optional final parameter.
916 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
917 (aarch64_legitimate_address_p): Likewise.
918 (aarch64_print_address_internal): Take an aarch64_addr_query_type
919 instead of an rtx code.
920 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
921 (aarch64_legitimate_address_hook_p): Likewise.
922 (aarch64_print_ldpstp_address): Likewise.
923 (aarch64_print_operand_address): Likewise.
924 (aarch64_address_cost): Likewise.
925 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
926 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
927 (aarch64_mem_pair_lanes_operand): Likewise.
929 2017-12-20 Richard Biener <rguenther@suse.de>
931 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
932 update_stmt_if_modified.
934 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
936 PR tree-optimization/83491
937 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
938 before walking uses. Improve coding style and comments.
940 2017-12-20 Tom de Vries <tom@codesourcery.com>
942 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
944 2017-12-20 Jakub Jelinek <jakub@redhat.com>
947 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
949 (pass_ipa_free_fn_summary): Add small_p private data member,
950 initialize to false in the ctor.
951 (pass_ipa_free_fn_summary::clone,
952 pass_ipa_free_fn_summary::set_pass_param,
953 pass_ipa_free_fn_summary::gate): New methods.
954 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
955 | TODO_dump_symtab if small_p.
956 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
957 entry and add another instance of the pass with false parm after
959 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
960 ipa_free_fn_summary here.
962 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
964 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
966 2017-12-20 Martin Sebor <msebor@redhat.com>
969 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
970 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
971 (expand_builtin_strncmp): Same.
973 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
976 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
977 allowed debug stmts before labels.
978 (expand_gimple_basic_block): Likewise.
979 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
980 * gimple-iterator.h (gsi_after_labels): Likewise.
981 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
982 rename reused variable, and simplify using gsi_move_before.
983 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
984 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
985 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
986 (gimple_verify_flow_info, gimple_block_label): Likewise.
987 (make_blocks): Move debug markers after adjacent labels.
988 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
989 allowed debug insns outside blocks.
990 * df-scan.c (df_insn_delete): Likewise.
991 * lra-constraints.c (update_ebb_live_info): Likewise.
992 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
993 (vt_initialize, delete_vta_debug_insns): Likewise.
994 (reemit_marker_as_note): Drop BB parm. Adjust callers.
996 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
997 Alan Hayward <alan.hayward@arm.com>
998 David Sherwood <david.sherwood@arm.com>
1000 * poly-int-types.h (round_down_to_byte_boundary): New macro.
1001 (round_up_to_byte_boundary): Likewise.
1002 * expr.h (get_bit_range): Add temporary shim.
1003 * gimple-ssa-store-merging.c (store_operand_info): Change the
1004 bitsize, bitpos, bitregion_start and bitregion_end fields from
1005 unsigned HOST_WIDE_INT to poly_uint64.
1006 (merged_store_group): Likewise load_align_base.
1007 (compatible_load_p, compatible_load_p): Update accordingly.
1008 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1009 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1010 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1011 bitregion_start and bitregion_end as poly_uint64s rather than
1012 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1013 (handled_load): Take the bitsize, bitpos,
1014 bitregion_start and bitregion_end as poly_uint64s rather than
1015 unsigned HOST_WIDE_INTs.
1016 (pass_store_merging::process_store): Update call to
1017 mem_valid_for_store_merging.
1019 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1020 Alan Hayward <alan.hayward@arm.com>
1021 David Sherwood <david.sherwood@arm.com>
1023 * builtins.c (get_object_alignment_2): Track polynomial offsets
1024 and sizes. Update the alignment handling.
1026 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1027 Alan Hayward <alan.hayward@arm.com>
1028 David Sherwood <david.sherwood@arm.com>
1030 * tree.h (get_inner_reference): Add a version that returns the
1031 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1032 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1033 the case in which bitpos is not associated with the first byte.
1035 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1036 Alan Hayward <alan.hayward@arm.com>
1037 David Sherwood <david.sherwood@arm.com>
1039 * tree-affine.h (get_inner_reference_aff): Return the size as a
1041 * tree-affine.c (get_inner_reference_aff): Likewise.
1042 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1043 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
1045 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1046 Alan Hayward <alan.hayward@arm.com>
1047 David Sherwood <david.sherwood@arm.com>
1049 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1050 HOST_WIDE_INT rather than a poly_int64.
1052 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1053 Alan Hayward <alan.hayward@arm.com>
1054 David Sherwood <david.sherwood@arm.com>
1056 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1057 HOST_WIDE_INT to poly_int64_pod.
1058 (perform_symbolic_merge): Update accordingly.
1059 (bswap_replace): Likewise.
1061 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1062 Alan Hayward <alan.hayward@arm.com>
1063 David Sherwood <david.sherwood@arm.com>
1065 * tree-affine.h (aff_tree::offset): Change from widest_int
1067 (wide_int_ext_for_comb): Delete.
1068 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1069 constants as poly_widest_int rather than widest_int.
1070 (aff_combination_constant_multiple_p): Return the multiplier
1071 as a poly_widest_int.
1072 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1074 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1075 version static and add an overload for poly_widest_int.
1076 (aff_combination_const, aff_combination_add_cst)
1077 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1078 the constants as poly_widest_int rather than widest_int.
1079 (tree_to_aff_combination): Generalize INTEGER_CST case to
1081 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1082 (aff_combination_add_product, aff_combination_mult): Handle
1084 (aff_combination_constant_multiple_p): Return the multiplier
1085 as a poly_widest_int.
1086 * tree-predcom.c (determine_offset): Return the offset as a
1088 (split_data_refs_to_components, suitable_component_p): Update
1090 (valid_initializer_p): Update call to
1091 aff_combination_constant_multiple_p.
1092 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1093 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1094 as a poly_int64 rather than a HOST_WIDE_INT.
1095 (get_address_cost): Handle polynomial offsets.
1096 (iv_elimination_compare_lt): Likewise.
1097 (rewrite_use_nonlinear_expr): Likewise.
1099 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1100 Alan Hayward <alan.hayward@arm.com>
1101 David Sherwood <david.sherwood@arm.com>
1103 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1104 as a poly_int64_pod rather than a HOST_WIDE_INT.
1105 (get_addr_base_and_unit_offset): Likewise.
1106 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1107 (get_addr_base_and_unit_offset): Likewise.
1108 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1109 to poly_int64 in example.
1110 * fold-const.c (fold_binary_loc): Update call to
1111 get_addr_base_and_unit_offset.
1112 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1113 (maybe_canonicalize_mem_ref_addr): Likewise.
1114 (gimple_fold_stmt_to_constant_1): Likewise.
1115 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1117 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
1118 * match.pd: Likewise.
1119 * omp-low.c (lower_omp_target): Likewise.
1120 * tree-sra.c (build_ref_for_offset): Likewise.
1121 (build_debug_ref_for_model): Likewise.
1122 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
1123 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
1124 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
1125 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1126 (constant_pointer_difference): Likewise.
1127 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
1128 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
1129 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
1130 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1131 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
1132 (set_ssa_val_to): Likewise.
1133 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
1134 (maybe_diag_stxncpy_trunc): Likewise.
1135 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1136 * tree.c (build_simple_mem_ref_loc): Likewise.
1137 (array_at_struct_end_p): Likewise.
1139 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1140 Alan Hayward <alan.hayward@arm.com>
1141 David Sherwood <david.sherwood@arm.com>
1143 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
1144 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1145 (get_ref_base_and_extent_hwi): Declare.
1146 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
1147 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1148 (get_ref_base_and_extent_hwi): New function.
1149 * cfgexpand.c (expand_debug_expr): Update call to
1150 get_ref_base_and_extent.
1151 * dwarf2out.c (add_var_loc_to_decl): Likewise.
1152 * gimple-fold.c (get_base_constructor): Return the offset as a
1153 poly_int64_pod rather than a HOST_WIDE_INT.
1154 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
1155 * ipa-polymorphic-call.c
1156 (ipa_polymorphic_call_context::set_by_invariant)
1157 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
1158 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
1159 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
1160 rather than get_ref_base_and_extent.
1161 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
1162 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
1163 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
1165 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
1166 call to get_ref_base_and_extent.
1167 * tree-sra.c (create_access, get_access_for_expr): Likewise.
1168 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
1169 (stmt_kills_ref_p): Likewise.
1170 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1171 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
1173 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
1174 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
1175 when calling native_encode_expr.
1176 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
1177 call to get_ref_base_and_extent.
1178 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
1179 get_ref_base_and_extent.
1180 * var-tracking.c (track_expr_p): Likewise.
1182 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1183 Alan Hayward <alan.hayward@arm.com>
1184 David Sherwood <david.sherwood@arm.com>
1186 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
1187 HOST_WIDE_INT to poly_int64_pod.
1188 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
1189 polynomail parameter offsets.
1191 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1192 Alan Hayward <alan.hayward@arm.com>
1193 David Sherwood <david.sherwood@arm.com>
1195 * gengtype.c (main): Handle poly_int64_pod.
1196 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
1197 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
1198 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
1199 from HOST_WIDE_INT to poly_int64_pod.
1200 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
1201 (copy_cfa): New function.
1202 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
1203 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
1204 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
1205 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
1206 (initial_return_save): Treat offsets as poly_ints.
1207 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
1208 offset is nonconstant.
1209 (reg_save): Take the offset as a poly_int64. Fall back to
1210 DW_CFA_expression for nonconstant offsets.
1211 (queue_reg_save): Take the offset as a poly_int64.
1212 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
1214 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1215 Alan Hayward <alan.hayward@arm.com>
1216 David Sherwood <david.sherwood@arm.com>
1218 * rtl.h (operand_subword, operand_subword_force): Take the offset
1219 as a poly_uint64 an unsigned int.
1220 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
1222 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1223 Alan Hayward <alan.hayward@arm.com>
1224 David Sherwood <david.sherwood@arm.com>
1226 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
1227 'p' format code. Use INT_LIST rather than SUBREG as the example of
1228 a code with an XINT and an XEXP. Remove the implication that
1229 accessing an rtx field using XINT is expected to work.
1230 * rtl.def (SUBREG): Change format from "ei" to "ep".
1231 * rtl.h (rtunion::rt_subreg): New field.
1232 (XCSUBREG): New macro.
1233 (SUBREG_BYTE): Use it.
1234 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
1235 Update constructor accordingly.
1236 (subreg_shape::operator ==): Update accordingly.
1237 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
1238 than an unsigned int.
1239 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
1240 a poly_uint64 rather than an unsigned int.
1241 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
1242 than an unsigned int.
1243 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
1244 (subreg_size_highpart_offset): Return a poly_uint64 rather than
1245 an unsigned int. Take the sizes as poly_uint64s.
1246 (subreg_offset_from_lsb): Return a poly_uint64 rather than
1247 an unsigned int. Take the shift as a poly_uint64 rather than
1249 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
1250 as a poly_uint64 rather than an unsigned int.
1251 (simplify_subreg_regno): Likewise.
1252 (byte_lowpart_offset): Return the memory offset as a poly_int64
1254 (subreg_memory_offset): Likewise. Take the subreg offset as a
1255 poly_uint64 rather than an unsigned int.
1256 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
1257 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
1258 poly_uint64 rather than an unsigned int.
1259 * rtl.c (rtx_format): Describe 'p' in comment.
1260 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
1261 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
1262 offset as a poly_uint64 rather than an unsigned int.
1263 (byte_lowpart_offset): Return the memory offset as a poly_int64
1265 (subreg_memory_offset): Likewise. Take the subreg offset as a
1266 poly_uint64 rather than an unsigned int.
1267 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
1268 mode sizes as poly_uint64s rather than unsigned ints. Return a
1269 poly_uint64 rather than an unsigned int.
1270 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
1271 (copy_insn_1): Handle 'p'.
1272 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
1273 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
1274 an unsigned int. Return the shift in the same way.
1275 (subreg_lsb): Return the shift as a poly_uint64 rather than an
1277 (subreg_size_offset_from_lsb): Take the sizes and shift as
1278 poly_uint64s rather than unsigned ints. Return the offset as
1280 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
1281 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
1283 * rtlhash.c (add_rtx): Handle 'p'.
1284 * genemit.c (gen_exp): Likewise.
1285 * gengenrtl.c (type_from_format, gendef): Likewise.
1286 * gensupport.c (subst_pattern_match, get_alternatives_number)
1287 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
1288 (subst_dup): Likewise.
1289 * gengtype.c (adjust_field_rtx_def): Likewise.
1290 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
1291 (match_pattern_2): Likewise.
1292 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
1293 (rtx_test::subreg_field): New function.
1294 (operator ==, safe_to_hoist_p, transition_parameter_type)
1295 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
1296 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
1297 * genpeep.c (match_rtx): Likewise.
1298 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
1299 (rtx_writer::print_rtx_operand): Handle 'p'.
1300 (print_value): Handle SUBREG.
1301 * read-rtl.c (apply_int_iterator): Likewise.
1302 (rtx_reader::read_rtx_operand): Handle 'p'.
1303 * alias.c (rtx_equal_for_memref_p): Likewise.
1304 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
1305 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
1307 * calls.c (expand_call): Likewise.
1308 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
1309 (make_extraction, gen_lowpart_for_combine): Likewise.
1310 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
1312 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
1313 rather than an unsigned int. Treat subreg offsets as poly_ints.
1314 (exp_equiv_p): Handle 'p'.
1315 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
1316 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
1317 * dse.c (find_shift_sequence): Likewise.
1318 * dwarf2out.c (rtl_for_decl_location): Likewise.
1319 * expmed.c (extract_low_bits): Likewise.
1320 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
1321 (expand_expr_real_2): Likewise.
1322 * final.c (alter_subreg): Likewise.
1323 (leaf_renumber_regs_insn): Handle 'p'.
1324 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
1325 Treat subreg offsets as poly_ints.
1326 * fwprop.c (forward_propagate_and_simplify): Likewise.
1327 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
1328 * ira.c (get_subreg_tracking_sizes): Likewise.
1329 * ira-conflicts.c (go_through_subreg): Likewise.
1330 * ira-lives.c (process_single_reg_class_operands): Likewise.
1331 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
1332 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
1333 as a poly_uint64 rather than an unsigned int.
1334 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
1335 subreg offsets as poly_ints.
1336 * lra-constraints.c (operands_match_p): Handle 'p'.
1337 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
1338 * lra-spills.c (assign_mem_slot): Likewise.
1339 * postreload.c (move2add_valid_value_p): Likewise.
1340 * recog.c (general_operand, indirect_operand): Likewise.
1341 * regcprop.c (copy_value, maybe_mode_change): Likewise.
1342 (copyprop_hardreg_forward_1): Likewise.
1343 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
1344 (record_subregs_of_mode): Likewise.
1345 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
1346 * reload.c (operands_match_p): Handle 'p'.
1347 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
1348 * reload1.c (alter_reg, choose_reload_regs): Likewise.
1349 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
1350 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
1351 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
1352 (simplify_const_poly_int_tests<N>::run): Likewise.
1353 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
1354 a poly_uint64 rather than an unsigned int.
1355 * valtrack.c (debug_lowpart_subreg): Likewise.
1356 * var-tracking.c (var_lowpart): Likewise.
1357 (loc_cmp): Handle 'p'.
1359 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1360 Alan Hayward <alan.hayward@arm.com>
1361 David Sherwood <david.sherwood@arm.com>
1363 * ira.c (get_subreg_tracking_sizes): New function.
1364 (init_live_subregs): Take an integer size rather than a register.
1365 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
1366 to init_live_subregs.
1368 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1369 Alan Hayward <alan.hayward@arm.com>
1370 David Sherwood <david.sherwood@arm.com>
1372 * expr.c (store_constructor_field): Change bitsize from a
1373 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
1374 HOST_WIDE_INT to a poly_int64.
1375 (store_constructor): Change size from a HOST_WIDE_INT to
1377 (store_field): Likewise bitsize and bitpos.
1379 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1380 Alan Hayward <alan.hayward@arm.com>
1381 David Sherwood <david.sherwood@arm.com>
1383 * expmed.h (store_bit_field): Change bitregion_start and
1384 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
1385 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
1386 (store_bit_field_1, store_integral_bit_field, store_bit_field)
1387 (store_fixed_bit_field, store_split_bit_field): Likewise.
1388 * expr.c (store_constructor_field, store_field): Likewise.
1389 (optimize_bitfield_assignment_op): Likewise. Make the same change
1390 to bitsize and bitpos.
1391 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
1392 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
1393 same change in the constructor arguments.
1394 (get_best_mode): Change bitregion_start and bitregion_end from
1395 unsigned HOST_WIDE_INT to poly_uint64.
1396 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1397 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
1399 (bit_field_mode_iterator::next_mode): Update for new types
1400 of m_bitregion_start and m_bitregion_end.
1401 (get_best_mode): Change bitregion_start and bitregion_end from
1402 unsigned HOST_WIDE_INT to poly_uint64.
1404 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1405 Alan Hayward <alan.hayward@arm.com>
1406 David Sherwood <david.sherwood@arm.com>
1408 * rtl.h (simplify_gen_subreg): Add a temporary overload that
1409 accepts poly_uint64 offsets.
1410 * expmed.h (extract_bit_field): Take bitsize and bitnum as
1411 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1412 * expmed.c (lowpart_bit_field_p): Likewise.
1413 (extract_bit_field_as_subreg): New function, split out from...
1414 (extract_bit_field_1): ...here. Take bitsize and bitnum as
1415 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
1416 extractions, check that BITSIZE matches the size of the extracted
1417 value and that BITNUM is an exact multiple of that size.
1418 If all else fails, try forcing the value into memory if
1419 BITNUM is variable, and adjusting the address so that the
1420 offset is constant. Split the part that can only handle constant
1421 bitsize and bitnum out into...
1422 (extract_integral_bit_field): ...this new function.
1423 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
1424 rather than unsigned HOST_WIDE_INTs.
1426 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1427 Alan Hayward <alan.hayward@arm.com>
1428 David Sherwood <david.sherwood@arm.com>
1430 * expmed.h (store_bit_field): Take bitsize and bitnum as
1431 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1432 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
1433 that returns the byte size.
1434 (store_bit_field_1): Take bitsize and bitnum as
1435 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
1436 to simple_mem_bitfield_p. Split the part that can only handle
1437 constant bitsize and bitnum out into...
1438 (store_integral_bit_field): ...this new function.
1439 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
1440 than unsigned HOST_WIDE_INTs.
1441 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
1443 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1444 Alan Hayward <alan.hayward@arm.com>
1445 David Sherwood <david.sherwood@arm.com>
1447 * lra-int.h (lra_reg): Change offset from int to poly_int64.
1448 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
1450 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1451 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
1452 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
1453 offset as a poly_int64 rather than an int.
1454 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
1455 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1456 * lra-constraints.c (equiv_address_substitution): Track offsets
1458 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
1459 (curr_insn_transform): Handle the new form of sp_offset.
1460 * lra-eliminations.c (lra_elim_table): Change previous_offset
1461 and offset from HOST_WIDE_INT to poly_int64.
1462 (print_elim_table, update_reg_eliminate): Update accordingly.
1463 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
1464 (get_elimination): Update accordingly.
1465 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
1466 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1467 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
1468 poly_int64 offsets generally.
1469 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
1470 (mark_not_eliminable, init_elimination): Update accordingly.
1471 (remove_reg_equal_offset_note): Return a bool and pass the new
1472 offset back by pointer as a poly_int64.
1473 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
1474 rather than a HOST_WIDE_INT.
1475 (do_remat): Track offsets poly_int64s.
1476 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
1478 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1479 Alan Hayward <alan.hayward@arm.com>
1480 David Sherwood <david.sherwood@arm.com>
1482 * rtl.h (mem_attrs): Add a default constructor. Change size and
1483 offset from HOST_WIDE_INT to poly_int64.
1484 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
1485 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
1486 (widen_memory_access): Take the sizes and offsets as poly_int64s
1487 rather than HOST_WIDE_INTs.
1488 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
1489 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
1491 (adjust_offset_for_component_ref): Change the offset from a
1492 HOST_WIDE_INT to a poly_int64.
1493 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
1494 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
1495 * dce.c (find_call_stack_args): Likewise.
1496 * dse.c (record_store): Likewise.
1497 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
1498 * print-rtl.c (rtx_writer::print_rtx): Likewise.
1499 * read-rtl-function.c (test_loading_mem): Likewise.
1500 * rtlanal.c (may_trap_p_1): Likewise.
1501 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1502 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
1503 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
1504 (mem_attrs::mem_attrs): New function.
1505 (set_mem_attributes_minus_bitpos): Change bitpos from a
1506 HOST_WIDE_INT to poly_int64.
1507 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
1508 (clear_mem_offset, clear_mem_size, change_address)
1509 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
1510 initialize mem_attrs.
1511 (set_mem_offset, set_mem_size, adjust_address_1)
1512 (adjust_automodify_address_1, offset_address, widen_memory_access):
1513 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
1515 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1516 Alan Hayward <alan.hayward@arm.com>
1517 David Sherwood <david.sherwood@arm.com>
1519 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
1520 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
1521 rather than 0 to represent an unknown size. Assert that the size
1522 is known when the mode isn't BLKmode.
1523 (may_trap_p_1): Use -1 for unknown sizes.
1524 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
1526 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1527 Alan Hayward <alan.hayward@arm.com>
1528 David Sherwood <david.sherwood@arm.com>
1530 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
1531 to poly_int64. Update commentary for positions_needed.large.
1532 (read_info_type): Change offset and width from HOST_WIDE_INT
1534 (set_usage_bits): Likewise.
1535 (canon_address): Return the offset as a poly_int64 rather than
1536 a HOST_WIDE_INT. Use strip_offset_and_add.
1537 (set_all_positions_unneeded, any_positions_needed_p): Use
1538 positions_needed.large to track stores with non-constant widths.
1539 (all_positions_needed_p): Likewise. Take the offset and width
1540 as poly_int64s rather than ints. Assert that rhs is nonnull.
1541 (record_store): Cope with non-constant offsets and widths.
1542 Nullify the rhs of an earlier store if we can't tell which bytes
1544 (find_shift_sequence): Take the access_size and shift as poly_int64s
1546 (get_stored_val): Take the read_offset and read_width as poly_int64s
1547 rather than HOST_WIDE_INTs.
1548 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
1549 non-constant offsets and widths.
1551 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1552 Alan Hayward <alan.hayward@arm.com>
1553 David Sherwood <david.sherwood@arm.com>
1555 * inchash.h (inchash::hash::add_poly_int): New function.
1556 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1557 Use poly_int64 rather than HOST_WIDE_INT.
1558 (ao_ref::max_size_known_p): New function.
1559 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1560 rather than HOST_WIDE_INT.
1561 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
1562 to temporaries until its interface is adjusted to match.
1563 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
1564 (aliasing_component_refs_p, decl_refs_may_alias_p)
1565 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
1566 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
1567 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
1569 * alias.c (ao_ref_from_mem): Likewise.
1570 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1571 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
1572 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
1573 (maybe_trim_complex_store, maybe_trim_constructor_store)
1574 (live_bytes_read, dse_classify_store): Likewise.
1575 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
1576 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
1577 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
1578 (vn_reference_lookup_3): Likewise.
1579 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
1581 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1582 Alan Hayward <alan.hayward@arm.com>
1583 David Sherwood <david.sherwood@arm.com>
1585 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
1586 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
1587 instead of ranges_overlap_p.
1589 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1590 Alan Hayward <alan.hayward@arm.com>
1591 David Sherwood <david.sherwood@arm.com>
1593 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
1594 sizes as poly_int64s rather than HOST_WIDE_INTs.
1596 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1597 Alan Hayward <alan.hayward@arm.com>
1598 David Sherwood <david.sherwood@arm.com>
1600 * gimple-fold.h (fold_ctor_reference): Take the offset and size
1601 as poly_uint64 rather than unsigned HOST_WIDE_INT.
1602 * gimple-fold.c (fold_ctor_reference): Likewise.
1604 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1605 Alan Hayward <alan.hayward@arm.com>
1606 David Sherwood <david.sherwood@arm.com>
1608 * target.def (dwarf_poly_indeterminate_value): New hook.
1609 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
1610 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
1611 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
1612 * doc/tm.texi: Regenerate.
1613 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
1614 offset as a poly_int64.
1615 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
1616 offset as a poly_int64.
1617 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
1618 Take the offset as a poly_int64.
1619 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1620 (frame_pointer_fb_offset): Change to a poly_int64.
1621 (int_loc_descriptor): Take the offset as a poly_int64. Use
1622 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1623 (based_loc_descr): Take the offset as a poly_int64.
1624 Use strip_offset_and_add to handle (plus X (const)).
1625 Use new_reg_loc_descr instead of an open-coded version of the
1626 previous implementation.
1627 (mem_loc_descriptor): Handle CONST_POLY_INT.
1628 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1629 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1631 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1632 Alan Hayward <alan.hayward@arm.com>
1633 David Sherwood <david.sherwood@arm.com>
1635 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1636 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1637 * inchash.h (inchash::hash::add_poly_hwi): New function.
1638 * gengtype.c (main): Register poly_int64.
1639 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1640 offset as a poly_int.
1641 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1642 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1643 offset as a poly_int64.
1644 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1645 * print-rtl.c (print_poly_int): New function.
1646 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1648 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1650 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1651 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1652 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1653 Enforce track_offset_p for parts of a PARALLEL.
1654 (vt_add_function_parameter): Use const_offset for the final
1655 offset to track. Use get_tracked_reg_offset for the parts
1658 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1659 Alan Hayward <alan.hayward@arm.com>
1660 David Sherwood <david.sherwood@arm.com>
1662 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1663 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1664 * doc/tm.texi: Regenerate.
1665 * hooks.h (hook_bool_uint_uint_true): Delete.
1666 (hook_bool_puint64_puint64_true): Declare.
1667 * hooks.c (hook_bool_uint_uint_true): Delete.
1668 (hook_bool_puint64_puint64_true): New function.
1669 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1670 instead of unsigned ints.
1671 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1672 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1674 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1675 Alan Hayward <alan.hayward@arm.com>
1676 David Sherwood <david.sherwood@arm.com>
1678 * optabs.h (expand_operand): Add an int_value field.
1679 (create_expand_operand): Add an int_value parameter and use it
1680 to initialize the new expand_operand field.
1681 (create_integer_operand): Replace with a declaration of a function
1682 that accepts poly_int64s. Move the implementation to...
1683 * optabs.c (create_integer_operand): ...here.
1684 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1685 the mode preserves the value of int_value, instead of calling
1686 const_int_operand on the rtx. Use gen_int_mode to generate
1689 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1690 Alan Hayward <alan.hayward@arm.com>
1691 David Sherwood <david.sherwood@arm.com>
1693 * dumpfile.h (dump_dec): Declare.
1694 * dumpfile.c (dump_dec): New function.
1695 * pretty-print.h (pp_wide_integer): Turn into a function and
1696 declare a poly_int version.
1697 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1699 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1700 Alan Hayward <alan.hayward@arm.com>
1701 David Sherwood <david.sherwood@arm.com>
1703 * doc/generic.texi (POLY_INT_CST): Document.
1704 * tree.def (POLY_INT_CST): New tree code.
1705 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1706 * tree-core.h (tree_poly_int_cst): New struct.
1707 (tree_node): Add a poly_int_cst field.
1708 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1709 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1710 instead of a wide_int_ref.
1711 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1713 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1714 instead of an unsigned HOST_WIDE_INT.
1715 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1716 (ptrdiff_tree_p): Declare.
1717 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1718 extern inline implementations if the target doesn't use POLY_INT_CST.
1719 (poly_int_tree_p): New function.
1720 (wi::unextended_tree): New class.
1721 (wi::int_traits <unextended_tree>): New override.
1722 (wi::extended_tree): Add a default constructor.
1723 (wi::extended_tree::get_tree): New function.
1724 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1725 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1726 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1727 (wi::tree_to_poly_wide_ref): New typedefs.
1728 (wi::ints_for): Provide overloads for extended_tree and
1730 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1731 (wi::to_wide): New functions.
1732 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1733 * tree.c (poly_int_cst_hasher): New struct.
1734 (poly_int_cst_hash_table): New variable.
1735 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1736 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1738 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1739 (init_ttree): Initialize poly_int_cst_hash_table.
1740 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1741 a poly_int64 instead of a HOST_WIDE_INT.
1742 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1743 instead of an unsigned HOST_WIDE_INT.
1744 (wide_int_to_tree): Rename to...
1745 (wide_int_to_tree_1): ...this.
1746 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1747 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1748 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1749 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1750 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1751 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1753 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1754 (streamer_read_tree_body): Likewise.
1755 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1756 (streamer_write_tree_body): Likewise.
1757 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1758 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1759 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1760 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1761 * gimple-expr.h (is_gimple_constant): Likewise.
1762 * gimplify.c (maybe_with_size_expr): Likewise.
1763 * print-tree.c (print_node): Likewise.
1764 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1765 * tree-pretty-print.c (dump_generic_node): Likewise.
1766 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1767 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1768 * tree-vrp.c (compare_values_warnv): Likewise.
1769 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1770 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1771 (force_expr_to_var_cost): Likewise.
1772 * tree-ssa-loop.c (for_each_index): Likewise.
1773 * fold-const.h (build_invariant_address, size_int_kind): Take a
1774 poly_int64 instead of a HOST_WIDE_INT.
1775 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1776 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1778 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1779 (int_const_binop_2): New function, split out from...
1780 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1781 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1782 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1783 * gimple-ssa-strength-reduction.c (slsr_process_add)
1784 (slsr_process_mul): Check for INTEGER_CSTs before using them
1786 * stor-layout.c (bits_from_bytes): New function.
1787 (bit_from_pos): Use it.
1788 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1789 by BITS_PER_UNIT to get the TYPE_SIZE.
1790 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1791 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1793 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1794 Alan Hayward <alan.hayward@arm.com>
1795 David Sherwood <david.sherwood@arm.com>
1797 * doc/rtl.texi (const_poly_int): Document. Also document the
1798 rtl sharing behavior.
1799 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1800 * rtl.h (const_poly_int_def): New struct.
1801 (rtx_def::u): Add a cpi field.
1802 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1803 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1804 (wi::rtx_to_poly_wide_ref): New typedef
1805 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1806 (poly_int_rtx_p): New functions.
1807 (trunc_int_for_mode): Declare a poly_int64 version.
1808 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1809 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1811 (strip_offset): Declare.
1812 (strip_offset_and_add): New function.
1813 * rtl.def (CONST_POLY_INT): New rtx code.
1814 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1815 (shared_const_p): Use poly_int_rtx_p.
1816 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1818 (gen_int_shift_amount): Likewise.
1819 * emit-rtl.c (const_poly_int_hasher): New class.
1820 (const_poly_int_htab): New variable.
1821 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1822 (const_poly_int_hasher::hash): New function.
1823 (const_poly_int_hasher::equal): Likewise.
1824 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1825 (immed_wide_int_const): Rename to...
1826 (immed_wide_int_const_1): ...this and make static.
1827 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1828 instead of a wide_int_ref.
1829 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1830 (gen_lowpart_common): Handle CONST_POLY_INT.
1831 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1832 * cselib.c (cselib_hash_rtx): Likewise.
1833 * dwarf2out.c (const_ok_for_output_1): Likewise.
1834 * expr.c (convert_modes): Likewise.
1835 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1836 * rtlhash.c (add_rtx): Likewise.
1837 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1838 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1839 Handle existing CONST_POLY_INT rtxes.
1840 * expmed.h (expand_shift): Take a poly_int64 instead of a
1842 * expmed.c (expand_shift): Likewise.
1843 * rtlanal.c (strip_offset): New function.
1844 (commutative_operand_precedence): Give CONST_POLY_INT the same
1845 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1847 * rtl-tests.c (const_poly_int_tests): New struct.
1848 (rtl_tests_c_tests): Use it.
1849 * simplify-rtx.c (simplify_const_unary_operation): Handle
1851 (simplify_const_binary_operation): Likewise.
1852 (simplify_binary_operation_1): Fold additions of symbolic constants
1853 and CONST_POLY_INTs.
1854 (simplify_subreg): Handle extensions and truncations of
1856 (simplify_const_poly_int_tests): New struct.
1857 (simplify_rtx_c_tests): Use it.
1858 * wide-int.h (storage_ref): Add default constructor.
1859 (wide_int_ref_storage): Likewise.
1860 (trailing_wide_ints): Use GTY((user)).
1861 (trailing_wide_ints::operator[]): Add a const version.
1862 (trailing_wide_ints::get_precision): New function.
1863 (trailing_wide_ints::extra_size): Likewise.
1865 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1866 Alan Hayward <alan.hayward@arm.com>
1867 David Sherwood <david.sherwood@arm.com>
1869 * emit-rtl.h (gen_int_shift_amount): Declare.
1870 * emit-rtl.c (gen_int_shift_amount): New function.
1871 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
1873 * calls.c (shift_return_value): Likewise.
1874 * cse.c (fold_rtx): Likewise.
1875 * dse.c (find_shift_sequence): Likewise.
1876 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
1877 (expand_shift, expand_smod_pow2): Likewise.
1878 * lower-subreg.c (shift_cost): Likewise.
1879 * optabs.c (expand_superword_shift, expand_doubleword_mult)
1880 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
1881 (expand_vec_perm_var): Likewise.
1882 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1883 (simplify_binary_operation_1): Likewise.
1884 * combine.c (try_combine, find_split_point, force_int_to_mode)
1885 (simplify_shift_const_1, simplify_shift_const): Likewise.
1886 (change_zero_ext): Likewise. Use simplify_gen_binary.
1888 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1890 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
1892 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1894 * doc/invoke.texi (ARM Options): Document accepted extension options
1895 for -march=armv8.3-a.
1897 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
1900 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
1901 or --with-float=softfp, set the default CPU to arm10e.
1903 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
1905 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
1906 * config/visium/predicates.md (const_shift_operand): Likewise.
1907 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
1908 (visium_legitimize_reload_address): Likewise.
1910 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1913 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
1915 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1918 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
1920 2017-12-20 Julia Koval <julia.koval@intel.com>
1922 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
1923 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
1924 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
1925 * config.gcc: Include vpclmulqdqintrin.h.
1926 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
1927 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
1929 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
1930 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
1931 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
1932 * config/i386/i386.c (isa2_opts): Add -mcx16.
1933 (isa_opts): Add -mpclmulqdq, remove -mcx16.
1934 (ix86_option_override_internal): Move mcx16 to flags2.
1935 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
1936 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
1937 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
1938 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
1939 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
1940 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
1941 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
1942 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
1943 * doc/invoke.texi: Add -mvpclmulqdq.
1945 2017-12-20 Tom de Vries <tom@codesourcery.com>
1948 * config/i386/i386.c (ix86_static_chain): Move
1949 DECL_STATIC_CHAIN test ...
1950 * calls.c (rtx_for_static_chain): ... here. New function.
1951 * calls.h (rtx_for_static_chain): Declare.
1952 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
1953 instead of targetm.calls.static_chain.
1954 * df-scan.c (df_get_entry_block_def_set): Same.
1956 2017-12-19 Tom de Vries <tom@codesourcery.com>
1958 PR tree-optimization/83493
1959 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
1962 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1964 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
1965 inform with hardcoded english plural handling.
1967 2017-12-18 Jeff Law <law@redhat.com>
1969 PR tree-optimization/83477
1970 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
1971 a non-virtual PHI, always push a new range.
1973 2017-12-19 Martin Sebor <msebor@redhat.com>
1976 * builtins.c (compute_objsize): Handle non-constant offsets.
1978 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1980 PR tree-optimization/83444
1981 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
1982 character load case, if get_stridx on MEM_REF's operand doesn't
1983 look usable, retry with get_addr_stridx.
1985 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
1988 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
1989 (variable_tracking_main_1): Keep markers even when VTA fails.
1992 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
1993 even if there are markers before it.
1994 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
1996 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1998 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
1999 typical order conditions.
2000 * sel-sched.c (extract_new_fences_from): Likewise.
2001 * config/visium/constraints.md (J, K, L): Likewise.
2002 * config/visium/predicates.md (const_shift_operand): Likewise.
2003 * config/visium/visium.c (visium_legitimize_address,
2004 visium_legitimize_reload_address): Likewise.
2005 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2006 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2007 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2008 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2009 SET_DUMP_DETAIL): Likewise.
2010 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2011 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2012 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2013 avr_set_core_architecture, avr_set_current_function,
2014 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2015 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2016 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2017 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2018 avr_map_decompose, avr_fold_builtin): Likewise.
2019 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2020 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2021 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2022 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2023 * config/m32c/m32c.c (m32c_conditional_register_usage,
2024 m32c_address_cost): Likewise.
2025 * config/m32c/predicates.md (shiftcount_operand,
2026 longshiftcount_operand): Likewise.
2027 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2028 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2029 can_use_cdx_ldstw): Likewise.
2030 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2031 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2033 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2034 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2035 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2036 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2037 * config/vax/vax.c (adjacent_operands_p): Likewise.
2038 * config/ft32/constraints.md (L, b, KA): Likewise.
2039 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2041 * cfgexpand.c (expand_stack_alignment): Likewise.
2042 * gcse.c (insert_expr_in_table): Likewise.
2043 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2044 * cgraphunit.c (cgraph_node::expand): Likewise.
2045 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2046 * emit-rtl.c (add_insn): Likewise.
2047 * input.c (dump_location_info): Likewise.
2048 * passes.c (NEXT_PASS): Likewise.
2049 * read-rtl-function.c (parse_note_insn_name,
2050 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2052 * sched-rgn.c (sched_rgn_init): Likewise.
2053 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2054 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2055 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2057 * explow.c (eliminate_constant_term): Likewise.
2058 * final.c (leaf_renumber_regs_insn): Likewise.
2059 * cfgrtl.c (print_rtl_with_bb): Likewise.
2060 * genhooks.c (emit_init_macros): Likewise.
2061 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2062 * tree-data-ref.c (conflict_fn): Likewise.
2063 * selftest.c (assert_streq): Likewise.
2064 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2065 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2066 fold_binary_loc, multiple_of_p): Likewise.
2067 * reload.c (push_reload, find_equiv_reg): Likewise.
2068 * et-forest.c (et_nca, et_below): Likewise.
2069 * dbxout.c (dbxout_symbol_location): Likewise.
2070 * reorg.c (relax_delay_slots): Likewise.
2071 * dojump.c (do_compare_rtx_and_jump): Likewise.
2072 * gengtype-parse.c (type): Likewise.
2073 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2074 simplify_const_relational_operation): Likewise.
2075 * reload1.c (do_output_reload): Likewise.
2076 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2077 * gengtype.c (type_for_name): Likewise.
2078 * gimple-ssa-sprintf.c (format_directive): Likewise.
2080 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2083 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2084 accessing it. Adjust comment.
2086 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2089 * predict.c (zero_one_minusone): New function.
2090 (apply_return_prediction): Avoid return prediction for functions
2091 returning only -1, 0 and 1 values, unless they only return -1 and 0
2094 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2096 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2097 fall-through warning.
2098 (arc_compute_frame_size): Remove unused variables.
2099 (arc_print_operand): Fix fprintif format.
2100 (arc_can_follow_jump): Clean fall-through warning.
2102 2017-12-19 Marek Polacek <polacek@redhat.com>
2105 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2108 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2110 * config/arc/arc.c (overriderregs): New variable.
2111 (arc_override_options): Track fixed/call saved/call options.
2112 (arc_conditional_register_usage): Check against overrideregs
2113 variable whenever we change register properties.
2115 2017-12-19 Nathan Sidwell <nathan@acm.org>
2117 * opts.c (finish_options): Don't prefix dump_base_name if it
2118 already contains directories.
2120 2017-12-19 Martin Liska <mliska@suse.cz>
2122 PR rtl-optimization/82675
2123 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
2124 more element in sbitmap.
2126 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
2128 * gimplify.c (gimplify_expr): Use error_operand_p.
2130 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
2133 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
2136 2017-12-19 Daniel Cederman <cederman@gaisler.com>
2138 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
2139 the jump is to a label.
2141 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2143 PR tree-optimization/83444
2144 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
2149 * ipa-inline.c (flatten_remove_node_hook): New function.
2150 (ipa_inline): Keep only nodes with flatten attribute at the end of
2151 the array in the order from ipa_reverse_postorder, only walk that
2152 portion of array for flattening, if there is more than one such
2153 node, temporarily register a removal hook and ignore removed nodes.
2155 PR tree-optimization/80631
2156 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
2157 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
2158 IFN_REDUC_MAX or IFN_REDUC_MIN.
2160 2017-12-18 Jeff Law <law@redhat.com>
2162 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
2163 record symbolic equivalences from backedges in the CFG.
2166 2017-11-19 Jeff Law <law@redhat.com>
2168 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2169 of degenerates resulting from ignoring an edge.
2171 2017-12-18 Martin Sebor <msebor@redhat.com>
2174 PR tree-optimization/78450
2175 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
2176 (handle_builtin_strlen): Call it.
2178 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
2180 PR rtl-optimization/83424
2181 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
2183 2017-12-18 Marek Polacek <polacek@redhat.com>
2186 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
2187 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
2190 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2192 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
2193 (umaddsidi4, umaddsidi_split): Likewise.
2195 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2197 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
2200 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2203 * config/s390/s390.c (s390_option_override): Avoid strncpy.
2205 2017-12-18 Richard Biener <rguenther@suse.de>
2207 PR tree-optimization/81877
2208 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
2209 (outermost_indep_loop): Adjust.
2210 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
2211 (can_sm_ref_p): Adjust.
2213 2017-12-18 Richard Biener <rguenther@suse.de>
2216 * tree.c (array_at_struct_end_p): Return true if the underlying
2217 object has space for at least one element in excess of what
2218 the array domain specifies.
2220 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2222 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
2223 example to avoid overfull hbox.
2224 * doc/invoke.texi (Option Summary): Add missing @gol.
2225 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
2228 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2229 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2231 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
2232 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
2233 (AArch64 Options): Document -mverbose-cost-dump.
2234 (ARM Options): Likewise, plus -mflip-thumb.
2236 2017-12-17 Martin Sebor <msebor@redhat.com>
2239 * gimple-ssa-warn-restrict.c
2240 (builtin_memref::offset_out_of_bounds): Correct the handling of
2243 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2245 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
2247 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
2249 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
2251 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
2253 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
2256 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
2258 * doc/invoke.texi: Fix some typos.
2260 2017-12-16 Martin Sebor <msebor@redhat.com>
2262 PR tree-optimization/78918
2263 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
2264 * builtins.c (check_sizes): Rename...
2265 (check_access): ...to this. Rename function arguments for clarity.
2266 (check_memop_sizes): Adjust names.
2267 (expand_builtin_memchr, expand_builtin_memcpy): Same.
2268 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
2269 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
2270 (check_strncat_sizes, expand_builtin_strncat): Same.
2271 (expand_builtin_strncpy, expand_builtin_memset): Same.
2272 (expand_builtin_bzero, expand_builtin_memcmp): Same.
2273 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
2274 (maybe_emit_sprintf_chk_warning): Same.
2275 (expand_builtin_strcpy): Adjust.
2276 (expand_builtin_stpcpy): Same.
2277 (expand_builtin_with_bounds): Detect out-of-bounds accesses
2278 in pointer-checking forms of memcpy, memmove, and mempcpy.
2279 (gcall_to_tree_minimal, max_object_size): Define new functions.
2280 * builtins.h (max_object_size): Declare.
2281 * calls.c (alloc_max_size): Call max_object_size instead of
2282 hardcoding ssizetype limit.
2283 (get_size_range): Handle new argument.
2284 * calls.h (get_size_range): Add a new argument.
2285 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
2286 * doc/invoke.texi (-Wrestrict): Adjust, add example.
2287 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
2289 (gimple_fold_builtin_memory_chk): Same.
2290 (gimple_fold_builtin_stxcpy_chk): New function.
2291 * gimple-ssa-warn-restrict.c: New source.
2292 * gimple-ssa-warn-restrict.h: New header.
2293 * gimple.c (gimple_build_call_from_tree): Propagate location.
2294 * passes.def (pass_warn_restrict): Add new pass.
2295 * tree-pass.h (make_pass_warn_restrict): Declare.
2296 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
2298 (handle_builtin_strcat): Same.
2299 (strlen_optimize_stmt): Rename...
2300 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
2301 stpncpy, strncpy, and their checking forms.
2303 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
2305 PR rtl-optimization/82849
2306 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
2307 and get_max_loop_iterations_int.
2309 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2310 Alan Hayward <alan.hayward@arm.com>
2311 David Sherwood <david.sherwood@arm.com>
2313 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
2314 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
2315 size as a poly_uint64.
2316 (mode_for_vector, mode_for_int_vector): Take the number of vector
2317 elements as a poly_uint64.
2318 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
2319 size as a poly_uint64.
2320 (mode_for_vector, mode_for_int_vector): Take the number of vector
2321 elements as a poly_uint64.
2323 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2324 Alan Hayward <alan.hayward@arm.com>
2325 David Sherwood <david.sherwood@arm.com>
2327 * machmode.h (MACRO_MODE): New macro.
2328 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
2329 * targhooks.c (default_libcall_value, default_secondary_reload)
2330 (default_memory_move_cost, default_register_move_cost)
2331 (default_class_max_nregs): Likewise.
2333 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2334 Alan Hayward <alan.hayward@arm.com>
2335 David Sherwood <david.sherwood@arm.com>
2337 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
2338 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
2340 * genautomata.c (main): Likewise.
2341 * genconditions.c (write_header): Likewise.
2342 * genemit.c (main): Likewise.
2343 * genextract.c (print_header): Likewise.
2344 * genopinit.c (main): Likewise.
2345 * genoutput.c (output_prologue): Likewise.
2346 * genpeep.c (main): Likewise.
2347 * genpreds.c (write_insn_preds_c): Likewise.
2348 * genrecog.c (writer_header): Likewise.
2349 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
2350 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
2351 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
2352 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
2353 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
2354 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
2355 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
2356 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
2357 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
2358 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
2359 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
2360 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
2361 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
2362 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
2363 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
2364 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
2365 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
2366 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
2367 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
2368 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
2369 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
2370 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
2371 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
2372 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
2373 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
2374 * config/darwin.c (IN_TARGET_CODE): Likewise.
2375 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
2376 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
2377 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
2378 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
2379 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
2380 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
2381 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
2382 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
2383 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
2384 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
2385 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
2386 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
2387 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
2388 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
2389 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
2390 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
2391 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
2392 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
2393 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
2394 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
2395 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
2396 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
2397 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
2398 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
2399 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
2400 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
2401 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
2402 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
2403 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
2404 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
2405 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
2406 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
2407 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
2408 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
2409 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
2410 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
2411 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
2412 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
2413 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
2414 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
2415 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
2416 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
2417 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
2418 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
2419 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
2420 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
2421 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
2422 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
2423 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
2424 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
2425 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
2426 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
2427 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
2428 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
2429 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
2430 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
2431 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
2432 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
2433 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
2434 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
2435 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2436 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
2437 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
2438 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
2439 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
2440 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
2441 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
2442 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
2443 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
2444 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
2445 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
2446 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2447 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
2448 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
2449 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
2450 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
2451 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
2452 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
2453 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
2454 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
2455 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
2456 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
2457 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
2458 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
2459 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
2460 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
2461 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
2462 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
2463 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
2464 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
2465 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
2466 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
2467 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
2468 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
2469 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
2470 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
2471 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
2472 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
2473 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
2474 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
2475 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
2476 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
2477 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
2478 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
2479 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
2480 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
2482 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2483 Alan Hayward <alan.hayward@arm.com>
2484 David Sherwood <david.sherwood@arm.com>
2486 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
2489 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2490 Alan Hayward <alan.hayward@arm.com>
2491 David Sherwood <david.sherwood@arm.com>
2493 * doc/generic.texi (VEC_SERIES_EXPR): Document.
2494 * doc/md.texi (vec_series@var{m}): Document.
2495 * tree.def (VEC_SERIES_EXPR): New tree code.
2496 * tree.h (build_vec_series): Declare.
2497 * tree.c (build_vec_series): New function.
2498 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
2499 * tree-pretty-print.c (dump_generic_node): Likewise.
2500 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
2501 * tree-inline.c (estimate_operator_cost): Likewise.
2502 * expr.c (expand_expr_real_2): Likewise.
2503 * optabs-tree.c (optab_for_tree_code): Likewise.
2504 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
2505 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
2506 * expmed.c (make_tree): Handle VEC_SERIES.
2507 * optabs.def (vec_series_optab): New optab.
2508 * optabs.h (expand_vec_series_expr): Declare.
2509 * optabs.c (expand_vec_series_expr): New function.
2510 * tree-vect-generic.c (expand_vector_operations_1): Check that
2511 the operands also have vector type.
2513 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2514 Alan Hayward <alan.hawyard@arm.com>
2515 David Sherwood <david.sherwood@arm.com>
2517 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
2518 (VEC_COND_EXPR): Add missing @tindex.
2519 * doc/md.texi (vec_duplicate@var{m}): Document.
2520 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
2521 * tree.c (build_vector_from_val): Add stubbed-out handling of
2522 variable-length vectors, using VEC_DUPLICATE_EXPR.
2523 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
2524 * cfgexpand.c (expand_debug_expr): Likewise.
2525 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2526 * tree-inline.c (estimate_operator_cost): Likewise.
2527 * tree-pretty-print.c (dump_generic_node): Likewise.
2528 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
2529 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
2530 (test_vec_duplicate_folding): New function.
2531 (fold_const_c_tests): Call it.
2532 * optabs.def (vec_duplicate_optab): New optab.
2533 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
2534 * optabs.h (expand_vector_broadcast): Declare.
2535 * optabs.c (expand_vector_broadcast): Make non-static. Try using
2536 vec_duplicate_optab.
2537 * expr.c (store_constructor): Try using vec_duplicate_optab for
2539 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
2541 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
2544 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
2545 div/mod latencies a bit.
2547 2017-12-15 Jeff Law <law@redhat.com>
2549 PR tree-optimization/36550
2550 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
2551 (mark_threaded_blocks): Rewrite code to avoid block copying when
2552 optimizing for size. Don't pessimize blocks which will be
2553 copied, but all the statements will be dead.
2555 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
2557 PR tree-optimization/81165
2558 * tree-ssa-threadupdate.c (uses_in_bb): New.
2559 (estimate_threading_killed_stmts): New.
2560 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
2561 * tree-ssa-threadedge.c
2562 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
2565 2017-12-15 Jeff Law <law@redhat.com>
2567 PR tree-optimization/83410
2568 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
2569 threads when parallelizing loops.
2571 2017-12-15 Jakub Jelinek <jakub@redhat.com>
2573 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
2575 * config/alpha/alpha.c (vms_attribute_table): Swap
2576 affects_type_identity and handler fields, adjust comments.
2577 * config/mips/mips.c (mips_attribute_table): Likewise.
2578 * config/visium/visium.c (visium_attribute_table): Likewise.
2579 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
2580 * config/microblaze/microblaze.c (microblaze_attribute_table):
2582 * config/spu/spu.c (spu_attribute_table): Likewise.
2583 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
2584 * config/arc/arc.c (arc_attribute_table): Likewise.
2585 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
2586 * config/v850/v850.c (v850_handle_interrupt_attribute,
2587 v850_handle_data_area_attribute): Formatting fixes.
2588 (v850_attribute_table): Swap affects_type_identity and handler
2589 fields, adjust comments.
2590 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
2591 * config/arm/arm.c (arm_attribute_table): Likewise.
2592 * config/avr/avr.c (avr_attribute_table): Likewise.
2593 * config/s390/s390.c (s390_attribute_table): Likewise.
2594 * config/sh/sh.c (sh_attribute_table): Likewise.
2595 * config/i386/i386.c (ix86_handle_cconv_attribute,
2596 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
2597 (ix86_attribute_table): Swap affects_type_identity and handler
2598 fields, adjust comments.
2599 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2600 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
2601 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
2602 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
2603 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
2604 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2605 * config/rx/rx.c (rx_attribute_table): Likewise.
2606 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
2607 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
2608 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
2609 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
2610 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2611 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
2612 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
2613 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
2614 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2615 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2616 * doc/plugins.texi (user_attr): Likewise. Add NULL for
2618 * attribs.c (empty_attribute_table): Swap affects_type_identity and
2620 (register_scoped_attributes, decl_attributes): Formatting fixes.
2622 PR tree-optimization/83269
2623 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2624 subtraction in arg0's type if type is signed and arg0 is unsigned.
2628 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2629 convert? on @0 instead of convert. Check type of @1, not @0.
2630 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2631 convert? on @0 instead of convert. Check type of @1, not @0.
2632 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2633 only optimize if either both @1 and @2 types are narrower
2634 precision, or both are wider or equal precision, and in the former
2635 case only if both have undefined overflow.
2637 2017-12-15 Richard Biener <rguenther@suse.de>
2640 * internal-fn.def (IFN_NOP): Add.
2641 * internal-fn.c (expand_NOP): Do nothing.
2642 * lto-streamer-in.c (input_function): Instead of removing
2643 sanitizer calls replace them with IFN_NOP calls.
2645 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2646 Alan Hayward <alan.hayward@arm.com>
2647 David Sherwood <david.sherwood@arm.com>
2649 * dse.c (store_info, read_info_type): Replace begin and end with
2651 (print_range): New function.
2652 (set_all_positions_unneeded, any_positions_needed_p)
2653 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2655 (record_store): Likewise. Optimize the case in which all positions
2657 (get_stored_val): Replace read_begin and read_end with read_offset
2659 (replace_read): Update call accordingly.
2661 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2663 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2664 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2665 (loop_cand::loop_cand): Initialize above members.
2666 (loop_cand::supported_operations): Delete.
2667 (loop_cand::can_interchange_p): Inline above function.
2668 (loop_cand::classify_simple_reduction): Record number of constant
2669 initialized simple reductions.
2670 (should_interchange_loops): New parameters. Check stmt cost of loops
2672 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2673 Update call to should_interchange_loops.
2674 (should_interchange_loop_nest): Update call to
2675 should_interchange_loops.
2677 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2680 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2681 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2682 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2684 2017-12-15 Julia Koval <julia.koval@intel.com>
2686 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2687 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
2689 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2690 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2691 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2693 2017-12-15 Julia Koval <julia.koval@intel.com>
2695 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2696 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2697 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2698 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2699 _mm_aesenc_epi128): New intrinsics.
2701 2017-12-15 Julia Koval <julia.koval@intel.com>
2703 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2704 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
2706 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2707 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2708 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2710 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2712 * gimple-ssa-strength-reduction.c (analyze_increments):
2713 Distinguish replacement costs for constant strides from those for
2716 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2718 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2720 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2722 * doc/invoke.texi: Document -Wcast-function-type.
2723 * recog.h (stored_funcptr): Change signature.
2724 * tree-dump.c (dump_node): Avoid warning.
2725 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2727 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2730 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2733 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2735 PR tree-optimization/83312
2736 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2737 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2738 "val" param, and to cope with arbitrary basic blocks.
2739 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2740 handle NULL_TREE for "val", dropping "bb" param.
2741 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2742 drop "bb" param. Handle NULL_TREE for "val".
2743 (find_case_label_for_value): Make "switch_stmt" param const.
2744 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2746 (vrp_prop::check_all_array_refs): Reimplement as...
2747 (check_array_bounds_dom_walker::before_dom_children): ...this new
2748 vfunc. Replace linear search through BB block list, excluding
2749 those with non-executable in-edges via dominator walk.
2751 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2753 * config/arm/arm.opt (mverbose-cost-dump): New option.
2754 * config/arm/arm.c (arm_rtx_costs): Use it.
2756 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2759 * reload1.c (emit_input_reload_insns): Skip debug markers.
2761 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2763 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2767 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2768 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2769 * config/sh/sh.c: Skip debug insns besides notes.
2770 * config/sh/sh.md: Likewise.
2771 * config/sh/sh_treg_combine.cc: Likewise.
2772 * config/sh/sync.md: Likewise.
2774 2017-12-14 Tom de Vries <tom@codesourcery.com>
2776 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2777 Add item for weak_undefined.
2779 2017-12-14 Richard Biener <rguenther@suse.de>
2781 PR tree-optimization/67842
2782 * sese.h (bb_in_region): Remove #if 0'ed code.
2784 2017-12-14 Richard Biener <rguenther@suse.de>
2786 PR tree-optimization/83326
2787 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2788 may_be_zero parameter and handle it by not marking the first
2789 peeled copy as not exiting the loop.
2790 (try_peel_loop): Likewise.
2791 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2792 to handle the case of constant or zero iterations and perform
2793 loop header copying on-the-fly.
2795 2017-12-14 Richard Biener <rguenther@suse.de>
2797 PR tree-optimization/83418
2798 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2799 Instead of asserting we don't get unfolded comparisons deal with them.
2801 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2804 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2805 basic blocks. Assert debug bind insns don't appear outside of bbs,
2806 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2809 PR tree-optimization/83198
2810 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2811 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2812 value if arg is a REAL_CST with incompatible type.
2814 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2815 Bin Cheng <bin.cheng@arm.com>
2818 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2820 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2823 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2825 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2826 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2828 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2829 Alan Hayward <alan.hayward@arm.com>
2830 David Sherwood <david.sherwood@arm.com>
2832 * poly-int.h: New file.
2833 * poly-int-types.h: Likewise.
2834 * coretypes.h: Include them.
2835 (POLY_INT_CONVERSION): Define.
2836 * target.def (estimated_poly_value): New hook.
2837 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2838 * doc/tm.texi: Regenerate.
2839 * doc/poly-int.texi: New file.
2840 * doc/gccint.texi: Include it.
2841 * doc/rtl.texi: Describe restrictions on subreg modes.
2842 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2843 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2844 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2845 * targhooks.h (default_estimated_poly_value): Declare.
2846 * targhooks.c (default_estimated_poly_value): New function.
2847 * target.h (estimated_poly_value): Likewise.
2848 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2849 (wi::unary_traits): Delete.
2850 (wi::binary_traits::signed_shift_result_type): Define for
2851 offset_int << HOST_WIDE_INT, etc.
2852 (generic_wide_int::operator <<=): Define for all types and use
2853 wi::lshift instead of <<.
2854 (wi::hwi_with_prec): Add a default constructor.
2855 (wi::ints_for): New class.
2856 (operator <<): Define for all wide-int types.
2857 (operator /): New function.
2858 (operator %): Likewise.
2859 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2860 (ASSERT_MAYBE_NE_AT): New macros.
2862 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2863 Dominik Vogt <vogt@linux.vnet.ibm.com>
2866 * emit-rtl.c (init_emit): Remove ??? comment.
2867 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2868 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
2869 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
2870 alignment of 3 virtual registers to BITS_PER_WORD.
2872 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
2874 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
2876 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
2877 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
2879 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2882 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
2884 2017-12-13 Jakub Jelinek <jakub@redhat.com>
2886 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
2887 with the exception of debug begin stmt markers appear before
2891 * final.c (rest_of_handle_final): Call variable_tracking_main only
2892 if !flag_var_tracking.
2894 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2895 Jakub Jelinek <jakub@redhat.com>
2899 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
2900 labels debug stmts that can only appear after labels.
2902 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
2904 PR rtl-optimization/82398
2905 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
2906 EXPR_USEFULNESS in priority comparison.
2908 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
2910 PR rtl-optimization/83393
2911 * combine.c (move_deaths): If reg_stat points to a too new insn in
2912 last_death, do not use it: find the proper insn instead.
2914 2017-12-12 Jeff Law <law@redhat.com>
2916 PR tree-optimization/83298
2917 PR tree-optimization/83362
2918 PR tree-optimization/83383
2919 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
2920 push_value_range a public interface. Add new argument to
2921 record_ranges_from_stmt.
2922 * gimple-ssa-evrp-analyze.c
2923 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
2924 Update comments. Handle recording temporary equivalences.
2925 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
2926 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
2927 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
2928 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
2929 gimple-ssa-evrp-analyze.h.
2930 (record_temporary_equivalences_from_phis): Add new argument. When
2931 the PHI arg is an SSA_NAME, set the result's range to the range
2933 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
2934 from statements too.
2935 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
2936 Pass it down to children as needed.
2937 (thread_outgoing_edges): Likewise.
2938 (thread_across_edge): Likewise. Push/pop range state as needed.
2939 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
2941 2017-12-12 Julia Koval <julia.koval@intel.com>
2943 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
2944 (PTA_CANNONLAKE): Remove PTA_CLWB.
2946 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2949 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
2950 (vcondv2div2df): Likewise.
2951 (vconduv2dfv2di): Likewise.
2953 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2955 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
2956 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
2957 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
2958 ATTR_MATHFN_FPROUNDING_ERRNO.
2960 2017-12-12 Richard Biener <rguenther@suse.de>
2962 PR tree-optimization/83385
2963 * graphite-scop-detection.c (get_order, order): Remove.
2964 (bb_to_rpo): New global.
2966 (build_scops): Sort pbbs in RPO order.
2968 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
2970 * combine.c (simplify_set): Do not transform subregs to zero_extends
2971 if the destination mode is a vector mode.
2973 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2975 PR tree-optimization/83359
2976 * tree-cfg.h (fold_loop_internal_call): Declare.
2977 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
2978 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
2979 (find_loop_dist_alias): New function.
2980 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
2981 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
2984 PR tree-optimization/80631
2985 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
2986 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
2987 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
2988 hardcoding zero as the value if COND_EXPR is never true. For
2989 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
2990 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
2991 hardcoding MAX_EXPR as the reduction operation.
2992 (is_nonwrapping_integer_induction): Allow negative step.
2993 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
2994 vect_create_epilog_for_reduction, if no value is suitable, don't
2995 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
2997 2017-12-12 Richard Biener <rguenther@suse.de>
2999 PR tree-optimization/81889
3000 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
3001 range info from the non-wrapping IV instead of just the range
3004 2017-12-12 Julia Koval <julia.koval@intel.com>
3006 * config.gcc: Add vaesintrin.h.
3007 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3009 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3010 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3012 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3013 * config/i386/immintrin.h: Include vaesintrin.h.
3014 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3015 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3016 _mm_aesdec_epi128): New intrinsics.
3018 2017-12-12 Julia Koval <julia.koval@intel.com>
3020 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3021 OPTION_MASK_ISA_VAES_UNSET): New.
3022 (ix86_handle_option): Handle -mvaes.
3023 * config/i386/cpuid.h: Define bit_VAES.
3024 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3025 * config/i386/i386-c.c (__VAES__): New.
3026 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3027 (ix86_valid_target_attribute_inner_p): Ditto.
3028 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3029 * config/i386/i386.opt: Add -mvaes.
3030 * doc/invoke.texi: Ditto.
3032 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3034 * debug.h (gcc_debug_hooks): Add inline_entry.
3035 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3036 * debug.c (do_nothing_debug_hooks): Likewise.
3037 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3038 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3039 (dwarf2_lineno_debug_hooks): Likewise.
3041 * common.opt (gstatement-frontiers): New, setting
3042 debug_nonbind_markers_p.
3043 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
3044 * toplev.c (process_options): Autodetect value for debug statement
3046 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3047 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3049 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3050 markers. Integrate source bind into debug stmt expand loop.
3051 (pass_expand::execute): Check debug marker limit. Avoid deep
3052 TER and expand debug locations for debug bind insns only.
3053 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3055 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3056 * final.c (reemit_insn_block_notes): Take current block from
3057 nonbind markers. Declare note where it's first set.
3058 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3059 begin stmt markers if enabled.
3060 (notice_source_line): Handle nonbind markers. Fail if their
3061 location is unknown or that of builtins.
3062 (rest_of_handle_final): Convert begin stmt markers to notes if
3063 var-tracking didn't run.
3064 (rest_of_clean_state): Skip begin stmt markers.
3065 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3067 * function.c (allocate_struct_function): Set begin_stmt_markers.
3068 * function.h (struct function): Add debug_marker_count counter
3069 and debug_nonbind_markers flag.
3070 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3071 * gimple-low.c (lower_function_body): Adjust
3072 debug_nonbind_markers.
3073 (lower_stmt): Drop or skip gimple debug stmts.
3074 (lower_try_catch): Skip debug stmts.
3075 * gimple.c (gimple_build_debug_begin_stmt): New.
3076 (gimple_copy): Increment debug_marker_count if copying one.
3077 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3078 * gimplify.c (rexpr_location): New.
3079 (rexpr_has_location): New.
3080 (warn_switch_unreachable_r): Handle gimple debug stmts.
3081 (shortcut_cond_r): Call expr_location.
3083 (find_goto_label): New.
3084 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3086 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3088 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3089 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3090 (LANG_HOOKS_INITIALIZER): ... this.
3091 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3092 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3094 (update_ebb_live_info): Skip debug insn markers.
3095 * lra.c (debug_insn_static_data): Rename to...
3096 (debug_bind_static_data): ... this.
3097 (debug_marker_static_data): New.
3098 (lra_set_insn_recog_data): Select one of the above depending
3100 (lra_update_isn_regno_info): Don't assume debug insns have
3102 (push_insns): Skip debug insns.
3103 * lto-streamer-in.c (input_function): Drop debug stmts
3104 depending on active options. Adjust debug_nonbind_markers.
3105 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3106 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3107 begin stmt marker notes.
3108 (print_insn): Likewise.
3109 * recog.c (extract_insn): Recognize rtl for debug markers.
3110 * rtl.def (DEBUG_MARKER): New.
3111 * tree-inline.c: Include params.h.
3112 (remap_gimple_stmt): Handle nonbind markers.
3113 (maybe_move_debug_stmts_to_successors): Likewise.
3114 (copy_debug_stmt): Likewise.
3115 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
3116 markers regardless of no side effects.
3117 (tsi_link_before): Don't update container's side effects when adding
3118 a begin stmt marker.
3119 (tsi_link_after): Likewise.
3120 (expr_first): Skip begin stmt markers.
3121 (expr_last): Likewise.
3122 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
3123 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3124 Disregard nonbind markers.
3125 * tree.c (make_node_stat): Don't set side effects for begin stmt
3127 (build1_stat): Likewise.
3128 * tree.def (DEBUG_BEGIN_STMT): New.
3129 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
3130 * var-tracking.c (delete_debug_insns): Renamed to...
3131 (delete_vta_debug_insns): ... this.
3132 (reemit_marker_as_note): New.
3133 (vt_initialize): Reemit markers.
3134 (delete_vta_debug_insns): Likewise.
3135 (vt_debug_insns_local): Reemit or delete markers.
3136 (variable_tracking_main_1): Likewise.
3137 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
3138 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
3139 (gimple_debug_nonbind_marker_p): New.
3140 (gimple_build_debug_bind): Adjust.
3141 (gimple_build_debug_begin_stmt): New.
3142 * doc/invoke.texi (max-debug-marker-count): New param.
3143 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
3144 (debug_parameter_ref, debug_marker): New.
3145 (NOTE_INSN_BEGIN_STMT): New.
3146 (DEBUG_INSN): Describe begin stmt markers.
3148 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
3149 without debug insns, we wouldn't, but clean up debug insns
3150 after a control flow insn nevertheless.
3152 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
3153 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
3154 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
3155 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
3156 (rtl_tidy_fallthru_edge): Likewise.
3157 (rtl_verify_fallthru): Likewise.
3158 (rtl_verify_bb_layout): Likewise.
3159 (skip_insns_after_block): Likewise.
3160 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
3161 * dwarf2out.c: Include print-rtl.h.
3162 (dwarf2out_next_real_insn): New.
3163 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
3164 Dump debug binds in asm comments.
3165 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
3166 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
3167 callers to use gsi_start_nondebug_bb instead.
3168 (gsi_after_labels): Skip gimple debug stmts.
3169 (gsi_start_nondebug): New.
3170 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
3171 (proper_loop_form_for_interchange): Adjust.
3172 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
3173 * gimple.h (gimple_seq_last_nondebug_stmt): New.
3174 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
3175 (collect_fallthrough_labels): Likewise.
3176 (should_warn_for_implicit_fallthrough): Likewise.
3177 (warn_implicit_fallthrough_r): Likewise.
3178 (expand_FALLTHROUGH_r): Likewise.
3179 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
3180 (graphite_copy_stmts_from_block): Skip nonbind markers.
3181 * haifa-sched.c (sched_extend_bb): Skip debug insns.
3182 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
3183 * jump.c (clean_barriers): Skip debug insns.
3184 * omp-expand.c (expand_parallel_call): Skip debug insns.
3185 (expand_task_call): Likewise.
3186 (remove_exit_barrier): Likewise.
3187 (expand_omp_taskreg): Likewise.
3188 (expand_omp_for_init_counts): Likewise.
3189 (expand_omp_for_generic): Likewise.
3190 (expand_omp_for_static_nochunk): Likewise.
3191 (expand_omp_for_static_chunk): Likewise.
3192 (expand_omp_simd): Likewise.
3193 (expand_omp_taskloop_for_outer): Likewise.
3194 (expand_omp_taskloop_for_inner): Likewise.
3195 (expand_oacc_for): Likewise.
3196 (expand_omp_sections): Likewise.
3197 (expand_omp_single): Likewise.
3198 (expand_omp_synch): Likewise.
3199 (expand_omp_atomic_load): Likewise.
3200 (expand_omp_atomic_store): Likewise.
3201 (expand_omp_atomic_fetch_op): Likewise.
3202 (expand_omp_atomic_pipeline): Likewise.
3203 (expand_omp_atomic_mutex): Likewise.
3204 (expand_omp_target): Likewise.
3205 (grid_expand_omp_for_loop): Likewise.
3206 (grid_expand_target_grid_body): Likewise.
3207 (build_omp_regions_1): Likewise.
3208 * omp-low.c (check_combined_parallel): Skip debug stmts.
3209 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
3210 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
3211 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
3213 * tree-cfg.c (make_blobs_1): Skip debug stmts.
3214 (make_edges): Likewise.
3215 (cleanup_dead_labels): Likewise.
3216 (gimple_can_merge_blocks_p): Likewise.
3217 (stmt_starts_bb_p): Likewise.
3218 (gimple_block_label): Likewise.
3219 (gimple_redirect_edge_and_branch): Likewise.
3220 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
3222 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
3224 * tree-pretty-print (print_declaration): Omit initializer in slim
3226 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
3228 (eliminate_unnecessary_stmts): Stabilize block removal order.
3229 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
3230 * var-tracking.c (get_first_insn): New.
3231 (vt_emit_notes): Call it.
3232 (vt_initialize): Walk any insns before the first BB.
3233 (delete_debug_insns): Likewise.
3235 * gimple.h (enum gimple_debug_subcode): Add
3236 GIMPLE_DEBUG_BEGIN_STMT.
3237 (gimple_debug_begin_stmt_p): New.
3238 (gimple_debug_nonbind_marker_p): New.
3239 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
3240 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
3241 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
3242 * insn-notes.def (BEGIN_STMT): New.
3243 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
3244 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
3245 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
3246 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
3247 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
3248 (INSN_DEBUG_MARKER_KIND): New.
3249 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
3250 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
3251 (INSN_VAR_LOCATION_PTR): New.
3252 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
3254 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
3255 deps only if debug bind insns are enabled.
3256 (pass_expand::execute): Avoid deep TER and expand
3257 debug locations for debug bind insns only.
3258 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
3259 debug stmts special handling down to debug bind stmts.
3260 * combine.c (try_combine): Narrow debug insns special handling
3261 down to debug bind insns.
3262 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
3263 Narrow debug insns preexisting special handling down to debug
3265 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
3266 handling down to debug bind insns.
3267 * function.c (instantiate_virtual_regs): Skip debug markers,
3268 adjust handling of debug binds.
3269 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
3270 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
3271 * haifa-sched.c (schedule_insn): Narrow special handling of debug
3272 insns to debug bind insns.
3273 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
3274 special handling of debug stmts to debug bind stmts.
3275 * ipa-split.c (split_function): Likewise.
3276 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
3277 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
3279 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
3280 * regrename.c (build_def_use): Likewise.
3281 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
3282 (pass_cprop_hardreg): Narrow special casing of debug insns to
3284 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
3285 * reload1.c (reload): Likewise.
3286 * sese.c (sese_insert_phis_for_liveouts): Narrow special
3287 casing of debug stmts to debug bind stmts.
3288 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
3289 * ssa-iterators.h (num_imm_uses): Likewise.
3290 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
3291 debug stmts to debug bind stmts.
3292 * tree-inline.c (tree_function_versioning): Narrow special casing
3293 of debug stmts to debug bind stmts.
3294 * tree-loop-distribution.c (generate_loops_for_partition):
3295 Narrow special casing of debug stmts to debug bind stmts.
3296 * tree-sra.c (analyze_access_subtree): Narrow special casing
3297 of debug stmts to debug bind stmts.
3298 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
3299 stmts to debug bind stmts.
3300 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
3301 casing of debug stmts to debug bind stmts.
3302 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
3303 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
3304 casing of debug stmts to debug bind stmts.
3305 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3307 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
3308 debug stmts to debug bind stmts.
3309 (gimple_replace_ssa_lhs): Likewise.
3310 (insert_debug_temp_for_var_def): Likewise.
3311 (insert_debug_temps_for_defs): Likewise.
3312 (reset_debug_uses): Likewise.
3313 * tree-ssanames.c (release_ssa_name_fn): Likewise.
3314 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
3315 (adjust_debug_stmts): Likewise.
3316 (adjust_phi_and_debug_stmts): Likewise.
3317 (vect_do_peeling): Likewise.
3318 * tree-vect-loop.c (vect_transform_loop): Likewise.
3319 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
3320 * var-tracking.c (adjust_mems): Narrow special casing of debug
3321 insns to debug bind insns.
3322 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
3323 (compute_bb_dataflow, vt_find_locations): Likewise.
3324 (vt_expand_loc, emit_notes_for_changes): Likewise.
3325 (vt_init_cfa_base): Likewise.
3326 (vt_emit_notes): Likewise.
3327 (vt_initialize): Likewise.
3328 (vt_finalize): Likewise.
3330 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
3331 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
3332 (next_nonnote_nondebug_insn_bb): New.
3333 (prev_nonnote_nondebug_insn_bb): New.
3334 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
3335 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
3336 (prev_nonnote_nondebug_insn_bb): Declare.
3337 (next_nonnote_nondebug_insn_bb): Declare.
3338 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
3339 * cfgrtl.c (get_last_bb_insn): Likewise.
3340 * lra.c (push_insns): Likewise.
3342 2017-12-11 David Malcolm <dmalcolm@redhat.com>
3345 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
3347 * selftest.c (selftest::test_runner::test_runner): New ctor.
3348 (selftest::test_runner::~test_runner): New dtor.
3349 * selftest.h (class selftest::test_runner): New class.
3351 2017-12-11 Carl Love <cel@us.ibm.com>
3353 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
3354 vec_extract_fp32_from_shortl]): Add #defines.
3355 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
3356 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
3357 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
3358 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
3359 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
3360 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
3361 * doc/extend.texi: Add documentation for the added builtins.
3363 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
3365 PR rtl-optimization/80693
3366 PR rtl-optimization/81019
3367 PR rtl-optimization/81020
3368 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
3369 are not mentioned in i3. Place the REG_UNUSED note on i2,
3370 possibly modified to REG_DEAD, if it did not originate in i3.
3372 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3374 * recog.c (store_data_bypass_p_1): New function.
3375 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
3376 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
3378 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
3380 PR rtl-optimization/83361
3381 * ifcvt.c (if_convert): Call fixup_partitions.
3383 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
3385 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3386 early folding of splat_u{8,16,32}.
3388 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3390 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
3391 output_operand_lossage first argument with capital letter.
3392 (aarch64_override_options): Don't start error and sorry first argument
3393 with capital letter.
3395 2017-12-11 Andi Kleen <ak@linux.intel.com>
3397 PR gcov-profile/83355
3398 * auto-profile.c (string_table::get_index_by_decl): Don't
3399 recurse when abstract origin points to itself.
3401 2017-12-11 Bin Cheng <bin.cheng@arm.com>
3403 PR tree-optimization/83320
3404 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
3405 (prune_datarefs_not_in_loop): Ditto.
3407 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
3409 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
3411 2017-12-10 Jakub Jelinek <jakub@redhat.com>
3413 PR tree-optimization/83337
3414 * gimple-loop-interchange.cc (compute_access_stride): Handle
3415 bitfield DRs properly.
3417 2017-12-09 Jakub Jelinek <jakub@redhat.com>
3419 PR tree-optimization/83338
3420 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
3421 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
3422 vector of unsigned integers to vector of signed integers.
3424 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
3426 PR rtl-optimization/83317
3427 * lra-constraints.c (process_address_1): Add insn code check.
3429 2017-12-08 Michael Matz <matz@suse.de>
3431 Fix PR tree-optimization/83323
3432 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
3433 head-controlled loops and loop BBs.
3434 * common.opt (funroll-and-jam): Remove, instead ...
3435 (floop-unroll-and-jam): ... reuse this option.
3436 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
3437 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
3438 (-floop-unroll-and-jam): ... this option.
3440 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3442 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
3443 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
3444 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
3446 2017-12-08 Richard Biener <rguenther@suse.de>
3449 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
3450 handle accesses outside of zero-sized vars.
3452 2017-12-08 Martin Jambor <mjambor@suse.cz>
3454 PR tree-optimization/83141
3455 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
3456 test for MEM_REFs implicitely changing types with padding. Remove
3458 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
3460 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3462 * config/arc/arc.c (arc_attribute_table): Add exclusions to
3464 * config/avr/avr.c (avr_attribute_table): Likewise.
3465 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3466 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3467 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3468 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
3469 of struct attribute_spec.
3470 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3472 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
3475 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
3477 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
3480 * profile-count.c (profile_count::from_gcov_type): Move from
3481 profile-count.h; handle overflow.
3482 * profile-count.h (profile_count::from_gcov_type): Move offline.
3484 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
3486 PR rtl-optimization/83304
3487 * combine.c (move_deaths): If we do not know where a register died,
3490 2017-12-08 Richard Biener <rguenther@suse.de>
3492 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
3493 Provide -fopt-info-loop feedback when we interchange in a nest.
3495 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3497 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
3498 for armv6 ARM CPU IDs.
3500 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3502 * common/config/arm/arm-common.c: Include <algorithm>.
3503 (INCLUDE_VECTOR): Define.
3504 (compare_opt_names): New function.
3505 (arm_rewrite_selected_arch): Only strip out extensions that can be
3506 expressed through -mfpu. Sort the remaining extensions
3509 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3511 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
3512 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
3513 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
3514 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
3516 2017-06-08 Tristan Gingold <gindold@adacore.com>
3519 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
3520 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
3522 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3524 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
3525 of struct attribute_spec.
3527 2017-12-08 Julia Koval <julia.koval@intel.com>
3529 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
3530 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
3531 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
3532 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
3533 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
3534 _mm_maskz_dpwssds_epi32): Ditto.
3536 2017-12-08 Richard Biener <rguenther@suse.de>
3538 PR tree-optimization/81303
3539 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
3540 conditions try to create a comparison vector type matching
3541 the data vector type.
3542 (vectorizable_condition): Adjust.
3543 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3544 Leave invariant conditions alone in case we can vectorize those.
3546 2017-12-08 Julia Koval <julia.koval@intel.com>
3548 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
3549 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
3550 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
3551 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
3552 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
3554 2017-12-08 Julia Koval <julia.koval@intel.com>
3556 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
3557 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
3558 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
3559 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
3560 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
3561 _mm_maskz_dpbusds_epi32): New intrinsics.
3563 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3565 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
3567 (ld<bh>io_signed): Likewise.
3568 (st<bhw>io): Likewise.
3569 * config/nios2/predicates.md (ldstio_memory_operand): Allow
3570 SMALL_INT12 constant integer operand.
3572 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3574 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
3576 (nios2_large_constant_allowed): ...this. Adjust uses.
3577 (nios2_plus_symbolic_constant_p): Rename to...
3578 (nios2_plus_large_constant_p): ...this. Adjust uses.
3579 (nios2_legitimate_address_p): Correct CONST_INT handling.
3580 (nios2_symbolic_memory_operand_p): Rename to...
3581 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
3582 (nios2_large_constant_p): Check for large constant integers too.
3583 (nios2_split_large_constant): Handle constant integers.
3584 (nios2_split_symbolic_memory_operand): Rename to...
3585 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
3586 (nios2_legitimize_constant_address): Handle constant integers.
3587 (r0rel_constant_p): Handle small constant integers.
3588 (nios2_print_operand_address): Handle r0-relative integer addresses.
3589 * config/nios2/nios2-protos.h: Adjust for renamed functions.
3590 * config/nios2/nios2.md: Adjust for renamed functions.
3592 2017-12-07 Andrew Waterman <andrew@sifive.com>
3594 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
3595 (TARGET_HAVE_SRODATA_SECTION): New define.
3596 (riscv_select_section): New function.
3598 2017-12-08 Joseph Myers <joseph@codesourcery.com>
3599 Alexander Monakov <amonakov@ispras.ru>
3600 Jakub Jelinek <jakub@redhat.com>
3603 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
3605 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3607 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
3608 a bool success value. Don't call output_operand_lossage here.
3609 (aarch64_print_ldpstp_address): Return a bool success value.
3610 (aarch64_print_operand_address): Call output_addr_const if
3611 aarch64_print_address_internal fails.
3612 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
3613 'y'; call output_operand_lossage instead. Call output_operand_lossage
3614 if aarch64_print_ldpstp_address fails.
3616 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3618 * tree-vector-builder.h
3619 (tree_vector_builder::binary_encoded_nelts): Declare.
3620 * tree-vector-builder.c
3621 (tree_vector_builder::binary_encoded_nelts): New function.
3622 * fold-const.c (negate_expr_p): Likewise.
3623 (operand_equal_p, fold_checksum_tree): Likewise.
3624 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3625 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3626 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3627 (uniform_vector_p): Likewise.
3628 * varasm.c (const_hash_1, compare_constant): Likewise.
3629 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3630 (valid_lattice_transition): Operate directly on the VECTOR_CST
3632 * ipa-icf.c: Include tree-vector-builder.h.
3633 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3634 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3636 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3638 * tree.c (build_vector): Delete.
3639 * tree.h (build_vector): Make static and move into the self-testing
3642 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3644 * vector-builder.h (vector_builder::derived): New const overload.
3645 (vector_builder::elt): New function.
3646 * tree-vector-builder.h (tree_vector_builder::type): New function.
3647 (tree_vector_builder::apply_step): Declare.
3648 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3650 * gimple-fold.h (tree_vector_builder): Declare.
3651 (gimple_build_vector): Take a tree_vector_builder instead of a
3652 type and vector of elements.
3653 * gimple-fold.c (gimple_build_vector): Likewise.
3654 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3656 (get_initial_defs_for_reduction): Likewise.
3657 (vectorizable_induction): Likewise.
3659 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3661 * tree-vector-builder.h
3662 (tree_vector_builder::new_binary_operation): Declare.
3663 * tree-vector-builder.c
3664 (tree_vector_builder::new_binary_operation): New function.
3665 * fold-const.c (fold_relational_const): Use it.
3666 (const_binop): Likewise. Check that both input vectors have
3667 the same number of elements, thus excluding things like WIDEN_SUM.
3668 Check whether it is possible to operate directly on the encodings
3671 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3673 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3674 new_unary_operation, operating only on the encoded elements.
3675 (const_unop): Likewise.
3676 (exact_inverse): Likewise.
3677 (distributes_over_addition_p): New function.
3678 (const_binop): Use tree_vector_builder and new_unary_operation
3679 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3680 on the encoded elements unless the encoding is strided and the
3681 operation does not distribute over addition.
3682 (fold_convert_const): Use tree_vector_builder and
3683 new_unary_operation. Operate only on the encoded elements
3684 for truncating integer conversions, or for non-stepped encodings.
3686 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3688 * config/sparc/sparc.c: Include tree-vector-builder.h.
3689 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3690 * expmed.c: Include tree-vector-builder.h.
3691 (make_tree): Use tree_vector_builder instead of build_vector.
3692 * fold-const.c: Include tree-vector-builder.h.
3693 (const_binop): Use tree_vector_builder instead of build_vector.
3694 (const_unop): Likewise.
3695 (native_interpret_vector): Likewise.
3696 (fold_vec_perm): Likewise.
3697 (fold_ternary_loc): Likewise.
3698 * gimple-fold.c: Include tree-vector-builder.h.
3699 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3701 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3702 (simplify_vector_constructor): Use tree_vector_builder instead
3704 * tree-vect-generic.c: Include tree-vector-builder.h.
3705 (add_rshift): Use tree_vector_builder instead of build_vector.
3706 (expand_vector_divmod): Likewise.
3707 (optimize_vector_constructor): Likewise.
3708 * tree-vect-loop.c: Include tree-vector-builder.h.
3709 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3710 of build_vector. Explicitly use a stepped encoding for
3712 * tree-vect-slp.c: Include tree-vector-builder.h.
3713 (vect_get_constant_vectors): Use tree_vector_builder instead
3715 (vect_transform_slp_perm_load): Likewise.
3716 (vect_schedule_slp_instance): Likewise.
3717 * tree-vect-stmts.c: Include tree-vector-builder.h.
3718 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3719 (vect_gen_perm_mask_any): Likewise.
3720 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3721 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3723 (build_vector_from_val): Likewise. Explicitly use a duplicate
3726 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3728 * doc/generic.texi (VECTOR_CST): Describe new representation of
3730 * vector-builder.h: New file.
3731 * tree-vector-builder.h: Likewise.
3732 * tree-vector-builder.c: Likewise.
3733 * Makefile.in (OBJS): Add tree-vector-builder.o.
3734 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3735 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3736 (tree_vector): Change the number of elements to
3737 vector_cst_encoded_nelts.
3738 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3739 (VECTOR_CST_ELTS): Delete.
3740 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3741 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3742 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3743 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3744 (VECTOR_CST_ENCODED_ELT): Likewise.
3745 (vector_cst_encoded_nelts): New function.
3746 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3747 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3748 (vector_cst_int_elt, vector_cst_elt): Declare.
3749 * tree.c: Include tree-vector-builder.h.
3750 (tree_code_size): Abort if passed VECTOR_CST.
3751 (tree_size): Update for new VECTOR_CST layout.
3752 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3753 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3754 (build_vector): Use tree_vector_builder.
3755 (vector_cst_int_elt, vector_cst_elt): New functions.
3756 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3757 encoded elements and then create the vector in the canonical form.
3758 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3759 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3760 (tree_c_tests): Call test_vector_cst_patterns.
3761 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3763 (hash_tree): Likewise.
3764 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3765 (streamer_write_tree_header): Likewise.
3766 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3767 (streamer_alloc_tree): Likewise. Update call to make_vector.
3768 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3770 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3772 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3773 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3774 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3776 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3777 Richard Biener <rguenther@suse.de>
3779 PR tree-optimization/81303
3780 * Makefile.in (gimple-loop-interchange.o): New object file.
3781 * common.opt (floop-interchange): Reuse the option from graphite.
3782 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3783 -floop-interchange and mention it for -O3.
3784 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3785 * gimple-loop-interchange.cc: New file.
3786 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3787 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3788 * passes.def (pass_linterchange): New pass.
3789 * timevar.def (TV_LINTERCHANGE): New time var.
3790 * tree-pass.h (make_pass_linterchange): New declaration.
3791 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3792 interchange. Record IV before/after increment in new parameters.
3793 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3794 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3796 (check_reduction_path): ...New function here.
3797 * tree-vectorizer.h (check_reduction_path): New declaration.
3799 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3802 PR rtl-optimization/80818
3803 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3804 always early clobbered.
3805 * lra-lives.c (process_bb_lives): Check input hard regs for early
3806 clobbered non-operand hard reg.
3808 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3811 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3812 types_compatible_p, just that TYPE_MODE is the same.
3814 2017-12-07 Martin Sebor <msebor@redhat.com>
3817 * attribs.c (empty_attribute_table): Initialize new member of
3818 struct attribute_spec.
3819 (decl_attributes): Add argument. Handle mutually exclusive
3820 combinations of attributes.
3821 (selftests::test_attribute_exclusions): New function.
3822 (selftests::attribute_c_tests): Ditto.
3823 * attribs.h (decl_attributes): Add default argument.
3824 * selftest.h (attribute_c_tests): Declare.
3825 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3826 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3828 * doc/extend.texi (Common Function Attributes): Update const and pure.
3829 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3830 of struct attribute_spec.
3831 * config/arc/arc.c (arc_attribute_table): Same.
3832 * config/arm/arm.c (arm_attribute_table): Same.
3833 * config/avr/avr.c ( avr_attribute_table): Same.
3834 * config/bfin/bfin.c (bfin_attribute_table): Same.
3835 * config/cr16/cr16.c (cr16_attribute_table): Same.
3836 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3837 * config/h8300/h8300.c (h8300_attribute_table): Same.
3838 * config/i386/i386.c (ix86_attribute_table): Same.
3839 * config/ia64/ia64.c (ia64_attribute_table): Same.
3840 * config/m32c/m32c.c (m32c_attribute_table): Same.
3841 * config/m32r/m32r.c (m32r_attribute_table): Same.
3842 * config/m68k/m68k.c (m68k_attribute_table): Same.
3843 * config/mcore/mcore.c (mcore_attribute_table): Same.
3844 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3845 * config/mips/mips.c (mips_attribute_table): Same.
3846 * config/msp430/msp430.c (msp430_attribute_table): Same.
3847 * config/nds32/nds32.c (nds32_attribute_table): Same.
3848 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3849 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3850 * config/rl78/rl78.c (rl78__attribute_table): Same.
3851 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3852 * onfig/rx/rx.c (rx_attribute_table): Same.
3853 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3854 * config/sh/sh.c (sh_attribute_table): Same.
3855 * config/sparc/sparc.c (sparc_attribute_table): Same.
3856 * config/spu/spu.c (spu_attribute_table): Same.
3857 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3858 * config/v850/v850.c (v850_attribute_table): Same.
3859 * config/visium/visium.c (visium_attribute_table): Same.
3861 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3864 * config/arm/arm.c (INCLUDE_STRING): Define.
3865 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3866 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
3869 2017-12-07 Michael Matz <matz@suse.de>
3871 Add unroll and jam pass
3873 * gimple-loop-jam.c: New file.
3874 * Makefile.in (OBJS): Add gimple-loop-jam.o.
3875 * common.opt (funroll-and-jam): New option.
3876 * opts.c (default_options_table): Add unroll-and-jam at -O3.
3877 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
3878 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
3879 * passes.def: Add pass_loop_jam.
3880 * timevar.def (TV_LOOP_JAM): Add.
3881 * tree-pass.h (make_pass_loop_jam): Declare.
3882 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
3883 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
3884 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
3885 to flow_loop_tree_node_add.
3886 (duplicate_subloops, copy_loops_to): Append to sibling list.
3887 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
3888 * doc/invoke.texi (-funroll-and-jam): Document new option.
3889 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
3891 2017-12-07 Richard Biener <rguenther@suse.de>
3893 PR tree-optimization/83296
3894 PR tree-optimization/67769
3895 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
3896 flow sensitive info in an unrelated BB.
3897 (value_replacement): Use reset_flow_sensitive_info.
3898 (minmax_replacement): Reset flow sensitive info on the def
3899 we move. Do not reset flow sensitive info in the whole BB
3900 we move the stmt to.
3901 (abs_replacement): Likewise.
3903 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3906 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3907 rs6000_cpu to the given -mcpu=, or to the default processor.
3909 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3911 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
3912 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
3913 instead of rs6000_cpu_attr.
3914 (rs6000_adjust_cost): Ditto.
3915 (is_microcoded_insn): Ditto.
3916 (rs6000_adjust_priority): Ditto.
3917 (rs6000_issue_rate): Ditto.
3918 (rs6000_use_sched_lookahead): Ditto.
3919 (rs6000_use_sched_lookahead_guard): Ditto.
3920 (rs6000_sched_reorder): Ditto.
3921 (force_new_group): Ditto.
3922 * config/rs6000/rs6000.md (cpu attribute): Ditto.
3923 (group_ending_nop): Ditto.
3925 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3927 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
3928 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
3929 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
3930 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
3932 (rs6000_reassociation_width): Ditto.
3933 (rs6000_emit_epilogue): Ditto.
3934 (rs6000_adjust_cost): Ditto.
3935 (is_microcoded_insn): Ditto.
3936 (is_cracked_insn): Ditto.
3937 (rs6000_adjust_priority): Ditto.
3938 (rs6000_sched_reorder): Ditto.
3939 (rs6000_sched_reorder2): Ditto.
3940 (insn_must_be_first_in_group): Ditto.
3941 (insn_must_be_last_in_group): Ditto.
3942 (rs6000_register_move_cost): Ditto.
3943 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
3946 2017-12-07 Julia Koval <julia.koval@intel.com>
3948 * config.gcc: Add vaesintrin.h.
3949 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3951 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3952 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3954 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3955 * config/i386/immintrin.h: Include vaesintrin.h.
3956 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3957 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3958 _mm_aesdec_epi128): New intrinsics.
3960 2017-12-06 David Malcolm <dmalcolm@redhat.com>
3962 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
3963 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
3964 c-family/c-spellcheck.cc.
3965 (best_macro_match::best_macro_match): Likewise.
3967 (struct edit_distance_traits<cpp_hashnode *>): Move to
3968 c-family/c-spellcheck.h.
3969 (class best_macro_match): Likewise.
3971 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3973 PR tree-optimization/83293
3974 * gimple-ssa-strength-reduction.c (insert_initializers): Use
3975 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
3976 might insert into empty bb.
3979 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
3980 simplify for plus with :c added, and pointer_plus without that.
3981 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
3982 with undefined overflow and the conversion is not widening,
3983 perform negation in utype and only convert to type afterwards.
3984 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
3985 simplify for plus with :c added, and pointer_plus without that.
3986 If type is integral with undefined overflow and the conversion is
3987 not widening, perform minus in utype and only convert to type
3988 afterwards. Move the last pointer_diff_expr simplify into the
3991 2017-12-06 Martin Sebor <msebor@redhat.com>
3993 PR tree-optimization/82646
3994 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
3995 strncpy, not maxlen.
3997 2017-12-06 Martin Sebor <msebor@redhat.com>
3999 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
4002 PR tree-optimization/83075
4003 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
4004 strncat/strncpy don't change length of source string.
4006 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4009 2017-11-29 Martin Aberg <maberg@gaisler.com>
4011 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4012 to prevent b2bst errata sequence.
4013 (sqrtdf2_fix): Likewise.
4015 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4017 PR tree-optimization/81945
4018 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4019 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4020 to dest_cfun has orig_loop_num set, either remap it to the new
4021 loop number if the loop got moved too, or clear it.
4023 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4025 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4027 (thunderx2t99_nothing): New insn reservation.
4028 (thunderx2t99_mrs): New insn reservation.
4029 (thunderx2t99_multiple): New insn reservation.
4030 (thunderx2t99_alu_basi): Add bfx to reservation.
4031 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4033 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4036 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4038 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4040 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4041 tree-ssa-pre.c::remove_dead_inserted_code.
4042 * tree-ssa-dce.h: New file.
4043 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
4044 (remove_dead_inserted_code): Move and rename to function
4045 tree-ssa-dce.c::simple_dce_from_worklist.
4046 (pass_pre::execute): Update use.
4048 2017-12-05 Richard Biener <rguenther@suse.de>
4050 PR tree-optimization/83277
4051 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4052 to code-gen liveout vars.
4054 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4056 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4057 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4058 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4061 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4063 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4065 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4066 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4069 2017-12-05 Richard Biener <rguenther@suse.de>
4071 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4072 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4073 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4074 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4076 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4078 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4080 2017-12-05 Martin Liska <mliska@suse.cz>
4081 Jakub Jelinek <jakub@redhat.com>
4083 * doc/invoke.texi: Document the options.
4084 * flag-types.h (enum sanitize_code): Add
4085 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4086 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4087 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4088 * opts.c: Define new sanitizer options.
4089 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4090 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4092 2017-12-05 Julia Koval <julia.koval@intel.com>
4094 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4095 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4096 (ix86_handle_option): Handle -mavx512vnni.
4097 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4098 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4099 * config/i386/i386-c (__AVX512VNNI__): New.
4100 * config/i386/i386.c (ix86_target_string): Handle new option.
4101 (ix86_valid_target_attribute_inner_p): Handle new option.
4102 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4103 * config/i386/i386.opt (mavx512vnni): New option.
4105 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4108 * athlon.md: Disable for generic.
4109 * haswell.md: Enable for generic.
4110 * i386.c (ix86_sched_init_global): Add core hooks for generic.
4111 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
4113 (ix86_adjust_cost): Move generic to haswell path.
4115 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
4117 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
4118 instead of MEM_P in a couple more places. Fix formatting issues.
4120 2017-12-04 Jim Wilson <jimw@sifive.com>
4122 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
4123 instead of GP_REG_LAST-1.
4124 (riscv_adjust_libcall_cfi_prologue): Likewise.
4125 (riscv_adjust_libcall_cri_epilogue): Likewise.
4126 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
4129 2017-12-04 Luis Machado <luis.machado@linaro.org>
4131 * ipa-pure-const.c (check_decl): Add missing newline.
4132 (state_from_flags): Likewise.
4134 2017-12-04 Jeff Law <law@redhat.com>
4136 PR tree-optimizatin/78496
4137 * gimple-ssa-evrp-analyze.h
4138 (evrp_range_analyzer::get_vr_values): Simplify.
4139 * gimple-ssa-evrp-analyze.c: Corresponding changes.
4140 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
4141 and gimple-ssa-evrp-analyze.h.
4142 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
4143 (simplify_stmt_for_jump_threading): Copy a blob of code from
4144 tree-vrp.c to use ranges to simplify statements.
4145 (dom_opt_dom_walker::before_dom_children): Call
4146 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
4147 (dom_opt_dom_walker::after_dom_children): Similarly for
4148 evrp_range_analyzer::leave.
4149 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
4152 * gimple-ssa-evrp-analyze.c
4153 (evrp_range_analyzer::extract_range_from_stmt): Always use
4154 vr_values::update_value_range so preexisting range info is
4155 medged with new range info, even if the new range is VR_VARYING.
4157 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4159 * combine.c: Adjust comment.
4160 (use_crosses_set_p): Delete.
4161 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
4162 (try_combine): Ditto.
4164 2017-12-04 Richard Biener <rguenther@suse.de>
4166 PR tree-optimization/83255
4167 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
4168 Re-add zero-iteration check.
4170 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4172 PR rtl-optimization/83245
4173 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
4174 hard registers as earlyclobber, also if not in an asm.
4176 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4180 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4183 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4184 rs6000_cpu based on cpu_index, not tune_index.
4186 2017-12-04 Richard Biener <rguenther@suse.de>
4188 PR tree-optimization/83238
4189 * graphite-scop-detection.c (scop_detection::merge_sese): Make
4190 code match comment, rejecting invalid SESE regions.
4192 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
4194 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
4195 require base operand is a REG_POINTER prior to reload on targets
4196 with non-equivalent space registers.
4198 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4200 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
4201 (update_specialized_profile): Fix updating of counts.
4202 (perhaps_add_new_callers): Likewise.
4204 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4207 * x86-tune.def: Remove obsolette FIXMEs.
4208 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
4209 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
4210 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
4212 (X86_TUNE_PAD_RETURNS): Disable for generic.
4214 2017-12-02 Jakub Jelinek <jakub@redhat.com>
4216 PR tree-optimization/83170
4217 PR tree-optimization/83241
4218 * gimple-ssa-store-merging.c
4219 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
4220 gimple_vuse (ins_stmt) in case it has changed.
4221 (imm_store_chain_info::output_merged_store): Likewise.
4223 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
4227 * tree-cfg.c (pass_warn_function_return::execute): Handle
4228 __builtin_ubsan_handle_missing_return like __builtin_unreachable
4229 with BUILTINS_LOCATION.
4233 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
4234 is BLKmode for vector field with vector raw mode, use TYPE_MODE
4235 instead of DECL_MODE.
4237 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
4238 last argument to rtx pointer.
4239 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
4240 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
4241 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
4242 depending on the chosen ISAs.
4243 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
4244 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
4245 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
4247 * config/i386/sse.md (mov<mode>_internal): Likewise.
4248 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
4250 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4252 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
4254 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4256 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
4257 parameters from prototype.
4258 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
4259 parameters. Don't print a comment.
4260 (emit_fusion_gpr_load): Adjust.
4261 (emit_fusion_load_store): Adjust.
4262 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
4263 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
4264 comment on the second line.
4266 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4269 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4270 rs6000_cpu based on cpu_index, not tune_index.
4272 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4274 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
4275 which_alternative instead of which_alternative + 1.
4276 (output_asm_insn): Print an extra tab if the template is short.
4278 2017-12-01 Jim Wilson <jimw@sifive.com>
4280 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
4282 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
4283 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
4285 * doc/tm.texi: Regenerate.
4287 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
4290 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
4291 whether we can allocate pseudos before trying to fix an address.
4292 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
4293 memory address is indexed or indirect.
4294 (floatuns_<mode>si2_hw2): Likewise.
4296 2017-12-01 Jason Merrill <jason@redhat.com>
4298 * Makefile.in (TAGS): Add c-family/*.cc.
4300 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
4302 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
4303 (call_value_insn): Likewise.
4304 (sibcall_insn): Likewise.
4305 (sibcall_value_insn): Likewise.
4306 (movsi_aarch64): Likewise.
4307 (movdi_aarch64): Likewise.
4308 (add_losym_): Likewise.
4309 (ldr_got_small_): Likewise.
4310 (ldr_got_small_sidi): Likewise.
4311 (ldr_got_small_28k_): Likewise.
4312 (ldr_got_small_28k_sidi): Likewise.
4313 * config/aarch64/aarch64.c (aarch64_print_address_internal):
4314 Move output_addr_const to symbolic case. Add error check.
4316 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4318 * config/s390/predicates.md (plus16_Q_operand): New predicate.
4319 * config/s390/s390.md: Disable MVC merging peephole if it would
4320 disable operand forwarding.
4321 (new peephole2): Split MVCs if it would turn them into up to 2
4324 2017-12-01 Richard Biener <rguenther@suse.de>
4326 PR tree-optimization/83232
4327 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
4328 detection of same access. Instead of breaking the group here
4329 do not consider the duplicate. Add comment explaining real fix.
4331 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
4333 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
4335 2017-12-01 Sudakshina Das <sudi.das@arm.com>
4337 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
4339 2017-12-01 Jakub Jelinek <jakub@redhat.com>
4341 * function.h (struct function): Remove cilk_frame_decl,
4342 is_cilk_function and calls_cilk_spawn fields.
4343 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
4345 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
4346 cilk_elemental field.
4347 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
4348 * target.def: Adjust comment.
4349 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
4350 Don't test cilk_elemental.
4352 PR tree-optimization/83233
4353 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
4354 bswap_stat name for the struct.
4357 * tree.h (SWITCH_BREAK_LABEL_P): Define.
4358 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
4359 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
4360 SWITCH_BREAK_LABEL_P set on the label.
4361 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
4362 added for default case if it was missing and not all cases covered.
4363 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
4364 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
4366 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
4367 to avoid -Wimplicit-fallthrough warning.
4368 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
4369 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
4371 PR tree-optimization/83221
4372 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
4374 (init_reassoc): Formatting fix.
4377 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
4378 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
4382 * tree-cfg.c: Include asan.h.
4383 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
4385 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4387 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
4389 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4391 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
4392 vpcmpeqd instruction.
4394 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4396 * config/i386/i386.c (standard_sse_constant_opcode): Fix
4397 registers type for 128bit mode.
4399 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4401 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
4403 * gengtype-state.c (read_state_pair): Likewise.
4404 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
4405 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
4406 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
4407 * ifcvt.c (noce_try_store_flag_constants): Likewise.
4408 * tree-ssa-ccp.c (ccp_finalize): Likewise.
4409 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
4410 * builtins.c (fold_builtin_3): Likewise.
4411 * graphite-scop-detection.c
4412 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
4413 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
4416 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
4419 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
4420 parameter. Return true if ignore_decl_rtl_set_p is true and other
4421 conditions are satisfied.
4422 * asan.h (asan_protect_global): Add new parameter.
4423 * varasm.c (categorize_decl_for_section): Pass true as second parameter
4424 to asan_protect_global calls.
4426 2017-11-30 Jim Wilson <jimw@sifive.com>
4428 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
4429 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
4430 -mstrict-align, add info on default value. Delete redundant lines for
4431 -mabi. Add missing -mexplicit-relocs docs.
4433 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4435 * config/arc/arc.md (trap): New pattern.
4437 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4439 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
4440 instruction to end into a delay slot.
4441 * config/arc/arc.md (cond_delay_insn): Check if the instruction
4442 can be placed into a delay slot against reg_note.
4444 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4446 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
4447 labels number of usages.
4449 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4451 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
4453 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
4455 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4458 * internal-fn.c (expand_mul_overflow): Optimize unsigned
4459 multiplication by power of 2 constant into two shifts + comparison.
4461 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
4464 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
4466 2017-11-30 Richard Biener <rguenther@suse.de>
4468 PR tree-optimization/83202
4469 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
4470 allow_peel argument and guard peeling.
4471 (canonicalize_loop_induction_variables): Likewise.
4472 (canonicalize_induction_variables): Pass false.
4473 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
4474 peeling from cunrolli.
4476 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4478 * combine.c (try_combine): Print a message to dump file whenever
4479 I0, I1, or I2 cannot be combined into I3.
4481 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4483 PR rtl-optimization/83156
4484 PR rtl-optimization/82621
4485 * combine.c (try_combine): Don't split an I2 if one of the dests is
4486 set again before I3. Allow unused dests.
4488 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4490 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
4492 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
4494 PR rtl-optimization/80818
4495 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
4496 recursively. Use insn code for clobber.
4497 (lra_set_insn_recog_data): Pass the new arg to
4498 collect_non_operand_hard_regs.
4499 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
4501 (lra_update_insn_regno_info): Pass insn to
4502 add_regs_to_insn_regno_info.
4504 2017-11-29 Jim Wilson <jimw@sifive.com>
4505 Andrew Waterman <andrew@sifive.com>
4507 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
4508 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
4509 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
4510 use SINGLE_SHIFT_COST.
4511 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
4512 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
4514 2017-11-29 Julia Koval <julia.koval@intel.com>
4516 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
4517 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
4518 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
4519 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
4520 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
4521 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
4522 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
4523 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
4524 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
4525 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
4526 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
4527 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
4528 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
4529 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
4530 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
4531 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
4532 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
4533 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
4534 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
4535 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
4536 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
4537 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
4538 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
4539 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
4540 __builtin_ia32_vpshldv_v2di_mask,
4541 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
4542 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
4543 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
4545 2017-11-29 Julia Koval <julia.koval@intel.com>
4547 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
4548 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
4549 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
4550 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
4551 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
4552 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
4553 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
4554 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
4555 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
4556 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
4557 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
4558 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
4559 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
4560 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
4561 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
4562 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
4563 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
4564 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
4565 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4566 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
4567 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
4569 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4571 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
4572 movsi_pic_gotdata_op instruction as a load for the UT699 errata
4575 2017-11-29 Martin Aberg <maberg@gaisler.com>
4577 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4578 to prevent b2bst errata sequence.
4579 (sqrtdf2_fix): Likewise.
4581 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4583 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
4584 (div_sqrt_insn_p): New function.
4585 (sparc_do_work_around_errata): Insert NOP instructions to
4586 prevent sequences that could trigger the TN-0013 errata for
4587 certain LEON3 processors.
4588 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
4589 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
4590 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
4591 (in_branch_delay): Prevent div and sqrt in delay slot if
4593 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
4595 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4597 * config/sparc/sparc.c (atomic_insn_p): New function.
4598 (sparc_do_work_around_errata): Insert NOP instructions to
4599 prevent sequences that could trigger the TN-0010 errata for
4601 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
4602 instruction referable in atomic_insns_p.
4604 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4606 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
4607 (atomic_compare_and_swap_leon3_1): Likewise.
4610 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4612 * config/sparc/sparc.c (fpop_insn_p): New function.
4613 (sparc_do_work_around_errata): Insert NOP instructions to
4614 prevent sequences that could trigger the TN-0012 errata for
4616 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
4617 * config/sparc/sparc.md (fix_gr712rc): New attribute.
4618 (in_branch_annul_delay): Prevent floating-point instructions
4619 in delay slot of annulled integer branch.
4621 2017-11-29 Richard Biener <rguenther@suse.de>
4623 PR tree-optimization/83202
4624 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4626 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4627 nodes vectorized to the same stmts multiple times.
4628 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4629 (vect_analyze_slp_instance): Adjust.
4630 (scalar_stmts_to_slp_tree_map_t): New typedef.
4631 (vect_schedule_slp_instance): Add a map recording the SLP node
4632 representing the vectorized stmts for a set of scalar stmts.
4633 Avoid code-generating redundancies.
4634 (vect_schedule_slp): Allocate map and pass it down.
4636 2017-11-29 Nathan Sidwell <nathan@acm.org>
4639 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4642 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4644 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4645 cases for printing LDP/STP memory addresses.
4646 (aarch64_print_address_internal): Renamed from
4647 aarch64_print_operand_address, added parameter, add Pmode check.
4648 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4649 (aarch64_print_operand_address): Indirect to
4650 aarch64_print_address_internal.
4651 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4652 'y' operand output specifier.
4654 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4657 * tree.c (build_simple_mem_ref_loc): Handle
4658 get_addr_base_and_unit_offset returning a MEM_REF.
4661 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4664 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4667 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4669 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4671 (layout::move_to_column): Likewise.
4673 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4676 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4677 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4679 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4680 Martin Jambor <mjambor@suse.cz>
4683 * tree.h (expr_type_first_operand_type_p): Declare
4684 * tree.c (expr_type_first_operand_type_p): New function.
4685 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4686 (ipa_value_from_jfunc): Adjust declaration.
4687 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4688 Use it as result type for arithmetics, unless it is NULL in which case
4689 be more conservative.
4690 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4691 ipa_get_jf_pass_through_result.
4692 (propagate_vals_across_pass_through): Likewise.
4693 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4694 is to propagate_vals_across_pass_through.
4695 (propagate_constants_across_call): Pass PARM_TYPE to
4696 propagate_scalar_across_jump_function.
4697 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4698 ipa_value_from_jfunc.
4699 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4700 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4701 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4702 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4703 target_type, pass it to ipa_value_from_jfunc.
4704 (update_indirect_edges_after_inlining): Pass parameter type to
4705 try_make_edge_direct_simple_call.
4707 2017-11-28 Jeff Law <law@redhat.com>
4709 * gimple-ssa-evrp-analyze.c
4710 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4711 refine ranges if scev_initialized_p returns true.
4712 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4714 2017-11-28 Julia Koval <julia.koval@intel.com>
4716 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4717 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4718 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4719 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4720 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4721 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4722 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4723 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4724 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4725 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4726 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4727 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4728 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4729 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4730 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4731 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4732 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4733 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4734 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4735 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4736 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4737 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4739 2017-11-28 Julia Koval <julia.koval@intel.com>
4741 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4742 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4743 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4744 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4745 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4746 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4747 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4748 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4749 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4750 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4751 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4752 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4753 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4754 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4755 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4756 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4757 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4758 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4759 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4760 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4761 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4762 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4763 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4764 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4765 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4766 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4767 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4768 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4769 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4770 __builtin_ia32_vpshld_v2di_mask): New builtins.
4771 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4772 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4774 2017-11-28 Richard Biener <rguenther@suse.de>
4776 PR tree-optimization/80776
4777 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4779 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4781 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4782 If the incoming edge is an effective fallthru because the other
4783 edge only reaches a __builtin_unreachable () then record ranges
4784 derived from the controlling condition in SSA info.
4785 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4786 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4788 2017-11-28 Olivier Hainque <hainque@adacore.com>
4790 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4791 on mingw build hosts.
4793 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4796 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4798 2017-11-28 Richard Biener <rguenther@suse.de>
4801 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4802 copies generated from memcpy use a character array as reference
4805 2017-11-28 Julia Koval <julia.koval@intel.com>
4806 Sebastian Peryt <sebastian.peryt@intel.com>
4808 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4809 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4810 cilk-common.o, cilk.h, cilk-common.c): Remove.
4812 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4813 * builtins.c (is_builtin_name): Remove cilkplus condition.
4814 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4815 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4816 cilk-builtins.def, cilkplus.def): Remove.
4817 * cif-code.def (CILK_SPAWN): Remove.
4818 * cilk-builtins.def: Delete.
4819 * cilk-common.c: Ditto.
4821 * cilkplus.def: Ditto.
4822 * config/darwin.h (fcilkplus): Delete.
4823 * cppbuiltin.c: Ditto.
4824 * doc/extend.texi: Remove cilkplus doc.
4825 * doc/generic.texi: Ditto.
4826 * doc/invoke.texi: Ditto.
4827 * doc/passes.texi: Ditto.
4828 * gcc.c (fcilkplus): Remove.
4829 * gengtype.c (cilk.h): Remove.
4830 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4832 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4834 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4836 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4837 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4838 cilkplus conditions.
4839 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4840 inline_read_section): Ditto.
4841 * ipa-inline-analysis.c (cilk.h): Remove.
4842 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4843 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4844 append_linker_options): Remove condition for fcilkplus.
4845 * lto/lto-lang.c (cilk.h): Remove.
4846 (lto_init): Remove condition for fcilkplus.
4847 * omp-expand.c (expand_cilk_for_call): Delete.
4848 (expand_omp_taskreg, expand_omp_for_static_chunk,
4849 expand_omp_for): Remove cilkplus
4851 (expand_cilk_for): Delete.
4852 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4853 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4854 execute_lower_omp, diagnose_sb_0): Ditto.
4855 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4856 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4857 * tree-nested.c: Ditto.
4858 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4859 (dump_generic_node): Ditto.
4860 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4861 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4863 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4865 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4867 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
4868 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
4869 for vector mode and !TARGET_SIMD.
4871 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4873 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
4875 * tree.h (SWITCH_LABELS): Remove.
4876 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
4877 assert SWITCH_BODY is non-NULL.
4878 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
4880 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
4882 PR tree-optimization/80788
4883 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
4884 has TREE_OVERFLOW set, call drop_tree_overflow.
4886 2017-11-28 Richard Biener <rguenther@suse.de>
4888 PR tree-optimization/83158
4889 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
4891 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
4894 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
4895 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
4897 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
4899 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
4900 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
4902 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4905 * varasm.c (compare_constant): Take the mode of the constants into
4906 account when comparing floating point constants.
4908 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
4910 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
4911 from explicit instantiation of debug_helper.
4912 * vec.h (DEFINE_DEBUG_VEC): Ditto.
4914 2017-11-27 Richard Biener <rguenther@suse.de>
4916 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
4919 2017-11-27 Richard Biener <rguenther@suse.de>
4921 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
4922 (type_hash_canon): Also clear int_cst_hash_table entry for
4924 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
4926 2017-11-27 Tamar Christina <tamar.christina@arm.com>
4928 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4929 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4930 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4932 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4934 * hash-map.h (gt_cleare_cache): Avoid UB.
4936 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
4938 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
4939 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
4940 (decide_unroll_runtime_iterations): Likewise.
4941 (decide_unroll_stupid): Likewise.
4943 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4946 * config/i386/i386.md: Add a loop with incssp.
4948 2017-11-27 Martin Jambor <mjambor@suse.cz>
4950 PR tree-optimization/81248
4951 * tree-sra.c (splice_param_accesses): Remove size check.
4952 (decide_one_param_reduction): Fix size check.
4953 * gimple-pretty-print.c (dump_profile): Silence warning.
4954 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
4956 2017-11-27 Jakub Jelinek <jakub@redhat.com>
4959 * dbxout.c (lastlineno): New variable.
4960 (dbx_debug_hooks): Use dbxout_switch_text_section as
4961 switch_text_section debug hook.
4962 (dbxout_function_end): Switch to current_function_section
4963 rather than function_section. If crtl->has_bb_partition,
4964 output just one N_FUN, depending on in_cold_section_p.
4965 (dbxout_source_line): Remember last lineno in lastlineno.
4966 (dbxout_switch_text_section): New function.
4967 (dbxout_function_decl): Adjust dbxout_block caller.
4968 (dbx_block_with_cold_children): New function.
4969 (dbxout_block): Return true if any LBRAC/RBRAC have been
4970 emitted. Use dbx_block_with_cold_children at depth == 0
4971 in second partition. Add PARENT_BLOCKNUM argument, pass
4972 it optionally adjusted to children. Output LBRAC/RBRAC
4973 around recursive call only if the block is in the current
4974 partition, if not and anything was output, emit empty
4976 * final.c (final_scan_insn): Compute cold_function_name
4977 before calling switch_text_section debug hook. Call
4978 that hook even if dwarf2out_do_frame if not emitting
4982 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
4983 TREE_READONLY decls.
4985 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4987 PR rtl-optimization/82488
4988 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
4990 2017-11-26 Julia Koval <julia.koval@intel.com>
4992 * config/i386/i386.c (processor_target_table): Add skylake_cost for
4994 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
4997 2017-11-26 Julia Koval <julia.koval@intel.com>
4999 * config/i386/driver-i386.c (host_detect_local_cpu):
5000 Detect skylake-avx512.
5002 2017-11-26 Julia Koval <julia.koval@intel.com>
5004 * config.gcc: Add -march=cannonlake.
5005 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5006 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5007 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5008 (PTA_CANNONLAKE): New.
5009 (processor_target_table): Add cannonlake.
5010 (ix86_option_override_internal): Ditto.
5011 (fold_builtin_cpu): Ditto.
5012 (get_builtin_code_for_version): Handle cannonlake.
5013 (M_INTEL_COREI7_CANNONLAKE): New.
5014 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5015 * doc/invoke.texi: Add -march=cannonlake.
5017 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5019 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5020 (try_init_one_plugin): Alternative implementation for MinGW.
5021 * Makefile.in (plugin_implib): New.
5022 (gengtype-lex.c): Fix broken AIX workaround.
5023 * configure: Regenerate.
5024 * doc/plugins.texi: Document support for MinGW.
5026 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5028 PR rtl-optimization/81553
5029 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5030 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5031 is a shift where C1 has different mode than the whole shift, use C1's
5032 mode for MULT rather than the shift's mode.
5035 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5036 builtins not enabled in the currently selected ISA.
5038 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5040 PR tree-optimization/71026
5041 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5042 (insert_reciprocals): Change to insert reciprocals before a division
5043 by a square and to insert the square of a reciprocal.
5044 (execute_cse_reciprocals_1): Change to consider division by a square.
5045 (register_division_in): Add importance parameter.
5047 2017-11-24 Richard Biener <rguenther@suse.de>
5049 PR tree-optimization/82402
5050 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5051 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5053 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5055 * match.pd (0-ptr): New transformation.
5057 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5060 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5061 when skipping checking.
5063 2017-11-24 Richard Biener <rguenther@suse.de>
5065 PR tree-optimization/83128
5066 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5067 (vn_reference_lookup_3): Likewise.
5069 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5072 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5073 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5076 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5077 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5078 check for the latter with gimple_call_builtin_p. Do not handle
5079 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5081 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5083 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5086 2017-11-23 Julia Koval <julia.koval@intel.com>
5088 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5089 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5090 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5091 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5092 _mm512_maskz_expandloadu_epi16): New intrinsics.
5093 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5094 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5095 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5096 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5097 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5098 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5099 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5100 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5101 _mm256_maskz_expandloadu_epi8): New intrinsics.
5102 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5103 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5104 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5105 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5106 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5107 * config/i386/sse.md (VI248_VLBW): New iterator.
5108 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5110 2017-11-23 Julia Koval <julia.koval@intel.com>
5112 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
5113 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
5114 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
5115 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
5116 _mm512_mask_compressstoreu_epi16): New.
5117 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
5118 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
5119 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
5120 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
5121 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
5122 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
5123 _mm256_mask_compressstoreu_epi8): New.
5124 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
5125 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
5126 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
5127 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
5128 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
5129 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
5130 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
5131 __builtin_ia32_compresshi128_mask,
5132 __builtin_ia32_compressstoreuqi512_mask,
5133 __builtin_ia32_compressstoreuhi512_mask,
5134 __builtin_ia32_compressstoreuqi256_mask,
5135 __builtin_ia32_compressstoreuqi128_mask,
5136 __builtin_ia32_compressstoreuhi256_mask,
5137 __builtin_ia32_compressstoreuhi128_mask): New builtins.
5138 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
5140 (ix86_expand_special_args_builtin): Handle new types.
5141 (s4fma_expand): Handle new builtin array.
5142 * config/i386/immintrin.h: Include new headers.
5143 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
5144 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
5146 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5148 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
5149 fixes. Declare temp and g variables at the top in order to avoid
5150 {} in most of the cases.
5152 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
5154 * match.pd (ptr-0): New transformation.
5156 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5158 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
5159 (struct addr_mode_cost_table): New.
5160 (struct tune_params): Add field addr_mode_costs.
5161 * config/arm/arm.c (generic_addr_mode_costs): New.
5162 (arm_slowmul_tune): Initialise addr_mode_costs field.
5163 (arm_fastmul_tune): Likewise.
5164 (arm_strongarm_tune): Likewise.
5165 (arm_xscale_tune): Likewise.
5166 (arm_9e_tune): Likewise.
5167 (arm_marvell_pj4_tune): Likewise.
5168 (arm_v6t2_tune): Likewise.
5169 (arm_cortex_tune): Likewise.
5170 (arm_cortex_a8_tune): Likewise.
5171 (arm_cortex_a7_tune): Likewise.
5172 (arm_cortex_a15_tune): Likewise.
5173 (arm_cortex_a35_tune): Likewise.
5174 (arm_cortex_a53_tune): Likewise.
5175 (arm_cortex_a57_tune): Likewise.
5176 (arm_exynosm1_tune): Likewise.
5177 (arm_xgene1_tune): Likewise.
5178 (arm_cortex_a5_tune): Likewise.
5179 (arm_cortex_a9_tune): Likewise.
5180 (arm_cortex_a12_tune): Likewise.
5181 (arm_cortex_a73_tune): Likewise.
5182 (arm_v7m_tune): Likewise.
5183 (arm_cortex_m7_tune): Likewise.
5184 (arm_v6m_tune): Likewise.
5185 (arm_fa726te_tune): Likewise.
5186 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
5188 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5190 * config/arm/arm.c (arm_mem_costs): New function.
5191 (arm_rtx_costs_internal): Use arm_mem_costs.
5193 2017-11-23 Mark Wielaard <mark@klomp.org>
5195 * dwarf2out.c (init_sections_and_labels): Use generation to create
5196 unique ranges_section_label and ranges_base_label. Return generation.
5197 (output_rnglists): Add generation argument. Use generation to create
5198 unique ranges labels.
5199 (dwarf2out_finish): Get generation from init_sections_and_labels
5200 and pass generation to output_rnglists.
5202 2017-11-23 Mike Stump <mikestump@comcast.net>
5203 Eric Botcazou <ebotcazou@adacore.com>
5205 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
5206 * cfgloop.h (struct loop): Add unroll field.
5207 * function.h (struct function): Add has_unroll bitfield.
5208 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
5209 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
5210 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
5211 (pass_rtl_unroll_loops::gate): Likewise.
5212 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
5213 for which loop->unroll==1.
5214 (decide_unroll_constant_iterations): Use note for consistency and
5215 take loop->unroll into account. Return early if loop->unroll is set.
5216 Fix thinko in existing test.
5217 (decide_unroll_runtime_iterations): Use note for consistency and
5218 take loop->unroll into account.
5219 (decide_unroll_stupid): Likewise.
5220 * lto-streamer-in.c (input_cfg): Read loop->unroll.
5221 * lto-streamer-out.c (output_cfg): Write loop->unroll.
5222 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
5224 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
5225 (print_loop): Print loop->unroll if set.
5226 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
5227 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
5228 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
5230 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
5231 loop->unroll is set and smaller than the trip count. Otherwise bypass
5232 entirely the heuristics if loop->unroll is set. Remove dead note.
5233 Fix off-by-one bug in other note.
5234 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
5235 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
5237 (tree_unroll_loops_completely): Make static.
5238 (pass_complete_unroll::execute): Use correct type for variable.
5239 (pass_complete_unrolli::execute): Fix formatting.
5240 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
5242 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
5244 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
5245 enable when TARGET_PREFER_AVX128 is set.
5247 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5249 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
5251 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
5252 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
5254 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5256 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
5257 frequencies. Fix estimation of aggregate parameters.
5259 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5261 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
5264 2017-11-23 Tom de Vries <tom@codesourcery.com>
5266 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
5268 2017-11-23 Tom de Vries <tom@codesourcery.com>
5270 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
5273 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
5276 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
5277 sibcall_value_pcrel_fdpic): Use local variable instead of
5279 (calli_tbr_rel): Add missing operand 2.
5280 (call_valuei_tbr_rel): Add missing operand 3.
5282 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5285 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
5286 bitpos/bitsize covering the whole destination, use store_expr only if
5287 the complex mode is the same. Otherwise, use expand_normal and if
5288 it returns CONCAT, subreg each part separately instead of trying to
5289 subreg the whole result.
5291 2017-11-23 Richard Biener <rguenther@suse.de>
5293 PR tree-optimization/23094
5294 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
5295 come here from walking over backedges in the first iteration.
5296 (vn_reference_lookup_3): Skip clobbers that store the same value.
5298 2017-11-23 Richard Biener <rguenther@suse.de>
5300 PR tree-optimization/81403
5301 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
5302 a block we need a leader relative to.
5303 (phi_translate_1): For nary processing require a leader from
5304 get_representative_for given we run expression simplification
5305 using match-and-simplify. Remove previous fix.
5307 2017-11-22 Jeff Law <law@redhat.com>
5309 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
5310 Use new method allocate_value_range rather than accessing the
5311 vrp_value_range_pool data member directly.
5312 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
5313 to use extract_range_from_stmt method to avoid need for
5314 extract_range_from_assignment method.
5315 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
5316 method rather than setting values_propgated data member directly.
5317 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
5318 and values propagated data members and extract_range_from_assignment
5319 method. Reorder private data members to conform to standards.
5320 Add new methods set_lattice_propagation_complete and
5321 allocate_value_range.
5323 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
5325 PR rtl-optimization/83030
5326 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
5327 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
5328 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
5329 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
5330 CROSSING_JUMP_P flag is already set before setting it.
5331 * cfgrtl.c (fixup_partition_crossing): Likewise.
5332 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
5335 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5337 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
5338 Handle the case where both arguments are using gen_const_vec_series.
5340 2017-11-22 David Malcolm <dmalcolm@redhat.com>
5343 * pretty-print.c (pp_format): Move quoting implementation to
5344 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
5345 to pass address of "quote" local.
5346 (pp_begin_quote): New function.
5347 (pp_end_quote): New function.
5348 * pretty-print.h (printer_fn): Convert penultimate param from bool
5350 (pp_begin_quote): New decl.
5351 (pp_end_quote): New decl.
5352 * tree-diagnostic.c (default_tree_printer): Convert penultimate
5353 param from bool to bool *.
5354 * tree-diagnostic.h (default_tree_printer): Likewise.
5356 2017-11-22 Jeff Law <law@redhat.com>
5358 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
5359 Initialize vr_values.
5360 (evrp_range_analyzer::try_find_new_range): Call methods attached to
5361 vr_values via vr_values class instance rather than delegators.
5362 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
5363 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5364 (evrp_range_analyzer::push_value_range): Likewise.
5365 (evrp_range_analyzer::pop_value_range): Likewise.
5366 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
5367 most delegators. Those remaining are exposed as public interfaces.
5368 Make vr_values a pointer and private.
5369 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
5371 (evrp_range_analyzer::get_vr_value): New method.
5372 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
5373 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
5374 (class evrp_dom_walker): Attach evrp_folder class, initialize
5375 it in the ctor. Remove temporary delegators.
5376 (evrp_dom_walker::before_dom_children): Call methods in attached
5377 evrp_range_analyzer class via class instance pointer. Use
5378 free value_range_constant_singleton to remove need for
5379 op_with_constant_singleton_value delegator method. Do not
5380 create a vrp_prop class instance for every call! Narrow
5381 scope of a couple variables.
5382 (evrp_dom_walker::cleanup): Call methods in attached
5383 evrp_range_analyzer class via class instance pointer.
5384 * vr-values.h (class vr_values): Privatize many methods and
5387 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5389 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
5390 * cfgexpand.c (expand_debug_expr): Remove handling for them.
5391 * expr.c (expand_expr_real_2): Likewise.
5392 * fold-const.c (const_unop): Likewise.
5393 * optabs-tree.c (optab_for_tree_code): Likewise.
5394 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
5395 * tree-inline.c (estimate_operator_cost): Likewise.
5396 * tree-pretty-print.c (dump_generic_node): Likewise.
5397 (op_code_prio): Likewise.
5398 (op_symbol_code): Likewise.
5399 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
5400 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
5401 * internal-fn.c (direct_internal_fn_optab): New function.
5402 (direct_internal_fn_array, direct_internal_fn_supported_p
5403 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
5404 * fold-const-call.c (fold_const_reduction): New function.
5405 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
5407 * tree-vect-loop.c: Include internal-fn.h.
5408 (reduction_code_for_scalar_code): Rename to...
5409 (reduction_fn_for_scalar_code): ...this and return an internal
5411 (vect_model_reduction_cost): Take an internal_fn rather than
5413 (vect_create_epilog_for_reduction): Likewise. Build calls rather
5415 (vectorizable_reduction): Use internal functions rather than tree
5416 codes for the reduction operation. Update calls to the functions
5418 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
5419 Use calls to internal functions rather than REDUC tree codes.
5420 * config/aarch64/aarch64-simd.md: Update comment accordingly.
5422 2017-11-22 Olivier Hainque <hainque@adacore.com>
5424 * config/vxworks.c (vxworks_override_options): Pick default
5425 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
5426 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
5427 DWARF_GNAT_ENCODINGS_DEFAULT.
5428 * config/vxworksae.h: Likewise.
5430 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
5432 PR tree-optimization/83104
5433 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
5436 2017-11-22 Marek Polacek <polacek@redhat.com>
5437 H.J. Lu <hongjiu.lu@intel.com>
5438 Jason Merrill <jason@redhat.com>
5443 * calls.c (initialize_argument_information): Call
5444 warn_parameter_passing_abi target hook.
5445 (store_one_arg): Use 0 for empty record size. Don't push 0 size
5446 argument onto stack.
5447 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
5448 * common.opt: Update -fabi-version description.
5449 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
5450 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
5452 (ix86_is_empty_record): New function.
5453 (ix86_warn_parameter_passing_abi): New function.
5454 (TARGET_EMPTY_RECORD_P): Redefine.
5455 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
5456 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
5457 * doc/tm.texi: Regenerated.
5458 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
5459 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
5460 * dwarf2out.c (get_ultimate_context): Move to tree.c.
5461 * explow.c (hard_function_value): Call arg_int_size_in_bytes
5462 instead of int_size_in_bytes.
5463 * expr.c (copy_blkmode_to_reg): Likewise.
5464 * function.c (aggregate_value_p): Return 0 for empty types.
5465 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
5467 (locate_and_pad_parm): Call arg size_in_bytes instead
5469 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
5470 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
5471 * target.def (empty_record_p, warn_parameter_passing_abi): New target
5473 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
5474 (std_gimplify_va_arg_expr): Skip empty records. Call
5475 arg_size_in_bytes instead size_in_bytes.
5476 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
5477 * tree-core.h (tree_type_common): Add empty_flag.
5478 (tree_decl_common): Update comments.
5479 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
5481 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5482 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
5484 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5485 * tree.c (default_is_empty_type): New function.
5486 (default_is_empty_record): New function.
5487 (arg_int_size_in_bytes): New function.
5488 (arg_size_in_bytes): New function.
5489 (get_ultimate_context): New function.
5490 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
5491 TRANSLATION_UNIT_WARN_EMPTY_P.
5492 (default_is_empty_record, arg_int_size_in_bytes,
5493 arg_size_in_bytes, get_ultimate_context): Declare.
5495 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5497 * config/arm/arm.c (cmse_clear_registers): New function.
5498 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
5499 code by call to cmse_clear_registers.
5500 (cmse_nonsecure_entry_clear_before_return): Likewise.
5502 2017-11-22 Tamar Christina <tamar.christina@arm.com>
5504 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
5505 (vdot_s32, vdotq_s32): New.
5506 (vdot_lane_u32, vdotq_lane_u32): New.
5507 (vdot_lane_s32, vdotq_lane_s32): New.
5510 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5513 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
5514 for unsigned values with fewer HWIs than the precision.
5515 (test_overflow): New function.
5516 (wide_int_cc_tests): Call it.
5518 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5520 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
5521 has the same mode class as Pmode.
5523 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5525 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
5526 padding_bits_to_clear_ptr.
5527 (cmse_nonsecure_entry_clear_before_return): Likewise.
5529 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5531 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
5532 auto_sbitap instead of integer bitfield to control register needing
5535 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5537 PR tree-optimization/83044
5538 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
5539 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
5540 that need the upper bound. Subtract offset from
5541 get_addr_base_and_unit_offset only if positive and subtract it
5542 before division by eltsize rather than after it.
5545 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
5546 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
5547 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
5551 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
5552 expand_binop with *mul_widen_optab, make sure at least one of the
5553 operands doesn't have VOIDmode.
5556 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
5558 PR rtl-optimization/82044
5559 PR tree-optimization/82042
5560 * dse.c (record_store): Check for overflow.
5561 (check_mem_read_rtx): Properly check for overflow if width == -1, call
5562 add_wild_read instead of clear_rhs_from_active_local_stores on
5563 overflow and log it into dump_file.
5565 2017-11-22 Richard Biener <rguenther@suse.de>
5567 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
5568 fake edges to exit when looking for a place to insert.
5569 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
5571 (insert_into_preds_of_block): Commit edge insertion immediately,
5572 assert that doesn't require new BBs.
5573 (fini_pre): Release expressions.
5574 (pass_pre::execute): Shuffle things around a bit, if the fn
5575 is too large do not compute AVAIL either as this is really the
5578 2017-11-22 Richard Biener <rguenther@suse.de>
5580 PR tree-optimization/83089
5581 * tree-if-conv.c (pass_if_conversion::execute): If anything
5582 changed reset SCEV and free the number of iteration estimates.
5584 2017-11-21 Martin Sebor <msebor@redhat.com>
5586 PR tree-optimization/82945
5587 * calls.h (warn_nonstring_bound): Remove unused function.
5589 2017-11-21 Martin Sebor <msebor@redhat.com>
5591 PR tree-optimization/82945
5592 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
5593 * calls.h (maybe_warn_nonstring_arg): Declare new function.
5594 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
5596 (initialize_argument_information): Call maybe_warn_nonstring_arg.
5597 * calls.h (get_attr_nonstring_decl): Declare new function.
5598 * doc/extend.texi (attribute nonstring): Update.
5599 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
5600 get_attr_nonstring_decl and handle it.
5601 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
5602 detection of nul-termination.
5603 (strlen_to_stridx): Change to a pointer.
5604 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
5605 (pass_strlen::execute): Same.
5607 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
5609 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
5610 for the new option -mprefer-vector-width=[none|128|256|512].
5611 * config/i386/i386.c (ix86_target_string): remove old style options
5612 -mprefer-avx256 and make -mprefer-avx128 as alias.
5613 (ix86_option_override_internal): Apply defaults for the
5614 -mprefer-vector-width=[128|256] option.
5615 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
5616 Implement macros to work with -mprefer-vector-width=.
5617 * config/i386/i386.opt: Implemented option
5618 -mprefer-vector-width=[none|128|256|512].
5619 * doc/invoke.texi: Documentation for
5620 -mprefer-vector-width=[none|128|256|512].
5622 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5624 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5626 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5628 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5630 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5632 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5634 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5635 * expr.c (expand_expr_real_2): Likewise.
5636 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5637 fold_binary_loc): Likewise.
5638 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5639 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5640 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5641 MINUS_EXPR transformations.
5642 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5643 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5644 * tree-inline.c (estimate_operator_cost): Likewise.
5645 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5646 op_symbol_code): Likewise.
5647 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5648 * vr-values.c (extract_range_from_binary_expr): Likewise.
5649 * varasm.c (initializer_constant_valid_p_1): Likewise.
5650 * tree.def: New tree code POINTER_DIFF_EXPR.
5652 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5654 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5655 integer suffix to movbe mnemonic.
5656 (*bswaphi2_movbe): Ditto.
5657 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5659 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5662 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5663 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5664 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5666 2017-11-21 Martin Liska <mliska@suse.cz>
5668 * tree-inline.c (expand_call_inline): Remove not needed
5671 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5672 Jakub Jelinek <jakub@redhat.com>
5675 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5676 Remove static keyword from f variable.
5678 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5680 PR tree-optimization/83086
5681 * gimple-ssa-store-merging.c
5682 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5683 rather than n.base_addr.
5685 2017-11-21 Martin Liska <mliska@suse.cz>
5687 PR rtl-optimization/82044
5688 PR tree-optimization/82042
5689 * dse.c (check_mem_read_rtx): Check for overflow.
5691 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5693 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5696 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5698 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5699 and its corresponding call arg location note.
5701 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5702 Andrew Burgess <andrew.burgess@embecosm.com>
5704 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5706 (arc_return_slot_offset): Likewise.
5707 (arc_eh_return_address_location): New declaration.
5708 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5709 (MUST_SAVE_REGISTER): Add exception handler case.
5710 (MUST_SAVE_RETURN_ADDR): Likewise.
5711 (arc_frame_pointer_required): Likewise.
5712 (arc_frame_pointer_needed): New function.
5713 (arc_compute_frame_size): Changed.
5714 (arc_expand_prologue): Likewise.
5715 (arc_expand_epilogue): Likewise.
5716 (arc_initial_elimination_offset): Likewise.
5717 (arc_return_slot_offset): Delete.
5718 (arc_eh_return_address_location): New function.
5719 (arc_builtin_setjmp_frame_value): Likewise.
5720 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5721 (EH_RETURN_STACKADJ_RTX): Define.
5722 (EH_RETURN_HANDLER_RTX): Likewise.
5723 * config/arc/arc.md (eh_return): Delete.
5725 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5727 * print-tree.h (debug_vec_tree): Remove prototype.
5728 * gdbinit.in (pvt): Remove macro.
5730 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5732 PR tree-optimization/83047
5733 * gimple-ssa-store-merging.c
5734 (imm_store_chain_info::output_merged_store): If the loads with the
5735 same vuse are in different basic blocks, for load_gsi pick a load
5736 location that is dominated by the other loads.
5739 * config/i386/i386.c (ix86_memmodel_check): Start
5740 -Winvalid-memory-model diagnostics with lowercase letter.
5743 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5744 set in_cold_section_p to the partition containing loc_list->first.
5745 When seeing loc_list->last_before_switch node, update secname and
5746 perform range_across_switch second partition handling only after that.
5749 * run-rtl-passes.c: Include debug.h.
5750 (run_rtl_passes): Call debug_hooks->assembly_start.
5751 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5755 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5756 OPTAB_DIRECT in calls to expand_simple_binop.
5758 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5761 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5763 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5766 * spellcheck.h (best_match::blithely_get_best_candidate): New
5769 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5771 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5772 with lower case letter.
5774 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5776 * config/i386/i386.md (bswaphi2): New expander.
5777 (*bswaphi2_movbe): New insn pattern.
5778 (bswaphi -> rorhi pepehole2): New peephole pattern.
5780 2017-11-20 Jeff Law <law@redhat.com>
5782 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5783 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5784 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5785 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5786 gimple-ssa-evrp-analyze.h.
5788 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5789 set BB_VISITED here.
5790 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5792 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5794 * tree-predcom.c: Add general comment on Store-Store chains.
5795 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5797 (get_chain_last_ref_at): Rename into...
5798 (get_chain_last_write_at): ...this.
5799 (get_chain_last_write_before_load): New function.
5800 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5801 CT_STORE_STORE when write reference is added.
5802 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5803 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5804 (initialize_root_vars_store_elim_1): Ditto.
5805 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5806 definition is created.
5807 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5808 chain by replacing it with dominant stored value.
5810 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5812 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5814 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5816 * vr-values.c (extract_range_from_binary_expr): Use a full range
5819 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5821 * config/arm/arm.md (R4_REGNUM): Define constant.
5822 (nonsecure_call_internal): Remove r4 clobber.
5823 (nonsecure_call_value_internal): Likewise.
5824 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5825 clobber and resequence match_operands.
5826 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5827 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5828 (nonsecure_call_value_reg_thumb2): Likewise.
5830 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5832 PR tree-optimization/78821
5833 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5834 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5835 to the address of the base rather than the base itself.
5836 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5837 (find_bswap_or_nop_finalize): New function.
5838 (find_bswap_or_nop): Use it.
5839 (bswap_replace): Return a tree rather than bool, change first
5840 argument from gimple * to gimple_stmt_iterator, allow inserting
5841 into an empty sequence, allow ins_stmt to be NULL - then emit
5842 all stmts into gsi. Fix up MEM_REF address gimplification.
5843 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5844 (struct store_immediate_info): Add N and INS_STMT non-static
5846 (store_immediate_info::store_immediate_info): Initialize them
5847 from newly added ctor args.
5848 (merged_store_group::apply_stores): Formatting fixes. Sort by
5850 (stmts_may_clobber_ref_p): For stores call also
5851 refs_anti_dependent_p.
5852 (gather_bswap_load_refs): New function.
5853 (imm_store_chain_info::try_coalesce_bswap): New method.
5854 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5855 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5856 (imm_store_chain_info::output_merged_store): Fail if number of
5857 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5858 and NOP_EXPR rhs_code.
5859 (pass_store_merging::process_store): Compute n and ins_stmt, if
5860 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5861 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5863 (pass_store_merging::execute): Calculate dominators.
5865 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5866 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5867 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
5868 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5869 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
5870 class pass_optimize_bswap, bswap_replace,
5871 pass_optimize_bswap::execute): Moved to ...
5872 * gimple-ssa-store-merging.c: ... this file.
5873 Include optabs-tree.h.
5874 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
5875 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5876 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
5877 anonymous namespace, remove static keywords.
5878 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
5879 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
5881 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
5884 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
5886 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
5888 * vec.h (debug_helper): New function.
5889 (DEFINE_DEBUG_VEC): New macro.
5890 * hash-set.h (debug_helper): New function.
5891 (DEFINE_DEBUG_HASH_SET): New macro.
5892 * cfg.c (debug_slim (edge)): New function.
5893 Call DEFINE_DEBUG_VEC for edges.
5894 Call DEFINE_DEBUG_HASH_SET for edges.
5895 * cfghooks.c (debug_slim (basic_block)): New function.
5896 Call DEFINE_DEBUG_VEC for basic blocks.
5897 Call DEFINE_DEBUG_HASH_SET for basic blocks.
5898 * print-tree.c (debug_slim): New function to handle trees.
5899 Call DEFINE_DEBUG_VEC for trees.
5900 Call DEFINE_DEBUG_HASH_SET for trees.
5901 (debug (vec<tree, va_gc>) &): Remove.
5902 (debug (<vec<tree, va_gc>) *): Remove.
5903 * print-rtl.c (debug_slim): New function to handle const_rtx.
5904 Call DEFINE_DEBUG_VEC for rtx_def.
5905 Call DEFINE_DEBUG_VEC for rtx_insn.
5906 Call DEFINE_DEBUG_HASH_SET for rtx_def.
5907 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
5908 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
5909 (debug (vec<rtx_insn *> *ptr): Remove.
5910 (debug_insn_vector): Remove.
5911 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
5913 2017-11-20 Tom de Vries <tom@codesourcery.com>
5915 PR rtl-optimization/82020
5916 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
5917 IF_THEN_ELSE condition.
5919 2017-11-19 Jeff Law <law@redhat.com>
5921 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
5922 of degenerates resulting from ignoring an edge.
5924 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5927 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
5929 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5932 * profile-count.c (profile_count::to_sreal_scale): Fix return value
5933 for uninitialied counts.
5935 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5938 * tree-inline.c (estimate_num_insns): Set to 1 at least.
5940 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5943 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
5945 2017-11-19 Tom de Vries <tom@codesourcery.com>
5947 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
5950 2017-11-19 Tom de Vries <tom@codesourcery.com>
5952 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
5955 2017-11-19 Tom de Vries <tom@codesourcery.com>
5957 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5960 2017-11-19 Tom de Vries <tom@codesourcery.com>
5962 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
5964 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
5966 2017-11-19 Tom de Vries <tom@codesourcery.com>
5968 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
5970 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
5971 semicolon after MCORE_EXPORT_NAME call.
5973 2017-11-19 Tom de Vries <tom@codesourcery.com>
5976 * vmsdbgout.c (vmsdbgout_early_finish): New function.
5977 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
5979 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
5981 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
5982 (cgraph_edge::clone): Cleanup updating of profile.
5983 * ipa-cp.c (update_profiling_info): Likewise.
5984 * ipa-inline-transform.c (inline_transform): Likewise.
5985 * ipa-inline.c (inline_small_functions): Add missing space to dump.
5986 * ipa-split.c (execute_split_functions): Do not split when function
5988 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
5989 * profile-count.c (profile_count::dump): Add global0.
5990 (profile_count::to_cgraph_frequency): Do not ICE when entry is
5992 (profile_count::to_sreal_scale): Likewise.
5993 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
5994 (profile_count::combine_with_ipa_count): New function.
5995 * profile-count.h (profile_guessed_global0adjusted): New.
5996 (profile_count::adjusted_zero): New.
5997 (profile_count::global0adjusted): New.
5998 (profile_count::combine_with_ipa_count): New.
5999 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
6000 correct profile of return block of split functions.
6001 (copy_cfg_body): Remove unused profile_count.
6002 (copy_body): Likewise.
6003 (expand_call_inline): Update.
6004 (tree_function_versioning): Update.
6006 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6008 * hash-set.h (hash_set::empty): New.
6009 * tree-ssa-threadbackward.h: Delete.
6010 * tree-ssa-threadbackward.c (class thread_jumps): New.
6011 Move max_threaded_paths into class.
6012 (fsm_find_thread_path): Remove arguments that are now in class.
6013 (profitable_jump_thread_path): Rename to...
6014 (thread_jumps::profitable_jump_thread_path): ...this.
6015 (convert_and_register_jump_thread_path): Rename to...
6016 (thread_jumps::convert_and_register_current_path): ...this.
6017 (check_subpath_and_update_thread_path): Rename to...
6018 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6019 (register_jump_thread_path_if_profitable): Rename to...
6020 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6021 (handle_phi): Rename to...
6022 (thread_jumps::handle_phi): ...this.
6023 (handle_assignment): Rename to...
6024 (thread_jumps::handle_assignment): ...this.
6025 (fsm_find_control_statement_thread_paths): Rename to...
6026 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6027 (find_jump_threads_backwards): Rename to...
6028 (thread_jumps::find_jump_threads_backwards): ...this.
6029 Initialize path local data.
6030 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6031 from within thread_jumps class.
6032 (pass_early_thread_jumps::execute): Same.
6034 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6036 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6038 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6040 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6041 to canonical location.
6043 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6046 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6048 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6050 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6052 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6053 rcrt1.o%s/grcrt1.o%s for -static-pie.
6055 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6057 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6058 ix86_shift_rotate_cost): Break out from ...
6059 (ix86_rtx_costs): ... here.
6060 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6063 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6065 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6066 when entry block was promoted unlikely.
6067 (estimate_bb_frequencies): Increase frequency scale.
6068 * profile-count.h (profile_count): Export precision info.
6070 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6072 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6073 disturbin profile of entry block.
6075 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6077 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6080 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6082 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6083 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6085 2017-11-17 Jeff Law <law@redhat.com>
6087 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6088 from evrp_dom_walker class. Various methods moved into new class.
6089 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6090 (evrp_range_analyzer::enter): New method.
6091 (evrp_range_analyzer::leave): New method.
6092 (evrp_dom_walker): Remove delegators no longer needed by this class.
6093 Replace vr_values data member with evrp_range_analyzer
6095 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6096 method extracted from evrp_dom_walker::before_dom_children.
6097 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6098 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6100 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6101 Add private copy constructor and move assignment operators.
6102 Privatize methods and class data where trivially possible.
6103 (evrp_dom_walker::cleanup): New function, extracted from
6104 execute_early_vrp. Simplify access to class data.
6106 * vr-values.h (get_output_for_vrp): Prototype.
6107 * vr-values.c (get_output_for_vrp): New function extracted from
6108 vrp_visit_assignment_or_call and extract_range_from_stmt.
6109 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
6111 2017-11-17 Luis Machado <luis.machado@linaro.org>
6113 * config/aarch64/aarch64.c
6114 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
6115 (qdf24xx_tunings) <autoprefetcher_model>: Set to
6116 tune_params::AUTOPREFETCHER_WEAK.
6118 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6121 * config/arm/arm.c (arm_valid_target_attribute_rec):
6122 Parse "arch=" and "+<ext>".
6123 (arm_valid_target_attribute_tree): Re-init global options.
6124 (arm_option_override): Make non-static.
6125 (arm_options_perform_arch_sanity_checks): Make errors fatal.
6126 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
6127 (__ARM_FEATURE_CRC32): Support undef.
6128 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
6129 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
6131 2017-11-17 David Malcolm <dmalcolm@redhat.com>
6133 * gdbinit.in (break-on-diagnostic): New command.
6135 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6137 * config/i386/i386.c (ix86_expand_epilogue): Change simple
6138 return to indirect jump for EH return if control-flow
6139 protection is enabled. Change explicit 'false' argument in
6140 pro_epilogue_adjust_stack with a value of
6142 * config/i386/i386.md (simple_return_indirect_internal):
6143 Remove SImode restriction to support 64-bit.
6145 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6147 * combine.c (added_notes_insn): New.
6148 (try_combine): Handle added_notes_insn like added_links_insn.
6149 Rewrite return value code.
6150 (distribute_notes): Set added_notes_insn to the earliest insn we added
6153 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6155 PR rtl-optimization/82621
6156 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
6157 dest of one of those SETs is unused.
6159 2017-11-17 Richard Biener <rguenther@suse.de>
6162 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
6163 * tree-pretty-print.c (dump_generic_node): Handle
6164 annot_expr_parallel_kind.
6165 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
6166 * gimplify.c (gimple_boolify): Likewise.
6168 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6170 * config.gcc (extra_headers): Add cet.h for x86 targets.
6171 * config/i386/cet.h: New file.
6172 * doc/install.texi: Add --enable-cet/--disable-cet.
6174 2017-11-17 Richard Biener <rguenther@suse.de>
6176 PR tree-optimization/83017
6177 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
6178 (gen_parallel_loop): Properly count iterations.
6179 (parallelize_loops): Handle loop->can_be_parallel independent
6180 of flag_loop_parallelize_all. Make static profitability test match
6182 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
6183 * invoke.texi (parloops-min-per-thread): Document.
6185 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
6187 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
6188 upstreaming review comments.
6190 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6192 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
6193 with fast unaligned access.
6194 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
6196 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
6198 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
6199 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
6201 2017-11-17 Richard Biener <rguenther@suse.de>
6203 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
6204 folding of references.
6206 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
6209 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
6210 of replacing call to strncmp with corresponding call to strcmp when
6213 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
6215 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
6216 option prefer-avx256 for skylake-avx512 configuration.
6217 * config/i386/i386.c (ix86_option_override_internal): Ditto.
6218 (get_builtin_code_for_version): Ditto.
6220 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6221 Monk Chiang <sh.chiang04@gmail.com>
6223 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
6224 (FIXED_REGISTERS): Reserve more register numbers.
6225 (CALL_USED_REGISTERS): Likewise.
6226 (REG_ALLOC_ORDER): Likewise.
6227 (REG_CLASS_CONTENTS): Likewise.
6228 (REGISTER_NAMES): Likewise.
6230 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6231 Kito Cheng <kito.cheng@gmail.com>
6233 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
6235 * config/nds32/iterators.md: Add vector mode iterators and attributes.
6237 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
6239 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
6240 parameter list for vec_splats.
6242 2017-11-16 Joseph Myers <joseph@codesourcery.com>
6244 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
6246 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
6248 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
6251 * doc/install.texi: Document incompatibility of Perl >=5.6.26
6252 with the required version of automake 1.11.6.
6254 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
6256 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
6258 (DU_C2_3_power9): Correct reservation combinations.
6259 (FP_DIV_power9, VEC_DIV_power9): New.
6260 (power9-alu): Split out rotate/shift...
6261 (power9-rot): ...to here, correct dispatch resource.
6262 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
6264 (power9-fp): Correct latency.
6265 (power9-sdiv): Add div/sqrt resource.
6266 (power9-ddiv): Correct latency, add div/sqrt resource.
6267 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
6268 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
6270 (power9-qpdiv, power9-qpmul): Adjust resource usage.
6272 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
6274 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
6275 switch statement mapping KF built-ins to TF built-ins if we don't
6276 have the proper ISA 3.0 assembler support.
6278 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6280 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
6281 (lower_emutls_function_body): Do not compute it.
6283 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6285 * ipa-split.c (split_bb_info): Turn time to sreal.
6286 (split_point): Likewise.
6287 (dump_split_point): Likewise.
6288 (fine_split_points): Likewise.
6289 (execute_split_functions): Only zero split_bbs; turn time to sreals.
6291 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6293 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
6296 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6298 * predict.c (combine_predictions_for_bb): Preserve zero predicted
6300 (expensive_function_p): Remove useless assert.
6301 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
6303 2017-11-16 Martin Sebor <msebor@redhat.com>
6305 PR tree-optimization/82588
6306 PR tree-optimization/82583
6307 * tree-vrp.c (check_array_ref): Handle flexible array members,
6308 string literals, and inner indices.
6309 (search_for_addr_array): Add detail to diagnostics.
6311 2017-11-16 Nathan Sidwell <nathan@acm.org>
6315 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
6316 (SET_DECL_ASSEMBLER_NAME): Forward to
6317 overwrite_decl_assembler_name.
6318 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
6319 (overwrite_decl_assembler_name): Declare.
6320 * tree.c (overwrite_decl_assembler_name): New.
6321 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
6322 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
6323 (LANG_HOOKS_INITIALIZER): Add it.
6324 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
6325 * langhooks.c (lhd_set_decl_assembler_name): Use
6326 SET_DECL_ASSEMBLER_NAME.
6327 (lhd_overwrite_decl_assembler_name): Default implementation.
6329 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6330 Jackson Woodruff <jackson.woodruff@arm.com>
6332 PR tree-optimization/71026
6333 * match.pd: Canonicalize constant multiplies in division.
6335 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6337 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
6338 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
6339 Remove OPT_fomit_frame_pointer entry.
6340 * common/config/arc/arc-common.c: Likewise.
6341 * common/config/arm/arm-common.c: Likewise.
6342 * common/config/avr/avr-common.c: Likewise.
6343 * common/config/c6x/c6x-common.c: Likewise.
6344 * common/config/cr16/cr16-common.c: Likewise.
6345 * common/config/cris/cris-common.c: Likewise.
6346 * common/config/epiphany/epiphany-common.c: Likewise.
6347 * common/config/fr30/fr30-common.c: Likewise.
6348 * common/config/frv/frv-common.c: Likewise.
6349 * common/config/ia64/ia64-common.c: Likewise.
6350 * common/config/iq2000/iq2000-common.c: Likewise.
6351 * common/config/lm32/lm32-common.c: Likewise.
6352 * common/config/m32r/m32r-common.c: Likewise.
6353 * common/config/mcore/mcore-common.c: Likewise.
6354 * common/config/microblaze/microblaze-common.c: Likewise.
6355 * common/config/mips/mips-common.c: Likewise.
6356 * common/config/mmix/mmix-common.c: Likewise.
6357 * common/config/mn10300/mn10300-common.c: Likewise.
6358 * common/config/nios2/nios2-common.c: Likewise.
6359 * common/config/pa/pa-common.c: Likewise.
6360 * common/config/pdp11/pdp11-common.c: Likewise.
6361 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
6362 * common/config/riscv/riscv-common.c: Likewise.
6363 * common/config/rs6000/rs6000-common.c: Likewise.
6364 * common/config/rx/rx-common.c: Likewise.
6365 * common/config/s390/s390-common.c: Likewise.
6366 * common/config/sh/sh-common.c: Likewise.
6367 * common/config/sparc/sparc-common.c: Likewise.
6368 * common/config/tilegx/tilegx-common.c: Likewise.
6369 * common/config/tilepro/tilepro-common.c: Likewise.
6370 * common/config/v850/v850-common.c: Likewise.
6371 * common/config/visium/visium-common.c: Likewise.
6372 * common/config/xstormy16/xstormy16-common.c: Likewise.
6373 * common/config/xtensa/xtensa-common.c: Likewise.
6374 * invoke.texi (-fomit-frame-pointer): Update documentation.
6376 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6378 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
6380 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6382 * cfg.c (scale_bbs_frequencies_int,
6383 cale_bbs_frequencies_gcov_type): Remove.
6384 * cfg.h (scale_bbs_frequencies_int,
6385 cale_bbs_frequencies_gcov_type): Remove.
6387 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6389 * tree-ssa-loop-manip.c
6390 (scale_dominated_blocks_in_loop): Update to profile counts.
6391 (tree_transform_and_unroll_loop): Likewise.
6393 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6395 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
6396 scale_bbs_frequencies_int.
6398 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6400 * final.c (compute_alignments): Use counts rather than frequencies.
6402 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6404 * cfgloopanal.c: Include sreal.h
6405 (average_num_loop_insns): Use counts and sreal for accounting.
6407 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6409 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
6412 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6413 Kito Cheng <kito.cheng@gmail.com>
6415 * config/nds32/constraints.md: Provide more constraints.
6416 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
6417 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
6418 support constraints usage.
6420 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6422 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
6424 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6425 Kito Cheng <kito.cheng@gmail.com>
6427 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
6428 * config/nds32/nds32.opt: Refine the layout.
6429 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
6430 TARGET_EXT_STRING): Support new options.
6431 * config/nds32/nds32.h: Likewise.
6432 * config/nds32/nds32.md: Likewise.
6433 * config/nds32/nds32-predicates.c: Likewise.
6434 * config/nds32/constraints.md: Likewise.
6435 * common/config/nds32/nds32-common.c: Likewise.
6437 2017-11-16 Julia Koval <julia.koval@intel.com>
6440 * config/i386/gfniintrin.h: Add sse check.
6441 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
6443 2017-11-16 Julia Koval <julia.koval@intel.com>
6445 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
6446 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
6447 (ix86_handle_option): Handle -mavx512vbmi2.
6448 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
6449 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
6450 * config/i386/i386-c.c (__AVX512VBMI2__): New.
6451 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
6452 (ix86_valid_target_attribute_inner_p): Ditto.
6453 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
6454 * config/i386/i386.opt (mavx512vbmi2): New option.
6455 * doc/invoke.texi: Add new option.
6457 2017-11-16 Julia Koval <julia.koval@intel.com>
6459 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
6460 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
6461 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
6462 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
6463 _mm512_gf2p8mul_epi8): New intrinsics.
6464 * config/i386/i386-builtin-types.def
6465 (V64QI_FTYPE_V64QI_V64QI): New type.
6466 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
6467 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
6468 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
6469 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
6470 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
6471 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
6473 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
6475 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
6476 explicitly as a stream of bytes.
6478 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6480 * config/rs6000/altivec.h (vec_xst_be): New #define.
6481 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
6482 and externalize from *altivec_vperm_<mode>_internal.
6483 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
6485 (XL_BE_V8HI): Likewise.
6486 (XL_BE_V4SI): Likewise.
6487 (XL_BE_V4SI): Likewise.
6488 (XL_BE_V2DI): Likewise.
6489 (XL_BE_V4SF): Likewise.
6490 (XL_BE_V2DF): Likewise.
6491 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
6492 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
6493 all array entries with these keys: VSX_BUILTIN_VEC_XL,
6494 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
6495 VSX_BUILTIN_VEC_XST_BE.
6496 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
6497 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
6499 (altivec_init_builtins): Replace conditional calls to def_builtin
6500 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
6501 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
6502 with unconditional calls. Remove calls to def_builtin for
6503 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
6504 __builtin_vec_xst_be.
6505 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
6506 to define_expand, and add alternate RTL generation for P8.
6507 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
6508 vsx_ld_elemrev_v8hi.
6509 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
6510 add alternate RTL generation for P8.
6511 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
6512 vsx_ld_elemrev_v16qi.
6513 (vsx_st_elemrev_v8hi): Convert define_insn
6514 to define_expand, and add alternate RTL generation for P8.
6515 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
6516 vsx_st_elemrev_v8hi.
6517 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
6518 add alternate RTL generation for P8.
6519 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
6520 vsx_st_elemrev_v16qi.
6522 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
6525 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
6526 TARGET_AVX512ER check.
6527 (ix86_option_override_internal): Set MASK_VZEROUPPER if
6528 neither -mzeroupper nor -mno-zeroupper is used and
6529 TARGET_EMIT_VZEROUPPER is set.
6530 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
6531 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
6533 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6535 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6536 folding of vector compares.
6537 (fold_build_vec_cmp): New helper function.
6538 (fold_compare_helper): New helper function.
6539 (builtin_function_type): Add compare builtins to the list of functions
6540 having unsigned arguments. Cosmetic updates to comment indentation.
6541 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
6542 the not+eq combination.
6544 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6546 PR tree-optimization/82726
6547 PR tree-optimization/70754
6548 * tree-predcom.c (order_drefs_by_pos): New function.
6549 (combine_chains): Move code setting has_max_use_after to...
6550 (try_combine_chains): ...here. New parameter. Sort combined chains
6551 according to position information.
6552 (tree_predictive_commoning_loop): Update call to above function.
6553 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
6555 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6557 PR tree-optimization/82726
6559 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6561 PR tree-optimization/70754
6562 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6563 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6564 combined stmt before it if not NULL.
6565 (combine_chains): Process refs reversely and compute dominance point
6569 2017-02-23 Bin Cheng <bin.cheng@arm.com>
6571 PR tree-optimization/79663
6572 * tree-predcom.c (combine_chains): Process refs in reverse order
6573 only for ZERO length chains, and add explaining comment.
6575 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6577 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
6578 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
6579 * doc/invoke.texi (ARM Options): Add armv8.3-a.
6581 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6583 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
6585 2017-11-15 Martin Liska <mliska@suse.cz>
6587 * tree-cfg.c (pass_warn_function_return::execute):
6588 Compare warn_return_type for greater than zero.
6590 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
6594 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
6595 to return true on Xeon and not on Xeon Phi.
6596 (ix86_check_avx256_register): Changed to ...
6597 (ix86_check_avx_upper_register): ... this. Add extra check for
6598 VALID_AVX512F_REG_OR_XI_MODE.
6599 (ix86_avx_u128_mode_needed): Changed
6600 ix86_check_avx256_register to ix86_check_avx_upper_register.
6601 (ix86_check_avx256_stores): Changed to ...
6602 (ix86_check_avx_upper_stores): ... this. Changed
6603 ix86_check_avx256_register to ix86_check_avx_upper_register.
6604 (ix86_avx_u128_mode_after): Changed
6605 avx_reg256_found to avx_upper_reg_found. Changed
6606 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
6607 (ix86_avx_u128_mode_entry): Changed
6608 ix86_check_avx256_register to ix86_check_avx_upper_register.
6609 (ix86_avx_u128_mode_exit): Ditto.
6610 * config/i386/i386.h: (host_detect_local_cpu): New define.
6612 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6614 * config/arm/xgene1.md (xgene1): Split into automatons
6615 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
6616 (xgene1_f_load): Adjust reservations and/or types.
6617 (xgene1_f_store): Likewise.
6618 (xgene1_load_pair): Likewise.
6619 (xgene1_store_pair): Likewise.
6620 (xgene1_fp_load1): Likewise.
6621 (xgene1_load1): Likewise.
6622 (xgene1_store1): Likewise.
6623 (xgene1_move): Likewise.
6624 (xgene1_alu): Likewise.
6625 (xgene1_simd): Likewise.
6626 (xgene1_bfm): Likewise.
6627 (xgene1_neon_load1): Likewise.
6628 (xgene1_neon_store1): Likewise.
6629 (xgene1_neon_logic): Likewise.
6630 (xgene1_neon_st1): Likewise.
6631 (xgene1_neon_ld1r): Likewise.
6632 (xgene1_alu_cond): Added.
6633 (xgene1_shift_reg): Likwise.
6634 (xgene1_bfx): Likewise.
6635 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6637 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6640 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6642 (can_widen_mult_without_libcall): New function.
6643 (expand_mul_overflow): If only checking unsigned mul overflow,
6644 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6645 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6646 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6647 (expand_DIVMOD): Formatting fix.
6648 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6649 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6650 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6652 PR tree-optimization/82977
6653 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6654 constructed temporary to strlen_to_stridx.put.
6656 2017-11-15 Martin Liska <mliska@suse.cz>
6658 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6659 * configure: Regenerate.
6661 2017-11-15 Martin Liska <mliska@suse.cz>
6664 * config/sh/sh-mem.cc: Use proper probability for
6667 2017-11-14 Jeff Law <law@redhat.com>
6669 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6670 the red zone for stack_clash_protection_final_dynamic_probe targets
6671 when the total dynamic stack size is zero bytes.
6673 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6674 blocks is post order.
6676 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6678 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6679 * final.c (rest_of_clean_state): Set it for the
6680 -fcompare-debug dump.
6681 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6682 class when TDF_COMPARE_DEBUG is set.
6684 * dwarf2out.c (gen_producer_string): Discard
6687 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6690 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6691 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6692 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6693 (__TGMATH_REAL_2_3): Remove macros.
6694 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6695 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6697 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6699 (remquo): Define using __TGMATH_REAL_3.
6701 2017-11-14 Jeff Law <law@redhat.com>
6703 * vr-values.c: New file with contents extracted from tree-vrp.c.
6704 * Makefile.in (OBJS): Add vr-values.o
6705 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6706 (set_value_range, set_and_canonicalize_value_range): Likewise.
6707 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6708 (value_range_constant_singleton, symbolic_range_p): Likewise.
6709 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6710 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6711 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6712 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6713 (find_case_label_range, find_case_label_index): Likewise.
6714 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6715 (range_int_cst_singleton_p, value_inside_range): Likewise.
6716 (get_single_symbol): Likewise.
6717 (switch_update): Move structure definition here.
6718 (to_remove_edges, to_update_switch_stmts): Provide externs.
6719 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6720 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6721 (vrp_val_is_min, set_value_range): Likewise.
6722 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6723 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6724 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6725 (range_is_nonnull, range_int_cst_p): Likewwise.
6726 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6727 (get_single_symbol, operand_less_p): Likewise
6728 (compare_values_warnv, compare_values): Likewise.
6729 (value_inside_range, value_range_constant_singleton): Likewise.
6730 (zero_nonzero_bitgs_from_vr): Likewise.
6731 (extract_range_from_binary_expr_1): Likewise.
6732 (overflow_comparison_p): Likewise.
6733 (to_remove_edges, to_update_switch_stmts): Likewise.
6734 (find_case_label-index, find_case_label_range): Likewise.
6735 (switch_update, set_value_range_to_nonnegative): Remove.
6736 (set_value_range_to_truthvalue): Likewise.
6737 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6738 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6739 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6740 (find_case_label_ranges, test_for_singularity): Likewise.
6741 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6742 (x_vr_values): Move to its remaining use site.
6744 2017-11-10 Jeff Law <law@redhat.com>
6746 * vr-values.h (VR_INITIALIZER): Move #define here.
6747 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6748 * Makefile.in (OBJS): Add tree-evrp.o
6749 * tree-vrp.h (assert_info): Move structure definition here.
6750 (set_value_range_to_varying): Prototype.
6751 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6752 (infer_value_range, register_edge_assert_for): Likewise.
6753 (stmt_interesting_for_vrp): Likewise.
6754 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6755 (set_value_range_to_varying): No longer static.
6756 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6757 (infer_value_range, register_edge_assert_for): Likewise.
6759 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6761 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6762 generate the XXBRD instruction.
6764 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6765 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6766 as being compatible if -mabi=ieeelongdouble.
6767 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6768 to setup float128 built-ins with hardware support.
6769 (BU_FLOAT128_HW_2): Likewise.
6770 (BU_FLOAT128_HW_3): Likewise.
6771 (BU_FLOAT128_HW_VSX_1): Likewise.
6772 (BU_FLOAT128_HW_VSX_2): Likewise.
6773 (scalar_extract_expq): Change float128 built-in functions to
6774 accommodate having both KFmode and TFmode functions. Use the
6775 KFmode variant as the default.
6776 (scalar_extract_sigq): Likewise.
6777 (scalar_test_neg_qp): Likewise.
6778 (scalar_insert_exp_q): Likewise.
6779 (scalar_insert_exp_qp): Likewise.
6780 (scalar_test_data_class_qp): Likewise.
6781 (sqrtf128_round_to_odd): Delete processing the round to odd
6782 built-in functions as special built-in functions, and define them
6783 as float128 built-ins. Use the KFmode variant as the default.
6784 (truncf128_round_to_odd): Likewise.
6785 (addf128_round_to_odd): Likewise.
6786 (subf128_round_to_odd): Likewise.
6787 (mulf128_round_to_odd): Likewise.
6788 (divf128_round_to_odd): Likewise.
6789 (fmaf128_round_to_odd): Likewise.
6790 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6791 support for KFmode and TFmode xststdcqp calls.
6792 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6793 point, switch the built-in handlers for the get/set float128
6794 exponent, get float128 mantissa, float128 test built-ins, and the
6795 float128 round to odd built-in functions. Eliminate creating the
6796 float128 round to odd built-in functions as special built-ins.
6797 (rs6000_init_builtins): Eliminate special creation of the float128
6798 round to odd built-in functions.
6799 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6800 function insns to support both TFmode and KFmode variants.
6801 (xsxsigqp_<mode>): Likewise.
6802 (xsiexpqpf_<mode>): Likewise.
6803 (xsiexpqp_<mode>): Likewise.
6804 (xststdcqp_<mode>): Likewise.
6805 (xststdcnegqp_<mode>): Likewise.
6806 (xststdcqp_<mode>): Likewise.
6808 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6810 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6811 unused path_in_freq_ptr parameter.
6812 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6814 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6816 * ipa-inline.c (edge_badness): Dump sreal frequency.
6817 (compute_inlined_call_time): Match natural implementaiton ...
6818 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6819 forgotten division by CGRAPH_FREQ_BASE.
6821 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6823 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6824 Solaris 11. Update comment.
6825 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6827 * config/sol2.h (STARTFILE_SPEC): Likewise.
6828 * configure: Regenerate.
6830 2017-11-14 Carl Love <cel@us.ibm.com>
6832 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6833 le_ and be_ prefixes to swap* variables. Remove
6834 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6837 2017-11-14 Jason Merrill <jason@redhat.com>
6839 Support GTY((cache)) on hash_map.
6840 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6841 (ggc_cache_remove): Override it instead of ggc_mx.
6842 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6843 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6844 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6845 (simple_cache_map_traits): Override maybe_mx.
6846 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6847 (hash_map): Friend gt_cleare_cache.
6848 (gt_cleare_cache): New.
6849 * tree.h (tree_cache_traits): New hash_map traits class.
6850 (tree_cache_map): New typedef.
6852 2017-11-14 Richard Biener <rguenther@suse.de>
6854 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6855 paramter and handling.
6856 (cleanup_control_flow_bb): Likewise.
6857 (cleanup_control_flow_pre): New helper performing a DFS walk
6858 to call cleanup_control_flow_bb in PRE order.
6859 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6860 via cleanup_control_flow_pre.
6862 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6864 * config/aarch64/aarch64-simd.md
6865 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6866 (*aarch64_simd_bsl<mode>_alt): Likewise.
6867 (aarch64_simd_bsldi_internal): New.
6868 (aarch64_simd_bsldi_alt): Likewise.
6870 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6872 * tracer.c (better_p): Do not compare frequencies.
6873 * reg-stack.c (better_edge): Likewise.
6874 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
6877 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6879 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
6880 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
6881 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
6882 * lto-streamer-in.c (input_function): Use update_max_bb_count.
6883 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
6884 * predict.c (maybe_hot_frequency_p): Inline to ...
6885 (maybe_hot_count_p): ... here; rewrite to counts.
6886 (counts_to_freqs): Rename to ...
6887 (update_max_bb_count): ... this one.
6888 (expensive_function_p): Use counts.
6889 (estimate_bb_frequencies): Update.
6890 (rebuild_frequencies): Update.
6891 * predict.h (counts_to_freqs): Rename to ...
6892 (update_max_bb_count): ... this one.
6893 * profile.c (compute_branch_probabilities): Add debug info
6894 * tree-inline.c (expand_call_inline): Update debug info.
6895 (optimize_inline_calls): Use update_max_bb_count..
6896 (tree_function_versioning): Use update_max_bb_count..
6897 * value-prof.c (gimple_value_profile_transformations):
6898 Do not use update_max_bb_count.
6900 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6902 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
6903 always use frequencies.
6905 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6907 * bb-reorder.c: Remove frequencies from comments.
6908 (better_edge_p): Use profile counts.
6909 (find_traces): Dump profile counts.
6910 (rotate_loop): Use profile counts.
6911 (find_traces_1_round): Likewise.
6912 (connect_better_edge_p): Use counts instead of probabilities for
6914 (copy_bb_p): Drop early check for non-0 frequency.
6915 (sanitize_hot_paths): Update comments.
6917 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6919 * ipa-split.c (struct split_point): Add count.
6920 (consider_split): Do not compute incoming frequency; compute incoming
6921 count and store it to split_point.
6922 (split_function): Set count of the call to split part correctly.
6924 2017-11-13 Carl Love <cel@us.ibm.com>
6926 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
6928 2017-11-13 Tom Tromey <tom@tromey.com>
6930 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
6932 2017-11-13 Carl Love <cel@us.ibm.com>
6934 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
6935 Add support for builtins:
6936 unsigned int vec_first_{,miss}_match_{,or_eos}index,
6937 vector {un,}signed {char,int,short},
6938 vector {un,}signed {char,int,short}) arguments.
6939 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
6940 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
6941 Add BU_P9V_AV_2 expansions for the builtins.
6942 * config/rs6000/altivec.h (vec_first_match_index,
6943 vec_first_mismatch_index, vec_first_match_or_eos_index,
6944 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
6945 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
6946 new extern declaration.
6947 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
6948 * config/rs6000/vsx.md (first_match_index_<mode>,
6949 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
6950 first_mismatch_or_eos_index_<mode>): Add define expand.
6951 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
6952 * doc/extend.texi: Update the built-in documenation file for the new
6955 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
6957 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
6958 into the min/max operations for _Float<N> and _Float<N>X types.
6960 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6963 * dwarf2out.c (do_eh_frame): New static variable.
6964 (dwarf2out_begin_prologue): Set it.
6965 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
6967 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6969 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
6971 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
6973 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6976 * debug.h (dwarf2out_do_eh_frame): Declare.
6977 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
6978 (dwarf2out_do_frame): Use it.
6979 (dwarf2out_do_cfi_asm): Likewise.
6980 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
6981 (dwarf2out_assembly_start): Likewise.
6982 (dwarf2out_begin_prologue): Fix comment.
6983 * toplev.c (compile_file): Always call dwarf2out_frame_finish
6984 if the target needs either debug or unwind DWARF2 info.
6985 * lto-opts.c (lto_write_options): Do not save -fexceptions,
6986 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
6987 -ftrapping-math, -ftrapv and -fwrapv.
6989 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6991 * cgraph.c (cgraph_edge::sreal_frequency): New function.
6992 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
6993 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
6994 (estimate_edge_size_and_time): Likewise.
6995 (ipa_merge_fn_summary_after_inlining): Likewise.
6996 * ipa-inline.c (cgraph_freq_base_rec): Remove.
6997 (compute_uninlined_call_time): Use sreal_frequency.
6998 (compute_inlined_call_time): Likewise.
6999 (ipa_inline): Do not initialize cgraph_freq_base_rec.
7000 * profile-count.c: Include sreal.h.
7001 (profile_count::to_sreal_scale): New.
7002 * profile-count.h: Forward declare sreal.
7003 (profile_count::to_sreal_scale): Declare.
7005 2017-11-13 Nathan Sidwell <nathan@acm.org>
7007 * diagnostic.c (maybe_line_and_column): New.
7008 (diagnostic_get_location_text): Use it.
7009 (diagnostic_report_current_module): Likewise.
7010 (test_diagnostic_get_location_text): Add tests.
7012 2017-11-13 Luis Machado <luis.machado@linaro.org>
7014 * doc/md.texi (Specifying processor pipeline description): Fix
7015 incorrect latency for the div instruction example.
7017 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7019 PR tree-optimization/78821
7020 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7021 that bit_not_p is the same.
7022 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7023 (split_group): Count precisely bit_not_p bits in each statement.
7024 (invert_op): New function.
7025 (imm_store_chain_info::output_merged_store): Use invert_op to
7026 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7027 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7029 2017-11-13 Martin Liska <mliska@suse.cz>
7031 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7032 (struct source_info): Likewise.
7033 (add_branch_counts): Likewise.
7034 (add_line_counts): Likewise.
7035 (function_summary): Likewise.
7036 (output_intermediate_line): Likewise.
7037 (generate_results): Likewise.
7039 2017-11-13 Martin Liska <mliska@suse.cz>
7041 * gcov.c (struct block_info): Remove typedef for block_t.
7042 (struct line_info): Likewise.
7043 (line_info::has_block): Likewise.
7044 (EXIT_BLOCK): Likewise.
7045 (unblock): Likewise.
7046 (circuit): Likewise.
7047 (get_cycles_count): Likewise.
7048 (process_file): Likewise.
7049 (read_graph_file): Likewise.
7050 (solve_flow_graph): Likewise.
7051 (find_exception_blocks): Likewise.
7052 (add_line_counts): Likewise.
7053 (accumulate_line_info): Likewise.
7054 (output_line_details): Likewise.
7056 2017-11-13 Martin Liska <mliska@suse.cz>
7058 * gcov.c (struct arc_info): Remove typedef for arc_t.
7059 (struct line_info): Likewise.
7060 (add_branch_counts): Likewise.
7061 (output_branch_count): Likewise.
7062 (function_info::~function_info): Likewise.
7063 (circuit): Likewise.
7064 (output_intermediate_line): Likewise.
7065 (read_graph_file): Likewise.
7066 (solve_flow_graph): Likewise.
7067 (find_exception_blocks): Likewise.
7068 (add_line_counts): Likewise.
7069 (accumulate_line_info): Likewise.
7070 (output_line_details): Likewise.
7071 (output_function_details): Likewise.
7073 2017-11-13 Martin Liska <mliska@suse.cz>
7075 * gcov.c (struct function_info): Remove typedef for function_t.
7076 (struct source_info): Likewise.
7077 (source_info::get_functions_at_location): Likewise.
7078 (solve_flow_graph): Likewise.
7079 (find_exception_blocks): Likewise.
7080 (add_line_counts): Likewise.
7081 (output_intermediate_file): Likewise.
7082 (process_file): Likewise.
7083 (generate_results): Likewise.
7084 (release_structures): Likewise.
7085 (read_graph_file): Likewise.
7086 (read_count_file): Likewise.
7087 (accumulate_line_counts): Likewise.
7088 (output_lines): Likewise.
7090 2017-11-13 Martin Liska <mliska@suse.cz>
7092 * gcov.c (function_info::function_info): Remove num_counts
7093 and add vector<gcov_type>.
7094 (function_info::~function_info): Use the vector.
7095 (process_file): Likewise.
7096 (read_graph_file): Likewise.
7097 (read_count_file): Likewise.
7098 (solve_flow_graph): Likewise.
7100 2017-11-13 Martin Liska <mliska@suse.cz>
7102 * gcov.c (function_info::is_artificial): New function.
7103 (process_file): Erase all artificial early.
7104 (generate_results): Skip as all artificial are already
7107 2017-11-13 Martin Liska <mliska@suse.cz>
7109 * gcov.c (read_graph_file): Store to global vector of functions.
7110 (read_count_file): Iterate the vector.
7111 (process_file): Likewise.
7112 (generate_results): Likewise.
7113 (release_structures): Likewise.
7115 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7117 PR tree-optimization/82954
7118 * gimple-ssa-store-merging.c
7119 (imm_store_chain_info::coalesce_immediate_stores): If
7120 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
7122 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
7124 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7125 Upddate call to ENDIAN_LANE_N.
7126 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
7127 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
7128 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
7129 and use aarch64_endian_lane_rtx.
7130 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
7132 2017-11-12 Tom de Vries <tom@codesourcery.com>
7134 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
7136 2017-11-12 Tom de Vries <tom@codesourcery.com>
7138 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
7140 2017-11-12 Tom de Vries <tom@codesourcery.com>
7142 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
7144 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7145 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7146 * defaults.h (ASM_OUTPUT_LABELREF): Same.
7148 2017-11-11 Martin Sebor <msebor@redhat.com>
7151 * doc/extend.texi (attribute nonstring): Remove spurious argument.
7154 * prefic.c (translate_name): Replace strncpy with memcpy to
7155 avoid -Wstringop-truncation.
7157 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7159 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
7161 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7163 * predict.c (maybe_hot_frequency_p): Do not use cfun.
7165 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7167 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
7170 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7172 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
7173 enable generating XXBRH if the value is in a vector register.
7174 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
7175 value is in a vector register.
7176 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
7177 register to register bswap64's instead of doing the GPR sequence
7178 used on previous machines.
7179 (bswapdi2_xxbrd): New insn.
7180 (bswapdi2_reg): Disallow on ISA 3.0.
7181 (register to register bswap64 splitter): Do not split the insn on
7182 ISA 3.0 systems that use XXBRD.
7184 2017-11-10 Martin Sebor <msebor@redhat.com>
7187 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
7189 (find_subframework_file): Same.
7191 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7193 * auto-profile.c (afdo_indirect_call): Drop frequency.
7194 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
7195 (cgraph_node::create_edge): Drop frequency argument.
7196 (cgraph_node::create_indirect_edge): Drop frequency argument.
7197 (cgraph_edge::make_speculative): Drop frequency arguments.
7198 (cgraph_edge::resolve_speculation): Do not update frequencies
7199 (cgraph_edge::dump_edge_flags): Do not dump frequency.
7200 (cgraph_node::dump): Check consistency in IPA mode.
7201 (cgraph_edge::maybe_hot_p): Use IPA counter.
7202 (cgraph_edge::verify_count_and_frequency): Rename to ...
7203 (cgraph_edge::verify_count): ... this one; drop frequency checking.
7204 (cgraph_node::verify_node): Update.
7205 * cgraph.h (struct cgraph_edge): Drop frequency.
7206 (cgraph_edge::frequency): New function.
7207 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
7209 (cgraph_edge::rebuild_edges): Likewise.
7210 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
7211 (duplicate_thunk_for_node): Do not pass frequency.
7212 (cgraph_node::create_clone): Scale only counts.
7213 (cgraph_node::create_virtual_clone): Do not pass frequency.
7214 (cgraph_node::create_edge_including_clones): Do not pass frequency.
7215 (cgraph_node::create_version_clone): Do not pass frequency.
7216 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
7217 (cgraph_node::expand_thunk): Do not pass frequency.
7218 (cgraph_node::create_wrapper): Do not pass frequency.
7219 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
7221 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
7222 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
7223 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
7224 (gather_caller_stats): Use frequency function.
7225 (ipcp_cloning_candidate_p): Use frequency function.
7226 (ipcp_propagate_stage): Use frequency function.
7227 (get_info_about_necessary_edges): Use frequency function.
7228 (update_profiling_info): Update only IPA profile.
7229 (update_specialized_profile): Use frequency functoin.
7230 (perhaps_add_new_callers): Update only IPA profile.
7231 * ipa-devirt.c (ipa_devirt): Use IPA profile.
7232 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
7233 (dump_ipa_call_summary): Use frequency function.
7234 (estimate_edge_size_and_time): Use frequency function.
7235 (ipa_merge_fn_summary_after_inlining): Use frequency function.
7236 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
7237 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
7238 (update_noncloned_counts): ... ths one; scale counts only.
7239 (clone_inlined_nodes): Do not scale frequency.
7240 (inline_call): Do not pass frequency.
7241 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
7242 (compute_inlined_call_time): Use IPA profile.
7243 (want_inline_small_function_p): Use IPA profile.
7244 (want_inline_self_recursive_call_p): Use IPA profile.
7245 (edge_badness): Use IPA profile.
7246 (lookup_recursive_calls): Use IPA profile.
7247 (recursive_inlining): Do not pass frequency.
7248 (resolve_noninline_speculation): Do not update frequency.
7249 (inline_small_functions): Collect max of IPA profile.
7250 (dump_overall_stats): Dump IPA porfile.
7251 (dump_inline_stats): Dump IPA porfile.
7252 (ipa_inline): Collect IPA stats.
7253 * ipa-inline.h (clone_inlined_nodes): Update prototype.
7254 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
7255 (ipa_propagate_frequency): Use frequency function.
7256 (ipa_profile): Cleanup.
7257 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
7258 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
7259 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
7260 (input_node): Do not stream frequency.
7261 (input_edge): Do not stream frequency.
7262 (merge_profile_summaries): Scale only IPA profiles.
7263 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
7264 * predict.c (drop_profile): Do not recompute frequency.
7265 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
7266 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
7267 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
7268 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
7269 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
7270 * tree-inline.c (copy_bb): Do not scale frequency.
7271 (expand_call_inline): Do not scale frequency.
7272 (tree_function_versioning): Do not scale frequency.
7273 * ubsan.c (ubsan_create_edge): Do not pass frequency.
7275 2017-11-10 Julia Koval <julia.koval@intel.com>
7277 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
7278 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
7279 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
7280 (_mm256_maskz_gf2p8affine_epi64_epi8)
7281 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
7282 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
7283 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
7284 (__builtin_ia32_vgf2p8affineqb_v32qi)
7285 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
7286 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
7288 2017-11-10 Tamar Christina <tamar.christina@arm.com>
7292 (arm_option_override): Refactor.
7293 (arm_option_reconfigure_globals): New.
7294 (arm_options_perform_arch_sanity_checks): New.
7295 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
7297 (arm_options_perform_arch_sanity_checks): Likewise
7299 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
7301 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
7302 (power9-qpmul): New.
7303 * rs6000/rs6000.md ("type" attr): Add qmul.
7304 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
7305 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
7306 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
7308 2017-11-10 Martin Sebor <msebor@redhat.com>
7311 * builtins.c (compute_objsize): Handle arrays that
7312 compute_builtin_object_size likes to fail for. Make extern.
7313 * builtins.h (compute_objsize): Declare.
7314 (check_strncpy_sizes): New function.
7315 (expand_builtin_strncpy): Call check_strncpy_sizes.
7316 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
7317 -Wstringop-truncation.
7318 (gimple_fold_builtin_strncat): Same.
7319 * gimple.c (gimple_build_call_from_tree): Set call location.
7320 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
7321 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
7322 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
7323 (handle_builtin_strlen): Use strlen_to_stridx.
7324 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
7326 Use strlen_to_stridx.
7327 (pass_strlen::execute): Release strlen_to_stridx.
7328 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
7329 (-Wstringop-truncation): Document new option.
7331 2017-11-10 Martin Liska <mliska@suse.cz>
7333 PR gcov-profile/82702
7334 * gcov.c (main): Handle intermediate files in a different
7336 (get_gcov_intermediate_filename): New function.
7337 (output_gcov_file): Remove support of intermediate files.
7338 (generate_results): Allocate intermediate file.
7339 (release_structures): Clean-up properly fn_end.
7340 (output_intermediate_file): Start iterating with line 1.
7342 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7344 PR tree-optimization/82929
7345 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7346 ops_swapped_p non-static data member.
7347 (store_immediate_info::store_immediate_info): Clear it.
7348 (imm_store_chain_info::coalesce_immediate_stores): If swapping
7349 ops set ops_swapped_p.
7350 (count_multiple_uses): Handle ops_swapped_p.
7352 2017-11-10 Martin Liska <mliska@suse.cz>
7354 * coverage.c (coverage_init): Stream information about
7355 support of has_unexecuted_blocks.
7356 * doc/gcov.texi: Document that.
7357 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
7358 * gcov.c (read_graph_file): Likewise.
7359 (output_line_beginning): Fix a small issue with
7362 2017-11-10 Bin Cheng <bin.cheng@arm.com>
7364 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
7365 reference of trivial component.
7367 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7370 * gimple-ssa-store-merging.c
7371 (pass_store_merging::terminate_all_aliasing_chains): For
7372 gimple_store_p stmts also call refs_output_dependent_p.
7374 PR rtl-optimization/82913
7375 * compare-elim.c (try_merge_compare): Punt if def_insn is not
7378 2017-11-09 Jeff Law <law@redhat.com>
7380 * vr-values.h: New file with vr_values class.
7381 * tree-vrp.c: Include vr-values.h
7382 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
7383 data objects into the vr_values class.
7384 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
7385 (get_value_range): Make it a member function within vr_values class.
7386 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
7387 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
7388 (op_with_constant_singleton_value_range): Likewise.
7389 (extract_range_for_var_from_comparison_expr): Likewise.
7390 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
7391 (extract_range_from_binary_expr): Likewise.
7392 (extract_range_from_unary_expr): Likewise.
7393 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
7394 (check_for_binary_op_overflow, extract_range_basic): Likewise.
7395 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
7396 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
7397 (compare_name_with_value, compare_names): Likewise.
7398 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
7399 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
7400 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
7401 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
7402 (extract_range_from_phi_node): Likewise.
7403 (simplify_truth_ops_using_ranges): Likewise.
7404 (simplify_div_or_mod_using_ranges): Likewise.
7405 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
7406 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
7407 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
7408 (simplify_float_conversion_using_ranges): Likewise.
7409 (simplify_internal_call_using_ranges): Likewise.
7410 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
7411 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
7412 poitner via x_vr_values for calls into gimple folder.
7413 (vrp_initialize_lattice): Make this the vr_values ctor.
7414 (vrp_free_lattice): Make this the vr_values dtor.
7415 (set_vr_value): New function.
7416 (class vrp_prop): Add vr_values data member. Add various member
7417 functions as well as member functions that delegate to vr_values.
7418 (check_array_ref): Make a member function within vrp_prop class.
7419 (search_for_addr_array, vrp_initialize): Likewise.
7420 (vrp_finalize): Likewise. Revamp to avoid direct access to
7421 vr_value, values_propagated, etc.
7422 (check_array_bounds): Extract vrp_prop class instance pointer from
7423 walk info structure. Use it to call member functions.
7424 (check_all_array_refs): Make a member function within vrp_prop class.
7425 Smuggle class instance pointer via walk info structure.
7426 (x_vr_values): New local static.
7427 (vrp_valueize): Use x_vr_values to get class instance.
7428 (vr_valueize_1): Likewise.
7429 (class vrp_folder): Add vr_values data member. Add various member
7430 functions as well as member functions that delegate to vr_values.
7431 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
7432 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
7433 class instance from vr_values. Use it to call member functions.
7434 (vrp_dom_walker): Add vr_values data member.
7435 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
7436 instance via x_vr_values.
7437 (identify_jump_threads): Accept vr_values as argument. Store
7438 it into the walker structure.
7439 (evrp_dom_walker): Add vr_values class data member. Add various
7441 (evrp_dom_walker::try_find_new_range): Use vr_values data
7442 member to access the memory allocator.
7443 (evrp_dom_walker::before_dom_children): Store vr_values class
7444 instance into the vrp_folder class.
7445 (evrp_dom_walker::push_value_range): Rework to avoid direct
7446 access to num_vr_values and vr_value.
7447 (evrp_dom_walker::pop_value_range): Likewise.
7448 (execute_early_vrp): Remove call to vrp_initialize_lattice.
7449 Use vr_values to get to dump_all_value_ranges member function.
7450 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
7451 and simplify_cond_using_ranges_2 via vrp_prop class instance.
7452 Pass vr_values class instance down to identify_jump_threads.
7453 Remove call to vrp_free_lattice.
7454 (debug_all_value_ranges): Remove.
7456 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
7457 (vrp_folder): Likewise.
7459 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
7460 Get it from the existing bitmap instead.
7461 (vrp_intersect_ranges_1): Likewise.
7463 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7465 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7467 (store_immediate_info::store_immediate_info): Add bitnotp argument,
7468 set bit_not_p to it.
7469 (imm_store_chain_info::coalesce_immediate_stores): Break group
7470 if bit_not_p is different.
7471 (count_multiple_uses, split_group,
7472 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
7473 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
7474 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
7475 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
7476 to store_immediate_info ctor.
7478 2017-11-09 Jim Wilson <jimw@sifive.com>
7480 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
7481 (scan_prog_file): Likewise.
7483 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7485 * bb-reorder.c (max_entry_frequency): Remove.
7486 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
7487 connect_traces, push_to_next_round_p): Remove prototypes.
7488 (find_traces_1_round): Use counts only.
7489 (push_to_next_round_p): Likewise.
7490 (find_traces): Likewise.
7491 (rotate_loop): Likewise.
7492 (find_traces_1_round): Likewise.
7493 (connect_traces): Likewise.
7494 (edge_order): Likewise.
7496 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
7498 * config/arm/arm.c (output_return_instruction): Add comments to
7499 indicate requirement for cmse_nonsecure_entry return to account
7500 for the size of clearing instruction output here.
7501 (thumb_exit): Likewise.
7502 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
7503 return in hardfloat mode.
7505 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
7507 * config/rs6000/rs6000.c (machine_function): Add a bool,
7508 "toc_is_wrapped_separately".
7509 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
7510 if it wasn't explicitly set or unset, we are optimizing for speed, and
7511 doing separate shrink-wrapping.
7512 (rs6000_get_separate_components): Enable the TOC component if
7513 saving the TOC register in the prologue.
7514 (rs6000_components_for_bb): Handle the TOC component.
7515 (rs6000_emit_prologue_components): Store the TOC register where needed.
7516 (rs6000_set_handled_components): Mark TOC as handled, if handled.
7517 (rs6000_emit_prologue): Don't save the TOC if that is already done.
7519 2017-11-09 Martin Jambor <mjambor@suse.cz>
7521 * ipa-param-manipulation.c: New file.
7522 * ipa-param-manipulation.h: Likewise.
7523 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
7524 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
7525 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
7526 (ipa_parm_adjustment): Likewise.
7527 (ipa_parm_adjustment_vec): Likewise.
7528 (ipa_get_vector_of_formal_parms): Moved declaration to
7529 ipa-param-manipulation.h.
7530 (ipa_get_vector_of_formal_parm_types): Likewise.
7531 (ipa_modify_formal_parameters): Likewise.
7532 (ipa_modify_call_arguments): Likewise.
7533 (ipa_combine_adjustments): Likewise.
7534 (ipa_dump_param_adjustments): Likewise.
7535 (ipa_modify_expr): Likewise.
7536 (ipa_get_adjustment_candidate): Likewise.
7537 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
7538 ipa-param-manipulation.c.
7539 (ipa_get_vector_of_formal_parm_types): Likewise.
7540 (ipa_modify_formal_parameters): Likewise.
7541 (ipa_modify_call_arguments): Likewise.
7542 (ipa_modify_expr): Likewise.
7543 (get_ssa_base_param): Likewise.
7544 (ipa_get_adjustment_candidate): Likewise.
7545 (index_in_adjustments_multiple_times_p): Likewise.
7546 (ipa_combine_adjustments): Likewise.
7547 (ipa_dump_param_adjustments): Likewise.
7548 * tree-sra.c: Also include ipa-param-manipulation.h
7549 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
7552 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7553 Alan Hayward <alan.hayward@arm.com>
7554 David Sherwood <david.sherwood@arm.com>
7556 * doc/sourcebuild.texi (vect_masked_store): Document.
7558 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7559 Alan Hayward <alan.hayward@arm.com>
7560 David Sherwood <david.sherwood@arm.com>
7562 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
7564 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7565 Alan Hayward <alan.hayward@arm.com>
7566 David Sherwood <david.sherwood@arm.com>
7568 * doc/sourcebuild.texi (vect_variable_length): Document.
7570 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7571 Alan Hayward <alan.hayward@arm.com>
7572 David Sherwood <david.sherwood@arm.com>
7574 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
7576 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7577 Alan Hayward <alan.hayward@arm.com>
7578 David Sherwood <david.sherwood@arm.com>
7580 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
7582 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7583 Alan Hayward <alan.hayward@arm.com>
7584 David Sherwood <david.sherwood@arm.com>
7586 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
7587 previously undocumented selectors.
7588 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
7590 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7592 * doc/rtl.texi (const_vector): Say that elements can be
7593 const_wide_ints too.
7594 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
7595 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
7596 (gen_vec_duplicate): Use it instead of CONSTANT_P.
7597 * optabs.c (expand_vector_broadcast): Likewise.
7599 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7600 Alan Hayward <alan.hayward@arm.com>
7601 David Sherwood <david.sherwood@arm.com>
7603 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
7604 scaled index even if the unscaled address was invalid.
7605 Don't increase the complexity of using a scale in that case.
7607 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7608 Alan Hayward <alan.hayward@arm.com>
7609 David Sherwood <david.sherwood@arm.com>
7611 * doc/rtl.texi: Rewrite the subreg rules so that they partition
7612 the inner register into REGMODE_NATURAL_SIZE bytes rather than
7613 UNITS_PER_WORD bytes.
7614 * emit-rtl.c (validate_subreg): Divide subregs into blocks
7615 based on REGMODE_NATURAL_SIZE of the inner mode.
7616 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
7617 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
7618 * expmed.c (lowpart_bit_field_p): Divide the value up into
7619 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
7620 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
7621 whether something is likely to occupy more than one register.
7623 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7626 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7627 profile_count::adjust_for_ipa_scaling.
7628 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7629 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7631 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7633 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7635 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7636 (split_group): Add total_orig and total_new arguments, estimate the
7637 number of statements related to the store group without store merging
7638 and with store merging.
7639 (imm_store_chain_info::output_merged_store): Adjust split_group
7640 callers, punt if estimated number of statements with store merging
7641 is not smaller than estimated number of statements without it.
7643 (handled_load): Remove has_single_use checks.
7644 (pass_store_merging::process_store): Likewise.
7646 2017-11-09 Richard Biener <rguenther@suse.de>
7648 PR tree-optimization/82902
7649 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7651 2017-11-09 Martin Liska <mliska@suse.cz>
7654 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7657 2017-11-09 Martin Liska <mliska@suse.cz>
7659 PR tree-optimization/82669
7660 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7662 2017-11-09 Martin Liska <mliska@suse.cz>
7664 PR gcov-profile/48463
7665 * coverage.c (coverage_begin_function): Output also end locus
7666 of a function and information whether the function is
7668 * gcov-dump.c (tag_function): Parse and print the information.
7669 * gcov.c (INCLUDE_MAP): Add include.
7670 (INCLUDE_SET): Likewise.
7671 (struct line_info): Move earlier in the source file because
7672 of vector<line_info> in function_info structure.
7673 (line_info::line_info): Likewise.
7674 (line_info::has_block): Likewise.
7675 (struct source_info): Add new member index.
7676 (source_info::get_functions_at_location): New function.
7677 (function_info::group_line_p): New function.
7678 (output_intermediate_line): New function.
7679 (output_intermediate_file): Use the mentioned function.
7680 (struct function_start): New.
7681 (struct function_start_pair_hash): Likewise.
7682 (process_file): Add code that identifies group functions.
7683 Assign lines either to global or function scope.
7684 (generate_results): Skip artificial functions.
7685 (find_source): Assign index for each source file.
7686 (read_graph_file): Read new flag artificial and end_line.
7687 (add_line_counts): Assign it either to global of function scope.
7688 (accumulate_line_counts): Isolate core of the function to
7689 accumulate_line_info and call it for both function and global
7691 (accumulate_line_info): New function.
7692 (output_line_beginning): Fix GNU coding style.
7693 (print_source_line): New function.
7694 (output_line_details): Likewise.
7695 (output_function_details): Likewise.
7696 (output_lines): Iterate both source (global) scope and function
7698 (struct function_line_start_cmp): New class.
7699 * doc/gcov.texi: Reflect changes in documentation.
7701 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7704 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7705 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7706 and similarly for not instead of neg.
7708 2017-11-08 Andi Kleen <ak@linux.intel.com>
7710 * config/i386/i386.opt: Add -mforce-indirect-call.
7711 * config/i386/predicates.md: Check for flag_force_indirect_call.
7712 * doc/invoke.texi: Document -mforce-indirect-call
7714 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7716 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7718 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7719 config/riscv/riscv.c (predict.h): New include.
7720 (riscv_slow_unaligned_access_p): No longer static.
7721 (riscv_block_move_straight): Add require.
7722 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7724 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7727 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7728 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7729 nonimmediate_operand predicate for operand 1 instead of
7732 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7734 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7736 * config/aarch64/constraints.md (Uml): New constraint.
7737 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7740 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7742 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7743 of two vec_duplicates into a vec_concat.
7745 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7747 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7748 Simplify vec_merge of vec_duplicate and vec_concat.
7749 * config/aarch64/constraints.md (Utq): New constraint.
7750 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7753 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7755 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7756 Simplify vec_merge of vec_duplicate and const_vector.
7757 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7759 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7760 mode iterator. Update predicate on operand 1 to
7761 handle non-const_vec constants. Delete constraints.
7762 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7764 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7766 PR tree-optimization/78821
7767 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7769 (store_operand_info::store_operand_info): Initialize it to false.
7770 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7771 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7772 store in the group, and if chain_info is non-NULL, to ignore altogether
7774 (compatible_load_p): Fail if bit_not_p does not match.
7775 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7776 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7777 (pass_store_merging::process_store): Adjust
7778 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7779 call terminate_all_aliasing_chains newly when adding a store into
7780 a chain with non-NULL chain_info.
7782 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7784 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7786 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7788 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7790 (aarch64_layout_frame): Initialise emit_frame_chain.
7791 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7792 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7794 2017-11-08 Martin Liska <mliska@suse.cz>
7796 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7797 of gimple_call_internal_p.
7799 2017-11-07 Tom de Vries <tom@codesourcery.com>
7801 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7803 2017-11-07 Tom de Vries <tom@codesourcery.com>
7805 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7808 2017-11-08 Martin Liska <mliska@suse.cz>
7811 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7813 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7815 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7817 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7820 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7821 (*cmp<mode>_ccz_1): New insn with $k alternative.
7824 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7825 RTX_COMM_COMPARE as commutative as well.
7826 (ix86_binary_operator_ok): Formatting fix.
7827 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7828 *<code><mode>3<mask_name><round_saeonly_name>,
7829 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7830 *<s>mul<mode>3_highpart<mask_name>,
7831 *vec_widen_umult_even_v16si<mask_name>,
7832 *vec_widen_umult_even_v8si<mask_name>,
7833 *vec_widen_umult_even_v4si<mask_name>,
7834 *vec_widen_smult_even_v16si<mask_name>,
7835 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7836 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7837 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7838 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7839 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7840 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7841 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7842 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7843 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7844 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7845 ix86_binary_operator_ok. Formatting fixes.
7846 (*<plusminus_insn><mode>3<mask_name><round_name>,
7847 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7850 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7852 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7853 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7854 eq and ne if TARGET_ISEL.
7855 (cmp): New code_iterator.
7856 (UNS, UNSU_, UNSIK): New code_attrs.
7857 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7858 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7860 ("eq<mode>3"): ... this.
7861 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7863 ("ne<mode>3"): ... this.
7865 2017-11-07 Julia Koval <julia.koval@intel.com>
7868 * common/config/i386/i386-common.c
7869 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
7870 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
7871 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
7872 * config/i386/i386.opt: Ditto.
7873 * config/i386/i386.c (ix86_target_string): Ditto.
7874 (ix86_option_override_internal): Ditto.
7875 (ix86_init_mpx_builtins): Move MPX to args2.
7876 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
7877 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7878 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7879 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7880 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7881 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7882 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
7884 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
7887 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
7888 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
7889 (zero-extendsidi peephole2): Remove peephole.
7891 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7894 * stor-layout.c: Include attribs.h.
7895 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
7896 explicit lookup of "aligned" attribute.
7898 2017-11-07 Andrew Waterman <andrew@sifive.com>
7900 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
7901 (riscv_expand_block_move): Likewise.
7902 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
7904 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
7905 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
7906 * config/riscv/riscv.c (riscv_block_move_straight): New function.
7907 (riscv_adjust_block_mem): Likewise.
7908 (riscv_block_move_loop): Likewise.
7909 (riscv_expand_block_move): Likewise.
7910 * config/riscv/riscv.md (movmemsi): New pattern.
7912 2017-11-07 Michael Clark <michaeljclark@mac.com>
7914 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
7915 (MUSL_DYNAMIC_LINKER): Likewise.
7917 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
7919 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
7920 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
7923 2017-11-07 Richard Biener <rguenther@suse.de>
7925 * match.pd: Fix build.
7927 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
7928 Jackson Woodruff <jackson.woodruff@arm.com>
7930 PR tree-optimization/71026
7931 * match.pd: Canonicalize negate in division.
7933 2017-11-07 Sudakshina Das <sudi.das@arm.com>
7936 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
7938 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7940 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
7941 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
7943 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7945 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
7946 non-scalar integral types.
7947 * match.pd (negate_expr_p): Handle MINUS_EXPR.
7948 (-(A-B), -(~A)): New transformations.
7950 2017-11-07 Tom de Vries <tom@codesourcery.com>
7952 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
7953 semicolon after "do {} while (0)".
7954 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7955 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7956 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7957 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7958 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7960 2017-11-07 Tom de Vries <tom@codesourcery.com>
7962 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
7963 after "do {} while (0)".
7964 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7965 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7966 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7967 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7968 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7970 2017-11-07 Tom de Vries <tom@codesourcery.com>
7973 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
7975 (arm_rtx_costs_internal): Add missing semicolon after
7976 HANDLE_NARROW_SHIFT_ARITH call.
7978 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7980 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7981 disable isel if it was not set explicitly.
7983 2017-11-06 James Bowman <james.bowman@ftdichip.com>
7985 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
7986 (add_type_attribute) likewise.
7988 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
7990 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
7991 of struct ix86_frame.
7992 (ix86_initial_elimination_offset): Likewise.
7993 (ix86_expand_split_stack_prologue): Likewise.
7995 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
7997 * tree-vrp.h (enum value_range_type): Update stale comment.
7999 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8000 Alan Hayward <alan.hayward@arm.com>
8001 David Sherwood <david.sherwood@arm.com>
8003 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
8004 (aarch64_expand_vec_perm_const): Take the number of units too.
8005 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8006 (aarch64_expand_vec_perm_const): Likewise.
8007 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8008 (vec_perm<mode>): Update accordingly.
8010 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8011 Alan Hayward <alan.hayward@arm.com>
8012 David Sherwood <david.sherwood@arm.com>
8014 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8015 Take the number of units too.
8016 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8017 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8018 but check for a vector mode before rather than after the call.
8019 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8020 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8021 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8022 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8023 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8024 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8025 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8026 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8027 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8028 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8029 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8030 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8031 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8032 (aarch64_sqdmull2_n<mode>): Update accordingly.
8034 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8035 Alan Hayward <alan.hayward@arm.com>
8036 David Sherwood <david.sherwood@arm.com>
8038 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8039 the number of units too.
8040 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8041 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8042 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
8043 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
8044 (vec_store_lanesxi<mode>): Update accordingly.
8046 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8047 Alan Hayward <alan.hayward@arm.com>
8048 David Sherwood <david.sherwood@arm.com>
8050 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8051 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8052 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8053 of units rather than the mode.
8054 * config/aarch64/iterators.md (nunits): New mode attribute.
8055 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8056 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8057 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8058 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8059 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8060 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8061 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8062 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8063 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8064 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8065 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8066 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8067 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8068 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8069 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8070 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8071 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8072 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8073 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8074 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8075 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8076 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8077 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8078 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8079 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8080 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8081 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8082 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8083 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8084 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8085 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8086 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8087 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8088 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8089 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8090 (aarch64_simd_vec_setv2di): Likewise.
8092 2017-11-06 Carl Love <cel@us.ibm.com>
8094 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8096 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8097 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8099 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8101 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8102 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8103 Add power 8 macro expansions.
8104 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8105 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8106 power 8 instructions. (VSX_XXBR): Add iterator.
8108 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8110 * config/arm/arm.md (predicable_short_it): Change default to "no",
8111 improve documentation, remove uses that are identical to the default.
8112 (enabled_for_depr_it): Rename to enabled_for_short_it.
8113 * gcc/config/arm/arm-fixed.md (predicable_short_it):
8114 Remove default uses.
8115 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
8116 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
8117 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
8118 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
8120 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
8123 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
8124 float128 helper macros, which are no longer used after deleting
8125 the old 'q' built-in functions, and moving the round to odd
8126 built-in functions to being special built-in functions.
8127 (BU_FLOAT128_2): Likewise.
8128 (BU_FLOAT128_1_HW): Likewise.
8129 (BU_FLOAT128_2_HW): Likewise.
8130 (BU_FLOAT128_3_HW): Likewise.
8131 (FABSQ): Delete old 'q' built-in functions.
8132 (COPYSIGNQ): Likewise.
8133 (SQRTF128_ODD): Move round to odd built-in functions to be
8134 special built-in functions, so that we can handle
8135 -mabi=ieeelongdouble.
8136 (TRUNCF128_ODD): Likewise.
8137 (ADDF128_ODD): Likewise.
8138 (SUBF128_ODD): Likewise.
8139 (MULF128_ODD): Likewise.
8140 (DIVF128_ODD): Likewise.
8141 (FMAF128_ODD): Likewise.
8142 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
8143 built-in names to 'f128'.
8144 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
8145 old 'q' built-in functions, as the machine independent code for
8146 'f128' built-in functions handles this.
8147 (rs6000_expand_builtin): Add expansion for float128 round to odd
8148 functions, keying off on -mabi=ieeelongdouble of whether to use
8149 the KFmode or TFmode variant.
8150 (rs6000_init_builtins): Initialize the _Float128 round to odd
8152 * doc/extend.texi (PowerPC Built-in Functions): Document the old
8153 _Float128 'q' built-in functions are now mapped into the new
8154 'f128' built-in functions.
8156 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
8158 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
8159 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
8161 2017-11-06 David Malcolm <dmalcolm@redhat.com>
8164 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
8165 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
8166 * toplev.c: Include "ipa-fnsummary.h".
8167 (toplev::finalize): Call ipa_fnsummary_c_finalize.
8169 2017-11-06 Jakub Jelinek <jakub@redhat.com>
8171 PR tree-optimization/82838
8172 * gimple-ssa-store-merging.c
8173 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
8174 on a separate gimple_seq which is then appended to seq.
8176 2017-11-06 Jeff Law <law@redhat.com>
8179 * config/i386/i386.c (PROBE_INTERVAL): Remove.
8180 (get_probe_interval): New functions.
8181 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
8182 (ix86_adjust_stack_and_probe): Likewise.
8183 (output_adjust_stack_and_probe): Likewise.
8184 (ix86_emit_probe_stack_range): Likewise.
8185 (ix86_expand_prologue): Likewise.
8187 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8189 PR tree-optimization/82816
8190 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
8191 if the modes of the two types are the same.
8192 (convert_plusminus_to_widen): Likewise.
8194 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8196 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
8198 (usadv16qi): New define_expand.
8199 (usadv8hi): New define_expand.
8201 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
8204 * ipa-inline-transform.c (update_noncloned_frequencies): Always
8206 (inline_transform): Likewise.
8207 * predict.c (counts_to_freqs): Remove useless conditional.
8208 * profile-count.h (profile_count::apply_scale): Move sanity check.
8209 * tree-inline.c (copy_bb): Always scale.
8210 (copy_cfg_body): Likewise.
8212 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
8215 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
8217 (*cmp_ite0): Add enabled_for_depr_it attribute.
8218 (*cmp_ite1): Likewise.
8220 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8222 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
8225 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8227 * tree-vrp.c (vrp_int_const_binop): Return true on success and
8228 return the value by pointer.
8229 (extract_range_from_multiplicative_op_1): Update accordingly.
8230 Return as soon as an operation fails.
8232 2017-11-05 Tom de Vries <tom@codesourcery.com>
8235 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
8236 (DEF_SANITIZER_BUILTIN): ... here.
8237 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
8238 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
8240 2017-11-05 Tom de Vries <tom@codesourcery.com>
8243 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
8245 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
8246 ASM_OUTPUT_BEFORE_CASE_LABEL call.
8247 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
8249 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8250 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8251 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8253 2017-11-05 Tom de Vries <tom@codesourcery.com>
8256 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
8259 2017-11-04 Michael Clark <michaeljclark@mac.com>
8261 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
8262 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
8264 (*addsi3_extended): Likewise.
8265 (*addsi3_extended2): Likewise.
8266 (<optab>si3): Likewise.
8267 (<optab>di3): Likewise.
8268 (<optab><mode>3): Likewise.
8269 (<*optabe>si3_internal): Likewise.
8270 (zero_extendqi<SUPERQI:mode>2): Likewise.
8271 (*add<mode>hi3): Likewise.
8272 (*xor<mode>hi3): Likewise.
8273 (<optab>di3): Likewise.
8274 (*<optab>si3_extend): Likewise.
8275 (*sge<u>_<X:mode><GPR:mode>): Likewise.
8276 (*slt<u>_<X:mode><GPR:mode>): Likewise.
8277 (*sle<u>_<X:mode><GPR:mode>): Likewise.
8279 2017-11-04 Andrew Waterman <andrew@sifive.com>
8281 * config/riscv/riscv.c (riscv_option_override): Conditionally set
8282 TARGET_STRICT_ALIGN based upon -mtune argument.
8284 2017-11-04 Andrew Waterman <andrew@sifive.com>
8286 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
8288 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
8290 * config/i386/i386.c (choose_basereg): Use optional scratch
8291 register and add assertion.
8292 (x86_emit_outlined_ms2sysv_save): Use scratch register when
8293 needed, and don't allocate stack.
8294 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
8295 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
8296 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
8298 2017-11-03 Jeff Law <law@redhat.com>
8300 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
8301 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
8302 to probe at the start of a noreturn function.
8304 2017-11-03 Jakub Jelinek <jakub@redhat.com>
8306 PR tree-optimization/78821
8307 * gimple-ssa-store-merging.c: Update the file comment.
8308 (MAX_STORE_ALIAS_CHECKS): Define.
8309 (struct store_operand_info): New type.
8310 (store_operand_info::store_operand_info): New constructor.
8311 (struct store_immediate_info): Add rhs_code and ops data members.
8312 (store_immediate_info::store_immediate_info): Add rhscode, op0r
8313 and op1r arguments to the ctor, initialize corresponding data members.
8314 (struct merged_store_group): Add load_align_base and load_align
8316 (merged_store_group::merged_store_group): Initialize them.
8317 (merged_store_group::do_merge): Update them.
8318 (merged_store_group::apply_stores): Pick the constant for
8319 encode_tree_to_bitpos from one of the two operands, or skip
8320 encode_tree_to_bitpos if neither operand is a constant.
8321 (class pass_store_merging): Add process_store method decl. Remove
8322 bool argument from terminate_all_aliasing_chains method decl.
8323 (pass_store_merging::terminate_all_aliasing_chains): Remove
8324 var_offset_p argument and corresponding handling.
8325 (stmts_may_clobber_ref_p): New function.
8326 (compatible_load_p): New function.
8327 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
8328 if there is overlap and rhs_code is not INTEGER_CST. For
8329 non-overlapping stores terminate group if rhs is not mergeable.
8330 (get_alias_type_for_stmts): Change first argument from
8331 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
8332 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
8333 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
8335 (get_location_for_stmts): Change first argument from
8336 auto_vec<gimple *> & to vec<gimple *> &.
8337 (struct split_store): Remove orig_stmts data member, add orig_stores.
8338 (split_store::split_store): Create orig_stores rather than orig_stmts.
8339 (find_constituent_stmts): Renamed to ...
8340 (find_constituent_stores): ... this. Change second argument from
8341 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
8342 to info structures rather than the statements.
8343 (split_group): Rename ALLOW_UNALIGNED argument to
8344 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
8345 it. Adjust find_constituent_stores caller.
8346 (imm_store_chain_info::output_merged_store): Handle rhs_code other
8347 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
8348 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
8349 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
8350 (mem_valid_for_store_merging): New function.
8351 (handled_load): New function.
8352 (pass_store_merging::process_store): New method.
8353 (pass_store_merging::execute): Use process_store method. Adjust
8354 terminate_all_aliasing_chains caller.
8356 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8358 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8359 Return true for more constants, symbols and label references.
8360 (aarch64_valid_floating_const): Remove unused function.
8362 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8365 * config/aarch64/aarch64.c (aarch64_layout_frame):
8366 Undo forcing of LR at bottom of frame.
8368 2017-11-03 Jeff Law <law@redhat.com>
8371 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
8372 for int_registers_saved.
8374 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
8375 extracted from tree-ssa-dom.c.
8376 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
8377 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
8378 (record_equivalences_from_incoming_edge): Add additional argument
8379 to single_pred_edge_ignoring_loop_edges call.
8380 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
8381 (uncprop_dom_walker::before_dom_children): Add additional argument
8382 to single_pred_edge_ignoring_loop_edges call.
8383 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
8384 single_pred_edge_ignoring_loop_edges rather than open coding.
8385 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
8387 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
8389 * match.pd (-(-A)): Rewrite.
8391 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8393 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
8394 (rs6000_emit_int_cmove): New declaration.
8395 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
8396 (rs6000_emit_sISEL): Delete.
8397 (rs6000_emit_int_cmove): Make non-static.
8398 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
8399 instead of rs6000_emit_sISEL.
8401 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
8403 * asan.c (create_cond_insert_point): Maintain profile.
8404 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
8406 * basic-block.h (struct basic_block_def): Remove frequency.
8407 (EDGE_FREQUENCY): Use to_frequency
8408 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
8410 (find_traces): Update to use to_frequency.
8411 (find_traces_1_round): Likewise; use only IPA counts.
8412 (bb_to_key): Likewise.
8413 (connect_traces): Use IPA counts only.
8414 (copy_bb_p): Update to use to_frequency.
8415 (fix_up_crossing_landing_pad): Likewise.
8416 (sanitize_hot_paths): Likewise.
8417 * bt-load.c (basic_block_freq): Likewise.
8418 * cfg.c (init_flow): Set count_max to uninitialized.
8419 (check_bb_profile): Remove frequencies; check counts.
8420 (dump_bb_info): Do not dump frequencies.
8421 (update_bb_profile_for_threading): Update counts only.
8422 (scale_bbs_frequencies_int): Likewise.
8423 (MAX_SAFE_MULTIPLIER): Remove.
8424 (scale_bbs_frequencies_gcov_type): Update counts only.
8425 (scale_bbs_frequencies_profile_count): Update counts only.
8426 (scale_bbs_frequencies): Update counts only.
8427 * cfg.h (struct control_flow_graph): Add count-max.
8428 (update_bb_profile_for_threading): Update prototype.
8429 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
8430 (find_many_sub_basic_blocks): Likewise.
8431 * cfgcleanup.c (try_forward_edges): Likewise.
8432 (try_crossjump_to_edge): Likewise.
8433 * cfgexpand.c (expand_gimple_cond): Likewise.
8434 (expand_gimple_tailcall): Likewise.
8435 (construct_init_block): Likewise.
8436 (construct_exit_block): Likewise.
8437 * cfghooks.c (verify_flow_info): Check consistency of counts.
8438 (dump_bb_for_graph): Do not dump frequencies.
8439 (split_block_1): Do not update frequencies.
8440 (split_edge): Do not update frequencies.
8441 (make_forwarder_block): Do not update frequencies.
8442 (duplicate_block): Do not update frequencies.
8443 (account_profile_record): Do not update frequencies.
8444 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
8445 for global heuristics.
8446 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
8447 (expected_loop_iterations_unbounded): Use counts only.
8448 * cfgloopmanip.c (scale_loop_profile): Simplify.
8449 (create_empty_loop_on_edge): Simplify
8451 (duplicate_loop_to_header_edge): Simplify
8452 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
8453 (update_br_prob_note): Take care of removing note when profile
8455 (relink_block_chain): Do not dump frequency.
8456 (rtl_account_profile_record): Use to_frequency.
8457 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
8458 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
8459 (cgraph_update_edges_for_call_stmt_node): Likewise.
8460 (cgraph_edge::verify_count_and_frequency): Update.
8461 (cgraph_node::verify_node): Temporarily disable frequency verification.
8462 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
8463 to_cgraph_frequency.
8464 (cgraph_edge::rebuild_edges): Convert to ipa counts.
8465 * cgraphunit.c (init_lowered_empty_function): Do not initialize
8467 (cgraph_node::expand_thunk): Update profile.
8468 * except.c (dw2_build_landing_pads): Do not update frequency.
8469 * final.c (compute_alignments): Use to_frequency.
8470 (dump_basic_block_info): Do not dump frequency.
8471 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
8472 (dump_gimple_bb_header): Do not dump frequency.
8473 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
8475 * gimple-streamer-in.c (input_bb): Do not stream frequency.
8476 * gimple-streamer-out.c (output_bb): Do not stream frequency.
8477 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
8478 (init_before_recovery): Do not update frequency.
8479 (sched_create_recovery_edges): Do not update frequency.
8480 * hsa-gen.c (convert_switch_statements): Do not update frequency.
8481 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
8482 (ipa_cp_c_finalize): Set max_count to uninitialized.
8483 * ipa-fnsummary.c (get_minimal_bb): Use counts.
8484 (param_change_prob): Use counts.
8485 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
8487 * ipa-split.c (consider_split): Use to_frequency.
8488 (split_function): Use to_frequency.
8489 * ira-build.c (loop_compare_func): Likewise.
8490 (mark_loops_for_removal): Likewise.
8491 (mark_all_loops_for_removal): Likewise.
8492 * loop-doloop.c (doloop_modify): Do not update frequency.
8493 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
8495 * lto-streamer-in.c (input_function): Update count_max.
8496 * omp-expand.c (expand_omp_taskreg): Update count_max.
8497 * omp-simd-clone.c (simd_clone_adjust): Update profile.
8498 * predict.c (maybe_hot_frequency_p): Use to_frequency.
8499 (maybe_hot_count_p): Use ipa counts only.
8500 (maybe_hot_bb_p): Simplify.
8501 (maybe_hot_edge_p): Simplify.
8502 (probably_never_executed): Do not take frequency argument.
8503 (probably_never_executed_bb_p): Do not pass frequency.
8504 (probably_never_executed_edge_p): Likewise.
8505 (combine_predictions_for_bb): Check that profile is nonzero.
8506 (propagate_freq): Do not set frequency.
8507 (drop_profile): Simplify.
8508 (counts_to_freqs): Simplify.
8509 (expensive_function_p): Use to_frequency.
8510 (propagate_unlikely_bbs_forward): Simplify.
8511 (determine_unlikely_bbs): Simplify.
8512 (estimate_bb_frequencies): Add hack to silence graphite issues.
8513 (compute_function_frequency): Use ipa counts.
8514 (pass_profile::execute): Update.
8515 (rebuild_frequencies): Use counts only.
8516 (force_edge_cold): Use counts only.
8517 * profile-count.c (profile_count::dump): Dump new count types.
8518 (profile_count::differs_from_p): Check compatiblity.
8519 (profile_count::to_frequency): New function.
8520 (profile_count::to_cgraph_frequency): New function.
8521 * profile-count.h (struct function): Declare.
8522 (enum profile_quality): Add profile_guessed_local and
8523 profile_guessed_global0.
8524 (class profile_proability): Decrease number of bits to 29;
8525 update from_reg_br_prob_note and to_reg_br_prob_note.
8526 (class profile_count: Update comment; decrease number of bits
8527 to 61. Check compatibility.
8528 (profile_count::compatible_p): New private member function.
8529 (profile_count::ipa_p): New member function.
8530 (profile_count::operator<): Handle global zero correctly.
8531 (profile_count::operator>): Handle global zero correctly.
8532 (profile_count::operator<=): Handle global zero correctly.
8533 (profile_count::operator>=): Handle global zero correctly.
8534 (profile_count::nonzero_p): New member function.
8535 (profile_count::force_nonzero): New member function.
8536 (profile_count::max): New member function.
8537 (profile_count::apply_scale): Handle IPA scalling.
8538 (profile_count::guessed_local): New member function.
8539 (profile_count::global0): New member function.
8540 (profile_count::ipa): New member function.
8541 (profile_count::to_frequency): Declare.
8542 (profile_count::to_cgraph_frequency): Declare.
8543 * profile.c (OVERLAP_BASE): Delete.
8544 (compute_frequency_overlap): Delete.
8545 (compute_branch_probabilities): Do not use compute_frequency_overlap.
8546 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
8547 * sched-ebb.c (rank): Use counts only.
8548 * shrink-wrap.c (handle_simple_exit): Use counts only.
8549 (try_shrink_wrapping): Use counts only.
8550 (place_prologue_for_one_component): Use counts only.
8551 * tracer.c (find_best_predecessor): Use to_frequency.
8552 (find_trace): Use to_frequency.
8553 (tail_duplicate): Use to_frequency.
8554 * trans-mem.c (expand_transaction): Do not update frequency.
8555 * tree-call-cdce.c: Do not update frequency.
8556 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
8557 (gimple_merge_blocks): Likewise.
8558 (gimple_split_edge): Likewise.
8559 (gimple_duplicate_sese_region): Likewise.
8560 (gimple_duplicate_sese_tail): Likewise.
8561 (move_sese_region_to_fn): Likewise.
8562 (gimple_account_profile_record): Likewise.
8563 (insert_cond_bb): Likewise.
8564 * tree-complex.c (expand_complex_div_wide): Likewise.
8565 * tree-eh.c (lower_resx): Update profile.
8566 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
8568 (initialize_cfun): Do not initialize frequencies
8569 (freqs_to_counts): Delete.
8570 (copy_cfg_body): Ignore count parameter.
8571 (copy_body): Update.
8572 (expand_call_inline): Update count_max.
8573 (optimize_inline_calls): Update count_max.
8574 (tree_function_versioning): Update count_max.
8575 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
8576 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
8578 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
8579 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
8580 (try_peel_loop): Likewise.
8581 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
8583 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
8584 (tree_transform_and_unroll_loop): Do not use frequencies
8585 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
8586 Use reliable prediction only.
8587 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
8588 * tree-ssa-sink.c (select_best_block): Use to_frequency.
8589 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
8590 probability scaling.
8591 * tree-ssa-threadupdate.c (create_block_for_threading): Do
8592 not update frequency
8593 (any_remaining_duplicated_blocks): Likewise.
8594 (update_profile): Likewise.
8595 (estimated_freqs_path): Delete.
8596 (freqs_to_counts_path): Delete.
8597 (clear_counts_path): Delete.
8598 (ssa_fix_duplicate_block_edges): Likewise.
8599 (duplicate_thread_path): Likewise.
8600 * tree-switch-conversion.c (gen_inbound_check): Use counts.
8601 * tree-tailcall.c (decrease_profile): Do not update frequency.
8602 (eliminate_tail_call): Likewise.
8603 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
8604 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
8605 (optimize_mask_stores): Likewise.
8606 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8607 * ubsan.c (ubsan_expand_null_ifn): Update profile.
8608 (ubsan_expand_ptr_ifn): Update profile.
8609 * value-prof.c (gimple_ic): Simplify.
8610 * value-prof.h (gimple_ic): Update prototype.
8611 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
8612 * ipa-inline.c (compute_uninlined_call_time): Be sure that
8614 (want_inline_self_recursive_call_p): Likewise.
8615 (resolve_noninline_speculation): Only cummulate defined counts.
8616 (inline_small_functions): Use nonzero_p.
8617 (ipa_inline): Do not access freed node.
8619 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8621 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8622 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8624 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8626 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8627 non-legitimate address.
8629 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8631 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8632 (*lt0_<mode>di, *lt0_<mode>si): New.
8634 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8636 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8637 TARGET_PAIRED_FLOAT.
8639 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8640 Jim Wilson <jim.wilson@linaro.org>
8642 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8643 * config/aarch64/aarch64-tune.md: Regenerated.
8644 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8645 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8647 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8649 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8650 (arc_expand_prologue): Restore blink for millicode.
8651 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8653 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8656 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8657 gen_vec_duplicate after forcing the scalar into a register.
8659 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8661 * combine (try_combine): Print the insns input to try_combine to the
8664 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8667 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8668 Remove second argument from aarch64_process_target_attr call.
8669 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8671 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8673 (aarch64_handle_attr_arch): Remove second argument.
8674 (aarch64_handle_attr_cpu): Ditto.
8675 (aarch64_handle_attr_tune): Ditto.
8676 (aarch64_handle_attr_isa_flags): Ditto.
8677 (aarch64_process_one_target_attr): Ditto.
8678 (aarch64_process_target_attr): Ditto.
8679 (aarch64_option_valid_attribute_p): Remove second argument.
8680 on aarch64_process_target_attr call.
8682 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8684 * diagnostic.c: Include "selftest-diagnostic.h".
8685 (selftest::assert_location_text): New function.
8686 (selftest::test_diagnostic_get_location_text): New function.
8687 (selftest::diagnostic_c_tests): Call it.
8689 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8691 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8692 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8693 (class selftest::test_diagnostic_context): Move to...
8694 * selftest-diagnostic.c: New file.
8695 * selftest-diagnostic.h: New file.
8697 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8699 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8700 offset range for FT32B.
8701 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8702 * config/ft32/ft32.md: Add TARGET_NOPM.
8703 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8704 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8706 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8708 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8710 2017-11-02 Jeff Law <law@redhat.com>
8712 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8713 virtual keyword on FINAL OVERRIDE members.
8715 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8716 virtuals together. Add virtual destructor.
8717 (substitute_and_fold_engine): Similarly.
8719 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8721 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8723 2017-11-02 Richard Biener <rguenther@suse.de>
8725 PR tree-optimization/82795
8726 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8728 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8730 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8732 (gcc_GAS_CHECK_FEATURE): Remove.
8733 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8734 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8735 * configure: Regenerate.
8737 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8739 * config/arc/arc.c (hwloop_optimize): Account for empty
8742 2017-11-02 Richard Biener <rguenther@suse.de>
8745 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8746 Truncate ARRAY_REF index and element size.
8748 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8750 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8752 2017-11-01 Jeff Law <law@redhat.com>
8754 * tree-ssa-ccp.c (ccp_folder): New class derived from
8755 substitute_and_fold_engine.
8756 (ccp_folder::get_value): New member function.
8757 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8758 (ccp_fold_stmt): Remove prototype.
8759 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8760 * tree-ssa-copy.c (copy_folder): New class derived from
8761 substitute_and_fold_engine.
8762 (copy_folder::get_value): Renamed from get_value.
8763 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8764 * tree-vrp.c (vrp_folder): New class derived from
8765 substitute_and_fold_engine.
8766 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8767 (vrp_folder::get_value): New member function.
8768 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8769 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8770 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8771 provide a class interface to folder/substitute routines.
8772 (ssa_prop_fold_stmt_fn): Remove typedef.
8773 (ssa_prop_get_value_fn): Likewise.
8774 (subsitute_and_fold): Remove prototype.
8775 (replace_uses_in): Likewise.
8776 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8777 Renamed from replace_uses_in. Call the virtual member function
8778 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8779 (substitute_and_fold_dom_walker): Remove initialization of
8780 data member entries for calbacks. Add substitute_and_fold_engine
8781 member and initialize it.
8782 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8783 member functions for get_value, replace_phi_args_in c
8784 replace_uses_in, and fold_stmt calls.
8785 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8786 substitute_and_fold. Remove assert. Update ctor call.
8788 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8789 (ssa_prop_visit_phi_fn): Likewise.
8790 (class ssa_propagation_engine): New class to provide an interface
8792 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8794 (ssa_prop_visit_phi): Likewise.
8795 (ssa_propagation_engine::simulate_stmt): Moved into class.
8796 Call visit_phi/visit_stmt from the class rather than via
8797 file scoped static variables.
8798 (ssa_propagation_engine::simulate_block): Moved into class.
8799 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8800 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8801 set file scoped statics for the visit_stmt/visit_phi callbacks.
8802 * tree-complex.c (complex_propagate): New class derived from
8803 ssa_propagation_engine.
8804 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8805 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8806 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8808 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8809 ssa_propagation_engine.
8810 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8811 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8812 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8813 * tree-ssa-copy.c (copy_prop): New class derived from
8814 ssa_propagation_engine.
8815 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8816 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8817 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8818 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8819 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8820 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8821 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8823 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8825 PR rtl-optimization/82778
8826 PR rtl-optimization/82597
8827 * compare-elim.c (struct comparison): Add in_a_setter field.
8828 (find_comparison_dom_walker::before_dom_children): Remove killed
8829 bitmap and df_simulate_find_defs call, instead walk the defs.
8830 Compute last_setter and initialize in_a_setter. Merge definitions
8831 with first initialization for a few variables.
8832 (try_validate_parallel): Use insn_invalid_p instead of
8833 recog_memoized. Return insn rather than just the pattern.
8834 (try_merge_compare): Fix up comment. Don't uselessly test if
8835 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8837 (execute_compare_elim_after_reload): Remove df_chain_add_problem
8840 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8841 Alan Hayward <alan.hayward@arm.com>
8842 David Sherwood <david.sherwood@arm.com>
8844 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
8845 aarch64_hard_regno_nregs to get the number of registers
8848 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8849 Alan Hayward <alan.hayward@arm.com>
8850 David Sherwood <david.sherwood@arm.com>
8852 * config/aarch64/constraints.md (Upl): Rename to...
8854 * config/aarch64/aarch64.md
8855 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
8858 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8859 Alan Hayward <alan.hayward@arm.com>
8860 David Sherwood <david.sherwood@arm.com>
8862 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
8863 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8866 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8867 Alan Hayward <alan.hayward@arm.com>
8868 David Sherwood <david.sherwood@arm.com>
8870 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
8871 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
8872 (aarch64_evpc_dup): Generate rtl direcly, rather than using
8874 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
8875 of a single element.
8876 * config/aarch64/iterators.md: Add a comment above the permute
8877 unspecs to say that they are generated directly by
8878 aarch64_expand_vec_perm_const.
8879 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
8881 2017-11-01 Nathan Sidwell <nathan@acm.org>
8883 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
8885 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8887 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
8888 and medany code models, and describe what they do.
8890 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8892 Revert accidental duplicate:
8894 * combine.c (can_change_dest_mode): Reject changes in
8895 REGMODE_NATURAL_SIZE.
8897 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
8899 PR rtl-optimization/64682
8900 PR rtl-optimization/69567
8901 PR rtl-optimization/69737
8902 PR rtl-optimization/82683
8903 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
8904 register mentioned in the note, drop the note, unless it came from I3,
8905 in which case it should go to I3 again.
8907 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8909 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
8910 and return false if not.
8911 (clear_bytes_written_by, live_bytes_read): Update accordingly.
8913 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8915 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
8916 range is known to be empty.
8918 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8919 Alan Hayward <alan.hayward@arm.com>
8920 David Sherwood <david.sherwood@arm.com>
8922 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
8923 and CONST_VECTOR_NUNITS instead of computing the number of units from
8924 the byte sizes of the vector and element.
8925 (simplify_binary_operation_1): Likewise.
8926 (simplify_const_binary_operation): Likewise.
8927 (simplify_ternary_operation): Likewise.
8929 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8930 Alan Hayward <alan.hayward@arm.com>
8931 David Sherwood <david.sherwood@arm.com>
8933 * var-tracking.c (INT_MEM_OFFSET): Replace with...
8934 (int_mem_offset): ...this new function.
8935 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
8936 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
8937 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
8940 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8941 Alan Hayward <alan.hayward@arm.com>
8942 David Sherwood <david.sherwood@arm.com>
8944 * lower-subreg.c (interesting_mode_p): New function.
8945 (compute_costs, find_decomposable_subregs, decompose_register)
8946 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
8947 (resolve_clobber, dump_choices): Use it.
8949 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8950 Alan Hayward <alan.hayward@arm.com>
8951 David Sherwood <david.sherwood@arm.com>
8953 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
8955 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8956 Alan Hayward <alan.hayward@arm.com>
8957 David Sherwood <david.sherwood@arm.com>
8959 * alias.c (find_base_value, find_base_term): Only process integer
8960 truncations. Check the precision rather than the size.
8962 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8963 Alan Hayward <alan.hayward@arm.com>
8964 David Sherwood <david.sherwood@arm.com>
8966 * machmode.h (is_narrower_int_mode): New function
8967 * optabs.c (expand_float, expand_fix): Use it.
8968 * dwarf2out.c (rotate_loc_descriptor): Likewise.
8970 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8971 Alan Hayward <alan.hayward@arm.com>
8972 David Sherwood <david.sherwood@arm.com>
8974 * rtl.h (narrower_subreg_mode): New function.
8975 * ira-color.c (update_costs_from_allocno): Use it.
8977 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8978 Alan Hayward <alan.hayward@arm.com>
8979 David Sherwood <david.sherwood@arm.com>
8981 * optabs-query.h (convert_optab_p): New function, split out from...
8982 (convert_optab_handler): ...here.
8983 (widening_optab_handler): Delete.
8984 (find_widening_optab_handler): Remove permit_non_widening parameter.
8985 (find_widening_optab_handler_and_mode): Likewise. Provide an
8986 override that operates on mode class wrappers.
8987 * optabs-query.c (widening_optab_handler): Delete.
8988 (find_widening_optab_handler_and_mode): Remove permit_non_widening
8989 parameter. Assert that the two modes are the same class and that
8990 the "from" mode is narrower than the "to" mode. Use
8991 convert_optab_handler instead of widening_optab_handler.
8992 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
8993 instead of widening_optab_handler.
8994 * expr.c (expand_expr_real_2): Update calls to
8995 find_widening_optab_handler.
8996 * optabs.c (expand_widen_pattern_expr): Likewise.
8997 (expand_binop_directly): Take the insn_code as a parameter.
8998 (expand_binop): Only call find_widening_optab_handler for
8999 conversion optabs; use optab_handler otherwise. Update calls
9000 to find_widening_optab_handler and expand_binop_directly.
9001 Use convert_optab_handler instead of widening_optab_handler.
9002 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
9003 find_widening_optab_handler and use scalar_mode rather than
9005 (convert_plusminus_to_widen): Likewise.
9007 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9008 Alan Hayward <alan.hayward@arm.com>
9009 David Sherwood <david.sherwood@arm.com>
9011 * machmode.h (fixed_size_mode): New class.
9012 * rtl.h (get_pool_mode): Return fixed_size_mode.
9013 * gengtype.c (main): Add fixed_size_mode.
9014 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9015 (get_raw_arg_mode): Likewise.
9016 * doc/tm.texi: Regenerate.
9017 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9018 * targhooks.c (default_get_reg_raw_mode): Likewise.
9019 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9020 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9021 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9022 (msp430_get_raw_result_mode): Likewise.
9023 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9024 * dbxout.c (dbxout_parms): Require fixed-size modes.
9025 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9026 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9027 * omp-low.c (lower_oacc_reductions): Likewise.
9028 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9029 (simplify_subreg): Update accordingly.
9030 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9031 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9032 that aren't fixed-size.
9033 (get_pool_mode): Return a fixed_size_mode.
9034 (output_constant_pool_2): Take a fixed_size_mode.
9036 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9037 Alan Hayward <alan.hayward@arm.com>
9038 David Sherwood <david.sherwood@arm.com>
9040 * doc/rtl.texi (vec_series): Document.
9041 (const): Say that the operand can be a vec_series.
9042 * rtl.def (VEC_SERIES): New rtx code.
9043 * rtl.h (const_vec_series_p_1): Declare.
9044 (const_vec_series_p): New function.
9045 * emit-rtl.h (gen_const_vec_series): Declare.
9046 (gen_vec_series): Likewise.
9047 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9048 (gen_vec_series): Likewise.
9049 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9050 * simplify-rtx.c (simplify_unary_operation): Handle negations
9052 (simplify_binary_operation_series): New function.
9053 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9054 (test_vector_ops_series): New function.
9055 (test_vector_ops): Call it.
9056 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9057 gen_const_vec_series.
9058 (altivec_lvsr): Likewise.
9059 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9061 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9062 Alan Hayward <alan.hayward@arm.com>
9063 David Sherwood <david.sherwood@arm.com>
9065 * doc/rtl.texi (const): Update description of address constants.
9066 Say that vector constants are allowed too.
9067 * common.md (E, F): Use CONSTANT_P instead of checking for
9069 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9070 checking for CONST_VECTOR.
9071 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9073 * expr.c (expand_expr_real_2): Check for vector modes instead
9074 of checking for CONST_VECTOR.
9075 * rtl.h (const_vec_p): New function.
9076 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9077 (unwrap_const_vec_duplicate): Handle them here too.
9079 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9080 David Malcolm <dmalcolm@redhat.com>
9081 Alan Hayward <alan.hayward@arm.com>
9082 David Sherwood <david.sherwood@arm.com>
9084 * rtl.h (vec_duplicate_p): New function.
9085 * selftest-rtl.c (assert_rtx_eq_at): New function.
9086 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9087 (assert_rtx_eq_at): Declare.
9088 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9089 * selftest-run-tests.c (selftest::run_tests): Call it.
9090 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9091 (simplify_unary_operation_1): Recursively handle vector duplicates.
9092 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9094 (simplify_subreg): Handle subregs of vector duplicates.
9095 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9096 (selftest::simplify_rtx_c_tests): New functions.
9098 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9099 Alan Hayward <alan.hayward@arm.com>
9100 David Sherwood <david.sherwood@arm.com>
9102 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9103 (gen_vec_duplicate): Likewise.
9104 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9106 (gen_const_vector): ...here.
9107 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9108 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9109 whose elements are all equal.
9110 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
9111 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
9112 (simplify_relational_operation): Likewise.
9113 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
9115 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
9116 (aarch64_expand_vector_init): Likewise.
9117 * config/arm/arm.c (neon_vdup_constant): Likewise.
9118 (neon_expand_vector_init): Likewise.
9119 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
9120 (arm_block_set_unaligned_vect): Likewise.
9121 (arm_block_set_aligned_vect): Likewise.
9122 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
9123 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
9124 (expand_vec_perm_even_odd_pack): Likewise.
9125 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
9126 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
9127 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
9128 gen_const_vec_duplicate.
9129 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
9130 * config/mips/mips.c (mips_gen_const_int_vector): Use
9131 gen_const_vec_duplicate.
9132 (mips_expand_vector_init): Use CONST0_RTX.
9133 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
9134 (define_split): Use gen_const_vec_duplicate.
9135 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
9136 (define_split): Use gen_const_vec_duplicate.
9137 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
9138 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
9139 * config/spu/spu.c (spu_const): Likewise.
9141 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9142 Alan Hayward <alan.hayward@arm.com>
9143 David Sherwood <david.sherwood@arm.com>
9145 * combine.c (can_change_dest_mode): Reject changes in
9146 REGMODE_NATURAL_SIZE.
9148 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
9150 * configure.ac (--enable-libssp): New.
9151 (gcc_cv_libc_provides_ssp): Check for explicit setting before
9152 trying to determine target-specific default. Adjust indentation.
9153 * configure: Regenerated.
9154 * doc/install.texi (Configuration): Expand --disable-libssp
9157 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
9159 config/i386/i386.c (ix86_expand_epilogue): Correct stack
9162 2017-10-31 Martin Jambor <mjambor@suse.cz>
9165 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
9167 2017-10-31 David Malcolm <dmalcolm@redhat.com>
9169 * auto-profile.c (autofdo_source_profile::read): Use
9170 UNKNOWN_LOCATION rather than 0.
9171 * diagnostic-core.h (warning_at_rich_loc): Rename to...
9172 (warning_at): ...this overload.
9173 (warning_at_rich_loc_n): Rename to...
9174 (warning_n): ...this overload.
9175 (error_at_rich_loc): Rename to...
9176 (error_at): ...this overload.
9177 (pedwarn_at_rich_loc): Rename to...
9178 (pedwarn): ...this overload.
9179 (permerror_at_rich_loc): Rename to...
9180 (permerror): ...this overload.
9181 (inform_at_rich_loc): Rename to...
9182 (inform): ...this overload.
9183 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
9184 (diagnostic_n_impl_richloc): Rename to...
9185 (diagnostic_n_impl): ...this rich_location *-based decl.
9186 (inform_at_rich_loc): Rename to...
9187 (inform): ...this, and add an assertion.
9188 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
9189 (warning_at_rich_loc): Rename to...
9190 (warning_at): ...this, and add an assertion.
9191 (warning_at_rich_loc_n): Rename to...
9192 (warning_n): ...this, and add an assertion.
9193 (warning_n): Update location_t-based implementation for removal of
9194 location_t-based diagnostic_n_impl.
9195 (pedwarn_at_rich_loc): Rename to...
9196 (pedwarn): ...this, and add an assertion.
9197 (permerror_at_rich_loc): Rename to...
9198 (permerror): ...this, and add an assertion.
9199 (error_n): Update for removal of location_t-based diagnostic_n_impl.
9200 (error_at_rich_loc): Rename to...
9201 (error_at): ...this, and add an assertion.
9202 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
9203 (driver::do_spec_on_infiles): Likewise.
9204 * substring-locations.c (format_warning_va): Update for renaming
9205 of inform_at_rich_loc.
9207 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
9209 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
9210 _Float<N>X built-in functions so that the variant without the
9211 "__builtin_" prefix is only enabled for the GNU C and Objective C
9212 languages when they are in non-strict ANSI/ISO mode.
9213 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
9214 * target.def (floatn_builtin_p): Add a target hook to control
9215 whether _Float<N> and _Float<N>X built-in functions without the
9216 "__builtin_" prefix are enabled, and return true for C and
9217 Objective C in the default hook. Include langhooks.h in
9219 * targhooks.h (default_floatn_builtin_p): Likewise.
9220 * targhooks.c (default_floatn_builtin_p): Likewise.
9221 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
9222 floatn_builtin_p target hook.
9223 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
9225 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
9226 Eric Botcazou <ebotcazou@adacore.com>
9228 PR rtl-optimization/81803
9229 * lra-constraints.c (curr_insn_transform): Also reload the whole
9230 register for a strict subreg no wider than a word if this is for
9231 a WORD_REGISTER_OPERATIONS target.
9233 2017-10-31 Jason Merrill <jason@redhat.com>
9235 * gdbinit.in: Skip over inlines from timevar.h.
9237 2017-10-31 Martin Liska <mliska@suse.cz>
9239 * doc/gcov.texi: Document new option.
9240 * gcov.c (print_usage): Likewise print it.
9241 (process_args): Support the argument.
9242 (format_count): New function.
9243 (format_gcov): Use the function.
9245 2017-10-31 Martin Liska <mliska@suse.cz>
9247 * gcov.c (struct name_map): do not use typedef.
9248 Define operator== and operator<.
9249 (name_search): Remove.
9250 (name_sort): Remove.
9251 (main): Do not allocate names.
9252 (process_file): Add vertical space.
9253 (generate_results): Use std::find.
9254 (release_structures): Do not release memory.
9255 (find_source): Use std::find.
9257 2017-10-31 Martin Liska <mliska@suse.cz>
9259 * gcov.c (struct line_info): Remove it's typedef.
9260 (line_info::line_info): Add proper ctor.
9261 (line_info::has_block): Do not use a typedef.
9262 (struct source_info): Do not use typedef.
9263 (circuit): Likewise.
9264 (get_cycles_count): Likewise.
9265 (output_intermediate_file): Iterate via vector iterator.
9266 (add_line_counts): Use std::vector methods.
9267 (accumulate_line_counts): Likewise.
9268 (output_lines): Likewise.
9270 2017-10-31 Martin Liska <mliska@suse.cz>
9272 * gcov.c (struct source_info): Remove typedef.
9273 (source_info::source_info): Add proper ctor.
9274 (accumulate_line_counts): Use struct, not it's typedef.
9275 (output_gcov_file): Likewise.
9276 (output_lines): Likewise.
9277 (main): Do not allocate an array.
9278 (output_intermediate_file): Use size of vector container.
9279 (process_file): Resize the vector.
9280 (generate_results): Do not preallocate, use newly added vector
9282 (release_structures): Do not release sources.
9283 (find_source): Use vector methods.
9284 (add_line_counts): Do not use typedef.
9286 2017-10-31 Martin Liska <mliska@suse.cz>
9288 * doc/gcov.texi: Document that.
9289 * gcov.c (add_line_counts): Mark lines with a non-executed
9291 (output_line_beginning): Handle such lines.
9292 (output_lines): Pass new argument.
9293 (output_intermediate_file): Print it in intermediate format.
9295 2017-10-31 Martin Liska <mliska@suse.cz>
9297 * color-macros.h: New file.
9298 * diagnostic-color.c: Factor out color related to macros to
9300 * doc/gcov.texi: Document -k option.
9301 * gcov.c (INCLUDE_STRING): Include string.h.
9302 (print_usage): Add -k option.
9303 (process_args): Parse it.
9304 (pad_count_string): New function.
9305 (output_line_beginning): Likewise.
9306 (DEFAULT_LINE_START): New macro.
9307 (output_lines): Support color output.
9309 2017-10-31 Martin Liska <mliska@suse.cz>
9311 PR gcov-profile/82633
9312 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
9313 their interaction with GCOV infrastructure.
9314 * configure.ac: Add -fkeep-{inline,static}-functions to
9316 * configure: Regenerate.
9318 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
9321 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
9323 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
9326 * config/rs6000/rs6000.md (allocate_stack): Force update interval
9327 into a register if it does not fit into an immediate offset field.
9329 2017-10-31 Olivier Hainque <hainque@adacore.com>
9331 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
9333 2017-10-31 Julia Koval <julia.koval@intel.com>
9335 * config.gcc: Add gfniintrin.h.
9336 * config/i386/gfniintrin.h: New.
9337 * config/i386/i386-builtin-types.def
9338 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
9339 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
9340 __builtin_ia32_vgf2p8affineinvqb_v32qi,
9341 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
9342 __builtin_ia32_vgf2p8affineinvqb_v16qi,
9343 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
9344 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
9345 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
9346 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
9347 V64QI_FTYPE_V64QI_V64QI_INT): New types.
9348 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
9349 * config/i386/immintrin.h: Include gfniintrin.h.
9350 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
9352 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
9354 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
9356 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9358 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
9359 (arm_ashldi3_1bit): Remove pattern.
9360 (ashrdi3): Remove shift by 1 expansion.
9361 (arm_ashrdi3_1bit): Remove pattern.
9362 (lshrdi3): Remove shift by 1 expansion.
9363 (arm_lshrdi3_1bit): Remove pattern.
9364 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
9365 cost of ashldi3 by 1.
9366 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
9367 (<shift>di3_neon): Likewise.
9369 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9371 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
9372 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
9373 and (*aarch64_simd_mov<VQ:mode>).
9374 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
9375 pattern alternative.
9376 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
9377 attributes to match pattern alternative.
9379 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
9381 * config.gcc (powerpc*-*-*): Add emmintrin.h.
9382 * config/rs6000/emmintrin.h: New file.
9383 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
9385 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9387 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
9388 * (movdi_vfp_cortexa8): Remove pattern.
9390 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9392 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
9394 (Specific, alpha*-dec-osf5.1): Remove.
9395 (Specific, mips-sgi-irix5): Remove.
9396 (Specific, mips-sgi-irix6): Remove.
9398 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9401 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
9402 arguments to clear_bit_region_be.
9404 2017-10-30 Jim Wilson <wilson@tuliptree.org>
9406 * gimplify.c: Include memmodel.h.
9408 2017-10-30 Martin Jambor <mjambor@suse.cz>
9410 * omp-grid.c (grid_attempt_target_gridification): Also insert a
9411 condition whether loop should be executed at all.
9413 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
9415 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9416 gimple folding of vec_madd() intrinsics.
9417 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
9418 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
9419 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
9421 2017-10-30 Richard Biener <rguenther@suse.de>
9423 PR tree-optimization/82762
9425 2017-10-23 Richard Biener <rguenther@suse.de>
9427 PR tree-optimization/82129
9429 2017-08-01 Richard Biener <rguenther@suse.de>
9431 PR tree-optimization/81181
9432 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9433 (compute_antic): ... end of iteration here.
9435 2017-10-30 Joseph Myers <joseph@codesourcery.com>
9437 * doc/invoke.texi (C Dialect Options): Document -std=c17,
9438 -std=iso9899:2017 and -std=gnu17.
9439 * doc/standards.texi (C Language): Document C17 support.
9440 * doc/cpp.texi (Overview): Mention -std=c17.
9441 (Standard Predefined Macros): Document C11 and C17 values of
9442 __STDC_VERSION__. Do not refer to C99 support as incomplete.
9443 * doc/extend.texi (Inline): Do not list individual options for
9444 standards newer than C99.
9445 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
9447 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
9450 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
9452 * asan.c (asan_finish_file): Align asan globals array by shadow
9455 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9458 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
9459 (struct store_immediate_info): Add bitregion_start and bitregion_end
9461 (store_immediate_info::store_immediate_info): Add brs and bre
9462 arguments and initialize bitregion_{start,end} from those.
9463 (struct merged_store_group): Add bitregion_start, bitregion_end,
9464 align_base and mask fields. Drop unnecessary struct keyword from
9465 struct store_immediate_info. Add do_merge method.
9466 (clear_bit_region_be): Use memset instead of loop storing zeros.
9467 (merged_store_group::do_merge): New method.
9468 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
9469 stores as long as the surrounding bitregions have no gaps.
9470 (merged_store_group::merge_overlapping): Use do_merge.
9471 (merged_store_group::apply_stores): Test that bitregion_{start,end}
9472 is byte aligned, rather than requiring that start and width are
9473 byte aligned. Drop unnecessary struct keyword from
9474 struct store_immediate_info. Allocate and populate also mask array.
9475 Make start of the arrays relative to bitregion_start rather than
9476 start and size them according to bitregion_{end,start} difference.
9477 (struct imm_store_chain_info): Drop unnecessary struct keyword from
9478 struct store_immediate_info.
9479 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
9480 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
9481 struct keyword from struct store_immediate_info.
9482 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
9483 between stores as long as the surrounding bitregions have no gaps.
9485 (struct split_store): Add orig non-static data member.
9486 (split_store::split_store): Initialize orig to false.
9487 (find_constituent_stmts): Return store_immediate_info *, non-NULL
9488 if there is exactly a single original stmt. Change stmts argument
9489 to pointer from reference, if NULL, don't push anything to it. Add
9490 first argument, use it to optimize skipping over orig stmts that
9491 are known to be before bitpos already. Simplify.
9492 (split_group): Return unsigned int count how many stores are or
9493 would be needed rather than a bool. Add allow_unaligned argument.
9494 Change split_stores argument from reference to pointer, if NULL,
9495 only do a dry run computing how many stores would be produced.
9496 Rewritten algorithm to use both alignment and misalign if
9497 !allow_unaligned and handle bitfield stores with gaps.
9498 (imm_store_chain_info::output_merged_store): Set start_byte_pos
9499 from bitregion_start instead of start. Compute allow_unaligned
9500 here, if true, do 2 split_group dry runs to compute which one
9501 produces fewer stores and prefer aligned if equal. Punt if
9502 new count is bigger or equal than original before emitting any
9503 statements, rather than during that. Remove no longer needed
9504 new_ssa_names tracking. Replace num_stmts with
9505 split_stores.length (). Use 32-bit stack allocated entries
9506 in split_stores auto_vec. Try to reuse original store lhs/rhs1
9507 if possible. Handle bitfields with gaps.
9508 (pass_store_merging::execute): Ignore bitsize == 0 stores.
9509 Compute bitregion_{start,end} for the stores and construct
9510 store_immediate_info with that. Formatting fixes.
9512 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
9515 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
9516 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
9518 2017-10-29 Jim Wilson <wilson@tuliptree.org>
9520 * gimplify.c: Include tm_p.h.
9522 * common.opt (gcoff): Re-add as ignored option.
9523 (gcoff1, gcoff2, gcoff3): Likewise.
9525 * Makefile.in (OBJS): Delete sdbout.o.
9526 (GTFILES): Delete $(srcdir)/sdbout.c.
9527 * debug.h: Delete sdb_debug_hooks.
9528 * final.c: Delete sdbout.h include.
9529 (final_scan_insn): Delete SDB_DEBUG check.
9530 (rest_of_clean_state): Likewise.
9531 * output.h: Delete sdb_begin_function_line.
9534 * toplev.c: Delete sdbout.h include.
9535 (process_options): Delete SDB_DEBUG check.
9536 * tree-core.h (tree_type_common): Delete pointer field of
9538 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
9539 TYPE_SYMTAB_POINTER.
9540 * tree.h (TYPE_SYMTAB_POINTER): Delete.
9541 (TYPE_SYMTAB_IS_POINTER): Delete.
9542 (TYPE_SYMTAB_IS_DIE): Renumber.
9543 * xcoffout.c: Refer to former sdbout.c file.
9544 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
9546 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
9547 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
9548 (Debugging Options): Delete -gcoff.
9549 (-gstabs): Delete SDB reference.
9551 (-gcoff@var{level}): Delete.
9552 * doc/passes.texi (Debugging information output): Delete SDB and
9553 sdbout.c references.
9554 * doc/tm.texi: Regenerate.
9555 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
9556 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
9558 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
9559 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
9560 (SDB_DEBUGGING_INFO): Delete.
9561 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
9562 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
9563 * target.def (output_source_filename): Delete COFF reference.
9565 * common.opt (gcoff): Delete.
9566 (gxcoff+): Update Negative chain.
9567 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
9569 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
9570 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
9571 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
9573 (expand_function_start): Change sdb reference to past tense.
9574 (expand_function_end): Change sdb reference to past tense.
9575 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
9576 * opts.c (debug_type_names): Delete coff entry.
9577 (common_handle_option): Delete OPT_gcoff case.
9578 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
9580 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
9581 * config/cris/cris.h: Delete SDB reference in comment.
9582 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
9583 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
9584 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
9585 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
9587 (ix86_expand_prologue): Likewise.
9588 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
9589 * config/ia64/ia64.h: Likewise.
9590 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
9591 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
9593 * config/mmix/mmix.h: Likewise.
9594 * config/nds32/nds32.c: Likewise.
9595 * config/stormy/storym16.h: Likewise.
9596 * config/visium/visium.h: Likewise.
9597 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
9599 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
9601 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
9602 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
9603 FRAME_POINTER_REGNUM point at high end of local var area.
9605 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
9607 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
9608 Move comment around. Do not reset best_edge for a copiable
9609 destination if the copy would cause a partition change.
9610 (better_edge_p): Remove redundant check.
9612 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9614 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
9616 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9618 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
9619 for math functions that have _Float<N> and _Float<N>X variants.
9620 (mathfn_built_in_2): Add support for math functions that have
9621 _Float<N> and _Float<N>X variants.
9622 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9623 (expand_builtin_mathfn_ternary): Add support for fma with
9624 _Float<N> and _Float<N>X variants.
9625 (expand_builtin): Likewise.
9626 (fold_builtin_3): Likewise.
9627 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9628 create math function _Float<N> and _Float<N>X variants as external
9630 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9631 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9632 the __builtin_ prefix and if not strict ansi, without the prefix.
9633 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9634 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9635 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9636 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9637 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9638 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9639 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9640 function signatures for fma _Float<N> and _Float<N>X variants.
9641 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9642 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9643 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9644 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9645 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9646 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9647 * gencfn-macros.c (print_case_cfn): Add support for math functions
9648 that have _Float<N> and _Float<N>X variants.
9649 (print_define_operator_list): Likewise.
9650 (fltfn_suffixes): Likewise.
9652 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9653 for math functions that have _Float<N> and _Float<N>X variants.
9654 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9655 and _Float<N>X variants.
9656 (COPYSIGN): Likewise.
9659 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9660 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9662 (integer_valued_read_call_p): Likewise.
9663 * fold-const-call.c (fold_const_call_ss): Likewise.
9664 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9665 _Float<N> and _Float<N>X variants.
9666 (fold_const_call_ssss): Add support for fma _Float<N> and
9667 _Float<N>X variants.
9668 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9669 support for copysign and fma _Float<N> and _Float<N>X variants.
9670 (backprop::process_builtin_call_use): Likewise.
9671 * tree-call-cdce.c (can_test_argument_range); Add support for
9672 sqrt _Float<N> and _Float<N>X variants.
9673 (edom_only_function): Likewise.
9674 (get_no_error_domain): Likewise.
9675 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9676 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9677 copysign _Float<N> and _Float<N>X variants.
9678 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9679 handled by machine independent code.
9680 (FMAF128): Likewise.
9681 * doc/cpp.texi (Common Predefined Macros): Document defining
9682 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9683 fma _Float<N> and _Float<N>X variants.
9685 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9688 * config/i386/i386-modes.def (CCFPU): Remove definition.
9689 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9690 (ix86_cc_modes_compatible): Ditto.
9691 (ix86_expand_carry_flag_compare): Ditto.
9692 (ix86_expand_int_movcc): Ditto.
9693 (ix86_expand_int_addcc): Ditto.
9694 (ix86_reverse_condition): Ditto.
9695 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9696 Return true/false for unordered/ordered fp comparisons.
9697 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9698 (ix86_prepare_fp_compare_args): Update for rename.
9699 (ix86_expand_fp_compare): Update for rename. Generate unordered
9700 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9701 (ix86_expand_sse_compare_and_jump): Ditto.
9702 * config/i386/predicates.md (fcmov_comparison_operator):
9703 Remove CCFPU mode handling.
9704 (ix86_comparison_operator): Ditto.
9705 (ix86_carry_flag_operator): Ditto.
9706 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9707 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9708 (*cmpu<mode>_cc_i387): Ditto.
9709 (FPCMP): Remove mode iterator.
9710 (unord): Remove mode attribute.
9711 (unord_subst): New define_subst transformation
9712 (unord): New define_subst attribute.
9714 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9715 (*cmpi<unord>xf_i387): Ditto.
9716 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9717 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9718 using unord_subst transformation.
9719 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9720 (round_saeonly): Also handle CCFP mode.
9721 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9722 Remove UNSPEC_SAHF unspec handling.
9724 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9726 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9728 2017-10-27 Jeff Law <law@redhat.com>
9730 * gimple-ssa-sprintf.c: Include domwalk.h.
9731 (class sprintf_dom_walker): New class, derived from dom_walker.
9732 (sprintf_dom_walker::before_dom_children): New function.
9733 (struct call_info): Moved into sprintf_dom_walker class
9734 (compute_formath_length, handle_gimple_call): Likewise.
9735 (sprintf_length::execute): Call the dominator walker rather
9736 than walking the statements.
9738 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9739 gimple statement locations.
9740 (check_array_bounds): Corresponding changes. Get the statement's
9741 location directly from wi->stmt.
9743 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9746 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9748 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9750 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9751 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9753 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9756 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9757 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9758 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9759 maybe_get_pool_constant.
9760 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9763 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9765 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9766 2.26 caveat. Update gas and gld versions.
9767 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9770 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9772 * cgraph.h (set_malloc_flag): Declare.
9773 * cgraph.c (set_malloc_flag_1): New function.
9774 (set_malloc_flag): Likewise.
9775 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9776 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9778 (read_ipa_call_summary): Add support for reading is_return_callee.
9779 (write_ipa_call_summary): Stream is_return_callee.
9780 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9781 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9782 ipa-prop.h, ipa-fnsummary.h.
9783 (pure_const_names): Change to static.
9784 (malloc_state_e): Define.
9785 (malloc_state_names): Define.
9786 (funct_state_d): Add field malloc_state.
9787 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9788 (check_retval_uses): New function.
9789 (malloc_candidate_p): Likewise.
9790 (analyze_function): Add support for malloc attribute.
9791 (pure_const_write_summary): Stream malloc_state.
9792 (pure_const_read_summary): Add support for reading malloc_state.
9793 (dump_malloc_lattice): New function.
9794 (propagate_malloc): New function.
9795 (warn_function_malloc): New function.
9796 (ipa_pure_const::execute): Call propagate_malloc and
9797 ipa_free_fn_summary.
9798 (pass_local_pure_const::execute): Add support for malloc attribute.
9799 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9800 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9802 2017-10-27 Martin Liska <mliska@suse.cz>
9804 PR gcov-profile/82457
9805 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9806 for fork-like functions to be properly instrumented.
9808 2017-10-27 Richard Biener <rguenther@suse.de>
9811 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9812 info when we redirected EH.
9814 2017-10-26 Michael Collison <michael.collison@arm.com>
9816 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9818 (<optab>_trunchf<GPI:mode>2: New pattern.
9819 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9820 * config/aarch64/iterators.md (wv): New mode attribute.
9821 (vf, VF): New mode attributes.
9822 (vgp, VGP): New mode attributes.
9823 (s): Update attribute with SImode and DImode prefixes.
9825 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9827 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9828 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9829 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9830 (nios2_option_overide): Initialize it. Don't allow R0-relative
9831 addressing with PIC.
9832 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9833 (nios2_symbolic_constant_p): Likewise.
9834 (nios2_legitimate_address_p): Likewise.
9835 (nios2_r0rel_section_name_p): New.
9836 (nios2_symbol_ref_in_r0rel_data_p): New.
9837 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9838 (r0rel_constant_p): New.
9839 (nios2_print_operand_address): Handle r0rel_constant_p.
9840 (nios2_cdx_narrow_form_p): Likewise.
9841 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9842 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9843 (Nios II Options): Document -mr0rel-sec.
9845 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9847 * config/nios2/nios2.c: Include xregex.h.
9848 (nios2_gprel_sec_regex): New.
9849 (nios2_option_overide): Initialize it. Don't allow GP-relative
9850 addressing with PIC.
9851 (nios2_small_section_name_p): Check for regex match.
9852 * config/nios2/nios2.opt (mgprel-sec=): New option.
9853 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9854 (Nios II Options): Document -mgprel-sec.
9856 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9858 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9860 2017-10-26 Tom de Vries <tom@codesourcery.com>
9862 PR tree-optimization/82707
9863 * gimple.c (gimple_copy): Fix unsharing of
9864 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9866 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
9868 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
9869 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
9870 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
9871 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
9872 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
9873 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
9874 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
9875 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
9876 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
9877 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
9878 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
9879 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
9880 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
9881 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
9882 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
9883 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
9884 _mm512_mask_cmpunord_ps_mask): New intrinsics.
9886 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9888 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
9890 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
9891 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
9892 warning to rs6000.c. Remove the Undocumented flag, since it has
9894 (-mabi=ibmlongdouble): Likewise.
9895 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
9896 already set, set the default format for long double.
9897 (rs6000_debug_reg_global): Print whether long double is IBM or
9899 (rs6000_option_override_internal): Rework setting long double
9900 format. Only warn if the user is changing the long double default
9901 and they did not use -Wno-psabi.
9902 * doc/invoke.texi (PowerPC options): Update the documentation for
9903 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
9905 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9906 Alan Hayward <alan.hayward@arm.com>
9907 David Sherwood <david.sherwood@arm.com>
9909 * rtl.h (wider_subreg_mode): New function.
9910 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
9911 rather than an unsigned int *.
9912 * ira-color.c (regno_max_ref_width): Replace with...
9913 (regno_max_ref_mode): ...this new variable.
9914 (coalesced_pseudo_reg_slot_compare): Update accordingly.
9915 Use wider_subreg_mode.
9916 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
9917 rather than an unsigned int *.
9918 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
9919 (process_alt_operands): Likewise.
9920 (invariant_p): Likewise.
9921 * lra-spills.c (assign_mem_slot): Likewise.
9922 (add_pseudo_to_slot): Likewise.
9923 * lra.c (collect_non_operand_hard_regs): Likewise.
9924 (add_regs_to_insn_regno_info): Likewise.
9925 * reload1.c (regno_max_ref_width): Replace with...
9926 (regno_max_ref_mode): ...this new variable.
9927 (reload): Update accordingly. Update call to
9928 ira_sort_regnos_for_alter_reg.
9929 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
9930 (init_eliminable_invariants): Update to use regno_max_ref_mode.
9931 (scan_paradoxical_subregs): Likewise.
9933 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9935 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
9936 (aarch64_frame): Add emit_frame_chain boolean.
9937 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9938 Move eh_return case to aarch64_layout_frame.
9939 (aarch64_layout_frame): Initialize emit_frame_chain.
9940 (aarch64_expand_prologue): Use emit_frame_chain.
9942 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9944 * config/aarch64/aarch64.c (aarch64_layout_frame):
9945 Ensure LR is always stored at the bottom of the callee-saves.
9946 Remove rarely used frame layout which saves callee-saves at top of
9947 frame, so the store of LR can be used as a valid probe in all cases.
9949 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9951 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
9952 Improve unaligned TImode/TFmode base/offset split.
9954 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9955 Alan Hayward <alan.hayward@arm.com>
9956 David Sherwood <david.sherwood@arm.com>
9958 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
9959 * combine.c (find_single_use_1): Likewise.
9960 (expand_field_assignment): Likewise.
9961 (move_deaths): Likewise.
9962 * lra-constraints.c (simplify_operand_subreg): Likewise.
9963 (curr_insn_transform): Likewise.
9964 * lra.c (collect_non_operand_hard_regs): Likewise.
9965 (add_regs_to_insn_regno_info): Likewise.
9966 * rtlanal.c (reg_referenced_p): Likewise.
9967 (covers_regno_no_parallel_p): Likewise.
9969 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9971 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
9972 Don't print any bits outside the precision of the value.
9973 * wide-int.cc (test_printing): Add some new tests.
9975 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9977 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
9978 supports -xbrace_comment option.
9979 * configure: Regenerate.
9980 * config.in: Regenerate.
9981 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
9982 (ASM_CPU_SPEC): Use it.
9984 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9986 * target.def (static_rtx_alignment): New hook.
9987 * targhooks.h (default_static_rtx_alignment): Declare.
9988 * targhooks.c (default_static_rtx_alignment): New function.
9989 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
9990 * doc/tm.texi: Regenerate.
9991 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
9992 instead of targetm.constant_alignment. Remove call to
9994 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9995 (cris_preferred_mininum_alignment): New function, split out from...
9996 (cris_constant_alignment): ...here.
9997 (cris_static_rtx_alignment): New function.
9998 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
10000 (ix86_constant_alignment): ...here.
10001 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10002 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10003 (mmix_static_rtx_alignment): New function.
10004 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10005 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10007 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10010 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10011 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10013 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10015 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10016 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10017 mark as a sign-extending load.
10018 (local_pic_load_u): Define.
10020 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10022 PR middle-end/82062
10023 * fold-const.c (operand_equal_for_comparison_p): Also return true
10024 if ARG0 is a simple variant of ARG1 with narrower precision.
10025 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10027 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10029 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10031 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10032 scatter_static, scatter_per_elt.
10033 * x86-tune-costs.h: Add new cost entries.
10035 2017-10-25 Richard Biener <rguenther@suse.de>
10037 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10038 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10039 class pass_fre): Move to ...
10040 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10041 class pass_fre): ... here and adjust for statistics.
10043 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10046 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10047 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10048 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10049 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10052 2017-10-25 Richard Biener <rguenther@suse.de>
10054 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10055 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10056 eliminate_push_avail, eliminate_insert): Move inside...
10057 (class eliminate_dom_walker): ... this class in preparation
10059 (fini_eliminate): Remove by merging with ...
10060 (eliminate): ... this function. Adjust for class changes.
10061 (pass_pre::execute): Remove fini_eliminate call.
10062 (pass_fre::execute): Likewise.
10064 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10067 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10068 (VPERMI2, VPERMI2I): New mode iterators.
10069 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10070 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10072 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10073 mode iterator. Remove 3 old define_insn patterns.
10074 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10075 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10076 VPERMI2 mode iterator, remove the other two expanders.
10077 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10078 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10079 instructions, remove the other two patterns.
10080 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10081 mode iterator, remove the other two patterns.
10082 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10083 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10084 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10085 and adjust argument order accordingly.
10086 (ix86_expand_vec_perm): Adjust caller.
10087 (expand_vec_perm_1): Likewise.
10088 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10089 (expand_vec_perm_vpermt2_vpshub2): ... this.
10090 (ix86_expand_vec_perm_const_1): Adjust caller.
10091 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10094 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10095 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10096 (vec_shl_<mode>): Remove unused expander.
10097 (avx512bw_<shift_insn><mode>3): New define_insn.
10098 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10099 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10101 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10104 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10107 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10109 PR rtl-optimization/82396
10110 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
10111 (autopref_multipass_init): Simplify initialization.
10112 (autopref_rank_data): Simplify sort order.
10113 * gcc/sched-int.h (autopref_multipass_data_): Remove
10114 multi_mem_insn_p, min_offset and max_offset.
10116 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10118 PR middle-end/60580
10119 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10120 Check special value of flag_omit_frame_pointer.
10121 (aarch64_can_eliminate): Likewise.
10122 (aarch64_override_options_after_change_1): Simplify handling of
10123 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
10125 2017-10-24 Richard Biener <rguenther@suse.de>
10127 PR tree-optimization/82697
10128 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
10129 zero for conditional load and unconditional store.
10131 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10133 * doc/install.texi: Document bootstrap-cet.
10135 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10138 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
10139 ENDBR instruction at function entrance if function is only
10142 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10145 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
10146 patterns to better describe from which operation the CF is computed.
10147 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
10148 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
10149 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
10150 is 0, use _0 suffixed expanders instead of emitting a comparison
10153 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
10155 * config/i386/i386.md(*movsf_internal, *movdf_internal):
10156 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
10158 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
10160 PR middle-end/82569
10161 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
10162 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
10163 * loop-iv.c (iv_get_reaching_def): Likewise.
10164 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
10165 variable is promoted and the partition contains undefined values.
10167 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10169 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
10171 (nios2_address_cost): Define.
10172 (nios2_legitimize_address): Recognize (exp + constant) directly.
10173 (TARGET_ADDRESS_COST): Define.
10175 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10177 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
10178 (nios2_symbolic_memory_operand_p): Declare.
10179 (nios2_split_large_constant): Declare.
10180 (nios2_split_symbolic_memory_operand): Declare.
10181 * config/nios2/nios2.c: Adjust includes.
10182 (nios2_symbolic_constant_allowed): New.
10183 (nios2_symbolic_constant_p): New.
10184 (nios2_plus_symbolic_constant_p): New.
10185 (nios2_valid_addr_expr_p): Recognize addresses involving
10186 symbolic constants.
10187 (nios2_legitimate_address_p): Likewise, also LO_SUM.
10188 (nios2_symbolic_memory_operand_p): New.
10189 (nios2_large_constant_p): New.
10190 (nios2_split_large_constant): New.
10191 (nios2_split_plus_large_constant): New.
10192 (nios2_split_symbolic_memory_operand): New.
10193 (nios2_legitimize_address): Code refactoring. Handle addresses
10194 involving symbolic constants.
10195 (nios2_emit_move_sequence): Likewise.
10196 (nios2_print_operand): Improve error output.
10197 (nios2_print_operand_address): Handle LO_SUM.
10198 (nios2_cdx_narrow_form_p): Likewise.
10199 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
10200 operands involving symbolic constants.
10201 (movhi_internal, movsi_internal): Likewise.
10202 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
10203 (extendhisi2, extendqi<mode>2): Likewise.
10205 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10207 * tree-pass.h (PROP_rtl_split_insns): Define.
10208 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
10210 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10212 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
10214 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10217 * target.def (const_not_ok_for_debug_p): Default to
10218 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
10219 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
10220 * targhooks.c (default_const_not_ok_for_debug_p): New function.
10221 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
10222 which targetm.const_not_ok_for_debug_p returned true.
10223 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
10225 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
10227 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
10228 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
10229 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
10231 (ix86_const_not_ok_for_debug_p): New function.
10232 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
10233 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
10235 2017-10-23 David Malcolm <dmalcolm@redhat.com>
10238 * system.h: Conditionally include "unique-ptr.h" if
10239 INCLUDE_UNIQUE_PTR is defined.
10240 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
10241 of defining INCLUDE_UNIQUE_PTR before including "system.h".
10243 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
10245 * config/rl78/rl78.md: New define_expand "subdi3".
10247 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
10250 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
10251 DF_REF_INSN if DF_REF_INSN_INFO is false.
10253 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
10255 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
10256 xmm_move instead of sse_move.
10257 (sse_store_index): New function.
10258 (ix86_register_move_cost): Be more sensible about mismatch stall;
10259 model AVX moves correctly; make difference between sse->integer and
10261 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
10262 moves; make difference between SSE and AVX.
10263 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
10264 and zmm_move. Increase size of sse load and store tables;
10265 add unaligned load and store tables; add ssemmx_to_integer.
10266 * x86-tune-costs.h: Update all entries according to real
10267 move latencies from Agner Fog's manual and chip documentation.
10269 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10272 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
10273 * config/i386/constraints.md (Wf): New constraint.
10274 * config/i386/i386.md (UNSPEC_SBB): New unspec.
10275 (cmp<dwi>_doubleword): Removed.
10276 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
10277 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
10278 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
10279 expand with cmp<dwi>_doubleword. For LTU and GEU use
10280 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
10282 * common.opt (gcolumn-info): Enable by default.
10283 * doc/invoke.texi (gcolumn-info): Document new default.
10285 2017-10-23 Richard Biener <rguenther@suse.de>
10287 PR tree-optimization/82672
10288 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
10289 Fold the stmt if we propagated into it.
10291 2017-10-23 Richard Biener <rguenther@suse.de>
10293 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
10294 (bitmap_remove_expr_from_set): ... this. All callers call this
10295 for non-constant values.
10296 (bitmap_set_subtract): Rename to...
10297 (bitmap_set_subtract_expressions): ... this. Adjust and
10299 (bitmap_set_contains_value): Remove superfluous check.
10300 (bitmap_set_replace_value): Inline into single caller ...
10301 (bitmap_value_replace_in_set): ... here and simplify.
10302 (dependent_clean): Merge into ...
10303 (clean): ... this using an overload. Adjust.
10304 (prune_clobbered_mems): Adjust.
10305 (compute_antic_aux): Likewise.
10306 (compute_partial_antic_aux): Likewise.
10308 2017-10-23 Richard Biener <rguenther@suse.de>
10310 PR tree-optimization/82129
10312 2017-08-01 Richard Biener <rguenther@suse.de>
10314 PR tree-optimization/81181
10315 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10316 (compute_antic): ... end of iteration here.
10318 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10320 * target.def (starting_frame_offset): New hook.
10321 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
10322 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
10323 * doc/tm.texi.in: Regenerate.
10324 * hooks.h (hook_hwi_void_0): Declare.
10325 * hooks.c (hook_hwi_void_0): New function.
10326 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
10327 STARTING_FRAME_OFFSET.
10328 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
10329 * reload1.c (reload): Likewise.
10330 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
10331 instead of STARTING_FRAME_OFFSET.
10332 * function.c (try_fit_stack_local): Likewise.
10333 (assign_stack_local_1): Likewise
10334 (instantiate_virtual_regs): Likewise.
10335 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
10336 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
10337 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
10338 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
10339 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
10340 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
10341 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
10342 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
10343 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
10344 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
10345 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
10346 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
10347 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
10348 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
10349 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
10350 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
10351 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
10352 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
10353 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
10354 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
10355 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
10356 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
10357 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
10358 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
10359 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
10360 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
10361 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
10362 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
10363 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
10364 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
10365 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
10366 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
10367 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
10368 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
10369 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
10370 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
10371 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
10372 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
10373 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
10374 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
10375 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
10376 * config/avr/avr.c (avr_starting_frame_offset): Make static and
10377 return a HOST_WIDE_INT.
10378 (avr_builtin_setjmp_frame_value): Use it instead of
10379 STARTING_FRAME_OFFSET.
10380 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10381 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
10382 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
10384 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10385 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
10386 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
10387 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10388 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
10389 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
10390 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10391 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
10392 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
10393 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10394 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
10395 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
10397 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10398 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
10399 * config/mips/mips.c (mips_compute_frame_info): Refer to
10400 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10401 (mips_starting_frame_offset): New function.
10402 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10403 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
10404 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
10405 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
10406 and return a HOST_WIDE_INT.
10407 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10408 (mmix_initial_elimination_offset): Refer to
10409 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10410 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
10411 * config/pa/pa.c (pa_starting_frame_offset): New function.
10412 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
10413 (pa_expand_prologue): Likewise.
10414 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10415 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
10416 !FRAME_GROWS_DOWNWARD handling to...
10417 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10418 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
10419 !FRAME_GROWS_DOWNWARD handling to...
10420 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10421 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
10422 !FRAME_GROWS_DOWNWARD handling to...
10423 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10424 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
10426 (rs6000_starting_frame_offset): New function.
10427 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
10428 !FRAME_GROWS_DOWNWARD handling to...
10429 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10430 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
10431 !FRAME_GROWS_DOWNWARD handling to...
10432 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10433 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
10434 !FRAME_GROWS_DOWNWARD handling to...
10435 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10436 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
10437 (rs6000_starting_frame_offset): New function.
10438 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
10439 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
10440 * config/vax/vax.c (vax_starting_frame_offset): New function.
10441 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
10442 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10443 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
10444 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
10445 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10446 * system.h (STARTING_FRAME_OFFSET): Poison.
10448 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10450 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10451 SCALAR_TYPE_MODE instead of TYPE_MODE.
10453 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10454 Alan Hayward <alan.hayward@arm.com>
10455 David Sherwood <david.sherwood@arm.com>
10457 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
10459 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10460 Alan Hayward <alan.hayward@arm.com>
10461 David Sherwood <david.sherwood@arm.com>
10463 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
10465 2017-10-23 Richard Biener <rguenther@suse.de>
10467 PR tree-optimization/82129
10468 * tree-ssa-pre.c (bitmap_set_and): Remove.
10469 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
10470 canonicalizing expressions in the set to those with lowest
10471 ID rather than taking that from the first edge.
10473 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10475 * combine.c (rtx_equal_for_field_assignment_p): Use
10476 byte_lowpart_offset.
10478 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10479 Alan Hayward <alan.hayward@arm.com>
10480 David Sherwood <david.sherwood@arm.com>
10482 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
10483 to a SUBREG_PROMOTED_VAR.
10485 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10486 Alan Hayward <alan.hayward@arm.com>
10487 David Sherwood <david.sherwood@arm.com>
10489 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
10490 (expand_debug_source_expr): Likewise.
10491 * combine.c (combine_simplify_rtx): Likewise.
10492 * cse.c (fold_rtx): Likewise.
10493 * optabs.c (expand_float): Likewise.
10494 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10495 (simplify_binary_operation_1): Likewise.
10497 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10498 Alan Hayward <alan.hayward@arm.com>
10499 David Sherwood <david.sherwood@arm.com>
10501 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
10502 (record_promoted_value): Likewise.
10503 * expr.c (expand_expr_real_2): Likewise.
10504 * ree.c (update_reg_equal_equiv_notes): Likewise.
10505 (combine_set_extension): Likewise.
10506 * rtlanal.c (low_bitmask_len): Likewise.
10507 * simplify-rtx.c (neg_const_int): Likewise.
10508 (simplify_binary_operation_1): Likewise.
10510 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10511 Alan Hayward <alan.hayward@arm.com>
10512 David Sherwood <david.sherwood@arm.com>
10514 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
10515 * regcprop.c (maybe_mode_change): Likewise.
10516 * reload1.c (alter_reg): Likewise.
10518 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10520 * inchash.h (inchash::hash::add_wide_int): New function.
10521 * lto-streamer-out.c (hash_tree): Use it.
10523 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10525 * inchash.h (inchash::hash::add_wide_int): Rename to...
10526 (inchash::hash::add_hwi): ...this.
10527 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
10528 (polymorphic_call_target_hasher::hash): Likewise.
10529 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
10530 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
10531 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
10532 * lto-streamer-out.c (hash_tree): Likewise.
10533 * optc-save-gen.awk: Likewise.
10534 * tree.c (add_expr): Likewise.
10536 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10539 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
10540 for ordered inequality comparisons even with TARGET_IEEE_FP.
10542 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10545 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
10546 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10547 Expand with cmp<dwi>_doubleword.
10549 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10551 * extend.texi: Add x86 specific to 'nocf_check' attribute.
10552 List CET intrinsics.
10553 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
10554 specific to -fcf-protection option.
10556 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10558 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
10559 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
10560 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
10561 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
10562 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
10563 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
10564 (extra_objs): Add cet.o for Linux/x86 targets.
10565 (tmake_file): Add i386/t-cet for Linux/x86 targets.
10566 * config/i386/cet.c: New file.
10567 * config/i386/cetintrin.h: Likewise.
10568 * config/i386/t-cet: Likewise.
10569 * config/i386/cpuid.h (bit_SHSTK): New.
10570 (bit_IBT): Likewise.
10571 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
10572 pass IBT and SHSTK bits.
10573 * config/i386/i386-builtin-types.def
10574 (VOID_FTYPE_UNSIGNED_PVOID): New.
10575 (VOID_FTYPE_UINT64_PVOID): Likewise.
10576 * config/i386/i386-builtin.def: Add CET intrinsics.
10577 * config/i386/i386-c.c (ix86_target_macros_internal): Add
10578 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
10579 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
10580 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
10582 * config/i386/i386.c (rest_of_insert_endbranch): New.
10583 (pass_data_insert_endbranch): Likewise.
10584 (pass_insert_endbranch): Likewise.
10585 (make_pass_insert_endbranch): Likewise.
10586 (ix86_notrack_prefixed_insn_p): Likewise.
10587 (ix86_target_string): Add -mibt, -mshstk flags.
10588 (ix86_option_override_internal): Add flag_cf_protection
10590 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
10591 (ix86_print_operand): Add 'notrack' prefix output.
10592 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
10593 (ix86_expand_builtin): Expand CET intrinsics.
10594 (x86_output_mi_thunk): Add 'endbranch' instruction.
10595 * config/i386/i386.h (TARGET_IBT): New.
10596 (TARGET_IBT_P): Likewise.
10597 (TARGET_SHSTK): Likewise.
10598 (TARGET_SHSTK_P): Likewise.
10599 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
10600 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
10601 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
10602 (builtin_setjmp_setup): New pattern.
10603 (builtin_longjmp): Likewise.
10604 (rdssp<mode>): Likewise.
10605 (incssp<mode>): Likewise.
10606 (saveprevssp): Likewise.
10607 (rstorssp): Likewise.
10608 (wrss<mode>): Likewise.
10609 (wruss<mode>): Likewise.
10610 (setssbsy): Likewise.
10611 (clrssbsy): Likewise.
10612 (nop_endbr): Likewise.
10613 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
10615 * config/i386/immintrin.h: Include <cetintrin.h>.
10616 * config/i386/linux-common.h
10617 (file_end_indicate_exec_stack_and_cet): New prototype.
10618 (TARGET_ASM_FILE_END): New.
10620 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10622 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10623 latencies instead of having separate table; make difference between
10624 integer and float costs.
10625 * i386.h (processor_costs): Remove scalar_stmt_cost,
10626 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10627 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10629 * x86-tune-costs.h: Remove entries which has been removed in
10630 procesor_costs from all tables; make cond_taken_branch_cost
10631 and cond_not_taken_branch_cost COST_N_INSNS based.
10633 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10635 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10637 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10639 * config/i386/i386.md (isa): Remove fma_avx512f.
10640 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10641 <avx512>_fmadd_<mode>_mask3<round_name>,
10642 <avx512>_fmsub_<mode>_mask<round_name>,
10643 <avx512>_fmsub_<mode>_mask3<round_name>,
10644 <avx512>_fnmadd_<mode>_mask<round_name>,
10645 <avx512>_fnmadd_<mode>_mask3<round_name>,
10646 <avx512>_fnmsub_<mode>_mask<round_name>,
10647 <avx512>_fnmsub_<mode>_mask3<round_name>,
10648 <avx512>_fmaddsub_<mode>_mask<round_name>,
10649 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10650 <avx512>_fmsubadd_<mode>_mask<round_name>,
10651 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10652 (*vec_widen_umult_even_v16si<mask_name>,
10653 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10654 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10656 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10658 * extend.texi: Add 'nocf_check' documentation.
10659 * gimple.texi: Add second parameter to
10660 gimple_build_call_from_tree.
10661 * invoke.texi: Add -fcf-protection documentation.
10662 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10664 2017-10-20 Richard Biener <rguenther@suse.de>
10666 PR tree-optimization/82473
10667 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10668 the largest input type.
10670 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10672 * c-attribs.c (handle_nocf_check_attribute): New function.
10673 (c_common_attribute_table): Add 'nocf_check' handling.
10674 * gimple-parser.c: Add second argument NULL to
10675 gimple_build_call_from_tree.
10676 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10677 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10679 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10680 * common.opt: Add fcf-protection flag.
10681 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10682 * flag-types.h: Add enum cf_protection_level.
10683 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10684 Add 'nocf_check' attribute propagation to gimple call.
10685 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10686 (gimple_build_call_from_tree): Update prototype.
10687 (gimple_call_nocf_check_p): New function.
10688 (gimple_call_set_nocf_check): Likewise.
10689 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10690 * ipa-icf.c: Add nocf_check attribute in statement hash.
10691 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10692 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10693 * toplev.c (process_options): Add flag_cf_protection handling.
10695 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10697 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10699 2017-10-20 Richard Biener <rguenther@suse.de>
10701 PR tree-optimization/82603
10702 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10703 remove false predicated stores.
10705 2017-10-20 Richard Biener <rguenther@suse.de>
10707 * graphite-isl-ast-to-gimple.c
10708 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10709 Remove return value and simplify, dump copied stmt after lhs
10711 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10712 Reduce dump verbosity.
10713 (gsi_insert_earliest): Likewise.
10714 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10715 * graphite.c (print_global_statistics): Adjust dumping.
10716 (print_graphite_scop_statistics): Likewise.
10717 (print_graphite_statistics): Do not dump loops here.
10718 (graphite_transform_loops): But here.
10720 2017-10-20 Nicolas Roche <roche@adacore.com>
10722 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10723 * configure: Regenerate.
10725 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10728 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10729 functions when optimizing replace GIMPLE_RETURN stmts with
10730 calls to __builtin_unreachable ().
10733 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10734 for -fsanitize=thread link of executables.
10735 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10736 link of executables.
10739 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10740 New mode iterators.
10741 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10742 define_insns for logical vector shifts to use VI248_AVX512BW
10743 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10744 condition, useless isa and prefix attributes. Change the first
10745 2 of these define_insns to ...
10746 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10747 define_insn for avx512vl.
10748 (<shift_insn><mode>3): ... and this, new define_insn without
10749 masking for non-avx512vl.
10752 * config/i386/sse.md (*andnot<mode>3,
10753 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10754 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10755 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10756 not applied use empty suffix even for TARGET_AVX512VL.
10757 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10758 is applied, supply evex,evex or evex,evex,evex instead of just
10761 2017-10-20 Julia Koval <julia.koval@intel.com>
10763 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10764 (OPTION_MASK_ISA_GFNI_UNSET): New.
10765 (ix86_handle_option): Handle OPT_mgfni.
10766 * config/i386/cpuid.h (bit_GFNI): New.
10767 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10768 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10769 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10770 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10771 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10772 * config/i386/i386.opt: Add mgfni.
10774 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10776 * config/msp430/msp430.c (msp430_option_override): Disable
10777 -fdelete-null-pointer-checks.
10778 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10780 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10782 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10783 of x87 and SSE instructions.
10785 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10787 * asan.c (create_cond_insert_point): Do not update edge count.
10788 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10789 (afdo_propagate_circuit): Likewise.
10790 (afdo_calculate_branch_prob): Likewise.
10791 (afdo_annotate_cfg): Likewise.
10792 * basic-block.h (struct edge_def): Remove count.
10793 (edge_def::count): New accessor.
10794 * bb-reorder.c (rotate_loop): Update.
10795 (find_traces_1_round): Update.
10796 (connect_traces): Update.
10797 (sanitize_hot_paths): Update.
10798 * cfg.c (unchecked_make_edge): Update.
10799 (make_single_succ_edge): Update.
10800 (check_bb_profile): Update.
10801 (dump_edge_info): Update.
10802 (update_bb_profile_for_threading): Update.
10803 (scale_bbs_frequencies_int): Update.
10804 (scale_bbs_frequencies_gcov_type): Update.
10805 (scale_bbs_frequencies_profile_count): Update.
10806 (scale_bbs_frequencies): Update.
10807 * cfganal.c (connect_infinite_loops_to_exit): Update.
10808 * cfgbuild.c (compute_outgoing_frequencies): Update.
10809 (find_many_sub_basic_blocks): Update.
10810 * cfgcleanup.c (try_forward_edges): Update.
10811 (try_crossjump_to_edge): Update
10812 * cfgexpand.c (expand_gimple_cond): Update
10813 (expand_gimple_tailcall): Update
10814 (construct_exit_block): Update
10815 * cfghooks.c (verify_flow_info): Update
10816 (redirect_edge_succ_nodup): Update
10817 (split_edge): Update
10818 (make_forwarder_block): Update
10819 (duplicate_block): Update
10820 (account_profile_record): Update
10821 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10822 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10823 * cfgloopmanip.c (scale_loop_profile): Update.
10825 (lv_adjust_loop_entry_edge): Update.
10826 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10827 (force_nonfallthru_and_redirect): Update.
10828 (purge_dead_edges): Update.
10829 (rtl_flow_call_edges_add): Update.
10830 * cgraphunit.c (init_lowered_empty_function): Update.
10831 (cgraph_node::expand_thunk): Update.
10832 * gimple-pretty-print.c (dump_probability): Update.
10833 (dump_edge_probability): Update.
10834 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10835 * haifa-sched.c (sched_create_recovery_edges): Update.
10836 * hsa-gen.c (convert_switch_statements): Update.
10837 * ifcvt.c (dead_or_predicable): Update.
10838 * ipa-inline-transform.c (inline_transform): Update.
10839 * ipa-split.c (split_function): Update.
10840 * ipa-utils.c (ipa_merge_profiles): Update.
10841 * loop-doloop.c (add_test): Update.
10842 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10843 * lto-streamer-in.c (input_cfg): Update.
10844 (input_function): Update.
10845 * lto-streamer-out.c (output_cfg): Update.
10846 * modulo-sched.c (sms_schedule): Update.
10847 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10848 * predict.c (maybe_hot_edge_p): Update.
10849 (unlikely_executed_edge_p): Update.
10850 (probably_never_executed_edge_p): Update.
10851 (dump_prediction): Update.
10852 (drop_profile): Update.
10853 (propagate_unlikely_bbs_forward): Update.
10854 (determine_unlikely_bbs): Update.
10855 (force_edge_cold): Update.
10856 * profile.c (compute_branch_probabilities): Update.
10857 * reg-stack.c (better_edge): Update.
10858 * shrink-wrap.c (handle_simple_exit): Update.
10859 * tracer.c (better_p): Update.
10860 * trans-mem.c (expand_transaction): Update.
10861 (split_bb_make_tm_edge): Update.
10862 * tree-call-cdce.c: Update.
10863 * tree-cfg.c (gimple_find_sub_bbs): Update.
10864 (gimple_split_edge): Update.
10865 (gimple_duplicate_sese_region): Update.
10866 (gimple_duplicate_sese_tail): Update.
10867 (gimple_flow_call_edges_add): Update.
10868 (insert_cond_bb): Update.
10869 (execute_fixup_cfg): Update.
10870 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
10871 * tree-complex.c (expand_complex_div_wide): Update.
10872 * tree-eh.c (lower_resx): Update.
10873 (unsplit_eh): Update.
10874 (cleanup_empty_eh_move_lp): Update.
10875 * tree-inline.c (copy_edges_for_bb): Update.
10876 (freqs_to_counts): Update.
10877 (copy_cfg_body): Update.
10878 * tree-ssa-dce.c (remove_dead_stmt): Update.
10879 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10880 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10881 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10882 (unloop_loops): Update.
10883 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10884 * tree-ssa-loop-split.c (connect_loops): Update.
10885 (split_loop): Update.
10886 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10887 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10888 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10889 * tree-ssa-reassoc.c (branch_fixup): Update.
10890 * tree-ssa-tail-merge.c (replace_block_by): Update.
10891 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10892 (compute_path_counts): Update.
10893 (update_profile): Update.
10894 (recompute_probabilities): Update.
10895 (update_joiner_offpath_counts): Update.
10896 (estimated_freqs_path): Update.
10897 (freqs_to_counts_path): Update.
10898 (clear_counts_path): Update.
10899 (ssa_fix_duplicate_block_edges): Update.
10900 (duplicate_thread_path): Update.
10901 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10902 (case_bit_test_cmp): Update.
10903 (collect_switch_conv_info): Update.
10904 (gen_inbound_check): Update.
10905 (do_jump_if_equal): Update.
10906 (emit_cmp_and_jump_insns): Update.
10907 * tree-tailcall.c (decrease_profile): Update.
10908 (eliminate_tail_call): Update.
10909 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10910 (vect_do_peeling): Update.
10911 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10912 * ubsan.c (ubsan_expand_null_ifn): Update.
10913 (ubsan_expand_ptr_ifn): Update.
10914 * value-prof.c (gimple_divmod_fixed_value): Update.
10915 (gimple_mod_pow2): Update.
10916 (gimple_mod_subtract): Update.
10917 (gimple_ic): Update.
10918 (gimple_stringop_fixed_value): Update.
10920 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10923 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
10925 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
10927 PR rtl-optimization/82395
10928 * ira-color.c (allocno_priority_compare_func): Fix comparison step
10929 based on non_spilled_static_chain_regno_p.
10931 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10933 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
10934 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
10935 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
10937 2017-10-19 Martin Sebor <msebor@redhat.com>
10939 PR tree-optimization/82596
10940 * tree.c (array_at_struct_end_p): Handle STRING_CST.
10942 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10944 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
10945 (get_mem_refs_of_builtin_call): Likewise.
10946 * builtins.c (expand_builtin_apply): Adjust call to
10947 allocate_dynamic_stack_space.
10948 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
10949 the third argument to allocate_dynamic_stack_space, otherwise -1.
10950 (expand_builtin): Deal with all alloca variants.
10951 (is_inexpensive_builtin): Likewise.
10952 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
10953 * calls.c (special_function_p): Deal with all alloca variants.
10954 (initialize_argument_information): Adjust call to
10955 allocate_dynamic_stack_space.
10956 (expand_call): Likewise.
10957 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
10958 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
10959 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
10960 use it for the stack usage computation.
10961 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
10962 * function.c (gimplify_parameters): Call build_alloca_call_expr.
10963 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
10964 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
10965 (in_loop_p): Remove first argument and useless check.
10966 (pass_walloca::execute): Remove useless test and adjust call to above.
10967 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
10968 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
10969 (gimplify_call_expr): Deal with all alloca variants.
10970 * hsa-gen.c (gen_hsa_alloca): Likewise.
10971 (gen_hsa_insns_for_call): Likewise.
10972 * ipa-pure-const.c (special_builtin_state): Likewise.
10973 * tree-chkp.c (chkp_build_returned_bound): Likewise.
10974 * tree-object-size.c (alloc_object_size): Likewise.
10975 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
10976 (call_may_clobber_ref_p_1): Likewise.
10977 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
10978 (ccp_fold_stmt): Likewise.
10979 (optimize_stack_restore): Likewise.
10980 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10981 (mark_all_reaching_defs_necessary_1): Likewise.
10982 (propagate_necessity): Likewise.
10983 (eliminate_unnecessary_stmts): Likewise.
10984 * tree.c (build_common_builtin_nodes): Build
10985 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
10986 (build_alloca_call_expr): New function.
10987 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
10988 (CASE_BUILT_IN_ALLOCA): Likewise.
10989 (build_alloca_call_expr): Declare.
10990 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
10992 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10995 * dwarf2out.c (new_die_raw): New static inline function.
10996 (new_die): Use it to create the DIE.
10997 (add_AT_external_die_ref): Likewise.
10998 (clone_die): Likewise.
10999 (clone_as_declaration): Likewise.
11000 (dwarf2out_vms_debug_main_pointer): Likewise.
11001 (base_type_die): Likewise. Remove early return for corner cases.
11002 Do not call add_pubtype on the DIE here.
11003 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
11004 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11005 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11006 native order exists for base types, attach the DIE manually and call
11007 add_pubtype on it. Do not equate a reverse order DIE to the type.
11009 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11011 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11012 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11014 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11016 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11018 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11019 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11020 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11021 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11022 * opts.c (sanitizer_opts): Add builtin.
11023 * ubsan.c (instrument_builtin): New function.
11024 (pass_ubsan::execute): Call it.
11025 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11026 * doc/invoke.texi: Document -fsanitize=builtin.
11028 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11029 builtins, store max (log2 (align), 0) into uchar field instead of
11030 align into uptr field.
11031 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11032 store uchar 0 field instead of uptr 0 field.
11033 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11034 instead of passing one address of struct with 2 locations pass
11035 two addresses of structs with 1 location each.
11036 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11037 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11038 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11039 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11040 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11041 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11042 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
11043 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
11045 2017-10-19 Martin Liska <mliska@suse.cz>
11048 * file-find.c (remove_prefix): Remove.
11049 * file-find.h (remove_prefix): Likewise.
11050 * gcc-ar.c: Remove smartness of lookup.
11052 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11054 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11055 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11056 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11058 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11061 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11062 (setcc + and to xor + setcc): New peephole2.
11064 2017-10-19 Tom de Vries <tom@codesourcery.com>
11066 * doc/sourcebuild.texi (Test Directives, Variants of
11067 dg-require-support): Add dg-require-stack-size.
11069 2017-10-19 Martin Liska <mliska@suse.cz>
11072 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11073 that have a large alignment.
11074 (gimplify_target_expr): Likewise.
11076 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11078 PR rtl-optimization/82602
11079 * ira.c (rtx_moveable_p): Return false for volatile asm.
11081 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11084 * config/i386/i386-modes.def (CCGZ): New CC mode.
11085 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11086 * config/i386/predicates.md (ix86_comparison_operator):
11088 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11089 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11090 with double-word subtraction.
11091 (put_condition_code): Handle CCGZmode.
11093 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11095 * wide-int.cc (debug (const wide_int &)): New.
11096 (debug (const wide_int *)): New.
11097 (debug (const widest_int &)): New.
11098 (debug (const widest_int *)): New.
11100 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11102 PR middle-end/82556
11103 * lra-constraints.c (curr_insn_transform): Use non-input operand
11104 instead of output one for matched reload.
11106 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11108 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11109 (tree-ssa-loop-ivopts.h): New header file.
11110 (struct builtin_info): New fields.
11111 (classify_builtin_1): Compute and record base and offset parts for
11112 memset builtin partition by calling strip_offset.
11113 (offset_cmp, fuse_memset_builtins): New functions.
11114 (finalize_partitions): Fuse adjacent memset partitions by calling
11116 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
11117 Expose the interface.
11118 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
11120 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11122 PR tree-optimization/82574
11123 * tree-loop-distribution.c (find_single_drs): New parameter. Check
11124 that data reference must be executed exactly once per iteration
11125 against the outermost loop in nest.
11126 (classify_partition): Update call to above function.
11128 2017-10-18 Richard Biener <rguenther@suse.de>
11130 PR tree-optimization/82591
11131 * graphite.c (graphite_transform_loops): Move code gen message
11133 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11134 Here. Handle scop_to_isl_ast failing.
11135 (scop_to_isl_ast): Limit the number of ISL operations.
11137 2017-10-18 Richard Biener <rguenther@suse.de>
11139 * graphite-isl-ast-to-gimple.c
11140 (translate_isl_ast_to_gimple::set_rename): Simplify.
11141 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
11142 (graphite_copy_stmts_from_block): ... here.
11143 (copy_bb_and_scalar_dependences): Simplify.
11144 (add_parameters_to_ivs_params): Canonicalize.
11145 (generate_entry_out_of_ssa_copies): Simplify.
11146 * graphite-sese-to-poly.c (extract_affine_name): Simplify
11147 by passing in ISL dimension.
11148 (parameter_index_in_region_1): Rename to ...
11149 (parameter_index_in_region): ... this.
11150 (extract_affine): Adjust assert, pass down parameter index.
11151 (add_param_constraints): Use range-info when available.
11152 (build_scop_context): Adjust.
11153 * sese.c (new_sese_info): Adjust.
11154 (free_sese_info): Likewise.
11155 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
11156 Remove unused typedefs.
11157 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
11159 2017-10-18 Martin Liska <mliska@suse.cz>
11161 * combine.c (simplify_compare_const): Add gcc_fallthrough.
11163 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11165 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
11166 (s390_sched_init): Do not reset s390_sched_state if we entered the
11167 current basic block via a fallthru edge and all others are unlikely.
11169 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11171 * config/s390/s390.c (NUM_SIDES): New variable.
11172 (LONGRUNNING_THRESHOLD): New variable.
11173 (LATENCY_FACTOR): New variable.
11174 (s390_sched_score): Decrease score for long-running instructions on
11176 (s390_sched_variable_issue): Perform bookkeeping for long-running
11179 2017-10-18 Richard Biener <rguenther@suse.de>
11181 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
11182 Simplify with removal of the parameter rename map.
11183 (set_rename): Likewise.
11184 (should_copy_to_new_region): Likewise.
11185 (graphite_copy_stmts_from_block): Likewise.
11186 (copy_bb_and_scalar_dependences): Remove initialization of
11187 unused copied_bb_map.
11188 (copy_def): Remove.
11189 (copy_internal_parameters): Likewise.
11190 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
11191 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
11192 Use INTEGRAL_TYPE_P.
11193 (parameter_index_in_region_1): Rename to ...
11194 (assign_parameter_index_in_region): ... this. Assert we have
11195 a parameter we handle.
11196 (scan_tree_for_params): Adjust.
11197 * sese.h (parameter_rename_map_t): Remove.
11198 (struct sese_info_t): Remove unused parameter_rename_map and
11199 copied_bb_map members.
11200 * sese.c (new_sese_info): Adjust.
11201 (free_sese_info): Likewise.
11203 2017-10-18 Martin Liska <mliska@suse.cz>
11206 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
11207 on an abnormal edge.
11209 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
11211 * doc/invoke.texi (ffunction-sections and fdata-sections):
11214 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11216 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
11217 the use statement can throw internally.
11219 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11221 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
11222 any RTX present on the RHS of a SET.
11223 * compare-elim.c (try_eliminate_compare): Restore comment.
11225 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11227 * langhooks.h (struct lang_hooks): Document that tree_size langhook
11228 may be also called on tcc_type nodes.
11229 * langhooks.c (lhd_tree_size): Likewise.
11231 2017-10-17 David Malcolm <dmalcolm@redhat.com>
11233 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
11234 format_warning_at_substring.
11235 (maybe_warn): Convert source_range * param to a location_t. Pass
11236 UNKNOWN_LOCATION rather than NULL to fmtwarn.
11237 (format_directive): Remove code to extract source_ranges and
11238 source_range * in favor of just a location_t.
11239 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
11241 * substring-locations.c (format_warning_va): Convert
11242 source_range * param to a location_t.
11243 (format_warning_at_substring): Likewise.
11244 * substring-locations.h (format_warning_va): Likewise.
11245 (format_warning_at_substring): Likewise.
11247 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
11249 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
11251 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
11252 and scatter/gather ops.
11254 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
11255 vec_gather_load and vec_scatter_store.
11256 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
11257 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
11258 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
11259 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
11260 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
11261 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
11263 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
11265 * reg-stack.c (compare_for_stack_reg): Add bool argument.
11266 Detect FTST instruction and handle its register pops. Only pop
11267 second operand if can_pop_second_op is true.
11268 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
11269 set can_pop_second_op to false in the compare_for_stack_reg call.
11271 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
11272 output_fp_compare for stack register operands.
11273 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
11274 instructions here. Do not emit stack register pops here. Assert
11275 that FCOMPP pops next to top stack register. Rewrite function.
11277 2017-10-17 Nathan Sidwell <nathan@acm.org>
11279 PR middle-end/82577
11280 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
11281 use DECL_ASSEMBLER_NAME_RAW.
11283 PR middle-end/82546
11284 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
11287 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
11288 Wilco Dijkstra <wilco.dijkstra@arm.com>
11290 * builtins.c (expand_builtin_update_setjmp_buf): Add a
11291 converstion to Pmode from the buf_addr.
11293 2017-10-17 Richard Biener <rguenther@suse.de>
11295 * graphite-dependences.c (scop_get_reads_and_writes): Change
11296 output parameters to references.
11298 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
11300 PR 71026/tree-optimization
11301 * fold-const.c (distribute_real_division): Removed.
11302 (fold_binary_loc): Remove calls to distribute_real_divison.
11304 2017-10-17 Richard Biener <rguenther@suse.de>
11306 * graphite-scop-detection.c
11307 (scop_detection::stmt_has_simple_data_refs_p): Always use
11308 the full nest as region.
11309 (try_generate_gimple_bb): Likewise.
11310 * sese.c (scalar_evolution_in_region): Simplify now that
11311 SCEV can handle instantiation in regions.
11312 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
11313 in the non-loop part of a function if requested.
11315 2017-10-17 Richard Biener <rguenther@suse.de>
11317 PR tree-optimization/82563
11318 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
11320 (graphite_regenerate_ast_isl): Call it.
11321 * graphite-scop-detection.c (build_scops): Remove entry edge split.
11323 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11325 PR tree-optimization/82549
11326 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
11327 Formatting fixes. Instead of calling make_bit_field_ref with negative
11330 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
11332 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
11333 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
11334 _mm_maskz_reduce_ss): New.
11335 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
11336 __builtin_ia32_reducess_mask): Ditto..
11337 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
11338 * config/i386/sse.md (reduces<mode>): Renamed to ...
11339 (reduces<mode><mask_scalar_name>): ... this.
11340 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
11342 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
11343 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
11345 2017-10-16 David Malcolm <dmalcolm@redhat.com>
11347 * Makefile.in (OBJS): Add unique-ptr-tests.o.
11348 * selftest-run-tests.c (selftest::run_tests): Call
11349 selftest::unique_ptr_tests_cc_tests.
11350 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
11351 * unique-ptr-tests.cc: New file.
11353 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
11356 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11358 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11359 (make_hard_regno_born, make_hard_regno_dead): Update
11360 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
11362 2017-10-16 Jeff Law <law@redhat.com>
11364 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11366 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
11368 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
11370 2017-10-16 Olivier Hainque <hainque@adacore.com>
11372 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
11373 with_cpu if we were configured for an e500v2 target cpu name.
11375 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
11377 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
11378 * doc/invoke.texi: Document +nodsp as a valid extension for
11381 2017-10-16 Martin Liska <mliska@suse.cz>
11383 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
11384 (test_set_range): Likewise.
11385 (test_range_functions): Rename to ...
11386 (test_bit_in_range): ... this.
11387 (sbitmap_c_tests): Add new test.
11389 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11391 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
11393 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
11394 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
11396 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11398 * config/aarch64/aarch64-builtins.c
11399 (aarch64_types_quadopu_lane_qualifiers): New.
11400 (TYPES_QUADOPU_LANE): New.
11401 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
11402 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
11403 (aarch64_<sur>dot_laneq<vsi2qi>): New.
11404 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
11405 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
11406 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
11407 (Vdottype, DOTPROD): New.
11408 (sur): Add SDOT and UDOT.
11410 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11412 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
11413 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
11414 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
11415 Add TARGET_DOTPROD.
11416 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
11417 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
11418 Enable TARGET_DOTPROD.
11419 (cortex-a75.cortex-a55): Likewise.
11420 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
11422 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11424 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
11425 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
11427 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
11429 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
11430 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
11431 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
11432 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
11433 * config/arm/types.md (neon_dot, neon_dot_q): New.
11434 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
11436 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11438 * config/arm/arm.h (TARGET_DOTPROD): New.
11439 * config/arm/arm.c (arm_arch_dotprod): New.
11440 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
11441 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
11442 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
11443 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
11444 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
11445 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
11446 * doc/invoke.texi (armv8.2-a): Document dotprod
11448 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11450 * i386.c (ix86_vec_cost): New function.
11451 (ix86_rtx_costs): Handle vector operations better.
11452 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
11453 * x86-tune-costs.h: Add new costs to all tables.
11455 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11457 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
11459 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
11460 divsd, sqrtss and sqrtsd
11461 * x86-tune-costs.h: Add new entries to all costs.
11462 (znver1_cost): Fix to match real instruction latencies.
11464 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11465 Michael Collison <michael.collison@arm.com>
11467 * compare-elim.c: Include emit-rtl.h.
11468 (can_merge_compare_into_arith): New function.
11469 (try_validate_parallel): Likewise.
11470 (try_merge_compare): Likewise.
11471 (try_eliminate_compare): Call the above when no previous clobber
11473 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
11476 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11478 PR middle-end/62263
11479 PR middle-end/82498
11480 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
11481 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
11483 PR middle-end/62263
11484 PR middle-end/82498
11485 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
11486 to be any operand_equal_p operands. For & (B - 1) require
11487 B to be power of 2. Recognize
11488 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
11490 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11493 * optabs.c (expand_memory_blockage): Fix call of
11494 targetm.have_memory_blockage.
11496 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11499 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
11500 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
11501 objects to extra_objs instead of overwriting it.
11503 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11505 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
11506 Use any_fp_register_operand as operand[3] predicate. Simplify
11507 equality test for operands[2] and operands[4] memory location.
11508 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
11509 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
11510 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
11511 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
11512 any_fp_register_operand as operand[1] predicate. Simplify
11513 equality test for operands[0] and operands[3] memory location.
11514 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
11515 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
11516 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
11518 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11520 * target-insns.def: Add memory_blockage.
11521 * optabs.c (expand_memory_blockage): New function.
11522 (expand_asm_memory_barrier): Rename ...
11523 (expand_asm_memory_blockage): ... to this.
11524 (expand_mem_thread_fence): Call expand_memory_blockage
11525 instead of expand_asm_memory_barrier.
11526 (expand_mem_singnal_fence): Ditto.
11527 (expand_atomic_load): Ditto.
11528 (expand_atomic_store): Ditto.
11529 * doc/md.texi (Standard Pattern Names For Generation):
11530 Document memory_blockage instruction pattern.
11532 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
11534 * config/rl78/rl78.c (rl78_emit_libcall): New function.
11535 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
11536 * config/rl78/rl78.md: New define_expand "adddi3".
11538 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11540 * cfghooks.c (verify_flow_info): Disable check that all probabilities
11543 2017-10-13 Jeff Law <law@redhat.com>
11545 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
11547 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11550 * internal-fn.c (expand_mul_overflow): If both operands have
11551 the same highpart of -1 or 0 and the topmost bit of lowpart
11552 is different, overflow is if res <= 0 rather than res < 0.
11554 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
11556 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
11557 TARGET_P9_VECTOR code for unaligned_load case.
11559 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11561 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
11563 2017-10-13 Nathan Sidwell <nathan@acm.org>
11565 * tree-core.h (tree_contains_struct): Make bool.
11566 * tree.c (tree_contains_struct): Likewise.
11567 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11568 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
11569 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
11570 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
11571 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
11573 2017-10-13 Richard Biener <rguenther@suse.de>
11575 * graphite-isl-ast-to-gimple.c
11576 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
11577 parameters and dominance check.
11578 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
11579 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
11580 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
11581 Do not update SSA form here or do intermediate IL verification.
11582 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
11583 (graphite_initialize): Remove check on the number of loops in
11584 the function and inline into graphite_transform_loops.
11585 (graphite_finalize): Inline into graphite_transform_loops.
11586 (graphite_transform_loops): Perform SSA update and IL verification
11588 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
11590 2017-10-13 Richard Biener <rguenther@suse.de>
11592 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
11593 graphite_expression_type_precision): Avoid global constructor
11595 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
11596 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
11597 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
11598 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
11599 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
11600 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
11602 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
11605 * config/i386/i386.h (ix86_red_zone_size): New.
11606 * config/i386/i386.md (push peephole2s): Replace
11607 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
11609 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11610 Alan Hayward <alan.hayward@arm.com>
11611 David Sherwood <david.sherwood@arm.com>
11613 * combine.c (can_change_dest_mode): Reject changes in
11614 REGMODE_NATURAL_SIZE.
11616 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11617 Alan Hayward <alan.hayward@arm.com>
11618 David Sherwood <david.sherwood@arm.com>
11620 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
11621 (expand_debug_source_expr): Likewise.
11622 * combine.c (combine_simplify_rtx): Likewise.
11623 * cse.c (fold_rtx): Likewise.
11624 * fwprop.c (canonicalize_address): Likewise.
11625 * targhooks.c (default_shift_truncation_mask): Likewise.
11627 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11628 Alan Hayward <alan.hayward@arm.com>
11629 David Sherwood <david.sherwood@arm.com>
11631 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11632 (widened_mode): Likewise.
11633 (expand_unop): Likewise.
11634 * ree.c (transform_ifelse): Likewise.
11635 (merge_def_and_ext): Likewise.
11636 (combine_reaching_defs): Likewise.
11637 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11639 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11640 Alan Hayward <alan.hayward@arm.com>
11641 David Sherwood <david.sherwood@arm.com>
11643 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11644 * combine.c (gen_lowpart_for_combine): Likewise.
11645 * dwarf2out.c (rtl_for_decl_location): Likewise.
11646 * final.c (alter_subreg): Likewise.
11647 * rtlhooks.c (gen_lowpart_general): Likewise.
11648 (gen_lowpart_if_possible): Likewise.
11650 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11651 Alan Hayward <alan.hayward@arm.com>
11652 David Sherwood <david.sherwood@arm.com>
11654 * calls.c (expand_call): Use subreg_lowpart_offset.
11655 * cse.c (cse_insn): Likewise.
11656 * regcprop.c (copy_value): Likewise.
11657 (copyprop_hardreg_forward_1): Likewise.
11659 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11662 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11663 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11664 =Q constraints to +Q and into insn condition add check
11665 that operands[0] and operands[1] are equal.
11666 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11667 =Q constraints to +Q and into insn condition add check
11668 that operands[0] is equal to either operands[1] or operands[2].
11671 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11672 instead of handling MINUS_EXPR twice (once for each argument),
11673 canonicalize operand order and handle just once, use rtype where
11674 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11677 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11678 any values of __C while still being pattern recognizable as a simple
11679 rotate instruction.
11681 2017-10-13 Richard Biener <rguenther@suse.de>
11683 PR tree-optimization/82451
11685 2017-10-02 Richard Biener <rguenther@suse.de>
11687 PR tree-optimization/82355
11688 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11689 a mapping for the enclosing loop but avoid generating one for
11690 the loop tree root.
11691 (copy_bb_and_scalar_dependences): Remove premature codegen
11692 error on PHIs in blocks duplicated into multiple places.
11693 * graphite-scop-detection.c
11694 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11695 in the region use it as loop and nest to analyze the DR in.
11696 (try_generate_gimple_bb): Likewise.
11697 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11698 (add_loop_constraints): For blocks in a loop not in the region
11699 create a dimension with a single iteration.
11700 * sese.h (gbb_loop_at_index): Remove assert.
11702 * cfgloop.c (loop_preheader_edge): For the loop tree root
11703 return the single successor of the entry block.
11704 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11705 Reset the SCEV hashtable and niters.
11706 * graphite-scop-detection.c
11707 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11708 assert that we only have POLYNOMIAL_CHREC that vary in loops
11709 contained in the region.
11710 (scop_detection::graphite_can_represent_expr): Adjust.
11711 (scop_detection::stmt_has_simple_data_refs_p): For loops
11712 not in the region set loop to NULL. The nest is now the
11713 entry edge to the region.
11714 (try_generate_gimple_bb): Likewise.
11715 * sese.c (scalar_evolution_in_region): Adjust for
11716 instantiate_scev change.
11717 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11718 Make nest parameter the edge into the region.
11719 (create_data_ref): Likewise.
11720 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11721 entry edge into a region and adjust instantiate_scev calls.
11722 (create_data_ref): Likewise.
11723 (graphite_find_data_references_in_stmt): Likewise.
11724 (find_data_references_in_stmt): Pass the loop preheader edge
11725 from the nest argument.
11726 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11727 parameter the edge into the region.
11728 (instantiate_parameters): Use the loop preheader edge as entry.
11729 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11731 (get_instantiated_value_entry): Make instantiate_below parameter
11732 the edge into the region.
11733 (instantiate_scev_name): Likewise. Adjust dominance checks,
11734 when we cannot use loop-based instantiation instantiate by
11735 walking use-def chains.
11736 (instantiate_scev_poly): Adjust.
11737 (instantiate_scev_binary): Likewise.
11738 (instantiate_scev_convert): Likewise.
11739 (instantiate_scev_not): Likewise.
11740 (instantiate_array_ref): Remove.
11741 (instantiate_scev_3): Likewise.
11742 (instantiate_scev_2): Likewise.
11743 (instantiate_scev_1): Likewise.
11744 (instantiate_scev_r): Do not blindly handle N-operand trees.
11745 Do not instantiate array-refs. Handle all constants and invariants.
11746 (instantiate_scev): Make instantiate_below parameter
11747 the edge into the region.
11748 (resolve_mixers): Use the loop preheader edge for the region
11749 parameter to instantiate_scev_r.
11750 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11752 2017-10-13 Richard Biener <rguenther@suse.de>
11754 PR tree-optimization/82525
11755 * graphite-isl-ast-to-gimple.c
11756 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11758 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11759 Fail code generation when we cannot represent the isl integer.
11760 (binary_op_to_tree): Elide modulo operations that are no-ops
11761 in the type we code generate. Remove now superfluous code
11764 2017-10-13 Richard Biener <rguenther@suse.de>
11766 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11767 (scop_detection::harmful_loop_in_region): Remove premature
11768 IV type restriction.
11769 (scop_detection::graphite_can_represent_scev): We can handle
11770 pointer IVs just fine.
11772 2017-10-13 Alan Modra <amodra@gmail.com>
11774 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11775 "Clobbers and Scratch Registers". Add paragraph on
11776 alternative to clobbers for scratch registers and OpenBLAS
11779 2017-10-13 Alan Modra <amodra@gmail.com>
11781 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11782 example of a memory input for a string of known length. Move
11783 commentary out of table. Add a number of new examples
11784 covering array memory inputs.
11786 2017-10-12 Martin Liska <mliska@suse.cz>
11788 PR tree-optimization/82493
11789 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11790 (test_range_functions): New function.
11791 (sbitmap_c_tests): Likewise.
11792 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11793 * selftest.h (sbitmap_c_tests): New function.
11795 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11797 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11799 * config/rs6000/amo.h: Fix spacing issue.
11801 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11804 * config/i386/i386.md (*ashl<mode>3_mask_1,
11805 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11806 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11809 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11811 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11812 (profile_probability): Set max_probability
11813 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11815 * profile-count.c (profile_probability::differs_from_p): Do not
11816 rely on max_probaiblity == 10000
11818 2017-10-12 Jeff Law <law@redhat.com>
11820 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11823 2017-10-12 Martin Sebor <msebor@redhat.com>
11827 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11828 (handle_alias_pairs): Call it.
11829 * common.opt (-Wattribute-alias): New option.
11830 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11831 * doc/invoke.texi (-Wattribute-alias): Document.
11833 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11836 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11838 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11840 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11841 (make_hard_regno_born, make_hard_regno_dead): Update
11842 bb_killed_pseudos and bb_gen_pseudos.
11844 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11846 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11848 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11850 * config/alpha/alpha.c (alpha_split_conditional_move):
11851 Use std::swap instead of manually swapping.
11852 (alpha_stdarg_optimize_hook): Ditto.
11853 (alpha_canonicalize_comparison): Ditto.
11855 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11857 * tree-loop-distribution.c (struct builtin_info): New struct.
11858 (struct partition): Refactor fields into struct builtin_info.
11859 (partition_free): Free struct builtin_info.
11860 (build_size_arg_loc, build_addr_arg_loc): Delete.
11861 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11862 information from struct builtin_info.
11863 (find_single_drs): New function refactored from classify_partition.
11864 Also moved builtin validity checks to this function.
11865 (compute_access_range, alloc_builtin): New functions.
11866 (classify_builtin_st, classify_builtin_ldst): New functions.
11867 (classify_partition): Refactor code into functions find_single_drs,
11868 classify_builtin_st and classify_builtin_ldst.
11869 (distribute_loop): Don't do runtime alias check when distributing
11871 (find_seed_stmts_for_distribution): New function.
11872 (pass_loop_distribution::execute): Refactor code finding seed
11873 stmts into above function. Support distribution for the innermost
11874 two-level loop nest. Adjust dump information.
11876 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11878 * tree-loop-distribution.c: Adjust the general comment.
11879 (NUM_PARTITION_THRESHOLD): New macro.
11880 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
11881 (classify_partition): Skip builtin pattern of loop nest's inner loop.
11882 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
11883 in call to build_partition_graph.
11884 (finalize_partitions): New parameter. Make loop distribution more
11885 conservative by fusing more partitions.
11886 (distribute_loop): Don't do runtime alias check in case of loop nest
11888 (find_seed_stmts_for_distribution): New function.
11889 (prepare_perfect_loop_nest): New function.
11890 (pass_loop_distribution::execute): Refactor code finding seed stmts
11891 and loop nest into above functions. Support loop nest distribution.
11892 Adjust dump information accordingly.
11894 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11896 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
11897 and set PTYPE_SEQUENTIAL for merged partition.
11899 2017-10-12 Richard Biener <rguenther@suse.de>
11901 PR tree-optimization/69728
11903 2017-09-19 Richard Biener <rguenther@suse.de>
11905 PR tree-optimization/69728
11906 * graphite-sese-to-poly.c (schedule_error): New global.
11907 (add_loop_schedule): Handle empty domain by failing the
11909 (build_original_schedule): Handle schedule_error.
11911 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
11912 domain by returning an unchanged schedule.
11914 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11916 * genrecog.c (validate_pattern): For VEC_SELECT verify that
11917 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
11919 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
11921 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
11924 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11927 * expr.c (store_field): Don't optimize away bitsize == 0 store
11928 from CALL_EXPR with addressable return type.
11930 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11932 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
11933 * config/rs6000/rs6000.md (sel): Delete mode attribute.
11934 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
11935 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
11936 TARGET_ISEL instead of TARGET_ISEL<sel>.
11938 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
11940 * config/rs6000/rs6000.c
11941 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
11943 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11945 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
11947 (reg_or_cint_operand): Fix comment.
11948 (reg_or_zero_operand): New predicate.
11949 * config/rs6000/rs6000-protos.h (output_isel): Delete.
11950 * config/rs6000/rs6000.c (output_isel): Delete.
11951 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
11952 instead of reg_or_cint_operand. Output instruction directly (not via
11954 (isel_unsigned_<mode>): Ditto.
11955 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
11956 gpc_reg_operand. Add an instruction alternative for this. Output
11957 instruction directly.
11958 (*isel_reversed_unsigned_<mode>): Ditto.
11960 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11962 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
11963 (TARGET_CANONICALIZE_COMPARISON): Define.
11965 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
11968 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11969 Check whether the dest is REG before adding REG_EQUIV note.
11971 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11974 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11976 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11977 (make_hard_regno_born, make_hard_regno_dead): Update
11978 bb_killed_pseudos and bb_gen_pseudos.
11980 2017-10-11 Nathan Sidwell <nathan@acm.org>
11982 * incpath.h (enum incpath_kind): Name enum, prefix values.
11983 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
11984 * incpath.c (heads, tails): Use INC_MAX.
11985 (add_env_var_paths, add_standard_paths): Use incpath_kind.
11986 (merge_include_chains, split_quote_chain,
11987 register_include_chains): Update incpath_kind names.
11988 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
11989 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
11991 (add_framework_path, darwin_register_objc_includes): Likewise.
11992 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
11994 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11996 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
11997 Do not use float_operator operator predicate.
11998 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
11999 * config/i386/predicates.md (float_operator): Remove predicate.
12001 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12003 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
12004 (*jccxf_i387): Ditto.
12005 (*jcc<mode>_i387): Ditto.
12006 (*jccu<mode>_i387): Ditto.
12007 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12008 (*jcc_*_i387 splitters): Remove.
12009 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12010 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12011 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12014 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12016 * profile-count.h (slow_safe_scale_64bit): New function.
12017 (safe_scale_64bit): New inline.
12018 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12019 * profile-count.c: Include wide-int.h
12020 (slow_safe_scale_64bit): New.
12022 2017-10-11 Nathan Sidwell <nathan@acm.org>
12024 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12025 HAS_DECL_ASSEMBLER_NAME_P.
12026 * gimple-expr.c (gimple_decl_printable_name: Check
12027 HAS_DECL_ASSEMBLER_NAME_P too.
12028 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12029 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12030 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12031 * passes.c (rest_of_decl_compilation): Check
12032 HAS_DECL_ASSEMBLER_NAME_P too.
12033 * recog.c (verify_changes): Likewise.
12034 * tree-pretty-print.c (dump_decl_name): Likewise.
12035 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12037 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12038 (SET_DECL_ASSEMBLER_NAME): Use it.
12039 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12040 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12041 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
12043 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12045 * config.gcc (i386, x86_64): Add extra objects.
12046 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12047 (ix86_min_insn_size): Declare.
12048 (ix86_issue_rate): Declare.
12049 (ix86_adjust_cost): Declare.
12050 (ia32_multipass_dfa_lookahead): Declare.
12051 (ix86_macro_fusion_p): Declare.
12052 (ix86_macro_fusion_pair_p): Declare.
12053 (ix86_bd_has_dispatch): Declare.
12054 (ix86_bd_do_dispatch): Declare.
12055 (ix86_core2i7_init_hooks): Declare.
12056 (ix86_atom_sched_reorder): Declare.
12057 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12058 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12059 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12060 (rip_relative_addr_p): Rename to ...
12061 (ix86_rip_relative_addr_p): ... this one; export.
12062 (memory_address_length): Update.
12063 (ix86_issue_rate): Move to x86-tune-sched.c.
12064 (ix86_flags_dependent): Move to x86-tune-sched.c.
12065 (ix86_agi_dependent): Move to x86-tune-sched.c.
12066 (exact_dependency_1): Move to x86-tune-sched.c.
12067 (exact_store_load_dependency): Move to x86-tune-sched.c.
12068 (ix86_adjust_cost): Move to x86-tune-sched.c.
12069 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12070 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12071 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12072 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12073 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12074 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12075 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12076 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12077 (min_insn_size): Rename to ...
12078 (ix86_min_insn_size): ... this one; export.
12079 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12080 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12081 (core2i7_first_cycle_multipass_backtrack): Move to
12082 x86-tune-sched-core.c.
12083 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12084 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12085 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12086 (ix86_avoid_jump_mispredicts): Update.
12087 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12088 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12089 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12090 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12091 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12092 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12093 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12094 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12095 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12096 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12097 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12098 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12099 (BIG): Move to ix86-tune-sched-bd.c.
12100 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12101 (enum insn_path): Move to ix86-tune-sched-bd.c.
12102 (get_mem_group): Move to ix86-tune-sched-bd.c.
12103 (is_cmp): Move to ix86-tune-sched-bd.c.
12104 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12105 (is_branch): Move to ix86-tune-sched-bd.c.
12106 (is_prefetch): Move to ix86-tune-sched-bd.c.
12107 (init_window): Move to ix86-tune-sched-bd.c.
12108 (allocate_window): Move to ix86-tune-sched-bd.c.
12109 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12110 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
12111 (process_end_window): Move to ix86-tune-sched-bd.c.
12112 (allocate_next_window): Move to ix86-tune-sched-bd.c.
12113 (find_constant): Move to ix86-tune-sched-bd.c.
12114 (get_num_immediates): Move to ix86-tune-sched-bd.c.
12115 (has_immediate): Move to ix86-tune-sched-bd.c.
12116 (get_insn_path): Move to ix86-tune-sched-bd.c.
12117 (get_insn_group): Move to ix86-tune-sched-bd.c.
12118 (count_num_restricted): Move to ix86-tune-sched-bd.c.
12119 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
12120 (add_insn_window): Move to ix86-tune-sched-bd.c.
12121 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
12122 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
12123 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
12124 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
12125 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
12126 (do_dispatch): Move to ix86-tune-sched-bd.c.
12127 (has_dispatch): Move to ix86-tune-sched-bd.c.
12128 * config/i386/t-i386: Add new object files.
12129 * config/i386/x86-tune-costs.h: New file.
12130 * config/i386/x86-tune-sched-atom.c: New file.
12131 * config/i386/x86-tune-sched-bd.c: New file.
12132 * config/i386/x86-tune-sched-core.c: New file.
12133 * config/i386/x86-tune-sched.c: New file.
12135 2017-10-11 Liu Hao <lh_mouse@126.com>
12137 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
12138 the generic version below instead.
12139 (should_colorize): Recognize Windows consoles as terminals
12141 * pretty-print.c [__MINGW32__] (write_all): New function.
12142 [__MINGW32__] (find_esc_head): Likewise.
12143 [__MINGW32__] (find_esc_terminator): Likewise.
12144 [__MINGW32__] (eat_esc_sequence): Likewise.
12145 [__MINGW32__] (mingw_ansi_fputs): New function that handles
12147 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
12150 2017-10-11 Richard Biener <rguenther@suse.de>
12152 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
12153 Properly call analyze_scalar_evolution with the loop of the stmt.
12155 2017-10-11 Richard Biener <rguenther@suse.de>
12157 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
12158 * tree-core.h (tree_base): Add chrec_var union member.
12159 * tree.h (CHREC_VAR): Remove.
12160 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
12161 * tree-chrec.h (build_polynomial_chrec): Adjust.
12162 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
12163 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
12165 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
12167 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
12168 * match.pd: ... here.
12169 ((T) X == (T) Y): Relax condition.
12171 2017-10-11 Bin Cheng <bin.cheng@arm.com>
12173 PR tree-optimization/82472
12174 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
12176 (break_alias_scc_partitions): Update postorder number.
12178 2017-10-11 Martin Liska <mliska@suse.cz>
12181 * opts.c (parse_no_sanitize_attribute): Do not use error_value
12183 * opts.h (parse_no_sanitize_attribute): Remove last argument.
12185 2017-10-11 Martin Liska <mliska@suse.cz>
12187 * print-rtl.c (print_insn): Move declaration of idbuf
12188 to same scope as name.
12190 2017-10-11 Martin Liska <mliska@suse.cz>
12195 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12196 volatile arguments.
12198 2017-10-11 Martin Liska <mliska@suse.cz>
12201 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12202 volatile arguments.
12204 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
12206 * config.gcc (default_gnu_indirect_function): Default to yes for
12207 arm*-*-linux* with glibc.
12209 2017-10-11 Richard Biener <rguenther@suse.de>
12211 * tree-scalar-evolution.c (get_scalar_evolution): Handle
12212 default-defs and types we do not want to analyze.
12213 (interpret_loop_phi): Replace unreachable code with an assert.
12214 (compute_scalar_evolution_in_loop): Remove and inline ...
12215 (analyze_scalar_evolution_1): ... here, replacing condition with
12216 what makes the intent clearer. Remove handling of cases
12217 get_scalar_evolution now handles.
12219 2017-10-10 Jim Wilson <wilson@tuliptree.org>
12221 PR rtl-optimization/81434
12222 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
12223 comment for main loop. In sched_group_found if, also add checks for
12224 pass and min_cost_group.
12226 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
12228 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
12229 (rs6000_insn_cost): New function.
12230 * config/rs6000/rs6000.md (cost): New attribute.
12232 2017-10-10 Jakub Jelinek <jakub@redhat.com>
12233 H.J. Lu <hongjiu.lu@intel.com>
12237 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
12238 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
12239 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
12240 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
12241 to OPTION_MASK_ISA_AVX512VL - builtins that have both
12242 OPTION_MASK_ISA_MMX and some other bit set require both
12243 mmx and the ISAs without the mmx bit.
12244 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
12245 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
12246 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
12247 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
12248 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
12249 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
12250 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
12251 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
12252 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
12253 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
12254 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
12255 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
12256 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
12257 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
12258 Add OPTION_MASK_ISA_MMX.
12260 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
12262 * config.gcc (armv7*-*-freebsd*): New target.
12263 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
12265 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
12267 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
12268 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
12271 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
12273 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
12274 a template parameter.
12275 (WIDE_INT_REF_FOR): Update accordingly.
12276 * tree.h (wi::int_traits <const_tree>): Delete.
12277 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
12278 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
12279 (wi::tree_to_wide_ref): New typedef.
12280 (wi::to_wide): New function.
12281 * calls.c (get_size_range): Use wi::to_wide when operating on
12282 trees as wide_ints.
12283 * cgraph.c (cgraph_node::create_thunk): Likewise.
12284 * config/i386/i386.c (ix86_data_alignment): Likewise.
12285 (ix86_local_alignment): Likewise.
12286 * dbxout.c (stabstr_O): Likewise.
12287 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
12288 * expr.c (const_vector_from_tree): Likewise.
12289 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
12290 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
12291 (fold_negate_expr_1, int_const_binop_1, const_binop)
12292 (fold_convert_const_int_from_real, optimize_bit_field_compare)
12293 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
12294 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
12295 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
12296 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
12297 (fold_not_const, round_up_loc): Likewise.
12298 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
12299 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
12300 (alloca_call_type): Likewise.
12301 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
12302 * godump.c (go_output_typedef): Likewise.
12303 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
12304 * internal-fn.c (get_min_precision): Likewise.
12305 * ipa-cp.c (ipcp_store_vr_results): Likewise.
12306 * ipa-polymorphic-call.c
12307 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
12308 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
12309 (ipa_modify_call_arguments): Likewise.
12310 * match.pd: Likewise.
12311 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
12312 * print-tree.c (print_node_brief, print_node): Likewise.
12313 * stmt.c (expand_case): Likewise.
12314 * stor-layout.c (layout_type): Likewise.
12315 * tree-affine.c (tree_to_aff_combination): Likewise.
12316 * tree-cfg.c (group_case_labels_stmt): Likewise.
12317 * tree-data-ref.c (dr_analyze_indices): Likewise.
12318 (prune_runtime_alias_test_list): Likewise.
12319 * tree-dump.c (dequeue_and_dump): Likewise.
12320 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
12321 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
12322 * tree-pretty-print.c (dump_generic_node): Likewise.
12323 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
12324 (simple_iv_with_niters): Likewise.
12325 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
12326 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
12327 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
12328 * tree-ssa-loop-niter.c (split_to_var_and_offset)
12329 (refine_value_range_using_guard, number_of_iterations_ne_max)
12330 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
12331 (get_cst_init_from_scev, record_nonwrapping_iv)
12332 (scev_var_range_cant_overflow): Likewise.
12333 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
12334 * tree-ssa-pre.c (compute_avail): Likewise.
12335 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
12336 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
12337 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
12338 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
12339 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
12340 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
12341 (dump_case_nodes, try_switch_expansion): Likewise.
12342 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
12343 (vect_do_peeling): Likewise.
12344 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
12345 * tree-vect-stmts.c (vectorizable_load): Likewise.
12346 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
12347 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
12348 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
12349 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
12350 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
12351 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
12352 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
12353 (evrp_dom_walker::before_dom_children): Likewise.
12354 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
12355 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
12356 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
12357 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
12358 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
12359 (get_range_pos_neg): Likewise.
12360 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
12361 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
12362 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
12363 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
12364 * config/avr/avr.c (avr_fold_builtin): Likewise.
12365 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
12366 * config/msp430/msp430.c (msp430_attr): Likewise.
12367 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
12368 * config/powerpcspe/powerpcspe-c.c
12369 (altivec_resolve_overloaded_builtin): Likewise.
12370 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
12371 (rs6000_expand_ternop_builtin): Likewise.
12372 * config/rs6000/rs6000-c.c
12373 (altivec_resolve_overloaded_builtin): Likewise.
12374 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
12375 (rs6000_expand_ternop_builtin): Likewise.
12376 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
12378 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12380 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
12381 when copying loop nest with only one inner loop.
12383 2017-10-10 Richard Biener <rguenther@suse.de>
12385 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
12386 blocks if SCEV is active.
12387 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
12389 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
12390 (scev_initialize): Assert we are not yet initialized.
12392 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12394 * tree-loop-distribution.c (generate_loops_for_partition): Remove
12395 inner loop's exit stmt by making it always exit the loop, otherwise
12396 we would generate an infinite empty loop.
12398 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12400 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
12401 renaming variables in new preheader if it's deleted.
12403 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12405 * tree-loop-distribution.c (struct partition): Remove unused field
12406 loops of the structure.
12407 (partition_alloc, partition_free): Ditto.
12408 (build_rdg_partition_for_vertex): Ditto.
12410 2017-10-09 Jeff Law <law@redhat.com>
12412 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
12413 return type to match prototype and documentation.
12415 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12417 * config/rs6000/rs6000.c (processor_costs): Move to ...
12418 * config/rs6000/rs6000.h: ... here.
12419 (rs6000_cost): Declare.
12421 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
12423 * except.c (setjmp_fn): New global variable.
12424 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
12425 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
12426 if DONT_USE_BUILTIN_SETJMP is defined.
12428 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12430 * target.def (insn_cost): New hook.
12431 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
12432 * doc/tm.texi: Regenerate.
12433 * rtlanal.c (insn_cost): Use the new hook.
12435 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12437 * combine.c (combine_validate_cost): Compute the new insn_cost,
12438 not just pattern_cost.
12439 (try_combine): Adjust comment.
12441 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12443 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
12445 * combine.c (uid_insn_cost): Adjust comment.
12446 (combine_validate_cost): Adjust comment. Use pattern_cost instead
12448 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
12449 * dse.c (find_shift_sequence): Ditto.
12450 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
12451 (bb_valid_for_noce_process_p): Use pattern_cost.
12452 * rtl.h (insn_rtx_cost): Delete.
12453 (pattern_cost): New prototype.
12454 (insn_cost): New prototype.
12455 * rtlanal.c (insn_rtx_cost): Rename to...
12456 (pattern_cost): ... this.
12459 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
12461 * config/i386/i386.md (*jcc_2): Remove insn pattern.
12462 (*jcc<mode>_0_r_i387): Ditto.
12463 (*jccxf_r_i387): Ditto.
12464 (*jcc<mode>_r_i387): Ditto.
12465 (*jccu<mode>_r_i387): Ditto.
12466 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
12467 (*jcc): Rename from *jcc_1.
12469 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12471 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
12472 deferred rescans after the lvx/stvx recombination pre-pass.
12474 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
12476 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
12477 memory operation instruction support.
12478 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
12479 (rs6000-ibm-aix[789]*): Likewise.
12480 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
12481 Document new functions.
12483 2017-10-09 Richard Biener <rguenther@suse.de>
12485 PR tree-optimization/82397
12486 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
12487 equality only for semantically equal trees.
12489 2017-10-09 Richard Biener <rguenther@suse.de>
12491 PR tree-optimization/82449
12492 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
12493 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
12494 allow constant addresses.
12495 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
12498 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12500 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
12503 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12506 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
12509 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12512 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
12514 2017-10-09 Jakub Jelinek <jakub@redhat.com>
12517 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
12518 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
12520 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
12522 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
12523 (WI_BINARY_PREDICATE_RESULT): Likewise.
12524 (wi::binary_traits::operator_result): New type.
12525 (wi::binary_traits::predicate_result): Likewise.
12526 (generic_wide_int::operator~, unary generic_wide_int::operator-)
12527 (generic_wide_int::operator==, generic_wide_int::operator!=)
12528 (generic_wide_int::operator&, generic_wide_int::and_not)
12529 (generic_wide_int::operator|, generic_wide_int::or_not)
12530 (generic_wide_int::operator^, generic_wide_int::operator+
12531 (binary generic_wide_int::operator-, generic_wide_int::operator*):
12533 (operator~, unary operator-, operator==, operator!=, operator&)
12534 (operator|, operator^, operator+, binary operator-, operator*): New
12536 * expr.c (get_inner_reference): Use wi::bit_and_not.
12537 * fold-const.c (fold_binary_loc): Likewise.
12538 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
12539 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
12540 (bit_value_binop): Likewise.
12541 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
12542 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
12543 (extract_range_from_binary_expr_1): Likewise.
12544 (masked_increment): Likewise.
12545 (simplify_bit_ops_using_ranges): Likewise.
12547 2017-10-09 Martin Jambor <mjambor@suse.cz>
12550 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
12551 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
12552 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
12554 (hsa_fixup_mov_insn_type): New function.
12555 (hsa_op_with_type::get_in_type): Use it.
12556 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
12557 immediates in an assert.
12558 (hsa_op_with_type::extend_int_to_32bit): New method.
12559 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
12560 types. Convert to dest type if necessary.
12561 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
12562 (reg_for_gimple_ssa): Pass false as min32int to
12563 hsa_type_for_scalar_tree_type.
12564 (gen_hsa_addr): Fixup type when creating addresable temporary.
12565 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
12566 (gen_hsa_unary_operation): Extend operands and convert to dest type if
12567 necessary. Call hsa_fixup_mov_insn_type.
12568 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
12569 extend operands and convert to dest type if necessary.
12570 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
12571 to dest type if necessary.
12572 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
12573 if conversion nt necessary and size matches.
12574 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
12575 to dest type if necessary.
12576 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
12577 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
12579 (gen_hsa_clrsb): Likewise.
12580 (gen_hsa_ffs): Likewise.
12581 (gen_hsa_divmod): Extend operands and convert to dest type if
12583 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
12585 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12587 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
12588 Remove empty default arguments. Use a brace block as output
12590 (conditional return): Ditto.
12592 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
12593 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
12595 (group_ending_nop): Ditto.
12596 (doloop_end): Ditto.
12597 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
12598 (splitters for those): Ditto.
12600 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12602 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
12603 a conditional jump (and the compare for it) so that pc_rtx is the
12605 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
12606 for the deleted and renamed ctr<mode>_internal[234] patterns.
12607 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
12608 Delete second conditional return pattern.
12609 (ctr<mode>_internal2): Delete this second bdnz pattern.
12610 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
12611 (ctr<mode>_internal4): Delete this second bdz pattern.
12613 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12615 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
12616 (always_initialized_rtx_for_ssa_name_p): New predicate.
12617 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
12618 (finish_out_of_ssa): Free new field of SA.
12619 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
12620 * tree-ssa-coalesce.c: Include tree-ssa.h.
12621 (get_parm_default_def_partitions): Remove extern keyword.
12622 (get_undefined_value_partitions): New function.
12623 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12624 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12625 uninitialized bits.
12626 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12628 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12630 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12632 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12634 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12635 for targets that preffer 128bit.
12637 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12639 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12641 2017-10-08 Olivier Hainque <hainque@adacore.com>
12643 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12644 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12646 (thumb_set_return_address): Likewise.
12648 2017-10-08 Olivier Hainque <hainque@adacore.com>
12650 * common/config/arm/arm-common.c (arm_except_unwind_info):
12651 Handle DWARF2_UNWIND_INFO.
12653 2017-10-07 Michael Collison <michael.collison@arm.com>
12655 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12658 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12660 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12661 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12662 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12663 defined, force the creation of a new block for a dispatch label.
12665 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12667 * invoke.texi (Wsuggest-attribute=cold): Document.
12668 * common.opt (Wsuggest-attribute=cold): New
12669 * ipa-pure-const.c (warn_function_cold): New function.
12670 * predict.c (compute_function_frequency): Use it.
12671 * predict.h (warn_function_cold): Declare.
12673 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12675 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12678 2017-10-06 Martin Liska <mliska@suse.cz>
12680 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12681 keyword for member functions.
12682 (struct sanopt_tree_couple): New struct.
12683 (struct sanopt_tree_couple_hash): New function.
12684 (struct sanopt_ctx): Add new hash_map.
12685 (has_dominating_ubsan_ptr_check): New function.
12686 (record_ubsan_ptr_check_stmt): Likewise.
12687 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12688 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12689 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12691 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12694 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12695 aarch64_simd_valid_immediate on CONST_VECTORs.
12696 (aarch64_reg_or_bic_imm): Likewise.
12698 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12700 PR rtl-optimization/82396
12701 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12703 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12705 * graphite-dependences.c (scop_get_reads): Move code to...
12706 (scop_get_must_writes): Move code to...
12707 (scop_get_may_writes): Move code to...
12708 (scop_get_reads_and_writes): ... here.
12709 (scop_get_dependences): Call scop_get_reads_and_writes.
12711 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12713 PR tree-optimization/82434
12714 * fold-const.h (can_native_encode_type_p,
12715 can_native_encode_string_p): Remove.
12716 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12717 don't encode anything, just return what would be otherwise returned.
12718 (native_encode_fixed, native_encode_complex, native_encode_vector):
12720 (native_encode_string): Likewise. Inline by hand
12721 can_native_encode_string_p.
12722 (can_native_encode_type_p): Remove.
12723 (can_native_encode_string_p): Remove.
12724 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12725 STRING_CSTs using can_native_encode_string_p, test all
12726 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12727 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12728 argument from native_encode_expr.
12729 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12730 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12733 2017-10-06 Richard Biener <rguenther@suse.de>
12735 PR tree-optimization/82397
12736 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12737 operand_equal_p but rely on data_ref_compare_tree for detecting
12739 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12740 to match up with dr_group_sort_cmp.
12742 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12745 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12747 * config/s390/s390-builtin-types.def: Regenerate.
12749 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12752 * config/s390/s390-builtin-types.def: Regenerate.
12753 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12754 Change flag from B_VXE to B_VX.
12755 (s390_vec_min_dbl): Remove B_VXE flag.
12757 2017-10-06 Richard Biener <rguenther@suse.de>
12759 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12760 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12761 translate_isl_ast_to_gimple::is_valid_rename,
12762 translate_isl_ast_to_gimple::get_rename,
12763 translate_isl_ast_to_gimple::get_def_bb_for_const,
12764 translate_isl_ast_to_gimple::get_new_name,
12765 translate_isl_ast_to_gimple::collect_all_ssa_names,
12766 translate_isl_ast_to_gimple::copy_loop_phi_args,
12767 translate_isl_ast_to_gimple::collect_all_ssa_names,
12768 translate_isl_ast_to_gimple::copy_loop_phi_args,
12769 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12770 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12771 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12772 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12773 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12774 translate_isl_ast_to_gimple::copy_cond_phi_args,
12775 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12776 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12777 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12778 translate_isl_ast_to_gimple::rename_uses,
12779 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12780 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12781 (set_rename_for_each_def): Likewise.
12782 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12783 here. Handle rewriting SCEV analyzable uses here.
12784 (copy_bb_and_scalar_dependences): Generate code for PHI
12786 (graphite_regenerate_ast_isl): Adjust.
12787 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12788 (add_write, add_read): New functions.
12789 (build_cross_bb_scalars_def): Use it and simplify.
12790 (build_cross_bb_scalars_use): Likewise.
12791 (graphite_find_cross_bb_scalar_vars): Inline into...
12792 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12793 simulating out-of-SSA. Compute liveout and add dependencies.
12794 (build_scops): Force an empty entry block.
12795 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12797 (sese_build_liveouts): Declare.
12798 (sese_trivially_empty_bb_p): Likewise.
12799 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12800 compute liveout and debug_liveout.
12801 (sese_bad_liveouts_use): Remove.
12802 (sese_reset_debug_liveouts_bb): Likewise.
12803 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12804 (sese_build_liveouts): Build liveout and debug_liveout and store
12806 (new_sese_info): Adjust.
12807 (free_sese_info): Likewise.
12808 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12809 do not build liveout here.
12810 (move_sese_in_condition): Adjust region entry.
12811 (scev_analyzable_p): Match up with chrec_apply requirements.
12812 (sese_trivially_empty_bb_p): New.
12813 * tree-into-ssa.c (get_reaching_def): Properly support generating
12814 default-defs for incremental rewrite of anonymous names.
12816 2017-10-06 Richard Biener <rguenther@suse.de>
12818 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12819 precision do not perform modulo reduction.
12821 2017-10-06 Richard Biener <rguenther@suse.de>
12823 PR tree-optimization/82436
12824 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12825 conservatively choose the vectorization factor when checking
12826 whether we can perform the required load permutation.
12827 (vect_transform_slp_perm_load): Assert when we may not fail.
12829 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12831 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12832 message for incompatible -msdata=* and -mcall-* options.
12834 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12836 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12837 rate for post-reload scheduling.
12839 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12841 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12843 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12845 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12846 to improve monte carlo in scimark.
12848 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12850 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12851 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12852 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12853 pentium4_cost, nocona_cost): Set reassociation width to 1.
12854 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12855 width to 2 for fp operations and 1 otherwise.
12856 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12858 (atom_cost): Set reassociation width to 2.
12859 (slm_cost, generic_cost): Set fp reassociation width
12860 to 2 and 1 otherwise.
12861 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12862 (core_cost): Set fp reassociation width to 4 and vector to 2.
12863 (ix86_reassociation_width): Rewrite using cost table; special case
12864 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12865 and TARGET_AVX128_OPTIMAL.
12866 * config/i386/i386.h (processor_costs): Add
12867 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
12868 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
12869 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
12870 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
12871 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
12872 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
12874 2017-10-05 Nathan Sidwell <nathan@acm.org>
12876 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
12878 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12880 * config/arm/arm.c (arm_test_fpu_data): New.
12881 (arm_run_selftests): Call arm_test_fpu_data.
12883 2017-10-04 Nathan Sidwell <nathan@acm.org>
12885 * toplev.c (toplev::main): Remove excess parens on pretty_printer
12887 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
12889 2017-10-04 Sudakshina Das <sudi.das@arm.com>
12891 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
12892 check type for aarch64_simd_valid_immediate.
12893 (aarch64_output_simd_mov_immediate): Update prototype.
12894 (aarch64_simd_valid_immediate): Update prototype.
12895 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
12896 support for ORR-immediate.
12897 (and<mode>3): modified pattern to add support for BIC-immediate.
12898 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
12899 now checks for valid immediate for BIC and ORR based on new enum
12901 (aarch64_output_simd_mov_immediate): Function now used to output
12902 BIC/ORR imm as well based on new enum argument.
12903 * config/aarch64/constraints.md (Do): New vector immediate constraint.
12905 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
12906 (aarch64_reg_or_bic_imm): Likewise.
12908 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12910 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
12911 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
12913 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12917 PR rtl-optimization/82396
12918 * haifa-sched.c (autopref_multipass_init): Simplify
12920 (autopref_rank_data): Simplify sort order.
12921 * sched-int.h (autopref_multipass_data_): Remove
12922 multi_mem_insn_p, min_offset and max_offset.
12924 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12926 * doc/sourcebuild.texi: Document vect_peeling_profitable.
12928 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12930 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
12931 vect_doubleint_cvt.
12933 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12935 * doc/sourcebuild.texi: Document vect_long_mult.
12937 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
12939 PR tree-optimization/82413
12940 * fold-const.c (build_range_check): Use widest_int when comparing
12941 the maximum ETYPE value with HIGH.
12943 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12945 PR rtl-optimization/82396
12946 * haifa-sched.c (autopref_multipass_init): Simplify
12948 (autopref_rank_data): Simplify sort order.
12949 * sched-int.h (autopref_multipass_data_): Remove
12950 multi_mem_insn_p, min_offset and max_offset.
12952 2017-10-04 Jakub Jelinek <jakub@redhat.com>
12954 PR tree-optimization/82381
12955 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
12956 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
12959 PR tree-optimization/82374
12960 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
12961 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
12962 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
12963 current_function_decl to the new decl.
12965 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
12967 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
12968 helper macro for IEEE float128 hardware built-in functions.
12969 (SQRTF128_ODD): Add built-in functions with the round-to-odd
12971 (TRUNCF128_ODD): Likewise.
12972 (ADDF128_ODD): Likewise.
12973 (SUBF128_ODD): Likewise.
12974 (MULF128_ODD): Likewise.
12975 (DIVF128_ODD): Likewise.
12976 (FMAF128_ODD): Likewise.
12977 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
12978 UNSPEC_TRUNC_ROUND_TO_ODD.
12979 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
12980 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
12981 floating point round to odd instructions.
12982 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
12983 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
12984 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
12985 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
12986 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
12987 (trunc<mode>sf2_hw): Change the truncate with round to odd
12988 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
12989 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
12990 to odd hardware instructions.
12991 (sub<mode>3_odd): Likewise.
12992 (mul<mode>3_odd): Likewise.
12993 (div<mode>3_odd): Likewise.
12994 (sqrt<mode>2_odd): Likewise.
12995 (fma<mode>4_odd): Likewise.
12996 (fms<mode>4_odd): Likewise.
12997 (nfma<mode>4_odd): Likewise.
12998 (nfms<mode>4_odd): Likewise.
12999 (trunc<mode>df2_odd): Change the truncate with round to odd
13000 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
13002 * doc/extend.texi (PowerPC built-in functions): Update documentation
13003 for existing IEEE float128-bit built-in functions. Add built-in
13004 functions that generate the IEEE 128-bit floating point round to
13007 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13009 PR rtl-optimization/77729
13010 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13011 to (X&(C1&~C2))|C2 transformations.
13013 2017-10-03 Martin Jambor <mjambor@suse.cz>
13015 PR tree-optimization/82363
13016 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13017 mismatch, mark lacc written regardless of racc.
13019 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13021 PR tree-optimization/82381
13022 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13023 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13024 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13025 fallthrough into reassoc_stmt_dominates_stmt_p.
13028 * combine.c (combine_instructions): Don't combine in unreachable
13031 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13034 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13035 function to not use the have_cpu variable. Do not set cpu_index,
13036 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13037 or the default cpu.
13038 (rs6000_valid_attribute_p): Remove duplicate initializations of
13039 old_optimize and func_optimize.
13040 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13041 (rs6000_activate_target_options): Make global.
13042 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
13045 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13047 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13048 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13049 Set *poffset to 0 and *psize and *pmax_size to -1 if
13050 *poffset + *psize overflows in HOST_WIDE_INT.
13052 PR tree-optimization/82387
13053 PR tree-optimization/82388
13054 PR tree-optimization/82389
13055 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13056 instead of live_bytes non-NULL.
13058 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13061 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13064 2017-10-02 Richard Biener <rguenther@suse.de>
13066 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13067 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13068 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13070 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13072 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13073 requested precision matches the type's.
13074 * calls.c (alloc_max_size): Calculate the new candidate size as
13075 a widest_int and use wi::to_widest when comparing it with the
13076 current candidate size.
13077 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13078 zero rather than integer_zero_node.
13079 * match.pd: Check for a no-op conversion before using wi::add
13080 rather than after. Use tree_to_uhwi when summing small shift
13081 counts into an unsigned int.
13083 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13084 Alan Hayward <alan.hayward@arm.com>
13085 David Sherwood <david.sherwood@arm.com>
13088 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13089 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13090 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13091 POINTER_AND_FP_REGS.
13093 2017-10-02 Richard Biener <rguenther@suse.de>
13095 PR tree-optimization/82355
13096 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13097 a mapping for the enclosing loop but avoid generating one for
13098 the loop tree root.
13099 (copy_bb_and_scalar_dependences): Remove premature codegen
13100 error on PHIs in blocks duplicated into multiple places.
13101 * graphite-scop-detection.c
13102 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13103 in the region use it as loop and nest to analyze the DR in.
13104 (try_generate_gimple_bb): Likewise.
13105 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13106 (add_loop_constraints): For blocks in a loop not in the region
13107 create a dimension with a single iteration.
13108 * sese.h (gbb_loop_at_index): Remove assert.
13110 2017-10-01 Kevin Buettner <kevinb@redhat.com>
13112 * omp-expand.c (adjust_context_scope): New function.
13113 (expand_parallel_call): Call adjust_context_scope.
13115 2017-10-01 Jeff Law <law@redhat.com>
13117 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
13118 dom_opt_dom_walker class with direct access to private members.
13119 Add comments. Call test_for_singularity.
13120 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
13121 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
13122 m_dummy_cond anymore.
13123 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
13125 (pass_dominator:execute): Build the dummy_cond here and pass it
13126 to the dom_opt_dom_walker ctor.
13127 (test_for_singularity): New function.
13129 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
13130 Maya Rashish <coypu@sdf.org>
13132 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
13133 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
13134 (alpha*-*-netbsd*) Use nbsd_tm_file.
13135 (arm*-*-netbsdelf*) Likewise.
13136 (i[34567]86-*-netbsdelf*) Likewise.
13137 (x86_64-*-netbsd*) Likewise.
13138 (mips*-*-netbsd*) Likewise.
13139 (powerpc-*-netbsd*) Likewise.
13140 (sh*-*-netbsd*) Likewise.
13141 (sparc-*-netbsdelf*) Likewise.
13142 (sparc64-*-netbsd*) Likewise.
13143 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
13145 (vax-*-netbsdelf*) Likewise.
13146 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
13147 (UINT_FAST8_TYPE) Likewise.
13148 (INT_FAST16_TYPE) Check CHAR_FAST16.
13149 (UINT_FAST16_TYPE) Likewise.
13151 2017-09-30 Jakub Jelinek <jakub@redhat.com>
13154 * config/i386/i386.md
13155 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
13156 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
13157 *divmodsi4_zext_2): New define_insn_and_split.
13158 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
13159 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
13160 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
13161 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
13162 New define_insn_and_split.
13163 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
13164 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
13165 operands[1] having DImode when mode is SImode.
13167 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
13168 always SImode for DIV and MOD in REG_EQUAL notes.
13170 2017-09-29 Yury Gribov <tetra2005@gmail.com>
13172 PR middle-end/82319
13173 * match.pd: Fix handling of NaNs in pattern.
13175 2017-09-29 Jeff Law <law@redhat.com>
13177 * sbitmap.c (bitmap_bit_in_range_p): New function.
13178 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
13179 * tree-ssa-dse.c (live_bytes_read): New function.
13180 (dse_classify_store): Ignore reads of dead bytes.
13182 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
13183 typos and whitespace errors.
13184 * config/i386/predicates.md (address_no_seg_operand): Likewise.
13185 * config/s390/s390.c (s390_emit_prologue): Likewise.
13187 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13190 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
13191 with a symbol for LRA.
13193 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13195 PR rtl-optimization/82338
13196 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
13198 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
13200 * genmodes.c (calc_wider_mode): Suppress qsort macro.
13201 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
13202 (qsort_chk): Declare.
13203 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
13204 (qsort_chk): New function.
13206 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13208 PR tree-optimization/82337
13209 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
13210 phi definition if the PHI result appears in an abnormal PHI.
13211 (find_basis_for_base_expr): Don't record a basis if the LHS of the
13212 basis appears in an abnormal PHI.
13214 2017-09-29 Richard Biener <rguenther@suse.de>
13216 * graphite-isl-ast-to-gimple.c
13217 (translate_isl_ast_to_gimple::set_codegen_error): New function.
13218 (binary_op_to_tree): Use it.
13219 (get_rename_from_scev): Likewise.
13220 (copy_loop_phi_nodes): Likewise.
13221 (copy_bb_and_scalar_dependences): Likewise.
13222 (translate_pending_phi_nodes): Likewise.
13224 2017-09-29 Jakub Jelinek <jakub@redhat.com>
13227 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
13228 for movabsq $(i32 << shift), r64.
13230 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
13232 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
13233 index when encoding %esp as %rsp to avoid 0x67 prefix.
13235 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
13237 * config/i386/i386.md (*movsf_internal, *movdf_internal):
13238 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
13240 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
13242 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
13243 Extensions with more than 16 double VFP registers.
13244 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
13245 to_clear_mask and all code related to it. Replace the remaining
13246 entry by a sbitmap and adapt code accordingly.
13248 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
13250 * brig-builtins.def: Change pure attributes to const.
13252 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13254 * config.gcc (default_gnu_indirect_function): Default to yes for
13255 sparc*-*-linux* with glibc.
13257 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13259 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
13260 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
13261 when creating .init_array and .fini_array sections with priority
13264 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
13267 * config/aarch64/aarch64.c
13268 (aarch64_builtin_support_vector_misalignment): Always return false
13269 when misalignment is unknown.
13271 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
13273 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
13274 this function to return false if the definition used by the swap
13275 instruction is artificial, or if the memory address from which the
13276 constant value is loaded is not represented by a base address held
13277 in a register or if the base address register is a frame or stack
13278 pointer. Additionally, return false if the base address of the
13279 loaded constant is a SYMBOL_REF but is not considered to be a
13281 (replace_swapped_load_constant): New function.
13282 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
13283 loaded constant vector with a load of a swapped constant vector.
13285 2017-09-27 Carl Love <cel@us.ibm.com>
13287 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
13288 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
13289 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
13292 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
13294 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
13295 first, always call autopref_rank_data otherwise.
13297 2017-09-27 Richard Biener <rguenther@suse.de>
13299 * graphite-scop-detection.c (find_scop_parameters): Move
13300 loop bound handling ...
13301 (gather_bbs::before_dom_children): ... here, avoiding the need
13302 to build scop_info->loop_nest.
13303 (record_loop_in_sese): Remove.
13304 * sese.h (sese_info_t::loop_nest): Remove.
13305 * sese.c (new_sese_info): Do not allocate loop_nest.
13306 (free_sese_info): Do not free loop_nest.
13308 2017-09-27 Jakub Jelinek <jakub@redhat.com>
13311 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
13312 lhs from calls if the lhs has addressable type.
13314 2017-09-27 Richard Biener <rguenther@suse.de>
13316 * graphite.h (scop::max_alias_set): New member.
13317 * graphite-scop-detection.c: Remove references to non-existing
13318 --param in comments.
13319 (build_alias_sets): Record the maximum alias set used for drs.
13320 (build_scops): Support zero as unlimited for
13321 --param graphite-max-arrays-per-scop.
13322 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
13323 and inline into ...
13324 (build_poly_sr_1): ... here. Compute alias set based on the
13325 maximum alias set used for drs rather than
13326 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
13328 2017-09-27 Richard Biener <rguenther@suse.de>
13330 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
13331 --param loop-block-tile-size=0 to disable tiling.
13333 2017-09-27 Richard Biener <rguenther@suse.de>
13335 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
13336 (graphite-max-nb-scop-params): Document special value zero.
13337 * domwalk.h (dom_walker::STOP): New symbolical constant.
13338 (dom_walker::dom_walker): Add optional parameter for bb to
13340 (dom_walker::~dom_walker): Declare.
13341 (dom_walker::before_dom_children): Document STOP return value.
13342 (dom_walker::m_user_bb_to_rpo): New member.
13343 (dom_walker::m_bb_to_rpo): Likewise.
13344 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
13345 mapping here if not provided by the user.
13346 (dom_walker::~dom_walker): Free bb to RPO mapping if not
13347 provided by the user.
13348 (dom_walker::STOP): Define.
13349 (dom_walker::walk): Do not compute bb to RPO mapping here.
13350 Support STOP return value from before_dom_children to stop
13352 * graphite-optimize-isl.c (optimize_isl): If the schedule
13353 is the same still generate code if -fgraphite-identity
13354 or -floop-parallelize-all are given.
13355 * graphite-scop-detection.c: Include cfganal.h.
13356 (gather_bbs::gather_bbs): Get and pass through bb to RPO
13358 (gather_bbs::before_dom_children): Return STOP for BBs
13360 (build_scops): Compute bb to RPO mapping and pass it to
13361 the domwalk. Treat --param graphite-max-nb-scop-params=0
13362 as not limiting the number of params.
13363 * graphite.c (graphite_initialize): Remove limit on the
13364 number of basic-blocks in a function.
13365 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
13366 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
13367 default value of 10.
13369 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13371 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
13372 Adjust code to eliminate needing to do the shift right 32-bits
13373 operation after XSCVDPSPN.
13375 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13377 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
13378 ((X / Y) != 0 -> X >= Y): Likewise.
13380 2017-09-26 Carl Love <cel@us.ibm.com>
13382 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
13383 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
13384 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
13385 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
13386 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
13387 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
13388 definitions and overloading.
13389 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
13390 statement for P9V_BUILTIN_XST_LEN_R.
13391 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
13392 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
13393 define_expand and define_insn for the instructions and builtins.
13394 * doc/extend.texi: Update the built-in documentation file for the new
13395 built-in functions.
13396 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
13397 define_insn for the instructions
13399 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
13402 * gcc/config/netbsd-protos.h: New file.
13403 * gcc/config/netbsd.c: New file.
13404 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
13405 * gcc/config/t-netbsd: New file.
13406 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
13407 (tmake_file) Add t-netbsd.
13408 (extra_objs) Add netbsd.o.
13410 2017-09-26 Janus Weil <janus@gcc.gnu.org>
13414 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
13416 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13418 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
13419 sign extension from a vector register to a GPR by doing a 32-bit
13420 direct move and then an EXTSW.
13421 (extendsi<mode>2 splitter): Likewise.
13422 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
13423 right or vector extract after doing XSCVDPSPN. Use
13424 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
13426 (movdi_from_sf_zero_ext): Likewise.
13427 (reload_gpr_from_vsxsf): Likewise.
13428 (p8_mfvsrd_4_disf): Delete, no longer used.
13429 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
13430 then needing to move the SFmode to a GPR to use the XSCVDPSP
13431 instruction instead of FRSP and XSCVDPSPN.
13432 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
13433 it is adjacent to the other XSCVSPDP insns.
13434 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
13435 SFmode to be in traditional Altivec registers.
13436 (vsx_xscvdpspn): Eliminate useless alternative constraint.
13437 (vsx_xscvspdpn): Likewise.
13438 (vsx_xscvspdpn_scalar): Likewise.
13440 2017-09-26 Martin Jambor <mjambor@suse.cz>
13442 * tree-sra.c (compare_access_positions): Put integral types first,
13443 stabilize sorting of integral types, remove conditions putting
13444 non-full-precision integers last.
13445 (sort_and_splice_var_accesses): Disable scalarization if a
13446 non-integert would be represented by a non-full-precision integer.
13448 2017-09-26 Joseph Myers <joseph@codesourcery.com>
13450 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
13451 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
13452 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
13453 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
13454 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
13455 conditionals inside the function instead of around it. Call
13456 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
13457 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
13459 2017-09-26 Richard Biener <rguenther@suse.de>
13461 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
13463 (scop_detection::build_scop_breadth): ... this. Removed.
13464 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
13465 (scop_detection::harmful_stmt_in_bb): Likewise.
13466 (scop_detection::graphite_can_represent_stmt): Likewise.
13467 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
13468 (scop_detection::can_represent_loop): Remove recursion, fold in ...
13469 (scop_detection::can_represent_loop_1): ... this. Removed.
13470 (scop_detection::harmful_loop_in_region): Simplify after inlining
13471 the above and remove more quadraticness.
13472 (build_scops): Adjust.
13473 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
13476 2017-09-26 Jakub Jelinek <jakub@redhat.com>
13479 * config/i386/i386.c (ix86_print_operand_address_as): Only test
13480 REGNO (base) == SP_REG if base is a REG.
13482 PR middle-end/35691
13483 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
13484 if it is different SSA_NAME.
13485 (optimize_range_tests_cmp_bitwise): New function.
13486 (optimize_range_tests): Call it.
13488 2017-09-26 Richard Biener <rguenther@suse.de>
13490 PR tree-optimization/82321
13491 * graphite.c (canonicalize_loop_closed_ssa): Properly check
13492 for the def being inside the loop.
13494 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13496 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
13498 * config/s390/s390-builtins.def: Fix constraint on op4.
13500 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13502 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
13503 independent expanders.
13504 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
13505 ("vec_ordered", "vec_unordered"): New expanders.
13507 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13509 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
13512 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13514 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
13515 vec_unpacks_lo_v16qi.
13516 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
13518 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13520 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
13521 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
13522 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
13524 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13526 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
13528 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
13530 ("*vec_slb<mode>"): New insn pattern.
13531 ("vec_shr_<mode>"): New expander.
13532 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
13533 and force the shift count operand to V16QImode.
13534 ("vec_srb<mode>"): Set shift count mode to V16QI.
13536 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13538 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
13539 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
13540 ("vec_widen_smult_hi_<mode>"): New expander definitions.
13542 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
13545 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
13547 2017-09-26 Richard Biener <rguenther@suse.de>
13549 PR tree-optimization/82320
13550 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
13553 2017-09-25 Jeff Law <law@redhat.com>
13555 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
13556 prototype for new argument.
13557 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
13558 mostly extracted from rs6000_emit_allocate_stack.
13559 (rs6000_emit_probe_stack_range_stack_clash): New function.
13560 (rs6000_emit_allocate_stack): Call
13561 rs6000_emit_probe_stack_range_stack_clash as needed.
13562 (rs6000_emit_probe_stack_range): Add additional argument
13563 to call to gen_probe_stack_range{si,di}.
13564 (output_probe_stack_range): New.
13565 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
13566 (output_probe_stack_range_stack_clash): New.
13567 (rs6000_emit_prologue): Emit notes into dump file as requested.
13568 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
13569 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
13570 Add additional operand and pass it to output_probe_stack_range.
13572 2017-09-25 Bin Cheng <bin.cheng@arm.com>
13574 PR tree-optimization/82163
13575 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
13576 (checking_verify_loop_closed_ssa): New parameter.
13577 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
13578 (check_loop_closed_ssa_stmt): Delete.
13579 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
13580 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
13581 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
13584 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
13586 * brig-builtins.def: Treat HSAIL barrier builtins as
13587 setjmp/longjump style functions.
13589 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13591 * target.def (constant_alignment): New hook.
13592 * defaults.h (CONSTANT_ALIGNMENT): Delete.
13593 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
13594 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
13595 * doc/tm.texi: Regenerate.
13596 * targhooks.h (default_constant_alignment): Declare.
13597 (constant_alignment_word_strings): Likewise.
13598 * targhooks.c (default_constant_alignment): New function.
13599 (constant_alignment_word_strings): Likewise.
13600 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
13601 instead of CONSTANT_ALIGNMENT.
13602 * varasm.c (align_variable, get_variable_align, build_constant_desc)
13603 (force_const_mem): Likewise.
13604 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
13605 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
13606 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
13607 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13608 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
13610 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
13611 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13612 constant_alignment_word_strings.
13613 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
13614 (CONSTANT_ALIGNMENT): Likewise.
13615 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13616 (arm_constant_alignment): New function.
13617 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
13618 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13619 constant_alignment_word_strings.
13620 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
13621 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13622 constant_alignment_word_strings.
13623 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13624 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13625 (cris_constant_alignment): New function.
13626 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13627 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13628 (epiphany_constant_alignment): New function.
13629 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13630 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13631 constant_alignment_word_strings.
13632 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13633 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13634 constant_alignment_word_strings.
13635 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13636 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13637 constant_alignment_word_strings.
13638 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13639 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13640 * config/i386/i386.c (ix86_constant_alignment): Make static.
13641 Use the same interface as the target hook.
13642 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13643 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13644 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13645 constant_alignment_word_strings.
13646 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13647 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13648 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13649 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13650 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13651 constant_alignment_word_strings.
13652 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13653 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13654 constant_alignment_word_strings.
13655 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13656 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13657 constant_alignment_word_strings.
13658 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13659 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13661 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13662 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13663 * config/mips/mips.c (mips_constant_alignment): New function.
13664 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13665 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13666 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13667 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13668 (mmix_constant_alignment): Make static. Use the same interface
13669 as the target hook.
13670 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13671 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13672 constant_alignment_word_strings.
13673 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13674 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13675 constant_alignment_word_strings.
13676 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13677 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13678 constant_alignment_word_strings.
13679 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13680 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13681 (rs6000_constant_alignment): New function.
13682 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13683 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13684 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13685 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13686 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13687 (rs6000_constant_alignment): New function.
13688 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13689 * config/s390/s390.c (s390_constant_alignment): New function.
13690 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13691 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13692 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13693 constant_alignment_word_strings.
13694 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13695 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13696 (sparc_constant_alignment): New function.
13697 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13698 * config/spu/spu.c (spu_constant_alignment): New function.
13699 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13700 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13701 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13702 constant_alignment_word_strings.
13703 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13704 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13705 constant_alignment_word_strings.
13706 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13707 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13708 constant_alignment_word_strings.
13709 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13710 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13711 (visium_constant_alignment): New function.
13712 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13713 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13714 (xtensa_constant_alignment): New function.
13715 * system.h (CONSTANT_ALIGNMENT): Poison.
13717 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13719 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13720 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13721 (rs6000_builtin_valid_without_lhs): New helper function.
13722 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13723 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13725 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13727 * target.h (vec_perm_indices): Use unsigned short rather than
13729 (auto_vec_perm_indices): Likewise.
13730 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13731 Use unsigned int rather than unsigned char.
13732 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13734 2017-09-25 Richard Biener <rguenther@suse.de>
13736 * cfgloop.h (sort_sibling_loops): Declare.
13737 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13738 (sort_sibling_loops): New function sorting the sibling loop list
13740 * graphite.c (graphite_transform_loops): Sort sibling loops.
13742 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13744 * target.def (vec_perm_const_ok): Change sel parameter to
13746 * optabs-query.c (can_vec_perm_p): Update accordingly.
13747 * doc/tm.texi: Regenerate.
13748 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13749 auto_vec_perm_indices and remove separate nelt field.
13750 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13751 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13752 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13753 (aarch64_expand_vec_perm_const): Update accordingly.
13754 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13755 to vec_perm_indices.
13756 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13757 auto_vec_perm_indices and remove separate nelt field.
13758 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13759 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13760 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13762 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13763 to vec_perm_indices.
13764 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13765 sel to vec_perm_indices.
13766 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13767 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13768 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13770 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13773 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13776 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13777 on the FUNCTION_DECL function context if it has a DIE that is a
13780 2017-09-25 Richard Biener <rguenther@suse.de>
13782 PR tree-optimization/82285
13783 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13786 2017-09-25 Tom de Vries <tom@codesourcery.com>
13790 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13793 2017-09-25 Richard Biener <rguenther@suse.de>
13795 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13796 ISL errors other than isl_error_quota happen. Dump if the
13797 schedule is the same.
13798 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13799 errors instead of aborting inside ISL.
13801 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13804 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13805 of libgcc_eh for m64.
13806 * config/i386/darwin64.h: Likewise.
13808 2017-09-25 Richard Biener <rguenther@suse.de>
13810 PR middle-end/82144
13811 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13812 attribute for incomplete types nor twice for complete ones.
13814 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13817 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13818 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13821 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13824 * config/i386/i386.opt (mprefer-avx256): Use
13825 ix86_target_flags variable.
13826 * config/i386/i386.c (ix86_target_string): Move
13827 -mprefer-avx256 to flag2_opts.
13829 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13831 PR middle-end/35691
13832 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13833 and x != -1 | y != -1 into (x & y) != -1.
13835 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13837 * config.gcc: Add new case statement to set
13838 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13839 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13840 s390x-*-linux* case statements. Added aarch64 to the list of
13841 supported architectures.
13843 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13845 PR tree-optimization/82289
13846 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13847 STMT_VINFO_RELEVANT_P.
13849 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13850 Alan Hayward <alan.hayward@arm.com>
13851 David Sherwood <david.sherwood@arm.com>
13853 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
13854 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13855 (extract_range_from_binary_expr_1): Don't call
13856 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
13858 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13859 Alan Hayward <alan.hayward@arm.com>
13860 David Sherwood <david.sherwood@arm.com>
13862 * target.def (preferred_vector_alignment): New hook.
13863 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13865 * doc/tm.texi: Regenerate.
13866 * targhooks.h (default_preferred_vector_alignment): Declare.
13867 * targhooks.c (default_preferred_vector_alignment): New function.
13868 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
13870 (DR_TARGET_ALIGNMENT): New macro.
13871 (aligned_access_p): Update commentary.
13872 (vect_known_alignment_in_bytes): New function.
13873 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
13875 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
13876 Calculate the misalignment based on the target alignment rather than
13878 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
13879 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
13880 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
13881 the target alignment, rather than masking the element misalignment
13882 with the number of elements in a vector. Also use the target
13883 alignment when calculating the maximum number of peels.
13884 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
13885 instead of TYPE_ALIGN_UNIT.
13886 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
13887 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
13888 (vect_create_addr_base_for_vector_ref): Update call accordingly.
13889 (vect_create_data_ref_ptr): Likewise.
13890 (vect_setup_realignment): Realign by ANDing with
13891 -DR_TARGET_MISALIGNMENT.
13892 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
13893 the number of peels based on DR_TARGET_ALIGNMENT.
13894 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
13895 with the guaranteed alignment boundary when deciding whether
13897 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
13898 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
13899 (ensure_base_align): Remove stmt_info parameter. Get the
13900 target base alignment from DR_TARGET_ALIGNMENT.
13901 (vectorizable_store): Update call accordingly. Interpret
13902 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
13904 (vectorizable_load): Likewise.
13906 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13907 Alan Hayward <alan.hayward@arm.com>
13908 David Sherwood <david.sherwood@arm.com>
13910 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
13911 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
13912 (vect_enhance_data_refs_alignment): Likewise.
13914 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13916 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13917 error. Only quit immediately if parsing is complete.
13918 (BEGIN): Initialize fatal_err and parse_done.
13919 (begin fpu, end fpu): Check number of arguments.
13920 (begin arch, end arch): Likewise.
13921 (begin cpu, end cpu): Likewise.
13922 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13923 (optalias): Likewise.
13925 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13927 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13928 * config/arm/arm-isa.h: Delete. Move definitions to ...
13929 * arm-cpus.in: ... here. Use new feature and fgroup values.
13930 * config/arm/arm.c (arm_option_override): Use lower case for feature
13932 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13933 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13934 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13936 (print_isa_bits_for): New function.
13937 (gen_isa): New function.
13938 (gen_comm_data): Use print_isa_bits_for.
13939 (define feature): New keyword.
13940 (define fgroup): New keyword.
13941 * config/arm/t-arm (TM_H): Remove.
13942 (GTM_H): Add arm-isa.h.
13943 (arm-isa.h): Add rule to generate file.
13944 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13945 case for feature bit names.
13947 2017-09-22 Richard Biener <rguenther@suse.de>
13949 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
13951 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
13952 print of no dependency loops ...
13953 * graphite.c (graphite_transform_loops): ... here.
13954 (canonicalize_loop_closed_ssa_form): Work from inner to outer
13956 (same_close_phi_node, remove_duplicate_close_phi,
13957 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
13958 (canonicalize_loop_closed_ssa): ... here and simplify.
13959 * graphite-optimize-isl.c: Include tree-vectorizer.h.
13960 (optimize_isl): Use dump_printf_loc to tell when we stopped
13961 optimizing because of an ISL timeout.
13963 2017-09-22 Richard Biener <rguenther@suse.de>
13965 PR tree-optimization/82291
13966 * tree-if-conv.c (predicate_mem_writes): Make sure to
13967 remove writes in blocks predicated with false.
13969 2017-09-22 Richard Biener <rguenther@suse.de>
13971 * sese.c: Include cfganal.h.
13972 (if_region_set_false_region): Remove.
13973 (create_if_region_on_edge): Likewise.
13974 (move_sese_in_condition): Re-implement without destroying
13977 2017-09-22 Richard Biener <rguenther@suse.de>
13979 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
13980 Verify both BBs contain loop PHI nodes before dispatching to
13981 copy_loop_phi_args.
13982 (graphite_regenerate_ast_isl): Do not recompute dominators,
13983 do not verify three times. Restructure for clarity.
13984 * graphite-scop-detection.c (same_close_phi_node,
13985 remove_duplicate_close_phi, make_close_phi_nodes_unique,
13986 defined_in_loop_p, canonicalize_loop_closed_ssa,
13987 canonicalize_loop_closed_ssa_form): Simplify, remove excess
13988 checking and SSA rewrite, move to ...
13989 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
13990 (graphite_initialize): Do not pass in ctx, do not reset the
13991 SCEV cache, compute only dominators.
13992 (graphite_transform_loops): Allocate ISL ctx after
13993 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
13994 Maintain post-dominators only around build_scops.
13995 * sese.c (if_region_set_false_region): Make static. Free
13996 and recompute dominators.
13997 (move_sese_in_condition): Assert we don't get called with
13998 post-dominators computed.
13999 * sese.h (if_region_set_false_region): Remove.
14001 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
14003 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
14004 mode attribute for TARGET_AVX512VL.
14006 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14008 * config/i386/i386.opt (mprefer-avx256): New option.
14009 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14011 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14012 for TARGET_PREFER_AVX256.
14013 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14015 2017-09-21 Jeff Law <law@redhat.com>
14017 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14018 Fix dump output if the only stack space is for pushed registers.
14020 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14022 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14025 2017-09-21 Martin Sebor <msebor@redhat.com>
14028 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14029 code (in C++) or code that triggers warnings.
14031 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14033 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14035 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14037 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14038 * sched-rgn.c: Ditto.
14039 * sel-sched-ir.c: Ditto.
14041 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
14043 * toplev.h (set_random_seed): Adjust return type.
14044 * toplev.c (init_local_tick): Move eager initialization of random_seed
14045 to get_random_seed. Adjust comment.
14046 (init_random_seed): Inline to get_random_seed, delete.
14047 (get_random_seed): Initialize random_seed lazily.
14048 (set_random_seed): Do not return previous value.
14049 (print_switch_value): Do not call get_random_seed.
14051 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14053 * cgraph.c (delete_function_version): New, broken out from...
14054 (cgraph_node::delete_function_version): ...here. Rename to
14055 cgraph_node::delete_function_version_by_decl. Update all uses.
14056 (cgraph_node::remove): Call delete_function_version.
14058 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14061 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14062 VAR_DECLs to which addressable non-volatile parameters are mapped
14063 and for id->retvar after the return value assignment. Clear
14064 id->retval and id->retbnd after inlining.
14066 2017-09-21 Richard Biener <rguenther@suse.de>
14068 PR tree-optimization/82276
14069 PR tree-optimization/82244
14070 * tree-vrp.c (build_assert_expr_for): Set
14071 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14073 (remove_range_assertions): Revert earlier change.
14075 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14078 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14080 2017-09-21 Richard Biener <rguenther@suse.de>
14082 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14083 Restore valid IL after code generation errors.
14084 * graphite.c (graphite_transform_loops): Diagnose code
14085 generation issues as MSG_MISSED_OPTIMIZATION and continue
14086 with processing SCOPs.
14088 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14089 Alan Hayward <alan.hayward@arm.com>
14090 David Sherwood <david.sherwood@arm.com>
14092 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14093 rather than choosing between plus_constant and gen_rtx_<CODE>.
14094 * expr.c (emit_push_insn): Likewise.
14095 (expand_expr_real_2): Likewise.
14097 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14098 Alan Hayward <alan.hayward@arm.com>
14099 David Sherwood <david.sherwood@arm.com>
14101 * loop-unroll.c (split_iv): Call copy_rtx on the step.
14103 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14104 Alan Hayward <alan.hayward@arm.com>
14105 David Sherwood <david.sherwood@arm.com>
14107 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14108 calling tree_to_uhwi.
14110 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14111 Alan Hayward <alan.hayward@arm.com>
14112 David Sherwood <david.sherwood@arm.com>
14114 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
14115 INTEGER_CST rather than a negative test for ADDR_EXPR.
14117 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14118 Alan Hayward <alan.hayward@arm.com>
14119 David Sherwood <david.sherwood@arm.com>
14121 * tree-vrp.c (extract_range_from_binary_expr_1): Check
14122 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
14124 2017-09-21 Richard Biener <rguenther@suse.de>
14126 PR tree-optimization/71351
14127 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
14128 graphite_create_new_loop_guard): Remove, fold remaining parts
14130 (translate_isl_ast_node_for): ... here and simplify.
14132 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14135 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
14136 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
14137 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
14138 alternative always use QI mode, for -Os imov (=R,R) alternative
14139 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
14142 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14143 Jeff Law <law@redhat.com>
14145 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
14146 (allocate_stack_space): New function, partially extracted from
14147 s390_emit_prologue.
14148 (s390_emit_prologue): Track offset to most recent stack probe.
14149 Code to allocate space moved into allocate_stack_space.
14150 Dump actions when no stack is allocated.
14151 (s390_prologue_plus_offset): New function.
14152 (s390_emit_stack_probe): Likewise.
14154 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
14156 * common.opt (Wa, Wl, Wp, g, gz=): Add
14158 (gno-column-info): Remove.
14159 (gcolumn-info): Drop RejectNegative.
14160 (gno-): New prefix.
14161 (gno-record-gcc-switches): Remove.
14162 (grecord-gcc-switches): Drop RejectNegative.
14163 (gno-split-dwarf): Remove.
14164 (gsplit-dwarf): Drop RejectNegative.
14165 (gno-strict-dwarf): Remove.
14166 (gstrict-dwarf): Drop RejectNegative.
14167 * config/darwin.opt (gfull, gused): Add RejectNegative.
14168 * dwarf2out.c (gen_producer_string): Drop
14169 gno-record-gcc-switches handler.
14170 * optc-gen.awk: Add g to prefixes with negative forms.
14171 * opts-common.c (remapping_prefix_p): New.
14172 (find_opt): Check it.
14173 (generate_canonical_option): Test g prefix.
14174 (option_map): Add -gno- mapping.
14175 (add_misspelling_candidates): Check remapping_prefix_p.
14177 2017-09-20 Jeff Law <law@redhat.com>
14179 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
14180 thinko in stack clash protection support.
14182 * explow.c (compute_stack_clash_protection_loop_data): Use
14183 CONST_INT_P instead of explicit test. Verify object is a
14184 CONST_INT_P before looking at INTVAL.
14185 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
14186 instead of explicit test.
14188 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
14191 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
14192 address instead of to r1 and r11.
14194 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
14196 * config.gcc: Support "knm".
14197 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
14198 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14200 * config/i386/i386.c (m_KNM): Define.
14201 (processor_target_table): Add "knm".
14203 (ix86_option_override_internal): Add "knm".
14204 (ix86_issue_rate): Add PROCESSOR_KNM.
14205 (ix86_adjust_cost): Ditto.
14206 (ia32_multipass_dfa_lookahead): Ditto.
14207 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
14208 (fold_builtin_cpu): Add M_INTEL_KNM.
14209 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
14210 (processor_type): Add PROCESSOR_KNM.
14211 * config/i386/x86-tune.def: Add m_KNM.
14212 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
14214 2017-09-20 Richard Biener <rguenther@suse.de>
14216 PR tree-optimization/80213
14217 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
14218 are allowed in empty BBs as well.
14219 (canonicalize_loop_closed_ssa): Also look for other complex
14221 (scop_detection::get_sese): Include the loop-closed PHI block
14223 (scop_detection::merge_sese): Remove code adding extra blocks.
14224 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
14225 (build_scops): Assert the final returned scop is invalid.
14227 2017-09-20 Richard Biener <rguenther@suse.de>
14229 PR tree-optimization/82264
14230 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
14232 (vn_phi_lookup): Likewise.
14233 (vn_phi_insert): Likewise.
14235 2017-09-20 Jakub Jelinek <jakub@redhat.com>
14237 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
14238 that fits into uhwi or shwi, add DW_AT_const_value regardless
14239 of early_dwarf without going through RTL, using add_AT_unsigned
14242 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
14243 (DEBUG_LTO_ABBREV_SECTION): Likewise.
14244 (DEBUG_LTO_MACINFO_SECTION): Likewise.
14245 (DEBUG_MACRO_SECTION): Likewise.
14246 (DEBUG_LTO_MACRO_SECTION): Likewise.
14247 (DEBUG_STR_DWO_SECTION): Likewise.
14248 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
14249 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
14250 (DEBUG_LTO_DWO_LINE_SECTION): Define.
14251 (DEBUG_LTO_LINE_STR_SECTION): Define.
14252 (init_sections_and_labels): Initialize debug_line_str_section
14253 variable. Initialize debug_loc_section for -gdwarf-5 to
14254 DEBUG_LOCLISTS_SECTION. Formatting fixes.
14256 2017-09-20 Richard Biener <rguenther@suse.de>
14258 * graphite-sese-to-poly.c (extract_affine): Properly handle
14259 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
14261 2017-09-20 Richard Biener <rguenther@suse.de>
14263 PR tree-optimization/81373
14264 * graphite-scop-detection.c (build_cross_bb_scalars_def):
14265 Force SESE live-out defs to be handled even if they are
14268 2017-09-19 Jeff Law <law@redhat.com>
14270 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
14271 nothing for stack adjustments with REG_STACK_CHECK.
14272 * sched-deps.c (parse_add_or_inc): Reject insns with
14273 REG_STACK_CHECK from dependency breaking.
14274 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
14275 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
14276 * reg-notes.def (STACK_CHECK): New note.
14278 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
14279 (ix86_expand_prologue): Dump stack clash info as needed.
14280 Call ix86_adjust_stack_and_probe_stack_clash as needed.
14282 * function.c (dump_stack_clash_frame_info): New function.
14283 * function.h (dump_stack_clash_frame_info): Prototype.
14284 (enum stack_clash_probes): New enum.
14286 * config/alpha/alpha.c (alpha_expand_prologue): Also check
14287 flag_stack_clash_protection.
14288 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
14289 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
14290 (arm_frame_pointer_required): Likewise.
14291 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
14292 (ia64_expand_prologue): Likewise.
14293 * config/mips/mips.c (mips_expand_prologue): Likewise.
14294 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
14295 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14296 (sparc_flat_expand_prologue): Likewise.
14297 * config/spu/spu.c (spu_expand_prologue): Likewise.
14299 * explow.c: Include "params.h".
14300 (anti_adjust_stack_and_probe_stack_clash): New function.
14301 (get_stack_check_protect): Likewise.
14302 (compute_stack_clash_protection_loop_data): Likewise.
14303 (emit_stack_clash_protection_loop_start): Likewise.
14304 (emit_stack_clash_protection_loop_end): Likewise.
14305 (allocate_dynamic_stack_space): Use get_stack_check_protect.
14306 Use anti_adjust_stack_and_probe_stack_clash.
14307 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
14308 (emit_stack_clash_protection_loop_start): Likewise.
14309 (emit_stack_clash_protection_loop_end): Likewise.
14310 * rtl.h (get_stack_check_protect): Prototype.
14311 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
14312 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
14313 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
14315 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
14317 * doc/tm.texi: Rebuilt.
14318 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
14319 get_stack_check_protect.
14320 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
14321 * config/arm/arm.c (arm_expand_prologue): Likewise.
14322 (arm_frame_pointer_required): Likewise.
14323 * config/i386/i386.c (ix86_expand_prologue): Likewise.
14324 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
14325 * config/mips/mips.c (mips_expand_prologue): Likewise.
14326 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
14327 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
14328 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14329 (sparc_flat_expand_prologue): Likewise.
14331 * common.opt (-fstack-clash-protection): New option.
14332 * flag-types.h (enum stack_check_type): Note difference between
14333 -fstack-check= and -fstack-clash-protection.
14334 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
14335 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
14336 * toplev.c (process_options): Issue warnings/errors for cases
14337 not handled with -fstack-clash-protection.
14338 * doc/invoke.texi (-fstack-clash-protection): Document new option.
14339 (-fstack-check): Note additional problem with -fstack-check=generic.
14340 Note that -fstack-check is primarily for Ada and refer users
14341 to -fstack-clash-protection for stack-clash-protection.
14342 Document new params for stack clash protection.
14344 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
14346 * config/i386/i386.c (ix86_split_long_move): Do not handle
14347 address used for LEA in a special way.
14349 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
14351 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
14353 2017-09-19 Martin Sebor <msebor@redhat.com>
14356 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
14357 of incompatible types.
14359 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
14361 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14362 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
14363 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14364 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
14366 2017-09-19 Richard Biener <rguenther@suse.de>
14368 PR tree-optimization/82244
14369 * tree-vrp.c (remove_range_assertions): Do not propagate
14370 a constant to abnormals but replace the assert with a copy.
14372 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
14374 PR rtl-optimization/57878
14375 PR rtl-optimization/68988
14376 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
14377 avoidance test involving non_reload_pseudos. Move frequency test
14378 below the general fragmentation avoidance test.
14380 2017-09-19 Richard Biener <rguenther@suse.de>
14382 PR tree-optimization/69728
14383 * graphite-sese-to-poly.c (schedule_error): New global.
14384 (add_loop_schedule): Handle empty domain by failing the
14386 (build_original_schedule): Handle schedule_error.
14388 2017-09-19 Richard Biener <rguenther@suse.de>
14390 * graphite-scop-detection.c (scop_detection::can_represent_loop):
14391 Do not iterate to sibling loops but only to siblings of inner
14394 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
14397 * config/m68k/m68k.md (moveq feeding equality comparison): Check
14398 that the registers are different.
14400 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
14402 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
14403 to processor_model and "amdfam17h" to arch_names_table.
14404 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
14406 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14409 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
14411 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14412 Alan Hayward <alan.hayward@arm.com>
14413 David Sherwood <david.sherwood@arm.com>
14415 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
14417 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
14419 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
14420 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
14421 vect_schedule_slp_instance.
14422 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
14423 Update call to vect_slp_analyze_node_operations. Simplify return
14425 (vect_slp_analyze_bb_1): Update call accordingly.
14426 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
14427 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
14428 (vect_schedule_slp): Update call accordingly.
14430 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14431 Alan Hayward <alan.hayward@arm.com>
14432 David Sherwood <david.sherwood@arm.com>
14434 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
14435 with types that aren't in fact scalar.
14437 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14439 * tree-vect-slp.c (vect_record_max_nunits): New function,
14441 (vect_build_slp_tree_1): ...here.
14442 (vect_build_slp_tree_2): Call it for phis too.
14444 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14446 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
14447 to vect_get_vec_def_for_operand when getting the mask operand.
14449 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14450 Alan Hayward <alan.hayward@arm.com>
14451 David Sherwood <david.sherwood@arm.com>
14453 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
14456 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14457 Alan Hayward <alan.hayward@arm.com>
14458 David Sherwood <david.sherwood@arm.com>
14460 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
14461 calculation for vector booleans.
14463 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14464 Alan Hayward <alan.hayward@arm.com>
14465 David Sherwood <david.sherwood@arm.com>
14467 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
14469 (vect_transform_stmt): ...here.
14470 (vect_analyze_stmt): Use it instead of calling
14471 vectorizable_live_operation directly.
14473 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
14475 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
14476 non-SIMT targets in acc vector loops.
14478 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
14480 * configure.ac: Add arc and check if assembler supports gdwarf2.
14481 * configure: Regenerate.
14483 2017-09-18 Richard Biener <rguenther@suse.de>
14485 PR tree-optimization/82220
14486 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
14487 epilogue niters from the min_profitable_iters compute.
14489 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14492 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
14493 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
14494 (ix86_init_pic_reg): Revert 2017-09-01 changes.
14496 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
14499 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
14500 switching to a new text section.
14502 2017-09-18 Richard Biener <rguenther@suse.de>
14504 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
14506 (build_alias_set): Reject aliases with no access function.
14508 2017-09-18 Richard Biener <rguenther@suse.de>
14510 PR tree-optimization/79622
14511 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
14513 (build_cross_bb_scalars_use): Likewise.
14515 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
14517 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
14519 2017-09-18 Alan Modra <amodra@gmail.com>
14522 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
14523 stack_pointer_rtx for count 0. Update comments. Break up
14524 large rtl expression.
14526 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
14528 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
14529 Increase to 20 bytes.
14530 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
14531 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
14532 or avx version of the stub.
14534 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
14537 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
14538 compute the minimum stack alignment. Also update preferred stack
14539 boundary for leaf functions.
14541 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
14543 PR tree-optimization/82228
14544 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
14547 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14549 * common/config/nds32/nds32-common.c
14550 (nds32_option_optimization_table): Refine formatting.
14551 (nds32_option_optimization_table): Use -fsched-pressure and
14552 -fomit-frame-pointer for specific optimization level.
14554 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14556 * config/nds32/nds32.c: Refine formatting and comments.
14557 * config/nds32/nds32.h: Likewise.
14558 * config/nds32/nds32.md: Likewise.
14559 * config/nds32/nds32-cost.c: Likewise.
14560 * config/nds32/nds32-isr.c: Likewise.
14561 * config/nds32/nds32-md-auxiliary.c: Likewise.
14562 * config/nds32/nds32-multiple.md: Likewise.
14563 * config/nds32/nds32-predicates.c: Likewise.
14565 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
14566 Jakub Jelinek <jakub@redhat.com>
14568 Add support for -std=c++2a.
14569 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
14571 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
14573 2017-09-15 Steve Ellcey <sellcey@cavium.com>
14576 * doc/extend.texi (Common Function Attributes): Add
14577 references to ARM, AArch64, and S/390 specific attributes.
14578 (Function Specific Option Pragmas): Add AArch64 and S/390
14579 to list of back ends that support the target pragma.
14581 2017-09-15 Nathan Sidwell <nathan@acm.org>
14583 * doc/standards.texi: Fix C++17 description. Update URLs for
14586 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14588 * common.opt (Wcast-align=strict): New warning option.
14589 * doc/invoke.texi: Document -Wcast-align=strict.
14591 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
14593 * cgraph.h (cgraph_thunk_info): Add comments.
14594 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
14595 assert for VIRTUAL_* arguments stricter.
14597 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
14599 PR tree-optimization/71026
14600 * match.pd: Move RDIV patterns from fold-const.c
14601 * fold-const.c (distribute_real_division): Removed.
14602 (fold_binary_loc): Remove calls to distribute_real_divison.
14604 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14606 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
14607 c++1z and gnu++1z as deprecated. Change other references to
14608 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
14609 Change -Wc++1z-compat to -Wc++17-compat.
14610 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
14611 * dwarf2out.c (highest_c_language): Handle C++17.
14612 (gen_compile_unit_die): Likewise.
14614 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14616 PR rtl-optimization/82192
14617 * combine.c (make_extraction): Don't look through non-paradoxical
14618 SUBREGs or TRUNCATE if pos + len is or might be bigger than
14621 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14622 Alan Hayward <alan.hayward@arm.com>
14623 David Sherwood <david.sherwood@arm.com>
14625 * target.def (function_arg_offset): New hook.
14626 * targhooks.h (default_function_arg_offset): Declare.
14627 * targhooks.c (default_function_arg_offset): New function.
14628 * function.c (locate_and_pad_parm): Use
14629 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14630 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14631 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14632 * doc/tm.texi: Regenerate.
14633 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14634 * config/spu/spu.c (spu_function_arg_offset): New function.
14635 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14636 * system.h (FUNCTION_ARG_OFFSET): Poison.
14638 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14639 Alan Hayard <alan.hayward@arm.com>
14640 David Sherwood <david.sherwood@arm.com>
14642 * target.def (truly_noop_truncation): New hook.
14643 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14644 than TRULY_NOOP_TRUNCATION.
14645 * hooks.h (hook_bool_uint_uint_true): Declare.
14646 * hooks.c (hook_bool_uint_uint_true): New function.
14647 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14648 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14649 * doc/tm.texi: Regenerate.
14650 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14651 rather than TRULY_NOOP_TRUNCATION in comments.
14652 (simplify_comparison): Likewise.
14653 (record_truncated_value): Likewise.
14654 * expmed.c (extract_bit_field_1): Likewise.
14655 (extract_split_bit_field): Likewise.
14656 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14657 instead of TRULY_NOOP_TRUNCATION.
14658 * function.c (assign_parm_setup_block): Likewise.
14659 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14660 * rtlhooks.c: Include target.h.
14661 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14662 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14663 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14664 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14665 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14666 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14667 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14668 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14669 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14670 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14671 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14672 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14673 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14674 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14675 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14676 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14677 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14678 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14679 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14680 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14681 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14682 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14683 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14684 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14685 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14686 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14687 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14688 rather than TRULY_NOOP_TRUNCATION in comments.
14689 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14690 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14691 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14692 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14693 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14694 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14695 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14696 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14697 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14698 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14699 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14700 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14701 rather than TRULY_NOOP_TRUNCATION in comments.
14702 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14703 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14704 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14705 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14706 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14707 TRULY_NOOP_TRUNCATION condition.
14708 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14709 (TRULY_NOOP_TRUNCATION): Delete.
14710 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14711 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14712 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14713 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14714 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14715 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14716 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14717 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14718 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14719 rather than TRULY_NOOP_TRUNCATION in comments.
14720 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14721 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14722 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14723 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14724 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14725 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14727 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14730 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14731 (*cmp_ior): Likewise.
14732 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14733 (*ior_scc_scc_cmp): Likewise.
14734 (*and_scc_scc): Likewise.
14735 (*and_scc_scc_cmp): Likewise.
14737 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14738 Alan Hayard <alan.hayward@arm.com>
14739 David Sherwood <david.sherwood@arm.com>
14741 * target.def (can_change_mode_class): New hook.
14742 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14743 (hard_regno_nregs): Likewise.
14744 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14745 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14746 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14747 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14748 (LOAD_EXTEND_OP): Update accordingly.
14749 * doc/tm.texi: Regenerate.
14750 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14751 CANNOT_CHANGE_MODE_CLASS.
14752 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14753 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14754 * combine.c (simplify_set): Update accordingly.
14755 * emit-rtl.c (validate_subreg): Likewise.
14756 * recog.c (general_operand): Likewise.
14757 * regcprop.c (mode_change_ok): Likewise.
14758 * reload1.c (choose_reload_regs): Likewise.
14759 (inherit_piecemeal_p): Likewise.
14760 * rtlanal.c (simplify_subreg_regno): Likewise.
14761 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14762 instead of CANNOT_CHANGE_MODE_CLASS.
14763 (reload_cse_simplify_operands): Likewise.
14764 * reload.c (push_reload): Use targetm.can_change_mode_class
14765 instead of CANNOT_CHANGE_MODE_CLASS.
14766 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14767 REG_CANNOT_CHANGE_MODE_P.
14768 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14769 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14770 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14771 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14772 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14773 (arm_can_change_mode_class): New function.
14774 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14775 than CANNOT_CHANGE_MODE_CLASS in comments.
14776 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14777 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14778 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14779 (ix86_can_change_mode_class): ...this new function, inverting the
14780 sense of the return value.
14781 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14782 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14783 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14784 (ia64_can_change_mode_class): New function.
14785 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14786 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14787 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14788 (m32c_can_change_mode_class): ...this new function, inverting the
14789 sense of the return value.
14790 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14791 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14792 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14793 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14794 (mips_can_change_mode_class): ...this new function, inverting the
14795 sense of the return value.
14796 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14797 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14798 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14799 (msp430_can_change_mode_class): New function.
14800 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14801 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14802 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14803 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14804 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14805 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14806 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14807 (pa_cannot_change_mode_class): Replace with...
14808 (pa_can_change_mode_class): ...this new function, inverting the
14809 sense of the return value.
14810 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14811 than CANNOT_CHANGE_MODE_CLASS in comments.
14812 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14813 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14814 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14815 (pdp11_cannot_change_mode_class): Replace with...
14816 (pdp11_can_change_mode_class): ...this new function, inverting the
14817 sense of the return value.
14818 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14819 * config/powerpcspe/powerpcspe-protos.h
14820 (rs6000_cannot_change_mode_class_ptr): Delete.
14821 * config/powerpcspe/powerpcspe.c
14822 (rs6000_cannot_change_mode_class_ptr): Delete.
14823 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14824 (rs6000_option_override_internal): Assign to
14825 targetm.can_change_mode_class instead of
14826 rs6000_cannot_change_mode_class_ptr.
14827 (rs6000_cannot_change_mode_class): Replace with...
14828 (rs6000_can_change_mode_class): ...this new function, inverting the
14829 sense of the return value.
14830 (rs6000_debug_cannot_change_mode_class): Replace with...
14831 (rs6000_debug_can_change_mode_class): ...this new function.
14832 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14833 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14834 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14835 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14836 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14838 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14839 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14840 (rs6000_option_override_internal): Assign to
14841 targetm.can_change_mode_class instead of
14842 rs6000_cannot_change_mode_class_ptr.
14843 (rs6000_cannot_change_mode_class): Replace with...
14844 (rs6000_can_change_mode_class): ...this new function, inverting the
14845 sense of the return value.
14846 (rs6000_debug_cannot_change_mode_class): Replace with...
14847 (rs6000_debug_can_change_mode_class): ...this new function.
14848 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14849 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14850 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14851 (s390_can_change_mode_class): ...this new function, inverting the
14852 sense of the return value.
14853 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14854 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14855 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14856 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14857 (sh_cannot_change_mode_class): Replace with...
14858 (sh_can_change_mode_class): ...this new function, inverting the
14859 sense of the return value.
14860 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14861 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14862 (sparc_can_change_mode_class): New function.
14863 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14864 * config/spu/spu.c (spu_can_change_mode_class): New function.
14865 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14866 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14867 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14868 (visium_can_change_mode_class): New function.
14869 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
14871 2017-09-15 Richard Biener <rguenther@suse.de>
14873 PR tree-optimization/82217
14874 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
14875 but not undefined case.
14877 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14880 * postreload.c (reload_cse_simplify_operands): Skip
14881 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
14883 2017-09-15 Richard Biener <rguenther@suse.de>
14885 PR tree-optimization/68823
14886 * graphite-scop-detection.c (build_alias_set): If we have a
14887 possible dependence check whether we can handle them by just
14888 looking at the DRs DR_ACCESS_FNs.
14889 (build_scops): If build_alias_set fails, fail the SCOP.
14891 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
14893 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
14894 to support float128 built-in functions that require the ISA 3.0
14896 (BU_FLOAT128_3_HW): Likewise.
14897 (SQRTF128): Add support for the IEEE 128-bit square root and fma
14898 built-in functions.
14899 (FMAF128): Likewise.
14901 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14902 support for built-in functions that need the ISA 3.0 IEEE 128-bit
14903 floating point instructions.
14904 (rs6000_invalid_builtin): Likewise.
14905 (rs6000_builtin_mask_names): Likewise.
14906 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
14907 (RS6000_BTM_FLOAT128_HW): Likewise.
14908 (RS6000_BTM_COMMON): Likewise.
14909 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
14911 * doc/extend.texi (RS/6000 built-in functions): Document the
14912 IEEE 128-bit floating point square root and fused multiply-add
14913 built-in functions.
14915 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
14917 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
14918 reg (r2) isn't in the set of registers defined in the prologue.
14920 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14921 Alan Hayward <alan.hayward@arm.com>
14922 David Sherwood <david.sherwood@arm.com>
14924 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
14925 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
14926 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
14927 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
14928 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
14930 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
14931 max_vectorization_factor.
14932 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
14934 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14935 Alan Hayward <alan.hayward@arm.com>
14936 David Sherwood <david.sherwood@arm.com>
14938 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
14939 (vect_worthwhile_without_simd_p): Declare.
14940 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
14941 (vectorizable_reduction): Use it.
14942 * tree-vect-stmts.c (vectorizable_shift): Likewise.
14943 (vectorizable_operation): Likewise.
14945 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14946 Alan Hayward <alan.hayward@arm.com>
14947 David Sherwood <david.sherwood@arm.com>
14949 * tree-vectorizer.h (vect_get_num_copies): New function.
14950 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
14951 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14952 (vectorizable_induction): Likewise.
14953 (vectorizable_live_operation): Likewise.
14954 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14955 (vectorizable_bswap): Likewise.
14956 (vectorizable_call): Likewise.
14957 (vectorizable_conversion): Likewise.
14958 (vectorizable_assignment): Likewise.
14959 (vectorizable_shift): Likewise.
14960 (vectorizable_operation): Likewise.
14961 (vectorizable_store): Likewise.
14962 (vectorizable_load): Likewise.
14963 (vectorizable_condition): Likewise.
14964 (vectorizable_comparison): Likewise.
14965 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
14967 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14968 Alan Hayward <alan.hayward@arm.com>
14969 David Sherwood <david.sherwood@arm.com>
14971 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
14972 of vect_init_vector.
14974 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14975 Alan Hayward <alan.hayward@arm.com>
14976 David Sherwood <david.sherwood@arm.com>
14978 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
14979 an inline wrapper that provides a location.
14980 (gimple_build_vector): Likewise.
14981 * gimple-fold.c (gimple_build_vector_from_val): New function.
14982 (gimple_build_vector): Likewise.
14983 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
14984 functions to build the initial value. Always return a gimple value.
14985 (get_initial_defs_for_reduction): Likewise. Only compute
14987 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
14988 vect_init_vector on the results from get_initial_def(s)_for_reduction.
14989 (vectorizable_induction): Use gimple_build_vector rather than
14992 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14993 Alan Hayward <alan.hayward@arm.com>
14994 David Sherwood <david.sherwood@arm.com>
14996 * target.h (vec_perm_indices): New typedef.
14997 (auto_vec_perm_indices): Likewise.
14998 * optabs-query.h: Include target.h
14999 (can_vec_perm_p): Take a vec_perm_indices *.
15000 * optabs-query.c (can_vec_perm_p): Likewise.
15001 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
15002 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15003 * tree-vect-generic.c (lower_vec_perm): Likewise.
15004 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15005 (vect_grouped_load_supported): Likewise.
15006 (vect_shift_permute_load_chain): Likewise.
15007 (vect_permute_store_chain): Use auto_vec_perm_indices.
15008 (vect_permute_load_chain): Likewise.
15009 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15010 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15011 Update uses of can_vec_perm_p.
15012 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15013 mode with a number of elements. Take a vec_perm_indices *.
15014 (vect_create_epilog_for_reduction): Update accordingly.
15015 Use auto_vec_perm_indices.
15016 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15017 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15018 (vect_transform_slp_perm_load): Likewise.
15019 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15020 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15021 (vect_gen_perm_mask_checked): Likewise.
15022 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15023 (vect_gen_perm_mask_checked): Likewise.
15024 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15025 (vectorizable_store): Likewise.
15026 (vectorizable_load): Likewise.
15027 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15028 (vectorizable_bswap): Likewise.
15030 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15031 Alan Hayward <alan.hayward@arm.com>
15032 David Sherwood <david.sherwood@arm.com>
15034 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15035 * tree.c (build_vector): Likewise.
15036 (build_vector_from_ctor): Update accordingly.
15037 (build_vector_from_val): Likewise.
15038 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15039 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15040 * tree-vect-generic.c (add_rshift): Likewise.
15041 (expand_vector_divmod): Likewise.
15042 (optimize_vector_constructor): Likewise.
15043 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15044 (vect_transform_slp_perm_load): Likewise.
15045 (vect_schedule_slp_instance): Likewise.
15046 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15047 (vectorizable_call): Likewise.
15048 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15049 * expmed.c (make_tree): Likewise.
15050 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15051 a vector passed to build_vector.
15052 (fold_convert_const): Likewise.
15053 (exact_inverse): Likewise.
15054 (fold_ternary_loc): Likewise.
15055 (fold_relational_const): Likewise.
15056 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15057 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15058 (const_unop): Likewise. Store the reduction accumulator in a
15059 variable rather than an array.
15060 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15061 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15062 the new vector, rather than constructing it after the input arrays.
15063 (native_interpret_vector): Use auto_vec<tree> when building
15064 a vector passed to build_vector. Add elements in order.
15065 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15066 auto_vec<tree> when building a vector passed to build_vector.
15067 (vect_create_epilog_for_reduction): Likewise.
15068 (vectorizable_induction): Likewise.
15069 (get_initial_def_for_reduction): Likewise. Fix indentation of
15071 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15073 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15074 passed to build_vector.
15076 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15077 Alan Hayward <alan.hayward@arm.com>
15078 David Sherwood <david.sherwood@arm.com>
15080 * tree-core.h (tree_base::u): Add an "nelts" field.
15081 (tree_vector): Use VECTOR_CST_NELTS as the length.
15082 * tree.c (tree_size): Likewise.
15083 (make_vector): Initialize VECTOR_CST_NELTS.
15084 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15085 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15086 TYPE_VECTOR_SUBPARTS.
15087 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15088 as the number of units, setting it from VECTOR_CST_NELTS.
15089 (const_vector_from_tree): Likewise.
15090 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15091 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15092 (fold_negate_expr_1): Likewise.
15093 (fold_convert_const): Likewise.
15094 (const_binop): Likewise. Differentiate the number of output and
15096 (const_unop): Likewise.
15097 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15098 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15099 in cases that did the opposite.
15101 2017-09-14 Richard Biener <rguenther@suse.de>
15103 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15106 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15108 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15110 2017-09-14 Jakub Jelinek <jakub@redhat.com>
15113 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
15114 if and where to split a bb, except for splitting before debug insn
15115 sequences followed by non-label real insn. Delete debug insns
15116 in between basic blocks.
15118 * combine.c (make_compound_operation_int): Formatting fixes.
15120 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
15121 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15122 * config/netbsd.h (LINK_EH_SPEC): Likewise.
15123 * config/sol2.h (LINK_EH_SPEC): Likewise.
15124 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15125 * config/s390/linux.h (LINK_SPEC): Likewise.
15126 * config/freebsd.h (LINK_EH_SPEC): Likewise.
15127 * config/openbsd.h (LINK_EH_SPEC): Likewise.
15128 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15129 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
15130 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
15131 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15132 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
15133 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
15135 2017-09-13 Jakub Jelinek <jakub@redhat.com>
15137 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
15139 (ENDFILE_LINUX_SPEC): Likewise.
15140 (LINK_EH_SPEC): Likewise.
15141 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
15142 (LINK_OS_LINUX_SPEC32): Likewise.
15143 (LINK_OS_LINUX_SPEC64): Likewise.
15144 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
15145 (LINK_OS_LINUX_SPEC): Likewise.
15147 2017-09-13 Martin Liska <mliska@suse.cz>
15149 PR middle-end/82154
15150 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
15151 CASE_HIGH is NULL_TREE.
15153 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15154 Alan Hayward <alan.hayward@arm.com>
15155 David Sherwood <david.sherwood@arm.com>
15157 * target.def (secondary_memory_needed): New hook.
15158 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
15159 instead of SECONDARY_MEMORY_NEEDED.
15160 (secondary_memory_needed_mode): Likewise.
15161 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
15162 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
15163 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
15164 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
15165 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
15166 * doc/tm.texi: Regenerate.
15167 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
15168 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
15169 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15170 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
15171 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
15172 * config/i386/i386.c (inline_secondary_memory_needed): Put the
15173 mode argument first and change the reg_class arguments to reg_class_t.
15174 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
15175 Make static. Update the call to inline_secondary_memory_needed.
15176 (ix86_register_move_cost): Update the call to
15177 inline_secondary_memory_needed.
15178 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15179 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
15181 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
15182 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15184 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
15185 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
15186 * config/mips/mips.c (mips_secondary_memory_needed): Make static
15187 and match hook interface. Add comment from mips.h.
15188 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15189 * config/mmix/mmix.md (truncdfsf2): Refer to
15190 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15192 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
15193 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
15194 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15195 (pa_secondary_memory_needed): New function.
15196 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
15197 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
15198 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15199 (pdp11_secondary_memory_needed): Make static and match hook interface.
15200 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
15201 * config/powerpcspe/powerpcspe-protos.h
15202 (rs6000_secondary_memory_needed_ptr): Delete.
15203 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
15205 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15206 (rs6000_option_override_internal): Assign to
15207 targetm.secondary_memory_needed rather than
15208 rs6000_secondary_memory_needed_ptr.
15209 (rs6000_secondary_memory_needed): Match hook interface.
15210 (rs6000_debug_secondary_memory_needed): Likewise.
15211 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
15212 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
15213 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
15214 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15215 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
15216 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
15218 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
15219 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15220 (rs6000_option_override_internal): Assign to
15221 targetm.secondary_memory_needed rather than
15222 rs6000_secondary_memory_needed_ptr.
15223 (rs6000_secondary_memory_needed): Match hook interface.
15224 (rs6000_debug_secondary_memory_needed): Likewise.
15225 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
15226 * config/s390/s390.c (s390_secondary_memory_needed): New function.
15227 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15228 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
15229 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15230 (sparc_secondary_memory_needed): New function.
15231 * lra-constraints.c (check_and_process_move): Refer to
15232 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15234 (curr_insn_transform): Likewise.
15235 (process_alt_operands): Use targetm.secondary_memory_needed
15236 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15237 (check_secondary_memory_needed_p): Likewise.
15238 (choose_split_class): Likewise.
15239 * reload.c: Unconditionally include code that was previously
15240 conditional on SECONDARY_MEMORY_NEEDED.
15241 (push_secondary_reload): Use targetm.secondary_memory_needed
15242 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15243 (push_reload): Likewise.
15244 * reload1.c: Unconditionally include code that was previously
15245 conditional on SECONDARY_MEMORY_NEEDED.
15246 (choose_reload_regs): Use targetm.secondary_memory_needed
15247 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15248 (gen_reload): Likewise.
15249 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
15251 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15252 Alan Hayward <alan.hayward@arm.com>
15253 David Sherwood <david.sherwood@arm.com>
15255 * target.def (secondary_memory_needed_mode): New hook:
15256 * targhooks.c (default_secondary_memory_needed_mode): Declare.
15257 * targhooks.h (default_secondary_memory_needed_mode): New function.
15258 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
15259 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
15260 * doc/tm.texi: Regenerate.
15261 * lra-constraints.c (check_and_process_move): Use
15262 targetm.secondary_memory_needed_mode instead of
15263 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
15264 (curr_insn_transform): Likewise.
15265 * reload.c (get_secondary_mem): Likewise.
15266 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15267 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
15269 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15270 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15271 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
15272 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15273 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
15275 * config/powerpcspe/powerpcspe-protos.h
15276 (rs6000_secondary_memory_needed_mode): Delete.
15277 * config/powerpcspe/powerpcspe.c
15278 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15279 (rs6000_secondary_memory_needed_mode): Make static.
15280 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15281 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
15283 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15285 (rs6000_secondary_memory_needed_mode): Make static.
15286 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15287 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
15288 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15289 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15290 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15292 (sparc_secondary_memory_needed_mode): New function.
15293 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
15295 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
15297 * config/aarch64/constraints.md (Umq): New constraint.
15298 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15300 (mov<mode>): Update condition.
15302 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15304 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
15305 when bitposition is the same.
15307 2017-09-13 Richard Biener <rguenther@suse.de>
15309 * dwarf2out.c (output_die_symbol): Remove.
15310 (output_die): Do not output a DIEs symbol.
15312 2017-09-13 Richard Biener <rguenther@suse.de>
15314 PR middle-end/82128
15315 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
15316 default-def to avoid breaking iterator update with the weird
15317 interaction with cgraph_update_edges_for_call_stmt_node.
15319 2017-09-13 Richard Biener <rguenther@suse.de>
15321 * tree-cfg.c (verify_gimple_assign_binary): Add verification
15322 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
15323 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
15324 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
15326 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
15328 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
15329 Disable pc relative literal load irrespective of
15330 TARGET_FIX_ERR_A53_84341 for default.
15332 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
15334 * config/sparc/sparc.c (output_return): Output the source location of
15335 the insn in the delay slot, if any.
15336 (output_sibcall): Likewise.
15338 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
15341 * common.opt (-static-pie): New alias.
15342 (shared): Negate static-pie.
15343 (-no-pie): Update help text.
15345 (static-pie): New option.
15346 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
15347 -static-pie support.
15348 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
15349 (LINK_EH_SPEC): Likewise.
15350 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15351 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15352 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15353 * gcc.c (LINK_COMMAND_SPEC): Likewise.
15354 (init_gcc_specs): Likewise.
15355 (init_spec): Likewise.
15356 (display_help): Update help message for -pie.
15357 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
15360 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
15362 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
15363 (movdi_aarch64): Likewise.
15364 (movti_aarch64): Likewise.
15366 2017-09-12 Simon Wright <simon@pushface.org>
15369 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
15370 calculation of the minor version, always output as 0.
15372 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15375 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15376 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
15377 on it early, rather than manual conversion late. For
15378 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
15379 instead of performing manual conversion.
15381 2017-09-12 Carl Love <cel@us.ibm.com>
15383 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
15384 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
15386 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15387 VMULOSW): Add definitions.
15388 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15389 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15390 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15391 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15392 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15394 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15396 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
15398 (movti_aarch64): Likewise.
15399 (movdf_aarch64): Likewise.
15400 (movtf_aarch64): Likewise.
15401 (load_pairdi): Likewise.
15402 (store_pairdi): Likewise.
15403 (load_pairdf): Likewise.
15404 (store_pairdf): Likewise.
15405 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
15406 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
15407 (ldr_got_small_<mode>): Likewise.
15408 (ldr_got_small_28k_<mode>): Likewise.
15409 (ldr_got_tiny): Likewise.
15410 * config/aarch64/iterators.md (ldst_sz): New.
15411 (ldpstp_sz): Likewise.
15412 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
15414 (thunderx_load): Split load_8 to load_16.
15415 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
15417 (thunderx2t99_storepair_basic): Split store_8 to store_16.
15418 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
15419 (xgene1_store_pair): Split store_8 to store_16.
15420 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
15421 (falkor_st_0_st_sd): Split store_8 to store_16.
15423 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15425 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
15426 and store1/2/3/4 to store_4/8/12/16.
15427 * config/aarch64/aarch64.md: Update for rename.
15428 * config/arm/arm.md: Likewise.: Likewise.
15429 * config/arm/arm.c: Likewise.
15430 * config/arm/thumb1.md: Likewise.
15431 * config/arm/thumb2.md: Likewise.
15432 * config/arm/vfp.md: Likewise.
15433 * config/arm/arm-generic.md: Likewise.
15434 * config/arm/arm1020e.md: Likewise.
15435 * config/arm/arm1026ejs.md: Likewise.
15436 * config/arm/arm1136jfs.md: Likewise.
15437 * config/arm/arm926ejs.md: Likewise.
15438 * config/arm/cortex-a15.md: Likewise.
15439 * config/arm/cortex-a17.md: Likewise.
15440 * config/arm/cortex-a5.md: Likewise.
15441 * config/arm/cortex-a53.md: Likewise.
15442 * config/arm/cortex-a57.md: Likewise.
15443 * config/arm/cortex-a7.md: Likewise.
15444 * config/arm/cortex-a8.md: Likewise.
15445 * config/arm/cortex-a9.md: Likewise.
15446 * config/arm/cortex-m4.md: Likewise.
15447 * config/arm/cortex-m7.md: Likewise.
15448 * config/arm/cortex-r4.md: Likewise.
15449 * config/arm/exynos-m1.md: Likewise.
15450 * config/arm/fa526.md: Likewise.
15451 * config/arm/fa606te.md: Likewise.
15452 * config/arm/fa626te.md: Likewise.
15453 * config/arm/fa726te.md: Likewise.
15454 * config/arm/fmp626.md: Likewise.
15455 * config/arm/iwmmxt.md: Likewise.
15456 * config/arm/ldmstm.md: Likewise.
15457 * config/arm/marvell-pj4.md: Likewise.
15458 * config/arm/xgene1.md: Likewise.
15459 * config/aarch64/thunderx.md: Likewise.
15460 * config/aarch64/thunderx2t99.md: Likewise.
15461 * config/aarch64/falkor.md: Likewise.
15463 2017-09-12 Martin Liska <mliska@suse.cz>
15465 * attribs.c (private_lookup_attribute): New function.
15466 * attribs.h (private_lookup_attribute): Declared here.
15467 (lookup_attribute): Called from this place.
15469 2017-09-12 Richard Biener <rguenther@suse.de>
15471 PR tree-optimization/82157
15472 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
15473 stmts with side-effects.
15475 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15476 Alan Hayward <alan.hayward@arm.com>
15477 David Sherwood <david.sherwood@arm.com>
15479 * target.def (hard_regno_nregs): New hook.
15480 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
15481 * targhooks.h (default_hard_regno_nregs): Declare.
15482 * targhooks.c (default_hard_regno_nregs): New function.
15483 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
15484 (TARGET_HARD_REGNO_NREGS): ...this hook.
15485 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
15486 (CLASS_MAX_NREGS): Likewise.
15487 * doc/tm.texi: Regenerate.
15488 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
15489 instead of HARD_REGNO_NREGS.
15490 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
15491 HARD_REGNO_NREGS in the comment.
15492 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
15493 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
15494 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
15495 Return an unsigned int.
15496 (TARGET_HARD_REGNO_NREGS): Redefine.
15497 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
15498 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
15499 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15500 (arc_hard_regno_nregs): New function.
15501 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
15502 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
15503 (arm_hard_regno_nregs): New function.
15504 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
15505 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
15506 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
15507 (TARGET_HARD_REGNO_NREGS): Redefine.
15508 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
15509 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
15510 (HARD_REGNO_NREGS): Delete.
15511 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
15512 (cr16_hard_regno_nregs): New function.
15513 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
15514 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
15515 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
15516 (cris_hard_regno_nregs): New function.
15517 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
15518 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
15519 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15520 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
15521 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
15522 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
15523 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
15524 (frv_hard_regno_nregs): Make static. Take and return an
15526 (frv_class_max_nregs): Remove outdated copy of documentation.
15527 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
15528 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
15529 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
15530 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
15531 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
15532 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
15533 (TARGET_HARD_REGNO_NREGS): Redefine.
15534 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
15535 (CLASS_MAX_NREGS): Update comment.
15536 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
15537 (ia64_hard_regno_nregs): New function.
15538 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
15539 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
15540 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
15541 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
15542 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
15544 (m32c_hard_regno_nregs): Likewise. Make static.
15545 (TARGET_HARD_REGNO_NREGS): Redefine.
15546 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
15547 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
15548 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
15549 (m68k_hard_regno_nregs): New function.
15550 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
15551 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
15552 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
15553 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
15554 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
15555 Take and return an unsigned int.
15556 (TARGET_HARD_REGNO_NREGS): Redefine.
15557 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
15558 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15559 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
15560 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
15561 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
15562 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
15563 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
15564 (msp430_hard_regno_nregs): Make static. Take and return an
15566 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
15567 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
15568 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
15569 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
15570 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
15571 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
15572 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
15573 (TARGET_HARD_REGNO_NREGS): Redefine.
15574 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
15575 (PA_HARD_REGNO_NREGS): ...this.
15576 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
15577 (PA_HARD_REGNO_NREGS): ...this.
15578 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15579 (pa_hard_regno_nregs): New function.
15580 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
15581 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
15582 (pdp11_hard_regno_nregs): New function.
15583 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
15584 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
15585 (rs6000_hard_regno_nregs_hook): New function.
15586 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
15587 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
15588 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
15589 Take and return an unsigned int. Move earlier in file.
15590 (TARGET_HARD_REGNO_NREGS): Redefine.
15591 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
15592 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
15593 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
15594 (rl78_hard_regno_nregs): Make static. Take and return an
15596 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
15597 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
15598 (rs6000_hard_regno_nregs_hook): New function.
15599 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
15600 * config/rx/rx.c (rx_hard_regno_nregs): New function.
15601 (TARGET_HARD_REGNO_NREGS): Redefine.
15602 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
15603 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
15604 instead of HARD_REGNO_NREGS.
15605 (s390_hard_regno_nregs): New function.
15606 (s390_hard_regno_mode_ok): Add comment from s390.h.
15607 (TARGET_HARD_REGNO_NREGS): Redefine.
15608 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
15609 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
15610 (sh_hard_regno_nregs): New function.
15611 (sh_pass_in_reg_p): Use it.
15612 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
15613 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15614 (sparc_hard_regno_nregs): New function.
15615 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
15616 * config/spu/spu.c (spu_hard_regno_nregs): New function.
15617 (spu_function_arg_advance): Use it, supplying a valid register number.
15618 (TARGET_HARD_REGNO_NREGS): Redefine.
15619 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
15620 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
15621 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
15622 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15623 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15624 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15625 (CLASS_MAX_NREGS): Remove copy of old documentation.
15626 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15627 (visium_hard_regno_nregs): New function.
15628 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15629 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15630 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15631 xtensa_hard_regno_nregs): New function.
15632 * system.h (HARD_REGNO_NREGS): Poison.
15634 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15636 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15637 hard_regno_nregs instead of HARD_REGNO_NREGS.
15638 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15639 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15640 (c6x_expand_epilogue): Likewise.
15641 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15642 (frv_read_iacc_argument): Likewise.
15643 * config/sh/sh.c: Include regs.h.
15644 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15645 (regs_used): Likewise.
15646 (output_stack_adjust): Likewise.
15647 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15648 * expmed.c: Include regs.h.
15649 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15650 * ree.c: Include regs.h.
15651 (combine_reaching_defs): Use hard_regno_nregs instead of
15653 (add_removable_extension): Likewise.
15655 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15657 * regs.h (hard_regno_nregs): Turn into a function.
15658 (end_hard_regno): Update accordingly.
15659 * caller-save.c (setup_save_areas): Likewise.
15660 (save_call_clobbered_regs): Likewise.
15661 (replace_reg_with_saved_mem): Likewise.
15662 (insert_restore): Likewise.
15663 (insert_save): Likewise.
15664 * combine.c (can_change_dest_mode): Likewise.
15665 (move_deaths): Likewise.
15666 (distribute_notes): Likewise.
15667 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15668 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15669 (rs6000_split_multireg_move): Likewise.
15670 (rs6000_register_move_cost): Likewise.
15671 (rs6000_memory_move_cost): Likewise.
15672 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15673 (rs6000_split_multireg_move): Likewise.
15674 (rs6000_register_move_cost): Likewise.
15675 (rs6000_memory_move_cost): Likewise.
15676 * cselib.c (cselib_reset_table): Likewise.
15677 (cselib_lookup_1): Likewise.
15678 * emit-rtl.c (set_mode_and_regno): Likewise.
15679 * function.c (aggregate_value_p): Likewise.
15680 * ira-color.c (setup_profitable_hard_regs): Likewise.
15681 (check_hard_reg_p): Likewise.
15682 (calculate_saved_nregs): Likewise.
15683 (assign_hard_reg): Likewise.
15684 (improve_allocation): Likewise.
15685 (calculate_spill_cost): Likewise.
15686 * ira-emit.c (modify_move_list): Likewise.
15687 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15688 (ira_hard_reg_in_set_p): Likewise.
15689 * ira.c (setup_reg_mode_hard_regset): Likewise.
15690 (clarify_prohibited_class_mode_regs): Likewise.
15691 (check_allocation): Likewise.
15692 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15693 (lra_setup_reg_renumber): Likewise.
15694 (setup_try_hard_regno_pseudos): Likewise.
15695 (spill_for): Likewise.
15696 (assign_hard_regno): Likewise.
15697 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15698 * lra-constraints.c (in_class_p): Likewise.
15699 (lra_constraint_offset): Likewise.
15700 (simplify_operand_subreg): Likewise.
15701 (lra_constraints): Likewise.
15702 (split_reg): Likewise.
15703 (split_if_necessary): Likewise.
15704 (invariant_p): Likewise.
15705 (inherit_in_ebb): Likewise.
15706 * lra-lives.c (process_bb_lives): Likewise.
15707 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15708 (get_hard_regs): Likewise.
15709 (do_remat): Likewise.
15710 * lra-spills.c (assign_spill_hard_regs): Likewise.
15711 * mode-switching.c (create_pre_exit): Likewise.
15712 * postreload.c (reload_combine_recognize_pattern): Likewise.
15713 * recog.c (peep2_find_free_register): Likewise.
15714 * regcprop.c (kill_value_regno): Likewise.
15715 (set_value_regno): Likewise.
15716 (copy_value): Likewise.
15717 (maybe_mode_change): Likewise.
15718 (find_oldest_value_reg): Likewise.
15719 (copyprop_hardreg_forward_1): Likewise.
15720 * regrename.c (check_new_reg_p): Likewise.
15721 (regrename_do_replace): Likewise.
15722 * reload.c (push_reload): Likewise.
15723 (combine_reloads): Likewise.
15724 (find_dummy_reload): Likewise.
15725 (operands_match_p): Likewise.
15726 (find_reloads): Likewise.
15727 (find_equiv_reg): Likewise.
15728 (reload_adjust_reg_for_mode): Likewise.
15729 * reload1.c (count_pseudo): Likewise.
15730 (count_spilled_pseudo): Likewise.
15731 (find_reg): Likewise.
15732 (clear_reload_reg_in_use): Likewise.
15733 (free_for_value_p): Likewise.
15734 (allocate_reload_reg): Likewise.
15735 (choose_reload_regs): Likewise.
15736 (reload_adjust_reg_for_temp): Likewise.
15737 (emit_reload_insns): Likewise.
15738 (delete_output_reload): Likewise.
15739 * rtlanal.c (subreg_get_info): Likewise.
15740 * sched-deps.c (sched_analyze_reg): Likewise.
15741 * sel-sched.c (init_regs_for_mode): Likewise.
15742 (mark_unavailable_hard_regs): Likewise.
15743 (choose_best_reg_1): Likewise.
15744 (verify_target_availability): Likewise.
15745 * valtrack.c (dead_debug_insert_temp): Likewise.
15746 * var-tracking.c (track_loc_p): Likewise.
15747 (emit_note_insn_var_location): Likewise.
15748 * varasm.c (make_decl_rtl): Likewise.
15749 * reginfo.c (choose_hard_reg_mode): Likewise.
15750 (init_reg_modes_target): Refer directly to
15751 this_target_regs->x_hard_regno_nregs.
15753 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15755 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15756 instead of hard_regno_nregs.
15758 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15760 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15761 end_hard_regno instead of hard_regno_nregs.
15762 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15763 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15764 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15765 * ira-color.c (improve_allocation): Likewise.
15766 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15767 * lra-lives.c (mark_regno_live): Likewise.
15768 (mark_regno_dead): Likewise.
15769 * lra-remat.c (operand_to_remat): Likewise.
15770 * lra.c (collect_non_operand_hard_regs): Likewise.
15771 * postreload.c (reload_combine_note_store): Likewise.
15772 (move2add_valid_value_p): Likewise.
15773 * reload.c (regno_clobbered_p): Likewise.
15775 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15777 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15779 * config/v850/v850.c (v850_reorg): Likewise.
15780 * reload.c (refers_to_regno_for_reload_p): Likewise.
15781 (find_equiv_reg): Likewise.
15782 * reload1.c (reload_reg_reaches_end_p): Likewise.
15784 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15786 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15788 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15789 * config/arm/arm.c (output_move_neon): Likewise.
15790 (arm_attr_length_move_neon): Likewise.
15791 (neon_split_vcombine): Likewise.
15792 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15793 (c6x_mark_reg_written): Likewise.
15794 (c6x_dwarf_register_span): Likewise.
15795 * config/i386/i386.c (ix86_save_reg): Likewise.
15796 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15797 (rws_access_reg): Likewise.
15798 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15799 * mode-switching.c (create_pre_exit): Likewise.
15800 * ree.c (combine_reaching_defs): Likewise.
15801 (add_removable_extension): Likewise.
15802 * regcprop.c (find_oldest_value_reg): Likewise.
15803 (copyprop_hardreg_forward_1): Likewise.
15804 * reload.c (reload_inner_reg_of_subreg): Likewise.
15805 (push_reload): Likewise.
15806 (combine_reloads): Likewise.
15807 (find_dummy_reload): Likewise.
15808 (reload_adjust_reg_for_mode): Likewise.
15809 * reload1.c (find_reload_regs): Likewise.
15810 (forget_old_reloads_1): Likewise.
15811 (reload_reg_free_for_value_p): Likewise.
15812 (reload_adjust_reg_for_temp): Likewise.
15813 (emit_reload_insns): Likewise.
15814 (delete_output_reload): Likewise.
15815 * sel-sched.c (choose_best_reg_1): Likewise.
15816 (choose_best_pseudo_reg): Likewise.
15818 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15819 Alan Hayward <alan.hayward@arm.com>
15820 David Sherwood <david.sherwood@arm.com>
15822 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15823 * target.def (slow_unaligned_access): New hook.
15824 * targhooks.h (default_slow_unaligned_access): Declare.
15825 * targhooks.c (default_slow_unaligned_access): New function.
15826 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15827 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15828 * doc/tm.texi: Regenerate.
15829 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15830 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15831 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15833 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15834 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15836 (rs6000_slow_unaligned_access): New function.
15837 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15838 (expand_block_compare): Likewise.
15839 (expand_strn_compare): Likewise.
15840 (rs6000_rtx_costs): Likewise.
15841 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15842 (riscv_slow_unaligned_access): Likewise.
15843 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15844 (riscv_slow_unaligned_access_p): ...this and make static.
15845 (riscv_option_override): Update accordingly.
15846 (riscv_slow_unaligned_access): New function.
15847 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15848 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15849 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15850 (rs6000_slow_unaligned_access): New function.
15851 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15852 (rs6000_rtx_costs): Likewise.
15853 * config/rs6000/rs6000-string.c (expand_block_compare)
15854 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15855 of SLOW_UNALIGNED_ACCESS.
15856 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15857 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15858 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15859 of SLOW_UNALIGNED_ACCESS.
15860 * expmed.c (simple_mem_bitfield_p): Likewise.
15861 * expr.c (alignment_for_piecewise_move): Likewise.
15862 (emit_group_load_1): Likewise.
15863 (emit_group_store): Likewise.
15864 (copy_blkmode_from_reg): Likewise.
15865 (emit_push_insn): Likewise.
15866 (expand_assignment): Likewise.
15867 (store_field): Likewise.
15868 (expand_expr_real_1): Likewise.
15869 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15870 * lra-constraints.c (simplify_operand_subreg): Likewise.
15871 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15872 * gimple-ssa-store-merging.c: Likewise in block comment at start
15874 * tree-ssa-strlen.c: Include target.h.
15875 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
15876 of SLOW_UNALIGNED_ACCESS.
15877 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
15879 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15881 PR rtl-optimization/82185
15882 * expmed.c (emit_store_flag_int): Only test tem if it has been
15885 2017-09-12 Richard Biener <rguenther@suse.de>
15887 PR middle-end/82149
15888 * match.pd ((FTYPE) N CMP CST): Fix typo.
15890 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
15892 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
15894 (mips_near_type_p): Add 'short_call' attribute as a synonym
15896 * doc/extend.texi (short_call): Document new function attribute.
15898 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15901 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
15902 assertion check that in the condition.
15903 (get_atomic_generic_size): Likewise. Before testing if parameter
15904 has pointer type, if it has array type, call for C++
15905 default_conversion to perform array-to-pointer conversion.
15907 2017-09-12 Richard Biener <rguenther@suse.de>
15909 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
15910 for operations we cannot scalarize.
15912 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
15914 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
15915 vectors heap vectors. Clean up comments.
15916 Make visited_bbs a reference.
15917 (profitable_jump_thread_path): Make GC
15918 vectors heap vectors. Clean up comments.
15920 (convert_and_register_jump_thread_path): Make GC vectors heap
15922 (check_subpath_and_update_thread_path): Same. Clean up comments.
15923 Make visited_bbs a reference.
15924 (handle_phi): Abstract common code to to
15925 register_jump_thread_path_if_profitable.
15926 Rename VAR_BB to DEF_BB.
15928 Make GC vectors heap vectors.
15929 Make visited_bbs a reference.
15930 (handle_assignment): Same.
15931 (register_jump_thread_path_if_profitable): New.
15932 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
15934 Make GC vectors heap vectors. Clean up comments.
15935 Make visited_bbs a reference.
15936 (find_jump_threads_backwards): Make visited_bbs live in the stack.
15937 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
15940 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
15943 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
15944 words of E_DImode object are reachable by xtensa_uimm8x4 access.
15946 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
15948 Revert r251800 and r251799.
15950 2017-09-11 Martin Jambor <mjambor@suse.cz>
15953 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
15954 arguments in advance.
15955 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
15956 use it to find predecessor edges.
15957 (naive_outof_ssa): Collect vector of predecessors.
15959 2017-09-08 Jason Merrill <jason@redhat.com>
15961 PR c++/70029 - ICE with ref-qualifier and -flto
15962 * langhooks.h (struct lang_hooks_for_types): Add
15963 copy_lang_qualifiers.
15964 * attribs.c (build_type_attribute_qual_variant): Use it.
15965 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
15967 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
15968 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
15970 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
15973 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
15974 (*mulsi3_sp64): New instruction.
15975 (mulsi3): New expander.
15977 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
15979 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
15981 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15983 * sancov.c: Include memmodel.h.
15985 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
15988 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
15991 2017-09-07 Carl Love <cel@us.ibm.com>
15993 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
15994 the sldi instruction.
15996 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
15998 * sancov.c: Include tm_p.h.
16000 2017-09-07 Jakub Jelinek <jakub@redhat.com>
16003 * output.h (switch_to_other_text_partition): New declaration.
16004 * varasm.c (switch_to_other_text_partition): New function.
16005 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16006 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16007 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16008 to the other text partition before emitting LCL label and switch back
16009 after emitting the word after it.
16011 2017-09-07 Richard Biener <rguenther@suse.de>
16013 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16014 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16015 critical edges here, after loop init.
16016 (pass_data_pre): Remove PROP_no_crit_edges flags.
16017 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16018 for valueization of call args to avoid leaking VN_TOP.
16019 (visit_use): Assert we do not visit default defs.
16020 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16021 Use error_mark_node to more easily detect leaking VN_TOP.
16022 All default-defs are varying, not VN_TOP. Mark them visited.
16023 (run_scc_vn): Make code match comment.
16025 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16027 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16028 OPTION_MASK_FLOAT128_KEYWORD.
16029 (POWERPC_MASKS): Likewise.
16030 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16031 support for the -mfloat128-type option, and make -mfloat128
16032 default on PowerPC Linux systems. Define or undefine
16033 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16034 Define __float128 to be __ieee128 if IEEE 128-bit support is
16035 enabled, or undefine it.
16036 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16037 Delete defining __FLOAT128_TYPE__.
16038 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16039 -mfloat128-type option and make -mfloat128 default on PowerPC
16041 (TARGET_FLOAT128_TYPE): Likewise.
16042 (-mfloat128-type): Likewise.
16043 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16044 Delete the -mfloat128-type option and make -mfloat128 default on
16045 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16046 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16047 128-bit floating point is enabled. Change tests from using
16048 -mfloat128-type to -mfloat128.
16049 (rs6000_mangle_type): Use the correct mangling for the __float128
16050 type even if normal long double is restricted to 64-bits.
16051 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16053 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16054 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16055 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16057 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16058 documentation for -mfloat128.
16060 2017-09-06 Olivier Hainque <hainque@adacore.com>
16062 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16064 2017-09-06 Wish Wu <wishwu007@gmail.com>
16065 Jakub Jelinek <jakub@redhat.com>
16067 * asan.c (initialize_sanitizer_builtins): Add
16068 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16069 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16070 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16071 BT_FN_VOID_UINT64_PTR variables.
16072 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16073 (BT_FN_VOID_UINT16_UINT16): Likewise.
16074 (BT_FN_VOID_UINT32_UINT32): Likewise.
16075 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16076 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16077 (BT_FN_VOID_UINT64_PTR): Likewise.
16078 * common.opt (flag_sanitize_coverage): New variable.
16079 (fsanitize-coverage=trace-pc): Remove.
16080 (fsanitize-coverage=): Add.
16081 * flag-types.h (enum sanitize_coverage_code): New enum.
16082 * fold-const.c (fold_range_test): Disable non-short-circuit
16083 optimization if flag_sanitize_coverage.
16084 (fold_truth_andor): Likewise.
16085 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16086 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16087 (coverage_sanitizer_opts): New array.
16088 (get_closest_sanitizer_option): Add OPTS argument, handle also
16089 OPT_fsanitize_coverage_.
16090 (parse_sanitizer_options): Adjusted to also handle
16091 OPT_fsanitize_coverage_.
16092 (common_handle_option): Add OPT_fsanitize_coverage_.
16093 * sancov.c (instrument_comparison, instrument_switch): New function.
16094 (sancov_pass): Add trace-cmp support.
16095 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16096 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16097 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16098 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16099 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16100 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16101 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16102 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16103 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16104 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16106 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16108 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16109 error. Only quit immediately if parsing is complete.
16110 (BEGIN): Initialize fatal_err and parse_done.
16111 (begin fpu, end fpu): Check number of arguments.
16112 (begin arch, end arch): Likewise.
16113 (begin cpu, end cpu): Likewise.
16114 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
16115 (optalias): Likewise.
16117 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16119 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
16120 * config/arm/arm-isa.h: Delete. Move definitions to ...
16121 * arm-cpus.in: ... here. Use new feature and fgroup values.
16122 * config/arm/arm.c (arm_option_override): Use lower case for feature
16124 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
16125 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
16126 * config/arm/parsecpu.awk (END): Add new command 'isa'.
16128 (print_isa_bits_for): New function.
16129 (gen_isa): New function.
16130 (gen_comm_data): Use print_isa_bits_for.
16131 (define feature): New keyword.
16132 (define fgroup): New keyword.
16133 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
16134 (arm-isa.h): Add rule to generate file.
16135 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
16136 case for feature bit names.
16138 2017-09-06 Richard Biener <rguenther@suse.de>
16140 * tree-ssa-pre.c (NECESSARY): Remove.
16141 (create_expression_by_pieces): Do not touch pass-local flags.
16142 (insert_into_preds_of_block): Likewise.
16143 (do_pre_regular_insertion): Likewise.
16144 (eliminate_insert): Likewise.
16145 (eliminate_dom_walker::before_dom_children): Likewise.
16146 (fini_eliminate): Do not look at inserted_exprs.
16147 (mark_operand_necessary): Remove.
16148 (remove_dead_inserted_code): Replace with simple work-list
16149 algorithm based on inserted_exprs and SSA uses.
16150 (pass_pre::execute): Re-order fini_eliminate and
16151 remove_dead_inserted_code.
16153 2017-09-06 Olivier Hainque <hainque@adacore.com>
16155 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
16156 for VxWorks 7. Adjust surrounding comments.
16158 2017-09-06 Richard Biener <rguenther@suse.de>
16160 * gimple-ssa-strength-reduction.c
16161 (find_candidates_dom_walker::before_dom_children): Also allow
16164 2017-09-06 Richard Biener <rguenther@suse.de>
16166 PR tree-optimization/82108
16167 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
16168 for gap in the non-permutation SLP case.
16170 2017-09-06 Martin Jambor <mjambor@suse.cz>
16172 PR tree-optimization/82078
16173 * tree-sra.c (sort_and_splice_var_accesses): Move call to
16174 add_access_to_work_queue...
16175 (build_accesses_from_assign): ...here.
16176 (propagate_all_subaccesses): Make sure racc is the group
16177 representative, if there is one.
16179 2017-09-06 Jakub Jelinek <jakub@redhat.com>
16181 PR middle-end/82095
16182 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
16185 2017-09-06 Richard Biener <rguenther@suse.de>
16187 * gimple-ssa-strength-reduction.c
16188 (find_candidates_dom_walker::before_doom_children): Use a
16189 type and not a mode check.
16191 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
16194 * config/arm/predicates.md (arm_general_adddi_operand): Create new
16196 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
16198 2017-09-05 Jeff Law <law@redhat.com>
16200 PR tree-optimization/64910
16201 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
16202 cases where we have 3 or more operands.
16204 2017-09-05 Jakub Jelinek <jakub@redhat.com>
16206 PR middle-end/81768
16207 * omp-low.c (lower_omp_for): Recompute tree invariant if
16208 gimple_omp_for_initial/final is ADDR_EXPR.
16210 PR middle-end/81768
16211 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
16212 into gimple val before gimplification fo the COND_EXPR.
16214 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
16216 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
16217 REGION_COPY argument.
16218 (thread_through_all_blocks): Remove unused argument to
16219 duplicate_thread_path.
16221 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16222 Alan Hayward <alan.hayward@arm.com>
16223 David Sherwood <david.sherwood@arm.com>
16225 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
16226 Take a scalar_mode rather than a machine_mode.
16227 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16228 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
16229 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16230 (aarch64_gen_adjusted_ldpstp): Likewise.
16231 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
16233 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16234 Alan Hayward <alan.hayward@arm.com>
16235 David Sherwood <david.sherwood@arm.com>
16237 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
16238 Take a scalar_int_mode instead of a machine_mode.
16239 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16240 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16241 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16242 (aarch64_simd_attr_length_rglist): Delete.
16243 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
16244 a scalar_int_mode instead of a machine_mode.
16245 (aarch64_add_offset): Likewise.
16246 (aarch64_internal_mov_immediate): Likewise
16247 (aarch64_add_constant_internal): Likewise.
16248 (aarch64_add_constant): Likewise.
16249 (aarch64_movw_imm): Likewise.
16250 (aarch64_rtx_arith_op_extract_p): Likewise.
16251 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16252 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16253 Remove assert that the mode isn't a vector.
16254 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16255 (aarch64_expand_mov_immediate): Update calls after above changes.
16256 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
16257 (aarch64_and_bitmask_imm): Check for scalar integer modes.
16258 (aarch64_move_imm): Likewise.
16259 (aarch64_can_const_movi_rtx_p): Likewise.
16260 (aarch64_strip_extend): Likewise.
16261 (aarch64_extr_rtx_p): Likewise.
16262 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
16263 a CONST_INT when the mode parameter is VOIDmode.
16264 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
16266 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16268 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
16269 * stor-layout.c (bitwise_mode_for_mode): Likewise.
16270 (bitwise_type_for_mode): Update accordingly.
16272 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16274 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
16275 * stor-layout.c (mode_for_size_tree): Likewise.
16276 (mode_for_array): Update accordingly.
16277 (layout_decl): Likewise.
16278 (compute_record_mode): Likewise. Only set the mode once.
16280 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16282 * target.def (get_mask_mode): Change return type to opt_mode.
16284 * doc/tm.texi: Regenerate.
16285 * targhooks.h (default_get_mask_mode): Return an opt_mode.
16286 * targhooks.c (default_get_mask_mode): Likewise.
16287 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16288 * optabs-query.c (can_vec_mask_load_store_p): Update use of
16289 targetm.get_mask_mode.
16290 * tree.c (build_truth_vector_type): Likewise.
16292 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16294 * machmode.h (mode_for_vector): Return an opt_mode.
16295 * stor-layout.c (mode_for_vector): Likewise.
16296 (mode_for_int_vector): Update accordingly.
16297 (layout_type): Likewise.
16298 * config/i386/i386.c (emit_memmov): Likewise.
16299 (ix86_expand_set_or_movmem): Likewise.
16300 (ix86_expand_vector_init): Likewise.
16301 (ix86_get_mask_mode): Likewise.
16302 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
16304 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
16305 * expmed.c (extract_bit_field_1): Likewise.
16306 * expr.c (expand_expr_real_2): Likewise.
16307 * optabs-query.c (can_vec_perm_p): Likewise.
16308 (can_vec_mask_load_store_p): Likewise.
16309 * optabs.c (expand_vec_perm): Likewise.
16310 * targhooks.c (default_get_mask_mode): Likewise.
16311 * tree-vect-stmts.c (vectorizable_store): Likewise.
16312 (vectorizable_load): Likewise.
16313 (get_vectype_for_scalar_type_and_size): Likewise.
16315 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16317 * machmode.h (mode_for_int_vector): New function.
16318 * stor-layout.c (mode_for_int_vector): Likewise.
16319 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
16320 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
16321 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16322 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16323 (s390_expand_vcond): Likewise.
16325 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16327 * machmode.h (opt_machine_mode): New type.
16328 (opt_mode<T>): Allow construction from anything that can be
16330 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
16331 (mode_for_size): Return an opt_machine_mode.
16332 * stor-layout.c (mode_for_size): Likewise.
16333 (mode_for_size_tree): Update call accordingly.
16334 (bitwise_mode_for_mode): Likewise.
16335 (make_fract_type): Likewise.
16336 (make_accum_type): Likewise.
16337 * caller-save.c (replace_reg_with_saved_mem): Update call
16339 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16340 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16341 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16342 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16343 * expmed.c (extract_bit_field_1): Likewise.
16344 * reload.c (get_secondary_mem): Likewise.
16345 * varasm.c (assemble_integer): Likewise.
16346 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
16349 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16351 * machmode.h (decimal_float_mode_for_size): New function.
16352 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
16353 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
16354 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
16355 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
16356 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
16357 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
16359 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16361 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
16362 (get_builtin_sync_mode): Likewise.
16363 (expand_ifn_atomic_compare_exchange): Likewise.
16364 (expand_builtin_atomic_clear): Likewise.
16365 (expand_builtin_atomic_test_and_set): Likewise.
16366 (fold_builtin_atomic_always_lock_free): Likewise.
16367 * calls.c (compute_argument_addresses): Likewise.
16368 (emit_library_call_value_1): Likewise.
16369 (store_one_arg): Likewise.
16370 * combine.c (combine_instructions): Likewise.
16371 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
16372 * config/arm/arm.c (arm_function_value): Likewise.
16373 (aapcs_allocate_return_reg): Likewise.
16374 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
16375 * config/i386/i386.c (construct_container): Likewise.
16376 (ix86_gimplify_va_arg): Likewise.
16377 (ix86_expand_sse_cmp): Likewise.
16378 (emit_memmov): Likewise.
16379 (emit_memset): Likewise.
16380 (expand_small_movmem_or_setmem): Likewise.
16381 (ix86_expand_pextr): Likewise.
16382 (ix86_expand_pinsr): Likewise.
16383 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
16384 * config/microblaze/microblaze.c (microblaze_block_move_straight):
16386 * config/mips/mips.c (mips_function_value_1) Likewise.
16387 (mips_block_move_straight): Likewise.
16388 (mips_expand_ins_as_unaligned_store): Likewise.
16389 * config/powerpcspe/powerpcspe.c
16390 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16391 (rs6000_darwin64_record_arg_flush): Likewise.
16392 * config/rs6000/rs6000.c
16393 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16394 (rs6000_darwin64_record_arg_flush): Likewise.
16395 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
16396 (sparc_function_value_1): Likewise.
16397 * config/spu/spu.c (adjust_operand): Likewise.
16398 (spu_emit_branch_or_set): Likewise.
16399 (arith_immediate_p): Likewise.
16400 * emit-rtl.c (gen_lowpart_common): Likewise.
16401 * expr.c (expand_expr_real_1): Likewise.
16402 * function.c (assign_parm_setup_block): Likewise.
16403 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
16404 * reload1.c (alter_reg): Likewise.
16405 * stor-layout.c (mode_for_vector): Likewise.
16406 (layout_type): Likewise.
16408 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16410 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
16411 (spu_convert_move): Likewise.
16412 * lower-subreg.c (resolve_simple_move): Likewise.
16414 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16417 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
16418 define_insn to a define_expand.
16419 (altivec_vsum2sws_direct): New define_insn.
16420 (altivec_vsumsws): Convert from a define_insn to a define_expand.
16422 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16424 * config/arm/arm.c (arm_option_params_internal): Improve setting of
16427 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
16432 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
16433 realign stack if stack alignment needed is less than incoming
16436 2017-09-05 Marek Polacek <polacek@redhat.com>
16439 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
16442 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16444 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
16446 2017-09-05 Richard Biener <rguenther@suse.de>
16448 PR tree-optimization/82084
16449 * fold-const.c (can_native_encode_string_p): Handle wide characters.
16451 2017-09-05 Richard Biener <rguenther@suse.de>
16453 PR tree-optimization/82102
16454 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
16456 2017-09-05 Martin Liska <mliska@suse.cz>
16458 PR tree-optimization/82032
16459 * tree-cfg.c (generate_range_test): New function.
16460 * tree-cfg.h (generate_range_test): Declared here.
16461 * tree-cfgcleanup.c (convert_single_case_switch): New function.
16462 (cleanup_control_expr_graph): Use it.
16463 * tree-switch-conversion.c (try_switch_expansion): Remove
16465 (emit_case_nodes): Use generate_range_test.
16467 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16470 * config/i386/i386.md (*<btsc><mode>_mask): Add
16471 TARGET_USE_BT to insn constraint.
16472 (*btr<mode>_mask): Ditto.
16474 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
16476 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
16477 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
16479 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
16482 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
16483 TARGET_NEON and TARGET_IWMMXT.
16484 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
16485 TARGET_NEON and TARGET_IWMMXT.
16486 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
16488 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16490 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
16491 (ix86_rewrite_tls_address): Ditto.
16492 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
16493 (ix86_rewrite_tls_address_1): Ditto.
16494 (ix86_rewrite_tls_address): Ditto.
16495 * config/i386/predicates.md (tls_address_pattern): New predicate.
16496 * config/i386/i386.md (TLS address splitter): New splitter.
16498 2017-09-04 Richard Biener <rguenther@suse.de>
16500 PR tree-optimization/82084
16501 * fold-const.h (can_native_encode_string_p): Declare.
16502 * fold-const.c (can_native_encode_string_p): Factor out from ...
16503 (native_encode_string): ... here.
16504 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
16505 vectorizing stores from constants we later cannot handle.
16507 2017-09-04 Marek Polacek <polacek@redhat.com>
16510 * doc/invoke.texi: Update -Wtautological-compare documentation.
16512 2017-09-04 Jeff Law <law@redhat.com>
16514 PR tree-optimization/64910
16515 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
16516 swap the first and last operand if the last is a constant.
16518 2017-09-04 Marek Polacek <polacek@redhat.com>
16521 * convert.c (do_narrow): When sanitizing signed integer overflows,
16522 bail out for signed types.
16523 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
16525 2017-09-04 Richard Biener <rguenther@suse.de>
16527 PR tree-optimization/82060
16528 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16529 Move devirtualization after stmt folding and before EH/AB/noreturn
16530 cleanup to get the stmt refs canonicalized. Use a bool instead
16531 of gimple_modified_p since that doesn't work for NOPs. Schedule
16532 NOPs generated by folding for removal.
16534 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16535 Alan Hayward <alan.hayward@arm.com>
16536 David Sherwood <david.sherwood@arm.com>
16538 * coretypes.h (pad_direction): New enum.
16539 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
16540 (FUNCTION_ARG_PADDING): Likewise.
16541 * target.def (function_arg_padding): New hook.
16542 * targhooks.h (default_function_arg_padding): Declare.
16543 * targhooks.c (default_function_arg_padding): New function.
16544 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
16545 (TARGET_FUNCTION_ARG_PADDING): ...this.
16546 * doc/tm.texi: Regenerate.
16547 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16548 instead of direction.
16549 (compute_argument_addresses): Likewise.
16550 (load_register_parameters): Likewise.
16551 (emit_library_call_value_1): Likewise.
16552 (store_one_arg): Use targetm.calls.function_arg_padding instead
16553 of FUNCTION_ARG_PADDING.
16554 (must_pass_in_stack_var_size_or_pad): Likewise.
16555 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
16556 (emit_group_store): Likewise.
16557 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
16558 instead of FUNCTION_ARG_PADDING.
16559 (emit_push_insn): Likewise, and propagate enum change throughout
16561 * function.h (direction): Delete.
16562 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
16564 * function.c (assign_parm_find_stack_rtl): Likewise.
16565 (assign_parm_setup_block_p): Likewise.
16566 (assign_parm_setup_block): Likewise.
16567 (gimplify_parameters): Likewise.
16568 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
16569 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
16571 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
16572 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16573 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
16574 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
16575 (aarch64_function_arg_padding): ...this new function.
16576 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
16577 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16578 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
16579 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16580 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
16581 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16582 (arm_pad_arg_upward): Replace with...
16583 (arm_function_arg_padding): ...this new function.
16584 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
16586 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
16587 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
16588 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16589 (ia64_hpux_function_arg_padding): Replace with...
16590 (ia64_function_arg_padding): ...this new function. Use pad_direction
16591 instead of direction. Check for TARGET_HPUX.
16592 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
16593 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16594 (iq2000_function_arg_padding): New function.
16595 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
16596 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
16597 (mips_function_arg_padding): ...this new function.
16598 (mips_pad_reg_upward): Update accordingly.
16599 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16600 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
16601 targetm.calls.function_arg_padding.
16602 (FUNCTION_ARG_PADDING): Delete.
16603 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16604 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
16605 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16606 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
16607 (nios2_block_reg_padding): Return pad_direction instead of direction.
16608 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
16609 instead of direction.
16610 (nios2_function_arg_padding): Likewise. Make static.
16611 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16612 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
16613 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
16614 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
16615 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16616 (pa_function_arg_padding): Make static. Return pad_direction instead
16618 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
16619 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16620 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
16621 instead of direction. Use targetm.calls.function_arg_padding.
16622 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16623 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16624 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16625 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16626 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16628 (function_arg_padding): Rename to...
16629 (rs6000_function_arg_padding): ...this. Make static. Return
16630 pad_direction instead of direction.
16631 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16632 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16633 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16634 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16635 instead of direction. Use targetm.calls.function_arg_padding.
16636 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16637 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16638 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16639 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16640 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16641 (function_arg_padding): Rename to...
16642 (rs6000_function_arg_padding): ...this. Make static. Return
16643 pad_direction instead of direction.
16644 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16645 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16646 * config/s390/s390.c (s390_function_arg_padding): New function.
16647 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16648 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16649 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16650 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16651 (function_arg_padding): Rename to...
16652 (sparc_function_arg_padding): ...this. Make static. Return
16653 pad_direction instead of direction.
16654 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16655 * config/spu/spu.c (spu_function_arg_padding): New function.
16656 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16657 * system.h (FUNCTION_ARG_PADDING): Poison.
16659 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16660 Alan Hayward <alan.hayward@arm.com>
16661 David Sherwood <david.sherwood@arm.com>
16663 * target.def (modes_tieable_p): New hook.
16664 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16665 (TARGET_MODES_TIEABLE_P): ...this.
16666 * doc/tm.texi.in: Regenerate.
16667 * hooks.h (hook_bool_mode_mode_true): Declare.
16668 * hooks.c (hook_bool_mode_mode_true): New function.
16669 * combine.c (subst): Use targetm.modes_tieable_p instead of
16671 * dse.c (find_shift_sequence): Likewise.
16672 * expmed.c (extract_low_bits): Likewise.
16673 * lower-subreg.c: Include target.h.
16674 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16676 * rtlanal.c (rtx_cost): Likewise.
16677 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16678 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16679 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16680 (TARGET_MODES_TIEABLE_P): Redefine.
16681 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16682 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16683 (TARGET_MODES_TIEABLE_P): Redefine.
16684 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16685 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16686 (arc_modes_tieable_p): New function.
16687 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16688 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16689 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16690 (arm_modes_tieable_p): Make static.
16691 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16692 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16693 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16694 (TARGET_MODES_TIEABLE_P): Redefine.
16695 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16696 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16697 (TARGET_MODES_TIEABLE_P): Redefine.
16698 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16699 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16700 (cr16_modes_tieable_p): New function.
16701 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16702 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16703 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16704 (TRULY_NOOP_TRUNCATION): Update comment.
16705 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16706 (TRULY_NOOP_TRUNCATION): Update comment.
16707 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16708 (frv_modes_tieable_p): New function.
16709 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16710 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16711 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16712 (TARGET_MODES_TIEABLE_P): Redefine.
16713 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16714 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16715 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16716 (TARGET_MODES_TIEABLE_P): Redefine.
16717 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16718 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16719 (ia64_modes_tieable_p): New function.
16720 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16721 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16722 (iq2000_modes_tieable_p): New function.
16723 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16724 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16725 (lm32_modes_tieable_p): New function.
16726 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16727 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16728 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16729 (TARGET_MODES_TIEABLE_P): Redefine.
16730 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16731 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16732 (m32r_modes_tieable_p): New function.
16733 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16734 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16735 (m68k_modes_tieable_p): New function.
16736 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16737 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16738 (mcore_modes_tieable_p): New function.
16739 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16740 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16742 (TARGET_MODES_TIEABLE_P): Redefine.
16743 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16744 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16745 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16746 (TARGET_MODES_TIEABLE_P): Redefine.
16747 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16748 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16749 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16750 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16751 (mn10300_modes_tieable_p): ...this and make static.
16752 (TARGET_MODES_TIEABLE_P): Redefine.
16753 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16754 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16755 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16756 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16757 (msp430_modes_tieable_p): Make static.
16758 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16759 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16760 (TARGET_MODES_TIEABLE_P): Redefine.
16761 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16762 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16763 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16764 (TARGET_MODES_TIEABLE_P): Redefine.
16765 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16766 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16767 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16768 (TARGET_MODES_TIEABLE_P): Redefine.
16769 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16770 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16771 (pdp11_modes_tieable_p): New function.
16772 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16773 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16774 (rs6000_modes_tieable_p): New function.
16775 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16776 * config/powerpcspe/powerpcspe.md: Update comment.
16777 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16778 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16779 (TARGET_MODES_TIEABLE_P): Redefine.
16780 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16781 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16782 (rl78_modes_tieable_p): New function.
16783 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16784 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16785 (rs6000_modes_tieable_p): New function.
16786 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16787 * config/rs6000/rs6000.md: Update comment.
16788 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16789 * config/rx/rx.c (rx_modes_tieable_p): New function.
16790 (TARGET_MODES_TIEABLE_P): Redefine.
16791 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16792 * config/s390/s390.c (s390_modes_tieable_p): New function.
16793 (TARGET_MODES_TIEABLE_P): Redefine.
16794 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16795 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16796 (sh_modes_tieable_p): New function.
16797 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16798 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16799 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16800 (sparc_modes_tieable_p): Make static.
16801 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16802 * config/spu/spu.c (spu_modes_tieable_p): New function.
16803 (TARGET_MODES_TIEABLE_P): Redefine.
16804 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16805 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16806 (TARGET_MODES_TIEABLE_P): Redefine.
16807 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16808 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16809 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16810 * config/v850/v850.c (v850_modes_tieable_p): New function.
16811 (TARGET_MODES_TIEABLE_P): Redefine.
16812 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16813 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16814 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16815 (visium_modes_tieable_p): New function.
16816 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16817 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16818 (xtensa_modes_tieable_p): New function.
16819 * system.h (MODES_TIEABLE_P): Poison.
16821 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16822 Alan Hayward <alan.hayward@arm.com>
16823 David Sherwood <david.sherwood@arm.com>
16825 * target.def (hard_regno_mode_ok): New hook.
16826 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16827 (TARGET_HARD_REGNO_MODE_OK): ...this.
16828 * doc/tm.texi.in: Regenerate.
16829 * hooks.h (hook_bool_uint_mode_true): Declare.
16830 * hooks.c (hook_bool_uint_mode_true): New function.
16831 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16832 HARD_REGNO_MODE_OK.
16833 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16834 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16835 instead of HARD_REGNO_MODE_OK.
16836 * caller-save.c: Include target.h.
16837 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16838 HARD_REGNO_MODE_OK.
16839 * combine.c (can_combine_p): Likewise.
16840 (combinable_i3pat): Likewise.
16841 (can_change_dest_mode): Likewise.
16842 * expr.c (init_expr_target): Likewise.
16843 (convert_move): Likewise.
16844 (convert_modes): Likewise.
16845 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16846 (setup_prohibited_mode_move_regs): Likewise.
16847 * ira.h (target_ira): Likewise.
16848 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16849 * lra-constraints.c (process_alt_operands): Likewise.
16850 (split_reg): Likewise.
16851 * recog.c (peep2_find_free_register): Likewise.
16852 * ree.c (combine_reaching_defs): Likewise.
16853 * regcprop.c (maybe_mode_change): Likewise.
16854 * reginfo.c (init_reg_sets_1): Likewise.
16855 (choose_hard_reg_mode): Likewise.
16856 (simplifiable_subregs): Likewise.
16857 * regrename.c (check_new_reg_p): Likewise.
16858 * reload.c (find_valid_class): Likewise.
16859 (find_valid_class_1): Likewise.
16860 (reload_inner_reg_of_subreg): Likewise.
16861 (push_reload): Likewise.
16862 (combine_reloads): Likewise.
16863 (find_dummy_reload): Likewise.
16864 (find_reloads): Likewise.
16865 * reload1.c (find_reg): Likewise.
16866 (set_reload_reg): Likewise.
16867 (allocate_reload_reg): Likewise.
16868 (choose_reload_regs): Likewise.
16869 (reload_adjust_reg_for_temp): Likewise.
16870 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
16871 (simplify_subreg_regno): Likewise.
16872 * sel-sched.c (init_regs_for_mode): Likewise.
16873 * varasm.c (make_decl_rtl): Likewise.
16874 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
16875 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
16876 HARD_REGNO_MODE_OK.
16877 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
16878 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
16879 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16880 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
16881 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
16882 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16883 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
16884 (arc_mode_class): Delete.
16885 (HARD_REGNO_MODE_OK): Delete.
16886 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16887 (arc_hard_regno_mode_ok): Rename old array to...
16888 (arc_hard_regno_mode_ok_modes): ...this.
16889 (arc_conditional_register_usage): Update accordingly.
16890 (arc_mode_class): Make static.
16891 (arc_hard_regno_mode_ok): New function.
16892 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
16893 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
16894 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16895 (arm_hard_regno_mode_ok): Make static.
16896 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
16897 HARD_REGNO_MODE_OK.
16898 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
16899 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
16900 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
16902 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16903 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
16904 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
16905 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
16906 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
16907 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16908 * config/bfin/predicates.md (valid_reg_operand): Use
16909 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
16910 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
16911 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
16912 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16913 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
16914 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
16915 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16916 (cr16_hard_regno_mode_ok): Make static and return a bool.
16917 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
16918 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16919 (cris_hard_regno_mode_ok): New function.
16920 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
16921 (epiphany_mode_class): Delete.
16922 (HARD_REGNO_MODE_OK): Delete.
16923 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
16924 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16925 (hard_regno_mode_ok): Rename to...
16926 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
16927 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
16928 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
16929 HARD_REGNO_MODE_OK.
16930 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
16931 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
16932 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16933 (frv_hard_regno_mode_ok): Make static and return a bool.
16934 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
16935 HARD_REGNO_MODE_OK.
16936 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
16937 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
16938 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
16939 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
16941 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16942 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
16943 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
16944 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
16946 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16947 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
16948 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16949 (ia64_hard_regno_mode_ok): New function.
16950 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
16951 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16952 (iq2000_hard_regno_mode_ok): New function.
16953 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
16954 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16955 (lm32_hard_regno_mode_ok): New function.
16956 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
16957 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
16958 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
16959 instead of HARD_REGNO_MODE_OK.
16960 (m32c_hard_regno_ok): Rename to...
16961 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
16962 (m32c_cannot_change_mode_class): Update accordingly.
16963 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16964 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
16965 (m32r_mode_class): Delete.
16966 (HARD_REGNO_MODE_OK): Delete.
16967 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16968 (m32r_hard_regno_mode_ok): Rename to...
16969 (m32r_hard_regno_modes): ...this.
16970 (m32r_mode_class): Make static.
16971 (m32r_hard_regno_mode_ok): New function.
16972 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
16973 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
16974 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16975 (m68k_hard_regno_mode_ok): Make static.
16976 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
16977 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16978 (mcore_hard_regno_mode_ok): New function.
16979 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
16980 (HARD_REGNO_MODE_OK): Delete.
16981 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
16983 (microblaze_hard_regno_mode_ok_p): ...this and make static.
16984 (microblaze_hard_regno_mode_ok): New function.
16985 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16986 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
16987 (mips_hard_regno_mode_ok): Delete.
16988 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
16989 (mips_hard_regno_mode_ok_p): ...this and make static.
16990 (mips_hard_regno_mode_ok_p): Rename to...
16991 (mips_hard_regno_mode_ok_uncached): ...this.
16992 (mips_hard_regno_mode_ok): New function.
16993 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
16994 of HARD_REGNO_MODE_OK.
16995 (mips_option_override): Update after above name changes.
16996 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16997 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
16998 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
16999 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
17000 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
17001 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17002 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
17003 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
17004 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17005 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17006 (msp430_hard_regno_mode_ok): Make static and return a bool.
17007 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17008 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17009 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17011 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17012 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17013 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17014 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17015 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17016 (PA_HARD_REGNO_MODE_OK): ...this
17017 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17018 (PA_HARD_REGNO_MODE_OK): ...this.
17019 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17020 (pa_hard_regno_mode_ok): New function.
17021 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17022 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17023 (pdp11_hard_regno_mode_ok): New function.
17024 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17025 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17027 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17029 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17030 (rs6000_hard_regno_mode_ok): Rename to...
17031 (rs6000_hard_regno_mode_ok_uncached): ...this.
17032 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17033 (rs6000_hard_regno_mode_ok): New function.
17034 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17035 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17036 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17037 (riscv_hard_regno_mode_ok): ...this and make static.
17038 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17039 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17040 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17041 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17042 (rl78_hard_regno_mode_ok): Make static and return bool.
17043 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
17044 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17046 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17047 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17048 (rs6000_hard_regno_mode_ok): Rename to...
17049 (rs6000_hard_regno_mode_ok_uncached): ...this.
17050 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17051 (rs6000_hard_regno_mode_ok): New function.
17052 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17053 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17054 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17055 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17056 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17057 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17058 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17059 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17060 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17061 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17062 (sh_hard_regno_mode_ok): Make static.
17063 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17064 instead of HARD_REGNO_MODE_OK.
17065 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17066 (sparc_mode_class): Delete.
17067 (HARD_REGNO_MODE_OK): Delete.
17068 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17069 (hard_regno_mode_classes): Make static.
17070 (sparc_mode_class): Likewise.
17071 (sparc_hard_regno_mode_ok): New function.
17072 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17073 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17074 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17076 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17077 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17078 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17079 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17080 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17081 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17082 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17083 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17084 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17085 (visium_hard_regno_mode_ok): New function.
17086 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17087 instead of HARD_REGNO_MODE_OK.
17088 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17089 (HARD_REGNO_MODE_OK): Delete.
17090 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17091 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17092 (xtensa_option_override): Update accordingly.
17093 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17094 (xtensa_hard_regno_mode_ok): New function.
17095 * system.h (HARD_REGNO_MODE_OK): Poison.
17097 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17098 Alan Hayward <alan.hayward@arm.com>
17099 David Sherwood <david.sherwood@arm.com>
17101 * target.def (hard_regno_call_part_clobbered): New hook.
17102 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17103 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17104 * doc/tm.texi: Regenerate.
17105 * hooks.h (hook_bool_uint_mode_false): Declare.
17106 * hooks.c (hook_bool_uint_mode_false): New function.
17107 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17108 * cselib.c (cselib_process_insn): Use
17109 targetm.hard_regno_call_part_clobbered instead of
17110 HARD_REGNO_CALL_PART_CLOBBERED.
17111 * ira-conflicts.c (ira_build_conflicts): Likewise.
17112 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17113 * lra-constraints.c (need_for_call_save_p): Likewise.
17114 * lra-lives.c: Include target.h.
17115 (check_pseudos_live_through_calls): Use
17116 targetm.hard_regno_call_part_clobbered instead of
17117 HARD_REGNO_CALL_PART_CLOBBERED.
17118 * regcprop.c: Include target.h.
17119 (copyprop_hardreg_forward_1): Use
17120 targetm.hard_regno_call_part_clobbered instead of
17121 HARD_REGNO_CALL_PART_CLOBBERED.
17122 * reginfo.c (choose_hard_reg_mode): Likewise.
17123 * regrename.c (check_new_reg_p): Likewise.
17124 * reload.c (find_equiv_reg): Likewise.
17125 * reload1.c (emit_reload_insns): Likewise.
17126 * sched-deps.c (deps_analyze_insn): Likewise.
17127 * sel-sched.c (init_regs_for_mode): Likewise.
17128 (mark_unavailable_hard_regs): Likewise.
17129 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
17130 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17131 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
17133 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17134 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17135 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
17137 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
17139 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17140 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17141 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
17143 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17144 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17145 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
17147 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17148 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
17150 * config/powerpcspe/powerpcspe.c
17151 (rs6000_hard_regno_call_part_clobbered): New function.
17152 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17153 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17154 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
17156 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17157 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17158 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
17160 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17161 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17162 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
17164 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17165 Alan Hayward <alan.hayward@arm.com>
17166 David Sherwood <david.sherwood@arm.com>
17168 * rtl.h (subreg_memory_offset): Declare.
17169 * emit-rtl.c (subreg_memory_offset): New function.
17170 * expmed.c (store_bit_field_1): Use it.
17171 * expr.c (undefined_operand_subword_p): Likewise.
17172 * simplify-rtx.c (simplify_subreg): Likewise.
17174 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
17176 PR rtl-optimization/57448
17179 * optabs.c (expand_atomic_load): Place compiler memory barriers if
17180 using atomic_load pattern.
17181 (expand_atomic_store): Likewise.
17183 2017-09-04 Jakub Jelinek <jakub@redhat.com>
17186 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
17187 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
17188 handling. Use replace_call_with_value with NULL instead of
17189 gsi_replace, unlink_stmt_vdef and release_defs.
17191 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
17194 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
17196 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17199 * rtl.h (emit_library_call_value_1): Declare.
17200 (emit_library_call): Replace declaration with a series of overloads.
17201 Remove the parameter count argument.
17202 (emit_library_call_value): Likewise.
17203 * calls.c (emit_library_call_value_1): Make global. Replace varargs
17204 with an "rtx_mode_t *".
17205 (emit_library_call_value): Delete.
17206 (emit_library_call): Likewise.
17207 * asan.c (asan_emit_stack_protection): Update calls accordingly.
17208 (asan_emit_allocas_unpoison): Likewise.
17209 * builtins.c (expand_builtin_powi): Likewise.
17210 (expand_asan_emit_allocas_unpoison): Likewise.
17211 * cfgexpand.c (expand_main_function): Likewise.
17212 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
17213 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
17214 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
17215 * config/arm/arm.c (arm_trampoline_init): Likewise.
17216 (arm_call_tls_get_addr): Likewise.
17217 (arm_expand_divmod_libfunc): Likewise.
17218 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
17219 (smulsi3_highpart): Likewise.
17220 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
17221 (c6x_expand_compare): Likewise.
17222 (c6x_expand_movmem): Likewise.
17223 * config/frv/frv.c (frv_trampoline_init): Likewise.
17224 * config/i386/i386.c (ix86_trampoline_init): Likewise.
17225 (ix86_expand_divmod_libfunc): Likewise.
17226 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
17227 (ia64_expand_compare): Likewise.
17228 (ia64_profile_hook): Likewise.
17229 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
17230 (nonlocal_goto): Likewise.
17231 (restore_stack_nonlocal): Likewise.
17232 * config/m32r/m32r.c (block_move_call): Likewise.
17233 (m32r_trampoline_init): Likewise.
17234 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
17235 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
17236 (m68k_call_m68k_read_tp): Likewise.
17237 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
17238 (microblaze_expand_divide): Likewise.
17239 * config/mips/mips.h (mips_args): Likewise.
17240 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
17241 (MIPS_ICACHE_SYNC): Likewise.
17242 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
17243 (nios2_trampoline_init): Likewise.
17244 * config/pa/pa.c (hppa_tls_call): Likewise.
17245 (pa_trampoline_init): Likewise.
17246 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
17247 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
17248 (expand_strn_compare): Likewise.
17249 (rs6000_generate_compare): Likewise.
17250 (rs6000_expand_float128_convert): Likewise.
17251 (output_profile_hook): Likewise.
17252 (rs6000_trampoline_init): Likewise.
17253 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
17254 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
17255 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
17256 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
17257 (rs6000_generate_compare): Likewise.
17258 (rs6000_expand_float128_convert): Likewise.
17259 (output_profile_hook): Likewise.
17260 (rs6000_trampoline_init): Likewise.
17261 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
17262 * config/sh/sh.c (sh_trampoline_init): Likewise.
17263 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
17264 (sparc_emit_float_lib_cmp): Likewise.
17265 (sparc32_initialize_trampoline): Likewise.
17266 (sparc64_initialize_trampoline): Likewise.
17267 (sparc_profile_hook): Likewise.
17268 * config/spu/spu.c (ea_load_store): Likewise.
17269 * config/spu/spu.md (floatunssidf2): Likewise.
17270 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
17271 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
17272 * config/visium/visium.c (expand_block_move_4): Likewise.
17273 (expand_block_move_2): Likewise.
17274 (expand_block_move_1): Likewise.
17275 (expand_block_set_4): Likewise.
17276 (expand_block_set_2): Likewise.
17277 (expand_block_set_1): Likewise.
17278 (visium_trampoline_init): Likewise.
17279 (visium_profile_hook): Likewise.
17280 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
17281 (xtensa_setup_frame_addresses): Likewise.
17282 (xtensa_trampoline_init): Likewise.
17283 * except.c (sjlj_emit_function_enter): Likewise.
17284 (sjlj_emit_function_exit): Likewise.
17285 * explow.c (allocate_dynamic_stack_space): Likewise.
17286 (probe_stack_range): Likewise.
17287 * expr.c (convert_mode_scalar): Likewise.
17288 * optabs.c (expand_binop): Likewise.
17289 (expand_twoval_binop_libfunc): Likewise.
17290 (expand_unop): Likewise.
17291 (prepare_cmp_insn): Likewise.
17292 (prepare_float_lib_cmp): Likewise.
17293 (expand_float): Likewise.
17294 (expand_fix): Likewise.
17295 (expand_fixed_convert): Likewise.
17296 (maybe_emit_sync_lock_test_and_set): Likewise.
17297 (expand_atomic_compare_and_swap): Likewise.
17298 (expand_mem_thread_fence): Likewise.
17299 (expand_atomic_fetch_op): Likewise.
17301 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
17303 * doc/generic.texi (OpenACC): Adjust URL.
17304 * doc/invoke.texi (C Dialect Options): Ditto.
17306 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
17308 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
17309 predicate for operand 1. Add (m,<S>) constraint.
17310 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
17311 Prevent memory operand 1 with register operand 2.
17313 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
17315 PR rtl-optimization/82024
17316 * combine.c (try_combine): If the combination result is a PARALLEL,
17317 and we only need to retain the SET in there that would be placed
17318 at I2, check that we can place that at I3 instead, before doing so.
17320 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17323 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
17325 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
17326 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
17329 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
17330 Jeff Law <law@redhat.com>
17332 * varasm.c (bss_initializer_p): Do not put constants into .bss
17333 (categorize_decl_for_section): Handle bss_initializer_p returning
17334 false when DECL_INITIAL is NULL.
17336 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17339 * config/s390/s390.c (s390_can_inline_p): New function.
17341 2017-09-01 Jeff Law <law@redhat.com>
17343 PR tree-optimization/82052
17344 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17345 Always initialize the returned slot after a hash table miss
17346 when INSERT is true.
17348 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
17350 * config/s390/s390.md (mem_signal_fence): Remove.
17351 * doc/md.texi (mem_signal_fence): Remove.
17352 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
17354 * target-insns.def (mem_signal_fence): Remove.
17356 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17359 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
17362 * asan.c (create_odr_indicator): Strip name encoding from assembler
17363 name before appending it after __odr_asan_.
17365 2017-09-01 Martin Liska <mliska@suse.cz>
17367 PR tree-optimization/82059
17368 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
17369 frequency only when an edge is redirected.
17371 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17373 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
17374 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
17375 (arc_conditional_register_usage): Remove ARC600 lp_count
17377 (arc_file_start): Emit Tag_ARC_CPU_variation.
17378 (arc_can_use_doloop_p): New conditions to use ZOLs.
17379 (hwloop_fail): New function.
17380 (hwloop_optimize): Likewise.
17381 (hwloop_pattern_reg): Likewise.
17382 (arc_doloop_hooks): New struct, to be used with reorg_loops.
17383 (arc_reorg_loops): New function, calls reorg_loops.
17384 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
17385 (arc600_corereg_hazard): Remove ZOL checking, case handled by
17387 (arc_loop_hazard): Remove function, functionality moved into
17389 (arc_hazard): Remove arc_loop_hazard call.
17390 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
17391 into hwloop_optimize.
17392 (arc_label_align): Remove ZOL handling.
17393 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
17394 * config/arc/arc.md (doloop_begin): Remove pattern.
17395 (doloop_begin_i): Likewise.
17396 (doloop_end_i): Likewise.
17397 (doloop_fallback): Likewise.
17398 (doloop_fallback_m): Likewise.
17399 (doloop_end): Reimplement expand.
17400 (arc_lp): New pattern for LP instruction.
17401 (loop_end): New pattern.
17402 (loop_fail): Likewise.
17403 (decrement_and_branch_until_zero): Likewise.
17404 * config/arc/arc.opt (mlpc-width): New option.
17405 * doc/invoke.texi (mlpc-width): Document option.
17407 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17409 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
17410 (arc_ccfsm_advance): Fix checking for delay slots.
17411 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
17413 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17415 * config/arc/arc.md (movqi_insn): Add stores to save constant long
17417 (movhi_insn): Update store instruction constraint which are saving
17418 6-bit short immediates.
17419 (movsi_insn): Consider also short scaled load operations.
17420 (zero_extendhisi2_i): Use Usd constraint instead of T.
17421 (extendhisi2_i): Add q constraint.
17422 (arc_clzsi2): Add type and length attributes.
17423 (arc_ctzsi2): Likewise.
17424 * config/arc/constraints.md (Usc): Update constraint, the
17425 assembler can parse two relocations for a single instruction.
17427 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17429 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
17430 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
17432 2017-08-31 Olivier Hainque <hainque@adacore.com>
17434 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
17435 match as powerpc-wrs-vxworks*.
17437 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
17439 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
17440 register constraint for by-element operand.
17441 (aarch64_mls_elt_merge<mode>): Likewise.
17443 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17445 * config/arc/arc.c (arc_can_follow_jump): Check for short
17448 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17450 * config.gcc: Use g.opt for arc.
17451 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
17452 functionality moved to ...
17453 (legitimate_scaled_address_p): New function, ...here.
17454 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
17455 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
17456 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
17458 (arc_override_options): Handle G option.
17459 (arc_output_pic_addr_const): Correct function definition.
17460 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
17461 (arc_decl_anon_ns_mem_p): Delete.
17462 (arc_in_small_data_p): Overhaul this function to take into
17463 consideration the value given via G option.
17464 (arc_rewrite_small_data_1): Renamed and corrected old
17465 arc_rewrite_small_data function.
17466 (arc_rewrite_small_data): New function.
17467 (small_data_pattern): Don't use pic_offset_table_rtx.
17468 * config/arc/arc.h (CC1_SPEC): Recognize G option.
17469 * config/arc/simdext.md (movmisalignv2hi): Use
17470 prepare_move_operands function.
17472 (movmisalign*): Likewise.
17473 * doc/invoke.texi (ARC options): Document -G option.
17475 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17477 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
17479 * config/arc/arc.c (arc_print_operand): Output scalled address for
17480 sdata whenever is possible.
17481 (arc_in_small_data_p): Allow sdata for 64bit datum when double
17482 load/stores are available.
17483 (compact_sda_memory_operand): Check for the alignment required by
17484 code density instructions.
17485 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
17487 * config/arc/constraints.md (Usd): Update constraint.
17488 (Us0): New constraint.
17489 (Usc): Update constraint.
17491 2017-08-31 Richard Biener <rguenther@suse.de>
17493 PR middle-end/82054
17494 * dwarf2out.c (dwarf2out_early_global_decl): Process each
17495 function only once.
17497 2017-08-31 Tamar Christina <tamar.christina@arm.com>
17499 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
17500 Resize type_signature.
17502 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17503 Alan Hayward <alan.hayward@arm.com>
17504 David Sherwood <david.sherwood@arm.com>
17506 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
17507 subregs whose inner modes can be stored in GPRs.
17508 (aarch64_classify_index): Likewise.
17510 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17511 Alan Hayward <alan.hayward@arm.com>
17512 David Sherwood <david.sherwood@arm.com>
17514 * config/aarch64/iterators.md (V_cmp_result): Rename to...
17515 (V_INT_EQUIV): ...this.
17516 (v_cmp_result): Rename to...
17517 (v_int_equiv): ...this.
17518 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
17519 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
17520 (copysign<mode>3): Likewise.
17521 (aarch64_simd_bsl<mode>_internal): Likewise.
17522 (aarch64_simd_bsl<mode>): Likewise.
17523 (vec_cmp<mode><mode>): Likewise.
17524 (vcond<mode><mode>): Likewise.
17525 (vcond<v_cmp_mixed><mode>): Likewise.
17526 (vcondu<mode><v_cmp_mixed>): Likewise.
17527 (aarch64_cm<optab><mode>): Likewise.
17528 (aarch64_cmtst<mode>): Likewise.
17529 (aarch64_fac<optab><mode>): Likewise.
17530 (vec_perm_const<mode>): Likewise.
17531 (vcond_mask_<mode><v_cmp_result>): Rename to...
17532 (vcond_mask_<mode><v_int_equiv>): ...this.
17533 (vec_cmp<mode><v_cmp_result>): Rename to...
17534 (vec_cmp<mode><v_int_equiv>): ...this.
17536 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17537 Alan Hayward <alan.hayward@arm.com>
17538 David Sherwood <david.sherwood@arm.com>
17540 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
17542 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
17543 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
17544 (UNSPEC_LD4_DREG): New unspecs.
17545 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
17546 (aarch64_ld2<mode>_dreg_be): Replace with...
17547 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
17549 (aarch64_ld3<mode>_dreg_le)
17550 (aarch64_ld3<mode>_dreg_be): Replace with...
17551 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
17553 (aarch64_ld4<mode>_dreg_le)
17554 (aarch64_ld4<mode>_dreg_be): Replace with...
17555 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
17558 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17560 PR tree-optimization/81987
17561 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
17562 insert an initializer in a location not dominated by the stride
17565 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17567 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
17568 on the entire header of the finally block in the fallthru case.
17570 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17572 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
17574 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
17576 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
17577 rs6000_emit_move_from_cr and call renamed function.
17578 (rs6000_emit_prologue): Call renamed functions.
17579 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
17580 movesi_from_cr, remove volatile CRs.
17582 2017-08-30 Jon Beniston <jon@beniston.com>
17583 Richard Biener <rguenther@suse.de>
17585 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
17586 of VECTOR_MODE_P check.
17587 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17588 element vector types.
17590 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17592 * df.h (df_read_modify_subreg_p): Remove in favor of...
17593 * rtl.h (read_modify_subreg_p): ...this new function. Take a
17594 const_rtx instead of an rtx.
17595 * cprop.c (local_cprop_find_used_regs): Update accordingly.
17596 * df-problems.c (df_word_lr_mark_ref): Likewise.
17597 * ira-lives.c (mark_pseudo_reg_live): Likewise.
17598 (mark_pseudo_reg_dead): Likewise.
17599 (mark_ref_dead): Likewise.
17600 * reginfo.c (init_subregs_of_mode): Likewise.
17601 * sched-deps.c (sched_analyze_1): Likewise.
17602 * df-scan.c (df_def_record_1): Likewise.
17603 (df_uses_record): Likewise.
17604 (df_read_modify_subreg_p): Remove in favor of...
17605 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
17606 const_rtx instead of an rtx.
17608 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17609 Alan Hayward <alan.hayward@arm.com>
17610 David Sherwood <david.sherwood@arm.com>
17612 * rtl.h (partial_subreg_p): New function.
17613 * caller-save.c (save_call_clobbered_regs): Use it.
17614 * calls.c (expand_call): Likewise.
17615 * combine.c (combinable_i3pat): Likewise.
17616 (simplify_set): Likewise.
17617 (make_extraction): Likewise.
17618 (make_compound_operation_int): Likewise.
17619 (gen_lowpart_or_truncate): Likewise.
17620 (force_to_mode): Likewise.
17621 (make_field_assignment): Likewise.
17622 (reg_truncated_to_mode): Likewise.
17623 (record_truncated_value): Likewise.
17624 (move_deaths): Likewise.
17625 * cse.c (record_jump_cond): Likewise.
17626 (cse_insn): Likewise.
17627 * cselib.c (cselib_lookup_1): Likewise.
17628 * expmed.c (extract_bit_field_using_extv): Likewise.
17629 * function.c (assign_parm_setup_reg): Likewise.
17630 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17631 * ira-build.c (create_insn_allocnos): Likewise.
17632 * lra-coalesce.c (merge_pseudos): Likewise.
17633 * lra-constraints.c (match_reload): Likewise.
17634 (simplify_operand_subreg): Likewise.
17635 (curr_insn_transform): Likewise.
17636 * lra-lives.c (process_bb_lives): Likewise.
17637 * lra.c (new_insn_reg): Likewise.
17638 (lra_substitute_pseudo): Likewise.
17639 * regcprop.c (mode_change_ok): Likewise.
17640 (maybe_mode_change): Likewise.
17641 (copyprop_hardreg_forward_1): Likewise.
17642 * reload.c (push_reload): Likewise.
17643 (find_reloads): Likewise.
17644 (find_reloads_subreg_address): Likewise.
17645 * reload1.c (alter_reg): Likewise.
17646 (eliminate_regs_1): Likewise.
17647 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17649 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17651 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17652 back to if statements, including unpack.
17654 2017-08-30 Martin Liska <mliska@suse.cz>
17656 PR inline-asm/82001
17657 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17659 (func_checker::compare_asm_inputs_outputs): ... this function.
17660 (func_checker::compare_gimple_asm): Use the function to compare
17661 also ASM constrains.
17662 * ipa-icf-gimple.h: Rename the function.
17664 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17665 Alan Hayward <alan.hayward@arm.com>
17666 David Sherwood <david.sherwood@arm.com>
17668 * coretypes.h (complex_mode): New type.
17669 * gdbhooks.py (build_pretty_printer): Handle it.
17670 * machmode.h (complex_mode): New class.
17671 (complex_mode::includes_p): New function.
17672 (is_complex_int_mode): Likewise.
17673 (is_complex_float_mode): Likewise.
17674 * genmodes.c (get_mode_class): Handle complex mode classes.
17675 * function.c (expand_function_end): Use is_complex_int_mode.
17677 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17678 Alan Hayward <alan.hayward@arm.com>
17679 David Sherwood <david.sherwood@arm.com>
17681 * coretypes.h (scalar_mode_pod): New typedef.
17682 * gdbhooks.py (build_pretty_printer): Handle it.
17683 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17684 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17685 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17686 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17687 as_a <scalar_mode>.
17689 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17690 Alan Hayward <alan.hayward@arm.com>
17691 David Sherwood <david.sherwood@arm.com>
17693 * machmode.h (mode_for_vector): Take a scalar_mode instead
17695 * stor-layout.c (mode_for_vector): Likewise.
17696 * explow.c (promote_mode): Use as_a <scalar_mode>.
17697 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
17699 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17700 Alan Hayward <alan.hayward@arm.com>
17701 David Sherwood <david.sherwood@arm.com>
17703 * target.def (preferred_simd_mode): Take a scalar_mode
17704 instead of a machine_mode.
17705 * targhooks.h (default_preferred_simd_mode): Likewise.
17706 * targhooks.c (default_preferred_simd_mode): Likewise.
17707 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17708 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17709 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17710 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17711 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17712 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17713 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17714 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17716 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17717 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17718 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17719 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17720 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17721 * doc/tm.texi: Regenerate.
17722 * optabs-query.c (can_vec_mask_load_store_p): Return false for
17725 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17726 Alan Hayward <alan.hayward@arm.com>
17727 David Sherwood <david.sherwood@arm.com>
17729 * target.def (scalar_mode_supported_p): Take a scalar_mode
17730 instead of a machine_mode.
17731 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17732 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17733 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17734 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17735 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17736 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17737 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17738 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17739 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17740 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17741 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17742 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17743 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17744 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17746 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17747 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17748 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17749 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17750 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17752 * doc/tm.texi: Regenerate.
17754 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17755 Alan Hayward <alan.hayward@arm.com>
17756 David Sherwood <david.sherwood@arm.com>
17758 * coretypes.h (opt_scalar_mode): New typedef.
17759 * gdbhooks.py (build_pretty_printers): Handle it.
17760 * machmode.h (mode_iterator::get_2xwider): Add overload for
17762 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17764 * expr.c (convert_mode_scalar): Likewise.
17765 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17766 * optabs.c (expand_float): Likewise.
17767 (expand_fix): Likewise.
17768 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17770 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17771 Alan Hayward <alan.hayward@arm.com>
17772 David Sherwood <david.sherwood@arm.com>
17774 * optabs.c (expand_float): Explicitly check for scalars before
17775 using a branching expansion.
17776 (expand_fix): Likewise.
17778 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17779 Alan Hayward <alan.hayward@arm.com>
17780 David Sherwood <david.sherwood@arm.com>
17782 * expr.c (convert_mode): Split scalar handling out into...
17783 (convert_mode_scalar): ...this new function. Treat the modes
17786 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17787 Alan Hayward <alan.hayward@arm.com>
17788 David Sherwood <david.sherwood@arm.com>
17790 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17792 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17794 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17795 Alan Hayward <alan.hayward@arm.com>
17796 David Sherwood <david.sherwood@arm.com>
17798 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17799 rather than a machine_mode.
17800 (fixed_from_string): Likewise.
17801 (fixed_convert): Likewise.
17802 (fixed_convert_from_int): Likewise.
17803 (fixed_convert_from_real): Likewise.
17804 (real_convert_from_fixed): Likewise.
17805 * fixed-value.c (fixed_from_double_int): Likewise.
17806 (fixed_from_string): Likewise.
17807 (fixed_convert): Likewise.
17808 (fixed_convert_from_int): Likewise.
17809 (fixed_convert_from_real): Likewise.
17810 (real_convert_from_fixed): Likewise.
17811 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17813 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17814 Alan Hayward <alan.hayward@arm.com>
17815 David Sherwood <david.sherwood@arm.com>
17817 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17818 of separate mode class checks. Do not allow vector modes here.
17819 (immed_wide_int_const): Use as_a <scalar_mode>.
17820 * explow.c (trunc_int_for_mode): Likewise.
17821 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17822 (wi::shwi): Likewise.
17823 (wi::min_value): Likewise.
17824 (wi::max_value): Likewise.
17825 * dwarf2out.c (loc_descriptor): Likewise.
17826 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17827 for CONST_WIDE_INT.
17829 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17830 Alan Hayward <alan.hayward@arm.com>
17831 David Sherwood <david.sherwood@arm.com>
17833 * tree.h (SCALAR_TYPE_MODE): New macro.
17834 * expr.c (expand_expr_addr_expr_1): Use it.
17835 (expand_expr_real_2): Likewise.
17836 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17837 (fold_convert_const_fixed_from_int): Likewise.
17838 (fold_convert_const_fixed_from_real): Likewise.
17839 (native_encode_fixed): Likewise
17840 (native_encode_complex): Likewise
17841 (native_encode_vector): Likewise.
17842 (native_interpret_fixed): Likewise.
17843 (native_interpret_real): Likewise.
17844 (native_interpret_complex): Likewise.
17845 (native_interpret_vector): Likewise.
17846 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17847 (simd_clone_adjust_argument_types): Likewise.
17848 (simd_clone_init_simd_arrays): Likewise.
17849 (simd_clone_adjust): Likewise.
17850 * stor-layout.c (layout_type): Likewise.
17851 * tree.c (build_minus_one_cst): Likewise.
17852 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17853 * tree-inline.c (estimate_move_cost): Likewise.
17854 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17855 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17856 (vectorizable_reduction): Likewise.
17857 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17858 (vect_recog_mixed_size_cond_pattern): Likewise.
17859 (check_bool_pattern): Likewise.
17860 (adjust_bool_pattern): Likewise.
17861 (search_type_for_mask_1): Likewise.
17862 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17863 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17864 (vectorizable_load): Likewise.
17865 (vectorizable_store): Likewise.
17866 * ubsan.c (ubsan_encode_value): Likewise.
17867 * varasm.c (output_constant): Likewise.
17869 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17870 Alan Hayward <alan.hayward@arm.com>
17871 David Sherwood <david.sherwood@arm.com>
17873 * coretypes.h (scalar_mode): New class.
17874 * machmode.h (scalar_mode): Likewise.
17875 (scalar_mode::includes_p): New function.
17876 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
17877 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
17878 * genmodes.c (get_mode_class): Handle remaining scalar modes.
17879 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
17880 * expmed.c (store_bit_field_1): Likewise.
17881 (extract_bit_field_1): Likewise.
17882 * expr.c (write_complex_part): Likewise.
17883 (read_complex_part): Likewise.
17884 (emit_move_complex_push): Likewise.
17885 (expand_expr_real_2): Likewise.
17886 * function.c (assign_parm_setup_reg): Likewise.
17887 (assign_parms_unsplit_complex): Likewise.
17888 * optabs.c (expand_binop): Likewise.
17889 * rtlanal.c (subreg_get_info): Likewise.
17890 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17891 * varasm.c (output_constant_pool_2): Likewise.
17893 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17894 Alan Hayward <alan.hayward@arm.com>
17895 David Sherwood <david.sherwood@arm.com>
17897 * expmed.c (extract_high_half): Use scalar_int_mode and remove
17899 (expmed_mult_highpart_optab): Likewise.
17900 (expmed_mult_highpart): Likewise.
17902 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17903 Alan Hayward <alan.hayward@arm.com>
17904 David Sherwood <david.sherwood@arm.com>
17906 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
17907 instead of a machine_mode.
17908 (builtin_memset_read_str): Likewise.
17909 * builtins.c (c_readstr): Likewise.
17910 (builtin_memcpy_read_str): Likewise.
17911 (builtin_strncpy_read_str): Likewise.
17912 (builtin_memset_read_str): Likewise.
17913 (builtin_memset_gen_str): Likewise.
17914 (expand_builtin_signbit): Use scalar_int_mode for local variables.
17915 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
17916 instead of a machine_mode.
17917 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
17919 (make_extraction): Likewise.
17920 (try_widen_shift_mode): Take and return scalar_int_modes instead
17922 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
17923 a scalar_int_mode instead of a machine_mode.
17924 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
17925 (avr_addr_space_pointer_mode): Likewise.
17926 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
17927 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
17928 (msp430_unwind_word_mode): Likewise.
17929 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
17930 (spu_addr_space_pointer_mode): Likewise.
17931 (spu_addr_space_address_mode): Likewise.
17932 (spu_libgcc_cmp_return_mode): Likewise.
17933 (spu_libgcc_shift_count_mode): Likewise.
17934 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
17935 (rl78_addr_space_pointer_mode): Likewise.
17936 (fl78_unwind_word_mode): Likewise.
17937 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
17939 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
17940 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
17941 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
17942 (mips_valid_pointer_mode): Likewise.
17943 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
17944 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
17945 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
17947 (ft32_addr_space_address_mode): Likewise.
17948 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
17949 scalar_int_mode instead of a machine_mode.
17950 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
17952 (m32c_addr_space_address_mode): Likewise.
17953 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
17954 (rs6000_eh_return_filter_mode): Likewise.
17955 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
17956 (rs6000_eh_return_filter_mode): Likewise.
17957 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
17958 (s390_libgcc_shift_count_mode): Likewise.
17959 (s390_unwind_word_mode): Likewise.
17960 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
17962 * target.def (mode_rep_extended): Likewise.
17963 (valid_pointer_mode): Likewise.
17964 (addr_space.valid_pointer_mode): Likewise.
17965 (eh_return_filter_mode): Return a scalar_int_mode rather than
17967 (libgcc_cmp_return_mode): Likewise.
17968 (libgcc_shift_count_mode): Likewise.
17969 (unwind_word_mode): Likewise.
17970 (addr_space.pointer_mode): Likewise.
17971 (addr_space.address_mode): Likewise.
17972 * doc/tm.texi: Regenerate.
17973 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
17975 (do_jump): Use scalar_int_mode for local variables.
17976 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
17977 rather than a machine_mode.
17978 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
17979 (scompare_loc_descriptor_wide): Likewise.
17980 (scompare_loc_descriptor_narrow): Likewise.
17981 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
17983 * except.c (sjlj_emit_dispatch_table): Likewise.
17984 (expand_builtin_eh_copy_values): Likewise.
17985 * explow.c (convert_memory_address_addr_space_1): Likewise.
17986 Take a scalar_int_mode rather than a machine_mode.
17987 (convert_memory_address_addr_space): Take a scalar_int_mode rather
17988 than a machine_mode.
17989 (memory_address_addr_space): Use scalar_int_mode for local variables.
17990 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
17991 rather than a machine_mode.
17992 * expmed.c (mask_rtx): Likewise.
17993 (init_expmed_one_conv): Likewise.
17994 (expand_mult_highpart_adjust): Likewise.
17995 (extract_high_half): Likewise.
17996 (expmed_mult_highpart_optab): Likewise.
17997 (expmed_mult_highpart): Likewise.
17998 (expand_smod_pow2): Likewise.
17999 (expand_sdiv_pow2): Likewise.
18000 (emit_store_flag_int): Likewise.
18001 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
18003 (extract_low_bits): Likewise.
18004 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18006 * expr.c (pieces_addr::adjust): Likewise.
18007 (can_store_by_pieces): Likewise.
18008 (store_by_pieces): Likewise.
18009 (clear_by_pieces_1): Likewise.
18010 (expand_expr_addr_expr_1): Likewise.
18011 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18012 (expand_expr_real_1): Likewise.
18013 (try_casesi): Likewise.
18014 * final.c (shorten_branches): Likewise.
18015 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18016 type of "mode" to machine_mode.
18017 * internal-fn.c (expand_arith_overflow_result_store): Take a
18018 scalar_int_mode rather than a machine_mode.
18019 (expand_mul_overflow): Use scalar_int_mode for local variables.
18020 * loop-doloop.c (doloop_modify): Likewise.
18021 (doloop_optimize): Likewise.
18022 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18023 than a machine_mode.
18024 (expand_doubleword_shift_condmove): Likewise.
18025 (expand_doubleword_shift): Likewise.
18026 (expand_doubleword_clz): Likewise.
18027 (expand_doubleword_popcount): Likewise.
18028 (expand_doubleword_parity): Likewise.
18029 (expand_absneg_bit): Use scalar_int_mode for local variables.
18030 (prepare_float_lib_cmp): Likewise.
18031 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18032 rather than a machine_mode.
18033 (convert_memory_address_addr_space): Likewise.
18034 (get_mode_bounds): Likewise.
18035 (get_address_mode): Return a scalar_int_mode rather than a
18037 * rtlanal.c (get_address_mode): Likewise.
18038 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18039 than a machine_mode.
18040 * targhooks.c (default_mode_rep_extended): Likewise.
18041 (default_valid_pointer_mode): Likewise.
18042 (default_addr_space_valid_pointer_mode): Likewise.
18043 (default_eh_return_filter_mode): Return a scalar_int_mode rather
18044 than a machine_mode.
18045 (default_libgcc_cmp_return_mode): Likewise.
18046 (default_libgcc_shift_count_mode): Likewise.
18047 (default_unwind_word_mode): Likewise.
18048 (default_addr_space_pointer_mode): Likewise.
18049 (default_addr_space_address_mode): Likewise.
18050 * targhooks.h (default_eh_return_filter_mode): Likewise.
18051 (default_libgcc_cmp_return_mode): Likewise.
18052 (default_libgcc_shift_count_mode): Likewise.
18053 (default_unwind_word_mode): Likewise.
18054 (default_addr_space_pointer_mode): Likewise.
18055 (default_addr_space_address_mode): Likewise.
18056 (default_mode_rep_extended): Take a scalar_int_mode rather than
18058 (default_valid_pointer_mode): Likewise.
18059 (default_addr_space_valid_pointer_mode): Likewise.
18060 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18062 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18063 rather than a machine_mode.
18064 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18065 for local variables.
18066 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18067 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18068 than a machine_mode.
18070 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18071 Alan Hayward <alan.hayward@arm.com>
18072 David Sherwood <david.sherwood@arm.com>
18074 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18075 the mode argument to scalar_int_mode.
18076 (do_jump_by_parts_zero_rtx): Likewise.
18077 (do_jump_by_parts_equality_rtx): Likewise.
18078 (do_jump_by_parts_greater): Take a mode argument.
18079 (do_jump_by_parts_equality): Likewise.
18080 (do_jump_1): Update calls accordingly.
18082 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18083 Alan Hayward <alan.hayward@arm.com>
18084 David Sherwood <david.sherwood@arm.com>
18086 * is-a.h (safe_dyn_cast): New function.
18087 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18088 (jump_table_for_label): Likewise.
18089 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18090 instead of an rtx_insn *.
18091 (shorten_branches): Use dyn_cast instead of LABEL_P and
18092 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18093 rtx_jump_table_data::get_data_mode.
18094 (final_scan_insn): Likewise.
18096 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18097 Alan Hayward <alan.hayward@arm.com>
18098 David Sherwood <david.sherwood@arm.com>
18100 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18101 trying to combine a full-register integer set with a subreg
18104 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18105 Alan Hayward <alan.hayward@arm.com>
18106 David Sherwood <david.sherwood@arm.com>
18108 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18109 that is always either address_mode or pointer_mode.
18111 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18112 Alan Hayward <alan.hayward@arm.com>
18113 David Sherwood <david.sherwood@arm.com>
18115 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
18116 when the two are known to be equal.
18118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18119 Alan Hayward <alan.hayward@arm.com>
18120 David Sherwood <david.sherwood@arm.com>
18122 * simplify-rtx.c (simplify_const_unary_operation): Use
18123 is_a <scalar_int_mode> instead of checking for a nonzero
18124 precision. Forcibly convert op_mode to a scalar_int_mode
18125 in that case. More clearly differentiate the operand and
18126 result modes and use the former when deciding what the value
18127 of a count-bits operation should be. Use is_int_mode instead
18128 of checking for a MODE_INT. Remove redundant check for whether
18129 this mode has a zero precision.
18131 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18132 Alan Hayward <alan.hayward@arm.com>
18133 David Sherwood <david.sherwood@arm.com>
18135 * optabs.c (widen_leading): Change the type of the mode argument
18136 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
18137 (widen_bswap): Likewise.
18138 (expand_parity): Likewise.
18139 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
18140 (expand_ffs): Likewise.
18141 (epand_unop): Check for scalar integer modes before calling the
18144 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18145 Alan Hayward <alan.hayward@arm.com>
18146 David Sherwood <david.sherwood@arm.com>
18148 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
18150 (expand_expr_real_1): Update call accordingly.
18152 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18153 Alan Hayward <alan.hayward@arm.com>
18154 David Sherwood <david.sherwood@arm.com>
18156 * expmed.c (store_bit_field_using_insv): Add op0_mode and
18157 value_mode arguments. Use scalar_int_mode internally.
18158 (store_bit_field_1): Rename the new integer mode from imode
18159 to op0_mode and use it instead of GET_MODE (op0). Update calls
18160 to store_split_bit_field, store_bit_field_using_insv and
18161 store_fixed_bit_field.
18162 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
18163 Use scalar_int_mode internally. Use a bit count rather than a mode
18164 when calculating the largest bit size for get_best_mode.
18165 Update calls to store_split_bit_field and store_fixed_bit_field_1.
18166 (store_fixed_bit_field_1): Add mode and value_mode arguments.
18167 Remove assertion that OP0 has a scalar integer mode.
18168 (store_split_bit_field): Add op0_mode and value_mode arguments.
18169 Update calls to extract_fixed_bit_field.
18170 (extract_bit_field_using_extv): Add an op0_mode argument.
18171 Use scalar_int_mode internally.
18172 (extract_bit_field_1): Rename the new integer mode from imode to
18173 op0_mode and use it instead of GET_MODE (op0). Update calls to
18174 extract_split_bit_field, extract_bit_field_using_extv and
18175 extract_fixed_bit_field.
18176 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
18177 to extract_split_bit_field and extract_fixed_bit_field_1.
18178 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
18179 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
18180 on the target mode.
18181 (extract_split_bit_field): Add an op0_mode argument. Update call
18182 to extract_fixed_bit_field.
18184 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18185 Alan Hayward <alan.hayward@arm.com>
18186 David Sherwood <david.sherwood@arm.com>
18188 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
18189 * explow.c (hard_function_value): Likewise.
18190 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
18191 convert_to_mode call outside the loop.
18192 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
18193 for the mode iterator. Require the mode specified by max_pieces
18195 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
18197 (copy_blkmode_to_reg): Likewise.
18198 (set_storage_via_setmem): Likewise.
18199 * optabs.c (prepare_cmp_insn): Likewise.
18200 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18201 * stor-layout.c (finish_bitfield_representative): Likewise.
18203 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18204 Alan Hayward <alan.hayward@arm.com>
18205 David Sherwood <david.sherwood@arm.com>
18207 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
18208 * expr.c (convert_move): Use them.
18209 (convert_modes): Likewise.
18210 (store_expr_with_bounds): Likewise.
18212 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18213 Alan Hayward <alan.hayward@arm.com>
18214 David Sherwood <david.sherwood@arm.com>
18216 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
18217 parameter for the mode of "x". Remove the "known_x", "known_mode"
18218 and "known_ret" arguments. Change the type of the mode argument
18219 to scalar_int_mode.
18220 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
18221 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
18222 (reg_num_sign_bit_copies_for_combine): Likewise.
18223 * rtlanal.c (nonzero_bits1): Likewise.
18224 (num_sign_bit_copies1): Likewise.
18225 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
18226 (reg_num_sign_bit_copies_general): Likewise.
18227 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
18228 (reg_nonzero_bits_general): Likewise.
18230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18231 Alan Hayward <alan.hayward@arm.com>
18232 David Sherwood <david.sherwood@arm.com>
18234 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
18235 than in subroutines. Return 1 for non-integer modes.
18236 (cached_num_sign_bit_copies): Change the type of the mode parameter
18237 to scalar_int_mode.
18238 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
18239 classes. Handle CONST_INT_P first and then check whether X also
18240 has a scalar integer mode. Check the same thing for inner registers
18241 of a SUBREG and for values that are being extended or truncated.
18243 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18244 Alan Hayward <alan.hayward@arm.com>
18245 David Sherwood <david.sherwood@arm.com>
18247 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
18248 in subroutines. Return the mode mask for non-integer modes.
18249 (cached_nonzero_bits): Change the type of the mode parameter
18250 to scalar_int_mode.
18251 (nonzero_bits1): Likewise. Remove early exit for other mode
18252 classes. Handle CONST_INT_P first and then check whether X
18253 also has a scalar integer mode.
18255 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18256 Alan Hayward <alan.hayward@arm.com>
18257 David Sherwood <david.sherwood@arm.com>
18259 * expr.c (widest_int_mode_for_size): Make the comment match the code.
18260 Return a scalar_int_mode and assert that the size is greater than
18262 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
18263 (op_by_pieces_d::op_by_pieces_d): Likewise.
18264 (op_by_pieces_d::run): Likewise.
18265 (can_store_by_pieces): Likewise.
18267 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18268 Alan Hayward <alan.hayward@arm.com>
18269 David Sherwood <david.sherwood@arm.com>
18271 * combine.c (extract_left_shift): Add a mode argument and update
18273 (make_compound_operation_int): Change the type of the mode parameter
18274 to scalar_int_mode and update the call to extract_left_shift.
18276 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18277 Alan Hayward <alan.hayward@arm.com>
18278 David Sherwood <david.sherwood@arm.com>
18280 * combine.c (simplify_and_const_int): Change the type of the mode
18281 parameter to scalar_int_mode.
18282 (simplify_and_const_int_1): Likewise. Update recursive call.
18284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18285 Alan Hayward <alan.hayward@arm.com>
18286 David Sherwood <david.sherwood@arm.com>
18288 * combine.c (simplify_compare_const): Check that the mode is a
18289 scalar_int_mode (rather than VOIDmode) before testing its
18291 (simplify_comparison): Move COMPARISON_P handling out of the
18292 loop and restrict the latter part of the loop to scalar_int_modes.
18293 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
18294 and when considering SUBREG_REGs. Use is_int_mode instead of
18295 checking GET_MODE_CLASS against MODE_INT.
18297 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18298 Alan Hayward <alan.hayward@arm.com>
18299 David Sherwood <david.sherwood@arm.com>
18301 * combine.c (try_widen_shift_mode): Move check for equal modes to...
18302 (simplify_shift_const_1): ...here. Use scalar_int_mode for
18303 shift_unit_mode and for modes involved in scalar shifts.
18305 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18306 Alan Hayward <alan.hayward@arm.com>
18307 David Sherwood <david.sherwood@arm.com>
18309 * combine.c (force_int_to_mode): New function, split out from...
18310 (force_to_mode): ...here. Keep xmode up-to-date and use it
18311 instead of GET_MODE (x).
18313 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18314 Alan Hayward <alan.hayward@arm.com>
18315 David Sherwood <david.sherwood@arm.com>
18317 * optabs-query.h (extraction_insn::struct_mode): Change type to
18318 opt_scalar_int_mode and update comment.
18319 (extraction_insn::field_mode): Change type to scalar_int_mode.
18320 (extraction_insn::pos_mode): Likewise.
18321 * combine.c (make_extraction): Update accordingly.
18322 * optabs-query.c (get_traditional_extraction_insn): Likewise.
18323 (get_optab_extraction_insn): Likewise.
18324 * recog.c (simplify_while_replacing): Likewise.
18325 * expmed.c (narrow_bit_field_mem): Change the type of the mode
18326 parameter to opt_scalar_int_mode.
18328 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18329 Alan Hayward <alan.hayward@arm.com>
18330 David Sherwood <david.sherwood@arm.com>
18332 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
18333 to a scalar_int_mode instead of a machine_mode.
18334 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
18335 (get_best_mode): Return a boolean and use a pointer argument to store
18336 the selected mode. Replace the limit mode parameter with a bit limit.
18337 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
18338 for the values returned by bit_field_mode_iterator::next_mode.
18339 (store_bit_field): Update call to get_best_mode.
18340 (store_fixed_bit_field): Likewise.
18341 (extract_fixed_bit_field): Likewise.
18342 * expr.c (optimize_bitfield_assignment_op): Likewise.
18343 * fold-const.c (optimize_bit_field_compare): Likewise.
18344 (fold_truth_andor_1): Likewise.
18345 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
18346 Update for new type of m_mode.
18347 (get_best_mode): As above.
18349 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18350 Alan Hayward <alan.hayward@arm.com>
18351 David Sherwood <david.sherwood@arm.com>
18353 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
18354 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
18355 (store_bit_field): Check is_a <scalar_int_mode> before calling
18356 strict_volatile_bitfield_p.
18357 (extract_bit_field): Likewise.
18359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18360 Alan Hayward <alan.hayward@arm.com>
18361 David Sherwood <david.sherwood@arm.com>
18363 * target.def (cstore_mode): Return a scalar_int_mode.
18364 * doc/tm.texi: Regenerate.
18365 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
18366 * targhooks.h (default_cstore_mode): Likewise.
18367 * targhooks.c (default_cstore_mode): Likewise, using a forced
18369 * expmed.c (emit_cstore): Expect the target of the cstore to be
18372 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18373 Alan Hayward <alan.hayward@arm.com>
18374 David Sherwood <david.sherwood@arm.com>
18376 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
18378 (niter_desc): Likewise mode.
18379 (iv_analyze): Add a mode parameter.
18381 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
18382 and change its type to scalar_int_mode.
18383 * loop-iv.c: Update commentary at head of file.
18384 (iv_constant): Pass the mode paraeter before the rtx it describes
18385 and change its type to scalar_int_mode. Remove VOIDmode handling.
18386 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
18387 (iv_extend): Likewise.
18388 (shorten_into_mode): Likewise.
18389 (iv_add): Use scalar_int_mode.
18390 (iv_mult): Likewise.
18391 (iv_shift): Likewise.
18392 (canonicalize_iv_subregs): Likewise.
18393 (get_biv_step_1): Pass the outer_mode parameter before the rtx
18394 it describes and change its mode to scalar_int_mode. Also change
18395 the type of the returned inner_mode to scalar_int_mode.
18396 (get_biv_step): Likewise, turning outer_mode from a pointer
18397 into a direct parameter. Update call to get_biv_step_1.
18398 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
18399 iv_constant and get_biv_step.
18400 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
18401 and change its type to scalar_int_mode. Don't initialise iv->mode
18402 to VOIDmode and remove later checks for its still being VOIDmode.
18403 Update calls to iv_analyze_op and iv_analyze_expr. Check
18404 is_a <scalar_int_mode> when changing the mode under consideration.
18405 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
18406 Update call to iv_analyze_expr.
18407 (iv_analyze_op): Add a mode parameter. Reject subregs whose
18408 inner register is not also a scalar_int_mode. Update call to
18410 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
18411 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
18412 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
18413 separate mode class checks. Update calls to iv_analyze. Remove
18414 fix-up of VOIDmodes after iv_analyze_biv.
18415 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
18416 don't have a scalar_int_mode. Update call to biv_p.
18418 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18419 Alan Hayward <alan.hayward@arm.com>
18420 David Sherwood <david.sherwood@arm.com>
18422 * cfgexpand.c (convert_debug_memory_address): Use
18423 as_a <scalar_int_mode>.
18424 * combine.c (expand_compound_operation): Likewise.
18425 (make_extraction): Likewise.
18426 (change_zero_ext): Likewise.
18427 (simplify_comparison): Likewise.
18428 * cse.c (cse_insn): Likewise.
18429 * dwarf2out.c (minmax_loc_descriptor): Likewise.
18430 (mem_loc_descriptor): Likewise.
18431 (loc_descriptor): Likewise.
18432 * expmed.c (init_expmed_one_mode): Likewise.
18433 (synth_mult): Likewise.
18434 (emit_store_flag_1): Likewise.
18435 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
18436 of a comparison with size.
18437 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
18438 (reduce_to_bit_field_precision): Likewise.
18439 * function.c (expand_function_end): Likewise.
18440 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
18441 * loop-doloop.c (doloop_modify): Likewise.
18442 * optabs.c (expand_binop): Likewise.
18443 (expand_unop): Likewise.
18444 (expand_copysign_absneg): Likewise.
18445 (prepare_cmp_insn): Likewise.
18446 (maybe_legitimize_operand): Likewise.
18447 * recog.c (const_scalar_int_operand): Likewise.
18448 * rtlanal.c (get_address_mode): Likewise.
18449 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18450 (simplify_cond_clz_ctz): Likewise.
18451 * tree-nested.c (get_nl_goto_field): Likewise.
18452 * tree.c (build_vector_type_for_mode): Likewise.
18453 * var-tracking.c (use_narrower_mode): Likewise.
18455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18456 Alan Hayward <alan.hayward@arm.com>
18457 David Sherwood <david.sherwood@arm.com>
18459 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
18460 * builtins.c (expand_builtin_signbit): Use it.
18461 * cfgexpand.c (expand_debug_expr): Likewise.
18462 * dojump.c (do_jump): Likewise.
18463 (do_compare_and_jump): Likewise.
18464 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
18465 * expmed.c (make_tree): Likewise.
18466 * expr.c (expand_expr_real_2): Likewise.
18467 (expand_expr_real_1): Likewise.
18468 (try_casesi): Likewise.
18469 * fold-const-call.c (fold_const_call_ss): Likewise.
18470 * fold-const.c (unextend): Likewise.
18471 (extract_muldiv_1): Likewise.
18472 (fold_single_bit_test): Likewise.
18473 (native_encode_int): Likewise.
18474 (native_encode_string): Likewise.
18475 (native_interpret_int): Likewise.
18476 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
18477 * internal-fn.c (expand_addsub_overflow): Likewise.
18478 (expand_neg_overflow): Likewise.
18479 (expand_mul_overflow): Likewise.
18480 (expand_arith_overflow): Likewise.
18481 * match.pd: Likewise.
18482 * stor-layout.c (layout_type): Likewise.
18483 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18484 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
18485 * tree-ssanames.c (get_range_info): Likewise.
18486 * tree-switch-conversion.c (array_value_type) Likewise.
18487 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
18488 (vect_recog_divmod_pattern): Likewise.
18489 (vect_recog_mixed_size_cond_pattern): Likewise.
18490 * tree-vrp.c (extract_range_basic): Likewise.
18491 (simplify_float_conversion_using_ranges): Likewise.
18492 * tree.c (int_fits_type_p): Likewise.
18493 * ubsan.c (instrument_bool_enum_load): Likewise.
18494 * varasm.c (mergeable_string_section): Likewise.
18495 (narrowing_initializer_constant_valid_p): Likewise.
18496 (output_constant): Likewise.
18498 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18499 Alan Hayward <alan.hayward@arm.com>
18500 David Sherwood <david.sherwood@arm.com>
18502 * machmode.h (NARROWEST_INT_MODE): New macro.
18503 * expr.c (alignment_for_piecewise_move): Use it instead of
18504 GET_CLASS_NARROWEST_MODE (MODE_INT).
18505 (push_block): Likewise.
18506 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
18508 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18510 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18511 Alan Hayward <alan.hayward@arm.com>
18512 David Sherwood <david.sherwood@arm.com>
18514 * postreload.c (move2add_valid_value_p): Change the type of the
18515 mode parameter to scalar_int_mode.
18516 (move2add_use_add2_insn): Add a mode parameter and use it instead
18518 (move2add_use_add3_insn): Likewise.
18519 (reload_cse_move2add): Update accordingly.
18521 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18522 Alan Hayward <alan.hayward@arm.com>
18523 David Sherwood <david.sherwood@arm.com>
18525 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
18527 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
18528 * optabs.c (expand_unop): Likewise.
18530 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18531 Alan Hayward <alan.hayward@arm.com>
18532 David Sherwood <david.sherwood@arm.com>
18534 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
18535 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
18536 (popcount_loc_descriptor): Likewise.
18537 (bswap_loc_descriptor): Likewise.
18538 (rotate_loc_descriptor): Likewise.
18539 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
18540 calling the functions above.
18542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18543 Alan Hayward <alan.hayward@arm.com>
18544 David Sherwood <david.sherwood@arm.com>
18546 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
18548 (try_combine): Likewise.
18549 (simplify_if_then_else): Likewise.
18550 * cse.c (cse_insn): Likewise.
18551 * dwarf2out.c (mem_loc_descriptor): Likewise.
18552 * emit-rtl.c (gen_lowpart_common): Likewise.
18553 * simplify-rtx.c (simplify_truncation): Likewise.
18554 (simplify_binary_operation_1): Likewise.
18555 (simplify_const_relational_operation): Likewise.
18556 (simplify_ternary_operation): Likewise.
18557 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
18559 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18560 Alan Hayward <alan.hayward@arm.com>
18561 David Sherwood <david.sherwood@arm.com>
18563 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
18564 * reload.c (push_reload): Likewise.
18565 (find_reloads): Likewise.
18567 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18568 Alan Hayward <alan.hayward@arm.com>
18569 David Sherwood <david.sherwood@arm.com>
18571 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
18572 (make_compound_operation_int): Likewise.
18573 (change_zero_ext): Likewise.
18574 * expr.c (convert_move): Likewise.
18575 (convert_modes): Likewise.
18576 * fwprop.c (forward_propagate_subreg): Likewise.
18577 * loop-iv.c (get_biv_step_1): Likewise.
18578 * optabs.c (widen_operand): Likewise.
18579 * postreload.c (move2add_valid_value_p): Likewise.
18580 * recog.c (simplify_while_replacing): Likewise.
18581 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18582 (simplify_binary_operation_1): Likewise. Remove redundant
18583 mode equality check.
18585 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18586 Alan Hayward <alan.hayward@arm.com>
18587 David Sherwood <david.sherwood@arm.com>
18589 * combine.c (combine_simplify_rtx): Add checks for
18590 is_a <scalar_int_mode>.
18591 (simplify_if_then_else): Likewise.
18592 (make_field_assignment): Likewise.
18593 (simplify_comparison): Likewise.
18594 * ifcvt.c (noce_try_bitop): Likewise.
18595 * loop-invariant.c (canonicalize_address_mult): Likewise.
18596 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18598 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18599 Alan Hayward <alan.hayward@arm.com>
18600 David Sherwood <david.sherwood@arm.com>
18602 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
18603 is_a <scalar_int_mode> instead of != BLKmode.
18605 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18606 Alan Hayward <alan.hayward@arm.com>
18607 David Sherwood <david.sherwood@arm.com>
18609 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
18610 instead of != VOIDmode.
18611 * combine.c (if_then_else_cond): Likewise.
18612 (change_zero_ext): Likewise.
18613 * dwarf2out.c (mem_loc_descriptor): Likewise.
18614 (loc_descriptor): Likewise.
18615 * rtlanal.c (canonicalize_condition): Likewise.
18616 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
18618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18619 Alan Hayward <alan.hayward@arm.com>
18620 David Sherwood <david.sherwood@arm.com>
18622 * simplify-rtx.c (simplify_binary_operation_1): Use
18623 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18625 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18626 Alan Hayward <alan.hayward@arm.com>
18627 David Sherwood <david.sherwood@arm.com>
18629 * wide-int.h (int_traits<unsigned char>) New class.
18630 (int_traits<unsigned short>) Likewise.
18631 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18632 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18634 * combine.c (set_nonzero_bits_and_sign_copies): Use
18635 is_a <scalar_int_mode>.
18636 (find_split_point): Likewise.
18637 (combine_simplify_rtx): Likewise.
18638 (simplify_logical): Likewise.
18639 (expand_compound_operation): Likewise.
18640 (expand_field_assignment): Likewise.
18641 (make_compound_operation): Likewise.
18642 (extended_count): Likewise.
18643 (change_zero_ext): Likewise.
18644 (simplify_comparison): Likewise.
18645 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18646 (ucompare_loc_descriptor): Likewise.
18647 (minmax_loc_descriptor): Likewise.
18648 (mem_loc_descriptor): Likewise.
18649 (loc_descriptor): Likewise.
18650 * expmed.c (init_expmed_one_mode): Likewise.
18651 * lra-constraints.c (lra_constraint_offset): Likewise.
18652 * optabs.c (prepare_libcall_arg): Likewise.
18653 * postreload.c (move2add_note_store): Likewise.
18654 * reload.c (operands_match_p): Likewise.
18655 * rtl.h (load_extend_op): Likewise.
18656 * rtlhooks.c (gen_lowpart_general): Likewise.
18657 * simplify-rtx.c (simplify_truncation): Likewise.
18658 (simplify_unary_operation_1): Likewise.
18659 (simplify_binary_operation_1): Likewise.
18660 (simplify_const_binary_operation): Likewise.
18661 (simplify_const_relational_operation): Likewise.
18662 (simplify_subreg): Likewise.
18663 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18664 * var-tracking.c (adjust_mems): Likewise.
18665 (prepare_call_arguments): Likewise.
18667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18668 Alan Hayward <alan.hayward@arm.com>
18669 David Sherwood <david.sherwood@arm.com>
18671 * machmode.h (is_int_mode): New fuction.
18672 * combine.c (find_split_point): Use it.
18673 (combine_simplify_rtx): Likewise.
18674 (simplify_if_then_else): Likewise.
18675 (simplify_set): Likewise.
18676 (simplify_shift_const_1): Likewise.
18677 (simplify_comparison): Likewise.
18678 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18679 * cse.c (notreg_cost): Likewise.
18680 (cse_insn): Likewise.
18681 * cselib.c (cselib_lookup_1): Likewise.
18682 * dojump.c (do_jump_1): Likewise.
18683 (do_compare_rtx_and_jump): Likewise.
18684 * dse.c (get_call_args): Likewise.
18685 * dwarf2out.c (rtl_for_decl_init): Likewise.
18686 (native_encode_initializer): Likewise.
18687 * expmed.c (emit_store_flag_1): Likewise.
18688 (emit_store_flag): Likewise.
18689 * expr.c (convert_modes): Likewise.
18690 (store_field): Likewise.
18691 (expand_expr_real_1): Likewise.
18692 * fold-const.c (fold_read_from_constant_string): Likewise.
18693 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18694 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18695 * optabs.c (expand_binop): Likewise.
18696 (expand_unop): Likewise.
18697 (expand_abs_nojump): Likewise.
18698 (expand_one_cmpl_abs_nojump): Likewise.
18699 * simplify-rtx.c (mode_signbit_p): Likewise.
18700 (val_signbit_p): Likewise.
18701 (val_signbit_known_set_p): Likewise.
18702 (val_signbit_known_clear_p): Likewise.
18703 (simplify_relational_operation_1): Likewise.
18704 * tree.c (vector_type_mode): Likewise.
18706 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18707 Alan Hayward <alan.hayward@arm.com>
18708 David Sherwood <david.sherwood@arm.com>
18710 * machmode.h (smallest_mode_for_size): Fix formatting.
18711 (smallest_int_mode_for_size): New function.
18712 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18713 instead of smallest_mode_for_size.
18714 * combine.c (make_extraction): Likewise.
18715 * config/arc/arc.c (arc_expand_movmem): Likewise.
18716 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18717 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18718 * config/s390/s390.c (s390_expand_insv): Likewise.
18719 * config/sparc/sparc.c (assign_int_registers): Likewise.
18720 * config/spu/spu.c (spu_function_value): Likewise.
18721 (spu_function_arg): Likewise.
18722 * coverage.c (get_gcov_type): Likewise.
18723 (get_gcov_unsigned_t): Likewise.
18724 * dse.c (find_shift_sequence): Likewise.
18725 * expmed.c (store_bit_field_1): Likewise.
18726 * expr.c (convert_move): Likewise.
18727 (store_field): Likewise.
18728 * internal-fn.c (expand_arith_overflow): Likewise.
18729 * optabs-query.c (get_best_extraction_insn): Likewise.
18730 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18731 * stor-layout.c (layout_type): Likewise.
18732 (initialize_sizetypes): Likewise.
18733 * targhooks.c (default_get_mask_mode): Likewise.
18734 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18736 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18737 Alan Hayward <alan.hayward@arm.com>
18738 David Sherwood <david.sherwood@arm.com>
18740 * machmode.h (opt_mode::else_blk): New function.
18741 (int_mode_for_mode): Declare.
18742 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18743 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18745 * cfgexpand.c (expand_debug_expr): Likewise.
18746 * combine.c (gen_lowpart_or_truncate): Likewise.
18747 (gen_lowpart_for_combine): Likewise.
18748 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18749 * config/avr/avr.c (avr_to_int_mode): Likewise.
18750 (avr_out_plus_1): Likewise.
18751 (avr_out_plus): Likewise.
18752 (avr_out_round): Likewise.
18753 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18754 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18755 (s390_expand_vcond): Likewise.
18756 * config/spu/spu.c (spu_split_immediate): Likewise.
18757 (spu_expand_mov): Likewise.
18758 * dse.c (get_stored_val): Likewise.
18759 * expmed.c (store_bit_field_1): Likewise.
18760 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18762 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18763 (extract_low_bits): Likewise.
18764 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18765 handling rather than repeating the check.
18766 (emit_group_store): Likewise.
18767 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18768 * optabs.c (expand_absneg_bit): Likewise.
18769 (expand_copysign_absneg): Likewise.
18770 (expand_copysign_bit): Likewise.
18771 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18772 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18773 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18774 * var-tracking.c (prepare_call_arguments): Likewise.
18775 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18776 int_mode_for_mode instead of mode_for_size.
18777 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18779 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18780 Alan Hayward <alan.hayward@arm.com>
18781 David Sherwood <david.sherwood@arm.com>
18783 * machmode.h (int_mode_for_size): New function.
18784 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18785 instead of mode_for_size.
18786 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18788 * combine.c (expand_field_assignment): Use int_mode_for_size
18789 instead of mode_for_size.
18790 (make_extraction): Likewise.
18791 (simplify_shift_const_1): Likewise.
18792 (simplify_comparison): Likewise.
18793 * dojump.c (do_jump): Likewise.
18794 * dwarf2out.c (mem_loc_descriptor): Likewise.
18795 * emit-rtl.c (init_derived_machine_modes): Likewise.
18796 * expmed.c (flip_storage_order): Likewise.
18797 (convert_extracted_bit_field): Likewise.
18798 * expr.c (copy_blkmode_from_reg): Likewise.
18799 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18800 * internal-fn.c (expand_mul_overflow): Likewise.
18801 * lower-subreg.c (simple_move): Likewise.
18802 * optabs-libfuncs.c (init_optabs): Likewise.
18803 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18804 * tree.c (vector_type_mode): Likewise.
18805 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18806 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18807 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18808 * tree-vect-stmts.c (vectorizable_load): Likewise.
18809 (vectorizable_store): Likewise.
18811 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18812 Alan Hayward <alan.hayward@arm.com>
18813 David Sherwood <david.sherwood@arm.com>
18815 * coretypes.h (pod_mode): New type.
18816 (scalar_int_mode_pod): New typedef.
18817 * machmode.h (pod_mode): New class.
18818 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18819 * genmodes.c (emit_mode_int_n): Update accordingly.
18820 * lower-subreg.h (target_lower_subreg): Change type to
18821 scalar_int_mode_pod.
18822 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18823 scalar_int_mode_pod.
18825 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18826 Alan Hayward <alan.hayward@arm.com>
18827 David Sherwood <david.sherwood@arm.com>
18829 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18830 machine_mode to scalar_int_mode.
18831 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18832 (rs6000_option_override_internal): Remove cast to int.
18833 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18834 machine_mode to scalar_int_mode.
18835 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18836 (rs6000_option_override_internal): Remove cast to int.
18837 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18838 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18840 * config/s390/s390.c (s390_expand_builtin): Likewise.
18841 * coretypes.h (scalar_int_mode): New type.
18842 (opt_scalar_int_mode): New typedef.
18843 * machmode.h (scalar_int_mode): New class.
18844 (scalar_int_mode::includes_p): New function.
18845 (byte_mode): Change type to scalar_int_mode.
18846 (word_mode): Likewise.
18847 (ptr_mode): Likewise.
18848 * emit-rtl.c (byte_mode): Likewise.
18849 (word_mode): Likewise.
18850 (ptr_mode): Likewise.
18851 (init_derived_machine_modes): Update accordingly.
18852 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18853 and MODE_PARTIAL_INT.
18854 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18855 opt_scalar_int_mode.
18857 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18858 Alan Hayward <alan.hayward@arm.com>
18859 David Sherwood <david.sherwood@arm.com>
18861 * target.def (libgcc_floating_mode_supported_p): Take a
18863 * doc/tm.texi: Regenerate.
18864 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18866 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18867 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
18870 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18871 Alan Hayward <alan.hayward@arm.com>
18872 David Sherwood <david.sherwood@arm.com>
18874 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
18875 * doc/tm.texi: Regenerate.
18876 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
18877 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
18878 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
18879 * targhooks.h (default_floatn_mode): Likewise.
18880 * targhooks.c (default_floatn_mode): Likewise.
18881 * tree.c (build_common_tree_nodes): Update accordingly.
18883 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18884 Alan Hayward <alan.hayward@arm.com>
18885 David Sherwood <david.sherwood@arm.com>
18887 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
18888 (mode_iterator::iterate_p): Likewise.
18889 (mode_iterator::get_wider): Likewise.
18890 * expr.c (init_expr_target): Use opt_scalar_float_mode.
18892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18893 Alan Hayward <alan.hayward@arm.com>
18894 David Sherwood <david.sherwood@arm.com>
18896 * coretypes.h (opt_scalar_float_mode): New typedef.
18897 * machmode.h (float_mode_for_size): New function.
18898 * emit-rtl.c (double_mode): Delete.
18899 (init_emit_once): Use float_mode_for_size.
18900 * stor-layout.c (layout_type): Likewise.
18901 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
18903 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18904 Alan Hayward <alan.hayward@arm.com>
18905 David Sherwood <david.sherwood@arm.com>
18907 * output.h (assemble_real): Take a scalar_float_mode.
18908 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
18909 * config/arm/arm.md (consttable_4): Likewise.
18910 (consttable_8): Likewise.
18911 (consttable_16): Likewise.
18912 * config/mips/mips.md (consttable_float): Likewise.
18913 * config/s390/s390.c (s390_output_pool_entry): Likewise.
18914 * varasm.c (assemble_real): Take a scalar_float_mode.
18915 (output_constant_pool_2): Update accordingly.
18916 (output_constant): Likewise.
18918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18919 Alan Hayward <alan.hayward@arm.com>
18920 David Sherwood <david.sherwood@arm.com>
18922 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
18923 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
18924 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
18925 (native_encode_real): Likewise.
18926 (native_interpret_real): Likewise.
18927 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
18928 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18930 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18931 Alan Hayward <alan.hayward@arm.com>
18932 David Sherwood <david.sherwood@arm.com>
18934 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
18935 <scalar_float_mode>. Simplify.
18936 (gen_extend_conv_libfunc): Likewise.
18938 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18939 Alan Hayward <alan.hayward@arm.com>
18940 David Sherwood <david.sherwood@arm.com>
18942 * coretypes.h (scalar_float_mode): New type.
18943 * machmode.h (mode_traits::from_int): Use machine_mode if
18944 USE_ENUM_MODES is defined.
18945 (is_a): New function.
18947 (dyn_cast): Likewise.
18948 (scalar_float_mode): New class.
18949 (scalar_float_mode::includes_p): New function.
18950 (is_float_mode): Likewise.
18951 * gdbhooks.py (MachineModePrinter): New class.
18952 (build_pretty_printer): Use it for scalar_float_mode.
18953 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
18954 (format_helper::format_helper): Turn into a template.
18955 * genmodes.c (get_mode_class): New function.
18956 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
18957 or machine_mode if none.
18958 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
18959 as_a <scalar_float_mode>.
18960 * dwarf2out.c (mem_loc_descriptor): Likewise.
18961 (insert_float): Likewise.
18962 (add_const_value_attribute): Likewise.
18963 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18964 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
18965 (expand_unop): Update accordingly.
18966 (expand_abs_nojump): Likewise.
18967 (expand_copysign_absneg): Take a scalar_float_mode.
18968 (expand_copysign_bit): Likewise.
18969 (expand_copysign): Update accordingly.
18971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18972 Alan Hayward <alan.hayward@arm.com>
18973 David Sherwood <david.sherwood@arm.com>
18975 * coretypes.h (opt_mode): New class.
18976 * machmode.h (opt_mode): Likewise.
18977 (opt_mode::else_void): New function.
18978 (opt_mode::require): Likewise.
18979 (opt_mode::exists): Likewise.
18980 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
18981 (GET_MODE_2XWIDER_MODE): Likewise.
18982 (mode_iterator::get_wider): Update accordingly.
18983 (mode_iterator::get_2xwider): Likewise.
18984 (mode_iterator::get_known_wider): Likewise, turning into a template.
18985 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
18986 forcing a wider mode to exist.
18987 * config/cr16/cr16.h (LONG_REG_P): Likewise.
18988 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18989 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
18990 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
18991 * lower-subreg.c (init_lower_subreg): Likewise.
18992 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
18993 on the final iteration.
18994 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
18995 a wider mode exists before asking for a move pattern.
18996 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
18997 forcing a wider mode to exist.
18998 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
18999 returning false if no such mode exists.
19000 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
19001 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
19002 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
19003 Avoid checking for a MODE_INT if we already know the mode is not a
19005 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19006 forcing a wider mode to exist.
19007 (expmed_mult_highpart_optab): Likewise.
19008 (expmed_mult_highpart): Likewise.
19009 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19011 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19012 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19013 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19014 * internal-fn.c (expand_mul_overflow): Update use of
19015 GET_MODE_2XWIDER_MODE.
19016 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19017 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19018 GET_MODE_WIDER_MODE.
19019 (convert_plusminus_to_widen): Likewise.
19020 * tree-switch-conversion.c (array_value_type): Likewise.
19021 * var-tracking.c (emit_note_insn_var_location): Likewise.
19022 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19023 Return false inside rather than outside the loop if no wider mode
19025 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19026 and GET_MODE_2XWIDER_MODE
19027 (can_compare_p): Use else_void.
19028 * gdbhooks.py (OptMachineModePrinter): New class.
19029 (build_pretty_printer): Use it for opt_mode.
19031 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19032 Alan Hayward <alan.hayward@arm.com>
19033 David Sherwood <david.sherwood@arm.com>
19035 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19036 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19038 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19039 Alan Hayward <alan.hayward@arm.com>
19040 David Sherwood <david.sherwood@arm.com>
19042 * machmode.h (mode_traits): New structure.
19043 (get_narrowest_mode): New function.
19044 (mode_iterator::start): Likewise.
19045 (mode_iterator::iterate_p): Likewise.
19046 (mode_iterator::get_wider): Likewise.
19047 (mode_iterator::get_known_wider): Likewise.
19048 (mode_iterator::get_2xwider): Likewise.
19049 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19050 (FOR_EACH_MODE): Likewise.
19051 (FOR_EACH_MODE_FROM): Likewise.
19052 (FOR_EACH_MODE_UNTIL): Likewise.
19053 (FOR_EACH_WIDER_MODE): Likewise.
19054 (FOR_EACH_2XWIDER_MODE): Likewise.
19055 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19056 * combine.c (simplify_comparison): Likewise
19057 * config/i386/i386.c (type_natural_mode): Likewise.
19058 * cse.c (cse_insn): Likewise.
19059 * dse.c (find_shift_sequence): Likewise.
19060 * emit-rtl.c (init_derived_machine_modes): Likewise.
19061 (init_emit_once): Likewise.
19062 * explow.c (hard_function_value): Likewise.
19063 * expmed.c (extract_fixed_bit_field_1): Likewise.
19064 (extract_bit_field_1): Likewise.
19065 (expand_divmod): Likewise.
19066 (emit_store_flag_1): Likewise.
19067 * expr.c (init_expr_target): Likewise.
19068 (convert_move): Likewise.
19069 (alignment_for_piecewise_move): Likewise.
19070 (widest_int_mode_for_size): Likewise.
19071 (emit_block_move_via_movmem): Likewise.
19072 (copy_blkmode_to_reg): Likewise.
19073 (set_storage_via_setmem): Likewise.
19074 (compress_float_constant): Likewise.
19075 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19076 * optabs-query.c (get_best_extraction_insn): Likewise.
19077 * optabs.c (expand_binop): Likewise.
19078 (expand_twoval_unop): Likewise.
19079 (expand_twoval_binop): Likewise.
19080 (widen_leading): Likewise.
19081 (widen_bswap): Likewise.
19082 (expand_parity): Likewise.
19083 (expand_unop): Likewise.
19084 (prepare_cmp_insn): Likewise.
19085 (prepare_float_lib_cmp): Likewise.
19086 (expand_float): Likewise.
19087 (expand_fix): Likewise.
19088 (expand_sfix_optab): Likewise.
19089 * postreload.c (move2add_use_add2_insn): Likewise.
19090 * reg-stack.c (reg_to_stack): Likewise.
19091 * reginfo.c (choose_hard_reg_mode): Likewise.
19092 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19093 * stor-layout.c (mode_for_size): Likewise.
19094 (smallest_mode_for_size): Likewise.
19095 (mode_for_vector): Likewise.
19096 (finish_bitfield_representative): Likewise.
19097 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19098 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19099 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19100 * var-tracking.c (prepare_call_arguments): Likewise.
19102 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19103 Alan Hayward <alan.hayward@arm.com>
19104 David Sherwood <david.sherwood@arm.com>
19106 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19107 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19108 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19109 * machmode.h (mode_size): Move earlier in file.
19110 (mode_precision): Likewise.
19111 (mode_inner): Likewise.
19112 (mode_nunits): Likewise.
19113 (mode_unit_size): Likewise.
19114 (unit_unit_precision): Likewise.
19115 (mode_wider): Likewise.
19116 (mode_2xwider): Likewise.
19117 (machine_mode): New class.
19118 (mode_to_bytes): New function.
19119 (mode_to_bits): Likewise.
19120 (mode_to_precision): Likewise.
19121 (mode_to_inner): Likewise.
19122 (mode_to_unit_size): Likewise.
19123 (mode_to_unit_precision): Likewise.
19124 (mode_to_nunits): Likewise.
19125 (GET_MODE_SIZE): Use mode_to_bytes.
19126 (GET_MODE_BITSIZE): Use mode_to_bits.
19127 (GET_MODE_PRECISION): Use mode_to_precision.
19128 (GET_MODE_INNER): Use mode_to_inner.
19129 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
19130 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
19131 (GET_MODE_NUNITS): Use mode_to_nunits.
19132 * system.h (ALWAYS_INLINE): New macro.
19133 * config/powerpcspe/powerpcspe-c.c
19134 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
19135 int for arg1_mode and arg2_mode.
19137 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19138 Alan Hayward <alan.hayward@arm.com>
19139 David Sherwood <david.sherwood@arm.com>
19141 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19142 Prefix mode names with E_ in case statements.
19143 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19144 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
19145 (aarch64_split_simd_move): Likewise.
19146 (aarch64_gen_storewb_pair): Likewise.
19147 (aarch64_gen_loadwb_pair): Likewise.
19148 (aarch64_gen_store_pair): Likewise.
19149 (aarch64_gen_load_pair): Likewise.
19150 (aarch64_get_condition_code_1): Likewise.
19151 (aarch64_constant_pool_reload_icode): Likewise.
19152 (get_rsqrte_type): Likewise.
19153 (get_rsqrts_type): Likewise.
19154 (get_recpe_type): Likewise.
19155 (get_recps_type): Likewise.
19156 (aarch64_gimplify_va_arg_expr): Likewise.
19157 (aarch64_simd_container_mode): Likewise.
19158 (aarch64_emit_load_exclusive): Likewise.
19159 (aarch64_emit_store_exclusive): Likewise.
19160 (aarch64_expand_compare_and_swap): Likewise.
19161 (aarch64_gen_atomic_cas): Likewise.
19162 (aarch64_emit_bic): Likewise.
19163 (aarch64_emit_atomic_swap): Likewise.
19164 (aarch64_emit_atomic_load_op): Likewise.
19165 (aarch64_evpc_trn): Likewise.
19166 (aarch64_evpc_uzp): Likewise.
19167 (aarch64_evpc_zip): Likewise.
19168 (aarch64_evpc_ext): Likewise.
19169 (aarch64_evpc_rev): Likewise.
19170 (aarch64_evpc_dup): Likewise.
19171 (aarch64_gen_ccmp_first): Likewise.
19172 (aarch64_gen_ccmp_next): Likewise.
19173 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
19174 (alpha_emit_xfloating_libcall): Likewise.
19175 (emit_insxl): Likewise.
19176 (alpha_arg_type): Likewise.
19177 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
19178 (arc_preferred_simd_mode): Likewise.
19179 (arc_secondary_reload): Likewise.
19180 (get_arc_condition_code): Likewise.
19181 (arc_print_operand): Likewise.
19182 (arc_legitimate_constant_p): Likewise.
19183 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19184 * config/arc/arc.md (casesi_load): Likewise.
19185 (casesi_compact_jump): Likewise.
19186 * config/arc/predicates.md (proper_comparison_operator): Likewise.
19187 (cc_use_register): Likewise.
19188 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19189 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19190 (arm_init_iwmmxt_builtins): Likewise.
19191 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
19192 (neon_expand_vector_init): Likewise.
19193 (arm_attr_length_move_neon): Likewise.
19194 (maybe_get_arm_condition_code): Likewise.
19195 (arm_emit_vector_const): Likewise.
19196 (arm_preferred_simd_mode): Likewise.
19197 (arm_output_iwmmxt_tinsr): Likewise.
19198 (thumb1_output_casesi): Likewise.
19199 (thumb2_output_casesi): Likewise.
19200 (arm_emit_load_exclusive): Likewise.
19201 (arm_emit_store_exclusive): Likewise.
19202 (arm_expand_compare_and_swap): Likewise.
19203 (arm_evpc_neon_vuzp): Likewise.
19204 (arm_evpc_neon_vzip): Likewise.
19205 (arm_evpc_neon_vrev): Likewise.
19206 (arm_evpc_neon_vtrn): Likewise.
19207 (arm_evpc_neon_vext): Likewise.
19208 (arm_validize_comparison): Likewise.
19209 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
19210 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
19211 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
19212 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
19213 (c6x_preferred_simd_mode): Likewise.
19214 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
19215 (epiphany_rtx_costs): Likewise.
19216 * config/epiphany/predicates.md (proper_comparison_operator):
19218 * config/frv/frv.c (condexec_memory_operand): Likewise.
19219 (frv_emit_move): Likewise.
19220 (output_move_single): Likewise.
19221 (output_condmove_single): Likewise.
19222 (frv_hard_regno_mode_ok): Likewise.
19223 (frv_matching_accg_mode): Likewise.
19224 * config/h8300/h8300.c (split_adds_subs): Likewise.
19225 (h8300_rtx_costs): Likewise.
19226 (h8300_print_operand): Likewise.
19227 (compute_mov_length): Likewise.
19228 (output_logical_op): Likewise.
19229 (compute_logical_op_length): Likewise.
19230 (compute_logical_op_cc): Likewise.
19231 (h8300_shift_needs_scratch_p): Likewise.
19232 (output_a_shift): Likewise.
19233 (compute_a_shift_length): Likewise.
19234 (compute_a_shift_cc): Likewise.
19235 (expand_a_rotate): Likewise.
19236 (output_a_rotate): Likewise.
19237 * config/i386/i386.c (classify_argument): Likewise.
19238 (function_arg_advance_32): Likewise.
19239 (function_arg_32): Likewise.
19240 (function_arg_64): Likewise.
19241 (function_value_64): Likewise.
19242 (ix86_gimplify_va_arg): Likewise.
19243 (ix86_legitimate_constant_p): Likewise.
19244 (put_condition_code): Likewise.
19245 (split_double_mode): Likewise.
19246 (ix86_avx256_split_vector_move_misalign): Likewise.
19247 (ix86_expand_vector_logical_operator): Likewise.
19248 (ix86_split_idivmod): Likewise.
19249 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
19250 (ix86_build_const_vector): Likewise.
19251 (ix86_build_signbit_mask): Likewise.
19252 (ix86_match_ccmode): Likewise.
19253 (ix86_cc_modes_compatible): Likewise.
19254 (ix86_expand_branch): Likewise.
19255 (ix86_expand_sse_cmp): Likewise.
19256 (ix86_expand_sse_movcc): Likewise.
19257 (ix86_expand_int_sse_cmp): Likewise.
19258 (ix86_expand_vec_perm_vpermi2): Likewise.
19259 (ix86_expand_vec_perm): Likewise.
19260 (ix86_expand_sse_unpack): Likewise.
19261 (ix86_expand_int_addcc): Likewise.
19262 (ix86_split_to_parts): Likewise.
19263 (ix86_vectorize_builtin_gather): Likewise.
19264 (ix86_vectorize_builtin_scatter): Likewise.
19265 (avx_vpermilp_parallel): Likewise.
19266 (inline_memory_move_cost): Likewise.
19267 (ix86_tieable_integer_mode_p): Likewise.
19268 (x86_maybe_negate_const_int): Likewise.
19269 (ix86_expand_vector_init_duplicate): Likewise.
19270 (ix86_expand_vector_init_one_nonzero): Likewise.
19271 (ix86_expand_vector_init_one_var): Likewise.
19272 (ix86_expand_vector_init_concat): Likewise.
19273 (ix86_expand_vector_init_interleave): Likewise.
19274 (ix86_expand_vector_init_general): Likewise.
19275 (ix86_expand_vector_set): Likewise.
19276 (ix86_expand_vector_extract): Likewise.
19277 (emit_reduc_half): Likewise.
19278 (ix86_emit_i387_round): Likewise.
19279 (ix86_mangle_type): Likewise.
19280 (ix86_expand_round_sse4): Likewise.
19281 (expand_vec_perm_blend): Likewise.
19282 (canonicalize_vector_int_perm): Likewise.
19283 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
19284 (expand_vec_perm_1): Likewise.
19285 (expand_vec_perm_interleave3): Likewise.
19286 (expand_vec_perm_even_odd_pack): Likewise.
19287 (expand_vec_perm_even_odd_1): Likewise.
19288 (expand_vec_perm_broadcast_1): Likewise.
19289 (ix86_vectorize_vec_perm_const_ok): Likewise.
19290 (ix86_expand_vecop_qihi): Likewise.
19291 (ix86_expand_mul_widen_hilo): Likewise.
19292 (ix86_expand_sse2_abs): Likewise.
19293 (ix86_expand_pextr): Likewise.
19294 (ix86_expand_pinsr): Likewise.
19295 (ix86_preferred_simd_mode): Likewise.
19296 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
19297 * config/i386/sse.md (*andnot<mode>3): Likewise.
19298 (<mask_codefor><code><mode>3<mask_name>): Likewise.
19299 (*<code><mode>3): Likewise.
19300 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
19301 (ia64_expand_atomic_op): Likewise.
19302 (ia64_arg_type): Likewise.
19303 (ia64_mode_to_int): Likewise.
19304 (ia64_scalar_mode_supported_p): Likewise.
19305 (ia64_vector_mode_supported_p): Likewise.
19306 (expand_vec_perm_broadcast): Likewise.
19307 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
19308 (iq2000_function_arg_advance): Likewise.
19309 (iq2000_function_arg): Likewise.
19310 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
19311 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
19312 (m68k_libcall_value): Likewise.
19313 (m68k_function_value): Likewise.
19314 (sched_attr_op_type): Likewise.
19315 * config/mcore/mcore.c (mcore_output_move): Likewise.
19316 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
19318 (microblaze_function_arg): Likewise.
19319 * config/mips/mips.c (mips16_build_call_stub): Likewise.
19320 (mips_print_operand): Likewise.
19321 (mips_mode_ok_for_mov_fmt_p): Likewise.
19322 (mips_vector_mode_supported_p): Likewise.
19323 (mips_preferred_simd_mode): Likewise.
19324 (mips_expand_vpc_loongson_even_odd): Likewise.
19325 (mips_expand_vec_unpack): Likewise.
19326 (mips_expand_vi_broadcast): Likewise.
19327 (mips_expand_vector_init): Likewise.
19328 (mips_expand_vec_reduc): Likewise.
19329 (mips_expand_msa_cmp): Likewise.
19330 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
19331 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
19332 (cc_flags_for_mode): Likewise.
19333 * config/msp430/msp430.c (msp430_print_operand): Likewise.
19334 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
19335 (nds32_output_casesi_pc_relative): Likewise.
19336 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19337 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
19338 (nvptx_gen_unpack): Likewise.
19339 (nvptx_gen_pack): Likewise.
19340 (nvptx_gen_shuffle): Likewise.
19341 (nvptx_gen_wcast): Likewise.
19342 (nvptx_preferred_simd_mode): Likewise.
19343 * config/pa/pa.c (pa_secondary_reload): Likewise.
19344 * config/pa/predicates.md (base14_operand): Likewise.
19345 * config/powerpcspe/powerpcspe-c.c
19346 (altivec_resolve_overloaded_builtin): Likewise.
19347 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
19349 (rs6000_preferred_simd_mode): Likewise.
19350 (output_vec_const_move): Likewise.
19351 (rs6000_expand_vector_extract): Likewise.
19352 (rs6000_split_vec_extract_var): Likewise.
19353 (reg_offset_addressing_ok_p): Likewise.
19354 (rs6000_legitimate_offset_address_p): Likewise.
19355 (rs6000_legitimize_address): Likewise.
19356 (rs6000_emit_set_const): Likewise.
19357 (rs6000_const_vec): Likewise.
19358 (rs6000_emit_move): Likewise.
19359 (spe_build_register_parallel): Likewise.
19360 (rs6000_darwin64_record_arg_recurse): Likewise.
19361 (swap_selector_for_mode): Likewise.
19362 (spe_init_builtins): Likewise.
19363 (paired_init_builtins): Likewise.
19364 (altivec_init_builtins): Likewise.
19365 (do_load_for_compare): Likewise.
19366 (rs6000_generate_compare): Likewise.
19367 (rs6000_expand_float128_convert): Likewise.
19368 (emit_load_locked): Likewise.
19369 (emit_store_conditional): Likewise.
19370 (rs6000_output_function_epilogue): Likewise.
19371 (rs6000_handle_altivec_attribute): Likewise.
19372 (rs6000_function_value): Likewise.
19373 (emit_fusion_gpr_load): Likewise.
19374 (emit_fusion_p9_load): Likewise.
19375 (emit_fusion_p9_store): Likewise.
19376 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
19377 (fusion_gpr_mem_load): Likewise.
19378 (fusion_addis_mem_combo_load): Likewise.
19379 (fusion_addis_mem_combo_store): Likewise.
19380 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
19381 (fusion_gpr_mem_load): Likewise.
19382 (fusion_addis_mem_combo_load): Likewise.
19383 (fusion_addis_mem_combo_store): Likewise.
19384 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19386 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
19387 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19388 (rs6000_preferred_simd_mode): Likewise.
19389 (output_vec_const_move): Likewise.
19390 (rs6000_expand_vector_extract): Likewise.
19391 (rs6000_split_vec_extract_var): Likewise.
19392 (reg_offset_addressing_ok_p): Likewise.
19393 (rs6000_legitimate_offset_address_p): Likewise.
19394 (rs6000_legitimize_address): Likewise.
19395 (rs6000_emit_set_const): Likewise.
19396 (rs6000_const_vec): Likewise.
19397 (rs6000_emit_move): Likewise.
19398 (rs6000_darwin64_record_arg_recurse): Likewise.
19399 (swap_selector_for_mode): Likewise.
19400 (paired_init_builtins): Likewise.
19401 (altivec_init_builtins): Likewise.
19402 (rs6000_expand_float128_convert): Likewise.
19403 (emit_load_locked): Likewise.
19404 (emit_store_conditional): Likewise.
19405 (rs6000_output_function_epilogue): Likewise.
19406 (rs6000_handle_altivec_attribute): Likewise.
19407 (rs6000_function_value): Likewise.
19408 (emit_fusion_gpr_load): Likewise.
19409 (emit_fusion_p9_load): Likewise.
19410 (emit_fusion_p9_store): Likewise.
19411 * config/rx/rx.c (rx_gen_move_template): Likewise.
19412 (flags_from_mode): Likewise.
19413 * config/s390/predicates.md (s390_alc_comparison): Likewise.
19414 (s390_slb_comparison): Likewise.
19415 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
19416 (s390_vector_mode_supported_p): Likewise.
19417 (s390_cc_modes_compatible): Likewise.
19418 (s390_match_ccmode_set): Likewise.
19419 (s390_canonicalize_comparison): Likewise.
19420 (s390_emit_compare_and_swap): Likewise.
19421 (s390_branch_condition_mask): Likewise.
19422 (s390_rtx_costs): Likewise.
19423 (s390_secondary_reload): Likewise.
19424 (__SECONDARY_RELOAD_CASE): Likewise.
19425 (s390_expand_cs): Likewise.
19426 (s390_preferred_simd_mode): Likewise.
19427 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
19428 * config/sh/sh.c (sh_print_operand): Likewise.
19429 (dump_table): Likewise.
19430 (sh_secondary_reload): Likewise.
19431 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19432 * config/sh/sh.md (casesi_worker_1): Likewise.
19433 (casesi_worker_2): Likewise.
19434 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
19435 (fcc_comparison_operator): Likewise.
19436 * config/sparc/sparc.c (sparc_expand_move): Likewise.
19437 (emit_soft_tfmode_cvt): Likewise.
19438 (sparc_preferred_simd_mode): Likewise.
19439 (output_cbranch): Likewise.
19440 (sparc_print_operand): Likewise.
19441 (sparc_expand_vec_perm_bmask): Likewise.
19442 (vector_init_bshuffle): Likewise.
19443 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
19444 (spu_vector_mode_supported_p): Likewise.
19445 (spu_expand_insv): Likewise.
19446 (spu_emit_branch_or_set): Likewise.
19447 (spu_handle_vector_attribute): Likewise.
19448 (spu_builtin_splats): Likewise.
19449 (spu_builtin_extract): Likewise.
19450 (spu_builtin_promote): Likewise.
19451 (spu_expand_sign_extend): Likewise.
19452 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
19453 (tilegx_simd_int): Likewise.
19454 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
19455 (tilepro_simd_int): Likewise.
19456 * config/v850/v850.c (const_double_split): Likewise.
19457 (v850_print_operand): Likewise.
19458 (ep_memory_offset): Likewise.
19459 * config/vax/vax.c (vax_rtx_costs): Likewise.
19460 (vax_output_int_move): Likewise.
19461 (vax_output_int_add): Likewise.
19462 (vax_output_int_subtract): Likewise.
19463 * config/visium/predicates.md (visium_branch_operator): Likewise.
19464 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
19465 (visium_print_operand_address): Likewise.
19466 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19467 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
19468 (xtensa_expand_conditional_branch): Likewise.
19469 (xtensa_copy_incoming_a7): Likewise.
19470 (xtensa_output_literal): Likewise.
19471 * dfp.c (decimal_real_maxval): Likewise.
19472 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19474 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19475 Alan Hayward <alan.hayward@arm.com>
19476 David Sherwood <david.sherwood@arm.com>
19478 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
19479 (mode_nunits_inline): Likewise.
19480 (mode_inner_inline): Likewise.
19481 (mode_unit_size_inline): Likewise.
19482 (mode_unit_precision_inline): Likewise.
19483 (emit_insn_modes_h): Likewise. Also emit a #define of the
19485 (emit_mode_wider): Add an E_ prefix to mode names.
19486 (emit_mode_complex): Likewise.
19487 (emit_mode_inner): Likewise.
19488 (emit_mode_adjustments): Likewise.
19489 (emit_mode_int_n): Likewise.
19490 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
19491 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
19492 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
19493 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
19494 (CRC32_BUILTIN, ENTRY): Likewise.
19495 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
19496 (aarch64_pop_regs): Likewise.
19497 (aarch64_process_components): Likewise.
19498 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
19499 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
19500 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
19501 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
19502 * config/arm/arm.c (arm_init_libfuncs): Likewise.
19503 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
19505 * config/i386/i386-builtin.def (pcmpestr): Likewise.
19506 (pcmpistr): Likewise.
19507 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
19508 * config/mmix/mmix.c (mmix_output_condition): Likewise.
19509 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19511 * config/rl78/rl78.c (mduc_regs): Likewise.
19512 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19513 (htm_expand_builtin): Likewise.
19514 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
19515 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
19516 * config/xtensa/xtensa.c (print_operand): Likewise.
19517 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
19518 (NUM_MODE_VECTOR_INT): Likewise.
19519 * genoutput.c (null_operand): Likewise.
19520 (output_operand_data): Likewise.
19521 * genrecog.c (print_parameter_value): Likewise.
19522 * lra.c (debug_operand_data): Likewise.
19524 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19526 * dwarf2out.c (scompare_loc_descriptor_wide)
19527 (scompare_loc_descriptor_narrow): New functions, split out from...
19528 (scompare_loc_descriptor): ...here.
19529 * expmed.c (emit_store_flag_int): New function, split out from...
19530 (emit_store_flag): ...here.
19532 2017-08-30 Richard Biener <rguenther@suse.de>
19534 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
19535 (dwarf2out_early_finish): Move setting of AT_pubnames from
19536 early debug output to early finish.
19538 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19540 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
19541 and -mdata-region to the assembler.
19543 2017-08-30 Richard Biener <rguenther@suse.de>
19545 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
19547 (gen_subprogram_die): Add DW_AT_object_pointer only early.
19548 (dwarf2out_early_global_decl): Only generate a DIE for the
19549 abstract origin if it doesn't already exist or is a declaration DIE.
19550 (resolve_addr): Do not add the linkage name twice when
19551 generating a stub DIE for the DW_TAG_GNU_call_site target.
19553 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19555 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19556 Use machine_mode rather than int for arg1_mode.
19558 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19561 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
19562 that the second argument of the built-in functions to unpack
19563 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
19564 switch statement instead a lot of if statements.
19565 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
19566 Allow 64-bit values to be in Altivec registers as well as
19567 traditional floating point registers.
19568 (pack<mode>, FMOVE128_VSX iterator): Likewise.
19570 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
19572 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
19573 MAX_REGS_PER_ADDRESS == 1.
19575 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
19577 * config/i386/i386.opt (flag_fentry): Do not init to -1.
19578 * config/i386/i386.c (ix86_option_override_internal): Simplify
19579 setting of opts->x_flag_entry.
19581 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19582 Jakub Jelinek <jakub@redhat.com>
19583 Richard Biener <rguenther@suse.de>
19585 PR tree-optimization/81503
19586 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
19587 folded constant fits in the target type; reorder tests for clarity.
19589 2017-08-29 Martin Liska <mliska@suse.cz>
19591 * passes.def: Include pass_lower_switch.
19592 * stmt.c (dump_case_nodes): Remove and move to
19593 tree-switch-conversion.
19594 (case_values_threshold): Likewise.
19595 (expand_switch_as_decision_tree_p): Likewise.
19596 (emit_case_decision_tree): Likewise.
19597 (expand_case): Likewise.
19598 (balance_case_nodes): Likewise.
19599 (node_has_low_bound): Likewise.
19600 (node_has_high_bound): Likewise.
19601 (node_is_bounded): Likewise.
19602 (emit_case_nodes): Likewise.
19603 (struct simple_case_node): New struct.
19604 (add_case_node): Remove.
19605 (emit_case_dispatch_table): Use vector instead of case_list.
19606 (reset_out_edges_aux): Remove.
19607 (compute_cases_per_edge): Likewise.
19608 (expand_case): Build list of simple_case_node.
19609 (expand_sjlj_dispatch_table): Use it.
19610 * tree-switch-conversion.c (struct case_node): Moved from
19611 stmt.c and adjusted.
19612 (emit_case_nodes): Likewise.
19613 (node_has_low_bound): Likewise.
19614 (node_has_high_bound): Likewise.
19615 (node_is_bounded): Likewise.
19616 (case_values_threshold): Likewise.
19617 (reset_out_edges_aux): Likewise.
19618 (compute_cases_per_edge): Likewise.
19619 (add_case_node): Likewise.
19620 (dump_case_nodes): Likewise.
19621 (balance_case_nodes): Likewise.
19622 (expand_switch_as_decision_tree_p): Likewise.
19623 (emit_jump): Likewise.
19624 (emit_case_decision_tree): Likewise.
19625 (try_switch_expansion): Likewise.
19626 (do_jump_if_equal): Likewise.
19627 (emit_cmp_and_jump_insns): Likewise.
19628 (fix_phi_operands_for_edge): New function.
19629 (record_phi_operand_mapping): Likewise.
19630 (class pass_lower_switch): New pass.
19631 (pass_lower_switch::execute): New function.
19632 (make_pass_lower_switch): Likewise.
19633 (conditional_probability):
19634 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19635 * tree-pass.h: Add make_pass_lower_switch.
19637 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19640 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19643 2017-08-29 Richard Biener <rguenther@suse.de>
19645 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19646 we do not add a DW_AT_inline attribute twice.
19647 (gen_subprogram_die): Remove code setting DW_AT_inline on
19648 DECL_ABSTRACT_P nodes.
19650 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19652 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19653 calls to internal functions.
19654 (gimplify_modify_expr): Likewise.
19655 * tree-call-cdce.c (use_internal_fn): Likewise.
19656 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19657 (convert_to_divmod): Set the nothrow flag.
19658 * tree-if-conv.c (predicate_mem_writes): Likewise.
19659 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19660 (vectorizable_call): Likewise.
19661 (vectorizable_store): Likewise.
19662 (vectorizable_load): Likewise.
19663 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19664 (vect_recog_mask_conversion_pattern): Likewise.
19666 2017-08-29 Martin Liska <mliska@suse.cz>
19669 * gcc.c (driver_handle_option): Add new argument.
19670 * opts-common.c (handle_option): Pass
19671 target_option_override_hook.
19672 * opts-global.c (lang_handle_option): Add new option.
19673 (set_default_handlers): Add new argument.
19674 (decode_options): Likewise.
19675 * opts.c (target_handle_option): Likewise.
19676 (common_handle_option): Call target_option_override_hook.
19677 * opts.h (struct cl_option_handler_func): Add hook for
19678 target option override.
19679 (struct cl_option_handlers): Likewise.
19680 (set_default_handlers): Add new argument.
19681 (decode_options): Likewise.
19682 (common_handle_option): Likewise.
19683 (target_handle_option): Likewise.
19684 * toplev.c (toplev::main): Pass targetm.target_option.override
19687 2017-08-29 Richard Biener <rguenther@suse.de>
19688 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19690 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19691 life to the active subtree.
19693 2017-08-28 Jeff Law <law@redhat.com>
19695 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19696 derive_equivalences.
19697 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19698 Code moved into....
19699 (edge_info::derive_equivalences): New private member function
19701 * tree-ssa-dom.c (class edge_info): Changed from a struct
19702 to a class. Add ctor/dtor, methods and data members.
19703 (edge_info::edge_info): Renamed from allocate_edge_info.
19704 Initialize additional members.
19705 (edge_info::~edge_info): New.
19706 (free_dom_edge_info): Delete the edge info.
19707 (record_edge_info): Use new class & associated member functions.
19708 Tighten forms for testing for edge equivalences.
19709 (record_temporary_equivalences): Iterate over the simple
19710 equivalences rather than assuming there's only one per edge.
19711 (cprop_into_successor_phis): Iterate over the simple
19712 equivalences rather than assuming there's only one per edge.
19713 (optimize_stmt): Use operand_equal_p rather than pointer
19714 equality for mini-DSE code.
19716 2017-08-28 Nathan Sidwell <nathan@acm.org>
19718 * gcc.c (execute): Fold SIGPIPE handling into switch
19719 statement. Adjust internal error message.
19721 2017-08-28 Richard Biener <rguenther@suse.de>
19724 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19725 Do nothing for removed DIEs.
19727 2017-08-28 Richard Biener <rguenther@suse.de>
19729 PR tree-optimization/81977
19730 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19733 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19736 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19737 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19738 using targetm.gen_mem_thread_fence.
19740 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19743 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19744 predicate to register_operand. Reorder operands.
19745 (*btr<mode>): Ditto.
19746 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19747 (*btr<mode>_mask): Ditto.
19749 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19751 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19752 * config/rs6000/xmmintrin.h: New file.
19753 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19755 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19758 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19759 parameter and_insn and return it.
19760 (recombine_lvx_pattern): Insert a copy to ensure availability of
19761 the base register of the copied masking operation at the point of
19762 the instruction replacement.
19763 (recombine_stvx_pattern): Likewise.
19765 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19767 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19768 undocumented switches.
19769 (-mpower9-dform-vector): Likewise.
19770 (-mpower9-dform): Likewise.
19771 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19772 comments to delete references to -mpower9-dform* switches.
19773 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19774 Delete reference to -mpower9-dform* switches, test for
19775 -mpower9-vector instead.
19776 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19777 (OTHER_P9_VECTOR_MASKS): Likewise.
19778 (POWERPC_MASKS): Likewise.
19779 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19780 tests against -mpower9-dform* to -mpower9-vector. Delete code
19781 that checked for -mpower9-dform* consistancy with other options.
19782 Add test for -mpower9-misc to enable other power9 switches.
19783 (rs6000_init_hard_regno_mode_ok): Likewise.
19784 (rs6000_option_override_internal): Likewise.
19785 (rs6000_emit_prologue): Likewise.
19786 (rs6000_emit_epilogue): Likewise.
19787 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19788 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19789 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19791 (emit_fusion_p9_store): Likewise.
19792 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19793 resetting these macros if the assembler does not support ISA 3.0
19795 (TARGET_P9_DFORM_VECTOR): Likewise.
19796 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19797 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19799 2017-08-25 Alan Modra <amodra@gmail.com>
19801 PR rtl-optimization/81747
19802 * cse.c (cse_extended_basic_block): Don't attempt to record
19803 equivalences for degenerate conditional branches that branch
19804 to their fall-through.
19806 2017-08-24 Martin Sebor <msebor@redhat.com>
19808 PR middle-end/81908
19809 * gimple-fold.c (size_must_be_zero_p): New function.
19810 (gimple_fold_builtin_memory_op): Call it.
19812 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19814 * config/rs6000/mm_malloc.h: New file.
19816 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19818 PR tree-optimization/81913
19819 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19820 analysis when either IVs in condition can wrap.
19822 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19824 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19825 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19827 2017-08-24 Richard Biener <rguenther@suse.de>
19830 * targhooks.c (default_target_can_inline_p): Properly
19831 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19832 is present and always compare.
19833 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19834 infer -mfpmath=sse from TARGET_SSE_P.
19835 (ix86_can_inline_p): Properly use target_option_default_node when
19836 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19838 2017-08-24 Richard Biener <rguenther@suse.de>
19841 * dwarf2out.c (output_die): Handle flag_generate_offload like
19843 (output_comp_unit): Likewise.
19844 (gen_array_type_die): Likewise.
19845 (dwarf2out_early_finish): Likewise.
19846 (note_variable_value_in_expr): Likewise.
19847 (dwarf2out_finish): Likewise. Adjust assert.
19848 * cgraphunit.c (symbol_table::compile): Move setting of
19849 flag_generate_offload earlier ...
19850 (symbol_table::finalize_compilation_unit): ... here, before
19851 early debug finalization.
19853 2017-08-24 Richard Biener <rguenther@suse.de>
19855 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19856 and ipa-fnsummary.h.
19857 (ix86_can_inline_p): When ix86_fpmath flags do not match
19858 check whether the callee uses FP math at all.
19860 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19862 PR middle-end/81931
19863 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19864 instead of TYPE_PRECISION.
19866 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19867 Alan Hayward <alan.hayward@arm.com>
19868 David Sherwood <david.sherwood@arm.com>
19870 * combine.c (make_extraction): Use subreg_offset_from_lsb.
19872 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
19874 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
19876 (ix86_frame::stack_realign_allocate): New field.
19877 (struct machine_frame_state): Modify comments.
19878 (machine_frame_state::sp_realigned_fp_end): New field.
19879 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
19880 layout calculation.
19881 (sp_valid_at): Add assertion to assure no attempt to access invalid
19882 offset of a realigned stack.
19883 (fp_valid_at): Likewise.
19884 (choose_baseaddr): Modify comments.
19885 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
19886 ix86_expand_prologue.
19887 (ix86_expand_prologue): Modify stack realignment and allocation.
19888 (ix86_expand_epilogue): Modify comments.
19889 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
19890 avx2_runtime, avx512f, and avx512f_runtime.
19892 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
19894 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
19895 (mstackrealign): Do not init to -1.
19896 * config/i386/i386.c (ix86_option_override_internal):
19897 Check opts_set, not opts when setting default value of
19898 opts->x_ix86_force_align_arg_pointer.
19900 2017-08-23 Richard Biener <rguenther@suse.de>
19902 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
19903 lang_hooks.decl_printable_name.
19904 * print-rtl-function.c (print_rtx_function): Likewise.
19905 * tree-pretty-print.c (dump_function_header): Likewise.
19907 2017-08-23 Richard Biener <rguenther@suse.de>
19910 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
19911 -g0 at compile-time.
19913 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19915 PR middle-end/19706
19916 * doc/sourcebuild.texi (Other hardware attributes):
19919 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19921 PR middle-end/19706
19922 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
19923 Fix single-use check.
19925 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
19927 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
19929 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19931 * doc/install.texi: Modify to add more details on running selected
19934 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19936 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
19937 is combined with -mabi=ms.
19938 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
19941 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19943 PR tree-optimization/81488
19944 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
19945 and cached_basis fields.
19946 (MAX_SPREAD): New constant.
19947 (alloc_cand_and_find_basis): Initialize new fields.
19948 (clear_visited): New function.
19949 (create_phi_basis_1): Rename from create_phi_basis, set visited
19950 and cached_basis fields.
19951 (create_phi_basis): New wrapper function.
19952 (phi_add_costs_1): Rename from phi_add_costs, add spread
19953 parameter, set visited field, short-circuit when limits reached.
19954 (phi_add_costs): New wrapper function.
19955 (record_phi_increments_1): Rename from record_phi_increments, set
19957 (record_phi_increments): New wrapper function.
19958 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
19959 (phi_incr_cost): New wrapper function.
19960 (all_phi_incrs_profitable_1): Rename from
19961 all_phi_incrs_profitable, set visited field.
19962 (all_phi_incrs_profitable): New wrapper function.
19964 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19965 Alan Hayward <alan.hayward@arm.com>
19966 David Sherwood <david.sherwood@arm.com>
19968 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
19969 that takes the outer and inner modes.
19970 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
19971 comparison as the canonical test for a paradoxical subreg.
19972 * combine.c (simplify_set): Use paradoxical_subreg_p.
19973 (make_extraction): Likewise.
19974 (force_to_mode): Likewise.
19975 (rtx_equal_for_field_assignment_p): Likewise.
19976 (gen_lowpart_for_combine): Likewise.
19977 (simplify_comparison): Likewise.
19978 * cse.c (equiv_constant): Likewise.
19979 * expmed.c (store_bit_field_1): Likewise.
19980 * final.c (alter_subreg): Likewise.
19981 * fwprop.c (propagate_rtx): Likewise.
19982 (forward_propagate_subreg): Likewise.
19983 * ira-conflicts.c (ira_build_conflicts): Likewise.
19984 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
19985 * lra-constraints.c (curr_insn_transform): Likewise.
19986 (split_reg): Likewise.
19987 * lra-eliminations.c (move_plus_up): Likewise.
19988 (lra_eliminate_regs_1): Likewise.
19989 * recog.c (general_operand): Likewise.
19990 * ree.c (combine_reaching_defs): Likewise.
19991 * reload.c (push_reload): Likewise.
19992 (find_reloads): Likewise.
19993 * reload1.c (elimination_effects): Likewise.
19994 (compute_reload_subreg_offset): Likewise.
19995 (choose_reload_regs): Likewise.
19996 * rtlanal.c (subreg_lsb_1): Likewise.
19997 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19998 (simplify_subreg): Likewise.
19999 * var-tracking.c (track_loc_p): Likewise.
20000 * emit-rtl.c (byte_lowpart_offset): Likewise.
20001 (paradoxical_subreg_p): Delete out-of-line definition.
20003 2017-08-22 Jeff Law <law@redhat.com>
20005 PR tree-optimization/81741
20006 PR tree-optimization/71947
20007 * tree-ssa-dom.c: Include tree-inline.h.
20008 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20009 equivalences if one is more expensive to compute than the other.
20010 * tree-ssa-scopedtables.h (class const_or_copies): Make
20011 record_const_or_copy_raw method private.
20012 (class avail_exprs_stack): New method simplify_binary_operation.
20013 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20014 avail_exprs_stack::simplify_binary_operation as needed.
20015 (avail_exprs_stack::simplify_binary_operation): New function.
20017 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20019 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20020 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20021 (DOT_SYMBOLS): Likewise.
20022 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20023 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20024 (RS6000_ABI_NAME): Likewise.
20025 (TARGET_CMODEL): Likewise.
20026 (TOC_SECTION_ASM_OP): Likewise.
20027 (SET_CMODEL): New macro.
20028 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20030 2017-08-22 Richard Biener <rguenther@suse.de>
20032 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20033 to allow for free-lang-data replacements similar to verify_type_variant.
20035 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20036 Alan Hayward <alan.hayward@arm.com>
20037 David Sherwood <david.sherwood@arm.com>
20039 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20040 VOIDmode for the LABEL_REF.
20042 2017-08-22 Richard Biener <rguenther@suse.de>
20044 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20046 2017-08-22 Richard Biener <rguenther@suse.de>
20048 * common.opt (feliminate-dwarf2-dups): Ignore.
20049 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20050 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20051 same_die_p_wrap, compute_section_prefix,
20052 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20053 (comdat_symbol_id, comdat_symbol_number): Likewise.
20054 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20056 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20057 (output_die): Mark unreachable path unreachable.
20058 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20059 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20060 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20061 (dwarf2out_early_finish): Likewise.
20063 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20065 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20067 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20070 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20071 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20072 (avr_attribute_table) <io, io_low, address>: Initialize
20073 .decl_required with true.
20075 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20077 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20078 undocumented debugging options.
20079 (-mvsx-scalar-double): Likewise.
20080 (-mallow-df-permute): Likewise.
20081 (-mvectorize-builtins): Likewise.
20082 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20083 (rs6000_builtin_vectorized_function): Likewise.
20084 (rs6000_builtin_md_vectorized_function): Likewise.
20085 (rs6000_opt_vars): Likewise.
20087 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20090 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20091 (*btrq_imm): Rename from *btrq.
20092 (*btcq_imm): Rename from *btcq.
20093 (btsc): New code attribute.
20094 (*<btsc><mode>): New insn pattern.
20095 (*btr<mode>): Ditto.
20096 (*<btsc><mode>_mask): New insn_and_split pattern.
20097 (*btr<mode>_mask): Ditto.
20099 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20100 Alan Hayward <alan.hayward@arm.com>
20101 David Sherwood <david.sherwood@arm.com>
20103 * function.c (pad_below): Simplify padding calculation.
20105 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20106 Alan Hayward <alan.hayward@arm.com>
20107 David Sherwood <david.sherwood@arm.com>
20109 * target.def (function_prologue): Remove frame size argument.
20110 (function_epilogue): Likewise.
20111 * doc/tm.texi: Regenerate.
20112 * final.c (final_start_function): Update call to function_prologue.
20113 (final_end_function): Update call to function_epilogue.
20114 (default_function_pro_epilogue): Remove frame size argument.
20115 * output.h (default_function_pro_epilogue): Likewise.
20116 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
20117 (arm_output_function_prologue): Likewise.
20118 * config/frv/frv.c (frv_function_prologue): Likewise.
20119 (frv_function_epilogue): Likewise.
20120 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
20121 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
20122 (ia64_output_function_epilogue): Likewise.
20123 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
20124 (m32r_output_function_epilogue): Likewise.
20125 * config/microblaze/microblaze.c (microblaze_function_prologue)
20126 (microblaze_function_epilogue): Likewise.
20127 * config/mips/mips.c (mips_output_function_prologue): Likewise.
20128 (mips_output_function_epilogue): Likewise.
20129 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
20130 (mmix_target_asm_function_epilogue): Likewise.
20131 * config/msp430/msp430.c (msp430_start_function): Likewise.
20132 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
20133 (nds32_asm_function_epilogue): Likewise.
20134 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
20135 * config/pa/pa.c (pa_output_function_prologue): Likewise.
20136 (pa_output_function_epilogue): Likewise.
20137 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
20138 (rs6000_output_function_epilogue): Likewise.
20139 * config/rl78/rl78.c (rl78_start_function): Likewise.
20140 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
20141 (rs6000_output_function_epilogue): Likewise.
20142 * config/rx/rx.c (rx_output_function_prologue): Likewise.
20143 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
20144 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
20145 (sparc_asm_function_epilogue): Likewise.
20147 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20149 * tree.h (type_has_mode_precision_p): New function.
20150 * convert.c (convert_to_integer_1): Use it.
20151 * expr.c (expand_expr_real_2): Likewise.
20152 (expand_expr_real_1): Likewise.
20153 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
20154 * match.pd: Likewise.
20155 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
20156 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
20157 * tree-tailcall.c (process_assignment): Likewise.
20158 * tree-vect-loop.c (vectorizable_reduction): Likewise.
20159 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
20160 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
20161 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
20162 (vectorizable_assignment): Likewise.
20163 (vectorizable_shift): Likewise.
20164 (vectorizable_operation): Likewise.
20165 * tree-vrp.c (register_edge_assert_for_2): Likewise.
20167 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
20169 * match.pd: Add pow (C, x) simplification.
20171 2017-08-21 Richard Biener <rguenther@suse.de>
20173 PR tree-optimization/81900
20174 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
20175 for blocks with abnormal predecessors.
20176 (compute_antic): Do not set visited flag prematurely.
20178 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
20181 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
20183 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20185 * stor-layout.h (vector_type_mode): Move to...
20186 * tree.h (vector_type_mode): ...here.
20187 * stor-layout.c (vector_type_mode): Move to...
20188 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
20190 2017-08-21 Richard Biener <rguenther@suse.de>
20192 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
20193 register_external_die hooks.
20194 (debug_false_tree_charstarstar_uhwistar): Declare.
20195 (debug_nothing_tree_charstar_uhwi): Likewise.
20196 * debug.c (do_nothing_debug_hooks): Adjust.
20197 (debug_false_tree_charstarstar_uhwistar): New do nothing.
20198 (debug_nothing_tree_charstar_uhwi): Likewise.
20199 * dbxout.c (dbx_debug_hooks): Adjust.
20200 (xcoff_debug_hooks): Likewise.
20201 * sdbout.c (sdb_debug_hooks): Likewise.
20202 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20203 * dwarf2out.c (macinfo_label_base): New global.
20204 (dwarf2out_register_external_die): New function for the
20205 register_external_die hook.
20206 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
20207 (dwarf2_debug_hooks): Use them.
20208 (dwarf2_lineno_debug_hooks): Adjust.
20209 (struct die_struct): Add with_offset flag.
20210 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
20211 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
20212 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
20213 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
20214 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
20215 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
20216 defining section names for the early LTO debug variants.
20217 (reset_indirect_string): New helper.
20218 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
20219 (print_dw_val): Add support for offsetted symbol references.
20220 (get_ultimate_context): Split out from is_cxx.
20221 (is_cxx): Use get_ultimate_context.
20222 (is_fortran): Add decl overload.
20223 (compute_comp_unit_symbol): Split out worker from
20224 compute_section_prefix.
20225 (compute_section_prefix): Call compute_comp_unit_symbol and
20226 set comdat_type_p here.
20227 (output_die): Skip DIE symbol output for the LTO added one.
20228 Handle DIE symbol references with offset.
20229 (output_comp_unit): Guard section name mangling properly.
20230 For LTO debug sections emit a symbol at the section beginning
20231 which we use to refer to its DIEs.
20232 (add_abstract_origin_attribute): For DIEs registered via
20233 dwarf2out_register_external_die directly refer to the early
20234 DIE rather than indirectly through the shadow one we created.
20235 Remove obsolete call to dwarf2out_abstract_function for
20236 non-function/block origins.
20237 (gen_array_type_die): When generating early LTO debug do
20238 not emit DW_AT_string_length.
20239 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
20240 late when in LTO. As suggested place a gcc_unreachable for
20241 the DECL_ABSTRACT_P case.
20242 (gen_subprogram_die): Avoid another specification DIE
20243 for early built declarations/definitions for the late LTO case.
20244 (gen_variable_die): Add type references for late duplicated VLA dies
20246 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
20247 we have the abstract instance already.
20248 (process_scope_var): Adjust decl DIE contexts in LTO which
20249 first puts them in limbo.
20250 (gen_decl_die): Do not generate type DIEs late apart from
20251 types for VLAs or for decls we do not yet have a DIE. Do not
20252 call dwarf2out_abstract_function late.
20253 (dwarf2out_early_global_decl): Make sure to create DIEs
20254 for abstract instances of a decl first.
20255 (dwarf2out_late_global_decl): Adjust comment.
20256 (output_macinfo_op): With multiple macro sections use
20257 macinfo_label_base to distinguish labels.
20258 (output_macinfo): Likewise. Update macinfo_label_base.
20259 Pass in the line info label.
20260 (note_variable_value_in_expr): When generating LTO resolve
20261 all variable values here by generating DIEs as needed.
20262 (init_sections_and_labels): Add early LTO debug flag parameter
20263 and generate different sections and names if set. Add generation
20264 counter for the labels so we can have multiple of them.
20265 (reset_dies): Helper to allow DIEs to be output multiple times.
20266 (dwarf2out_finish): When outputting DIEs to the fat part of an
20267 LTO object first reset DIEs.
20268 (dwarf2out_early_finish): Output early DIEs when generating LTO.
20269 (modified_type_die): Check for decl_ultimate_origin being self
20271 (gen_type_die_with_usage): Likewise.
20272 (gen_typedef_die): Allow decl_ultimate_origin being self.
20273 (set_decl_abstract_flags): Remove.
20274 (set_block_abstract_flags): Likewise.
20275 (dwarf2out_abstract_function): Treat the early generated DIEs
20276 as the abstract copy and only add DW_AT_inline and
20277 DW_AT_artificial here and call set_decl_origin_self.
20278 If the DIE has an abstract origin don't do anything.
20279 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
20280 if we have none yet (Go fails to build one, PR78628).
20281 (variably_modified_type_p): Prevent endless recursion for Ada
20282 cyclic pointer types.
20283 * lto-streamer-in.c: Include debug.h.
20284 (dref_queue): New global.
20285 (lto_read_tree_1): Stream in DIE references.
20286 (lto_input_tree): Register DIE references.
20287 (input_function): Stream DECL_DEBUG_ARGS.
20288 * lto-streamer-out.c: Include debug.h.
20289 (lto_write_tree_1): Output DIE references.
20290 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
20291 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
20292 (output_function): Stream DECL_DEBUG_ARGS.
20293 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
20294 Stream DECL_ABSTRACT_ORIGIN.
20295 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
20296 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
20297 DECL_CONTEXT for file-scope decls.
20298 * lto-streamer.h (struct dref_entry): Declare.
20299 (dref_queue): Likewise.
20300 * cfgexpand.c (pass_expand::execute): Do not call the
20301 outlining_inline_function hook here.
20302 * lto-wrapper.c (debug_obj): New global.
20303 (tool_cleanup): Unlink it if required.
20304 (debug_objcopy): New function.
20305 (run_gcc): Handle early debug sections in the IL files by
20306 extracting them to separate files, partially linkin them and
20307 feeding the result back as result to the linker.
20308 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
20309 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
20310 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
20311 sections into a separate segment.
20312 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
20314 (darwin_asm_dwarf_section): Likewise.
20315 (darwin_asm_output_dwarf_offset): Likewise.
20316 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
20318 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20319 Alan Hayward <alan.hayward@arm.com>
20320 David Sherwood <david.sherwood@arm.com>
20322 * read-md.h (md_reader::record_potential_iterator_use): Replace
20323 pointer argument with an rtx and an index.
20324 * read-rtl.c (iterator_group::apply_iterator): Likewise.
20325 (apply_mode_iterator): Likewise.
20326 (apply_code_iterator): Likewise.
20327 (apply_int_iterator): Likewise.
20328 (apply_subst_iterator): Likewise.
20329 (record_iterator_use): Likewise.
20330 (record_attribute_use): Likewise.
20331 (md_reader::record_potential_iterator_use): Likewise. Update calls
20332 to record_iterator_use and apply_iterator.
20333 (iterator_use): Replace ptr with x and index.
20334 (attribute_use): Likewise.
20335 (apply_attribute_uses): Update calls to apply_iterator.
20336 (apply_iterators): Likewise. Update initialization of iterator_use.
20337 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
20338 and record_potential_iterator_use.
20339 (rtx_reader::read_rtx_operand): Likewise.
20341 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20342 Alan Hayward <alan.hayward@arm.com>
20343 David Sherwood <david.sherwood@arm.com>
20345 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
20348 2017-08-21 Richard Biener <rguenther@suse.de>
20350 PR middle-end/81884
20351 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
20352 at struct end conservatively when comparing common bases.
20354 2017-08-21 Richard Biener <rguenther@suse.de>
20356 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
20357 (mem_ref_in_stmt): Remove.
20358 (determine_max_movement): Use ref index to get at the reference.
20359 (invariantness_dom_walker::before_dom_children): Deal with
20360 lim data already initialized.
20361 (gather_mem_refs_stmt): Initialize lim data and record ref index.
20363 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20365 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
20366 (TARGET_ISA_ROUND): Ditto.
20367 (TARGET_ROUND): Ditto.
20368 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
20369 * config/i386/i386.md: Ditto.
20370 * config/i386/sse.md: Ditto.
20371 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
20372 with OPTION_MASK_ISA_SSE4_1.
20374 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20377 * doc/extend.texi (x86 Built-in Functions): Correct the name of
20378 __builtin_ia32_lzcnt_u16.
20380 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
20383 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
20384 (rs6000_set_current_function): Rewrite function to use it.
20386 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
20389 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
20390 and TYPE_WARN_IF_NOT_ALIGN.
20391 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
20392 (handle_warn_if_not_align): New.
20393 (place_union_field): Call handle_warn_if_not_align.
20394 (place_field): Call handle_warn_if_not_align.
20395 Copy TYPE_WARN_IF_NOT_ALIGN.
20396 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
20397 (layout_type): Likewise.
20398 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
20400 (tree_decl_common): Add warn_if_not_align.
20401 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
20402 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
20403 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
20404 (DECL_WARN_IF_NOT_ALIGN): Likewise.
20405 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
20406 * doc/extend.texi: Document warn_if_not_aligned attribute.
20407 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
20409 2017-08-17 Martin Liska <mliska@suse.cz>
20412 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
20413 (get_data_dependence): Use it as pointer type.
20414 (distribute_loop): Likewise.
20416 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20418 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
20419 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
20420 (p8_vmrgow_<mode>_direct): New define_insn.
20421 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
20422 handle endianness for vmrgew and vmrgow permute patterns.
20424 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20426 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
20427 * config/rs6000/rs6000-cpus.def: Remove comment.
20428 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
20429 (POWERPC_MASKS): Likewise.
20430 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
20431 use of TARGET_VSX_TIMODE.
20432 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
20433 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
20434 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
20435 (rs6000_option_override_internal): Remove dead code.
20436 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
20437 (rs6000_legitimize_reload_address): Likewise.
20438 (rs6000_legitimate_address_p): Likewise.
20439 (rs6000_opt_masks): Delete "vsx-timode".
20440 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
20441 from function comment.
20442 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
20443 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
20444 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
20446 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
20447 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
20448 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
20449 (VSX_TI): Likewise.
20451 (define_peephole2): Likewise.
20453 2017-08-17 Martin Sebor <msebor@redhat.com>
20456 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
20457 past the end of an array.
20458 (test_pp_format): Add test cases.
20460 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
20462 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
20463 missing ECF_NOTHROW flags.
20465 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20468 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
20469 operands residing in integer registers.
20470 (*vsx_le_perm_load_<mode>): Likewise.
20471 (*vsx_le_perm_store_<mode>): Likewise.
20472 (define_peephole2): Add peepholes to optimize the above.
20474 2017-08-17 Marek Polacek <polacek@redhat.com>
20476 PR middle-end/81814
20477 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
20478 to mimic what shorten_compare did. Change the return type to bool.
20479 (fold_cond_expr_with_comparison): Update call to
20480 operand_equal_for_comparison_p.
20481 (fold_ternary_loc): Likewise.
20483 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
20485 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
20487 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
20489 2017-08-17 Richard Biener <rguenther@suse.de>
20491 * tree-ssa-structalias.c (solve_graph): When propagating
20492 to successors update the graphs succ edges and avoid duplicate work.
20494 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
20497 * config/i386/i386.c (ix86_option_override_internal): Save target
20498 specific options after ix86_stack_protector_guard_reg was changed.
20500 2017-08-17 Richard Biener <rguenther@suse.de>
20502 PR tree-optimization/81827
20503 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
20504 (new_var_info): Initialize it conservatively.
20505 (get_call_vi): Mark register vars.
20506 (new_scalar_tmp_constraint_exp): Likewise.
20507 (handle_rhs_call): Likewise.
20508 (handle_const_call): Likewise.
20509 (create_function_info_for): Likewise.
20510 (solve_constraints): Sort varinfos to separate register from
20511 non-register vars to pack points-to solution bitmaps during
20514 2017-08-17 Marek Polacek <polacek@redhat.com>
20516 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
20518 2017-08-17 Richard Biener <rguenther@suse.de>
20520 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
20521 to true when overflow is undefined and we saturated the result.
20523 2017-08-17 Alan Modra <amodra@gmail.com>
20526 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
20527 Don't use store multiple if only one reg needs saving.
20528 (interesting_frame_related_regno): New function.
20529 (rs6000_frame_related): Don't emit frame info for regs that
20531 (rs6000_emit_epilogue): Likewise.
20533 2017-08-16 Nathan Sidwell <nathan@acm.org>
20535 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
20536 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
20537 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
20538 * tree.c (free_lang_data_in_type): Use else-if chain. Always
20539 clear TYPE_LANG_1. Remove obsolete member-function stripping.
20540 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
20541 (verify_type): Adjust for TYPE_BINFO move.
20542 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
20543 process TYPE_BINFO directly.
20544 (hash_tree): Likewise.
20545 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20547 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20550 2017-08-16 David Malcolm <dmalcolm@redhat.com>
20552 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
20554 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
20557 * config/i386/i386.md (*anddi_1_btr): Change predicates of
20558 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
20559 Add ix86_binary_operator_ok to insn constraint.
20560 (*iordi_1_bts): Ditto.
20561 (*xordi_1_btc): Ditto.
20562 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
20563 Update corresponding peephole2 pattern.
20567 2017-08-16 Bin Cheng <bin.cheng@arm.com>
20569 PR tree-optimization/81832
20570 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
20571 copy loop header which has IFN_LOOP_DIST_ALIAS call.
20573 2017-08-16 Marek Polacek <polacek@redhat.com>
20576 * fold-const.c (fold_indirect_ref_1): Restore original behavior
20577 regarding size_zero_node.
20579 2017-08-16 Martin Liska <mliska@suse.cz>
20582 * config.gcc: Respect previously set extra_objs in case
20585 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
20587 PR tree-optimization/81835
20588 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
20589 the phi SSA_NAME. Check that the condition in a COND_EXPR does
20590 not depend on the phi.
20592 2017-08-16 Alan Modra <amodra@gmail.com>
20594 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
20597 2017-08-16 Alan Modra <amodra@gmail.com>
20599 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
20600 (save_reg_p): ..into this. Update all callers.
20601 (first_reg_to_save): Simplify.
20603 2017-08-16 Alan Modra <amodra@gmail.com>
20605 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
20608 2017-08-15 Joseph Myers <joseph@codesourcery.com>
20612 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
20613 constant count if that count is less than 32.
20615 2017-08-15 Nathan Sidwell <nathan@acm.org>
20617 * gcc.c (execute): Emit friendlier message if inferior is killed
20618 by an external cause.
20620 2017-08-15 Richard Biener <rguenther@suse.de>
20622 PR tree-optimization/81790
20623 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20624 CONSTRUCTORs from simplifying and VN.
20626 2017-08-14 Martin Sebor <msebor@redhat.com>
20628 * builtin-attrs.def: Add comments.
20630 2017-08-14 Martin Sebor <msebor@redhat.com>
20633 * doc/extend.texi (attribute nonstring): Document new attribute.
20635 2017-08-14 Martin Sebor <msebor@redhat.com>
20638 * tree-diagnostic.c (default_tree_printer): Handle %G.
20639 * gimple-pretty-print.h (percent_G_format): Declare new function.
20640 * gimple-pretty-print.c (percent_G_format): Define.
20641 * tree-pretty-print.c (percent_K_format): Add argument.
20643 2017-08-14 Martin Sebor <msebor@redhat.com>
20645 PR translation/79998
20646 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20647 Remove a stray space.
20649 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20652 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20653 (*iordi_1_bts): Ditto.
20654 (*xordi_1_btc): Ditto.
20656 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20659 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20660 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20662 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20663 quoted strings, and make more translator-friendly.
20664 (darwin_rs6000_override_options): Likewise.
20665 (rs6000_option_override_internal): Likewise.
20666 (rs6000_return_in_memory): Fix overlong line.
20667 (init_cmulative_args): Use quoted strings, and make more
20668 translator-friendly.
20669 (rs6000_pass_by_reference): Fix overlong line.
20670 (def_builtin): Use quoted strings.
20671 (altivec_expand_predicate_builtin): Use quoted strings, and make
20672 more translator-friendly.
20673 (htm_expand_builtin): Use quoted strings.
20674 (cpu_expand_builtin): Use quoted strings, and make more
20675 translator-friendly.
20676 (altivec_expand_builtin): Likewise.
20677 (paired_expand_predicate_builtin): Likewise.
20678 (rs6000_invalid_builtin): Likewise.
20679 (builtin_function_type): Use quoted strings.
20680 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20681 more translator-friendly.
20682 (rs6000_trampoline_init): Likewise.
20683 (rs6000_handle_altivec_attribute): Likewise.
20684 (rs6000_inner_target_options): Use quoted strings.
20685 (rs6000_disable_incompatible_switches): Likewise.
20686 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20687 strings, and make more translator-friendly.
20688 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20690 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20692 PR tree-optimization/81799
20693 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20694 cond_expr to simple gimple operand.
20696 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20698 PR middle-end/46932
20699 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20701 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20705 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20706 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20707 TARGET_GASISR_PROLOGUES.
20708 * config/avr/avr.c (avr_option_override): Same.
20709 (avr_pass_pre_proep::execute): Same.
20711 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20714 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20715 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20717 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20719 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20720 define_insn_and_split. Split to a memory load from 0 in
20721 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20722 using PTR mode iterator.
20723 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20724 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20725 (*add_tp_<mode>): Redefine as define_insn_and_split.
20726 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20727 address space. Merge with *add_tp_x32 using PTR mode iterator.
20728 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20729 Split to an add with a memory load from 0 in
20730 DEFAULT_TLS_SEG_REG address space.
20732 2017-08-12 Andrew Pinski <apinski@cavium.com>
20734 * config/aarch64/aarch64-option-extensions.def (rdma):
20735 Fix feature string to what Linux prints out in /proc/cpuinfo.
20737 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20740 * dwarf2out.c (modified_type_die): For C typedef types that have
20741 an ultimate origin, process the ultimate origin instead of the
20743 (gen_typedef_die): Assert that input DECLs have no ultimate
20745 (gen_type_die_with_usage): For typedef variants that have an
20746 ultimate origin, just call gen_decl_die on the original DECL.
20747 (process_scope_var): Avoid creating DIEs for local typedefs and
20748 concrete static variables.
20750 2017-08-12 Alan Modra <amodra@gmail.com>
20754 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20755 match gnu-user.h startfile.
20756 (ENDFILE_LINUX_SPEC): Similarly.
20758 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20761 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20763 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20765 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20766 * config/aarch64/aarch64.md (mov<mode>): Change.
20767 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20768 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20769 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20771 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20773 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20774 for storage order barriers.
20776 2017-08-11 Martin Liska <mliska@suse.cz>
20779 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20780 variables of void type.
20782 2017-08-11 Martin Liska <mliska@suse.cz>
20784 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20785 TARGET_SUPPORTS_ALIASES.
20786 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20787 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20788 (optimize_weakref): Likewise.
20789 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20790 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20791 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20793 2017-08-11 Martin Liska <mliska@suse.cz>
20796 * config/i386/i386.c (make_resolver_func): Do complete
20797 refactoring of the function.
20799 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20802 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20803 * config/i386/i386.c (ix86_stack_protect_guard): Use
20804 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20805 * doc/invoke.texi (x86 Options): Document
20806 -mstack-protector-guard-symbol= option.
20808 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20810 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20811 * config/i386/i386.c (ix86_split_stack_guard): New function.
20812 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20813 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20814 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20815 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20816 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20817 (split_stack_space_check): Call ix86_split_stack_guard.
20819 2017-08-10 Martin Sebor <msebor@redhat.com>
20821 * print-tree.c (print_node): Print location using the established
20823 Replace spaces with colons.
20824 (debug_raw, debug): Ditto.
20826 2017-08-10 Martin Sebor <msebor@redhat.com>
20829 * pretty-print.c (pp_format): Correct the handling of %s precision.
20831 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20834 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20836 (ix86_finalize_stack_frame_flags): This. Also clear
20837 frame_pointer_needed if -fno-omit-frame-pointer is used without
20839 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20840 with ix86_finalize_stack_frame_flags.
20841 (ix86_expand_epilogue): Likewise.
20842 (ix86_expand_split_stack_prologue): Likewise.
20843 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20845 2017-08-10 Martin Liska <mliska@suse.cz>
20848 * c-attribs.c (handle_target_attribute):
20849 Report warning for an empty string argument of target attribute.
20851 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20854 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20856 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20857 or DECL_NONLOCAL labels.
20858 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20859 or DECL_NONLOCAL labels here.
20861 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20863 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20864 to indicate when early gimple folding has been disabled.
20865 (rs6000_gimple_fold_builtin): Add debug content.
20866 (rs6000_invalid_builtin): Fix whitespace.
20867 (rs6000_expand_builtin): Fix whitespace.
20868 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
20870 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
20873 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
20874 SAVE_MULTIPLE if not all the registers that saves, should be saved.
20876 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
20878 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
20879 (qdf24xx): Likewise.
20880 * config/aarch64/aarch64.md: Include falkor.md.
20881 * config/aarch64/falkor.md: New.
20883 2017-08-09 Marek Polacek <polacek@redhat.com>
20886 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
20887 * diagnostic.c (emit_diagnostic): Add a comment.
20888 (emit_diagnostic_valist): New function.
20890 2017-08-09 Marek Polacek <polacek@redhat.com>
20893 * input.c (make_location): New overload.
20894 * input.h (make_location): Declare.
20896 2017-08-08 Alan Modra <amodra@gmail.com>
20897 H.J. Lu <hongjiu.lu@intel.com>
20900 * gcc.c (NO_PIE_SPEC): Delete.
20901 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
20903 (LINK_PIE_SPEC): ..to here.
20904 (LINK_COMMAND_SPEC): Support -no-pie.
20905 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
20906 chain of crtbegin*.o selection, update for PIE_SPEC changes and
20908 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
20909 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
20910 (ENDFILE_CRTEND_SPEC): Similarly.
20912 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
20915 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
20916 (mstack-protector-guard-offset=): Ditto.
20917 * config/i386/i386.c (ix86_option_override): Handle
20918 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
20920 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
20921 ix86_stack_protect_guard_offset variables.
20922 (TARGET_STACK_PROTECT_GUARD): Always define.
20923 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
20924 and -mstack-protector-guard-offset= options.
20926 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20928 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
20929 boundary case for the last candidate.
20931 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20933 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
20934 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
20936 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20938 PR middle-end/19706
20939 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
20940 * config/aarch64/aarch64-builtins.c
20941 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
20942 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
20943 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
20945 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20946 Andrew Pinski <pinskia@gmail.com>
20948 PR middle-end/19706
20949 * internal-fn.def (XORSIGN): New.
20950 * optabs.def (xorsign_optab): New.
20951 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
20952 (convert_expand_mult_copysign): New.
20953 (pass_optimize_widening_mul::execute): Call
20954 convert_expand_mult_copysign.
20956 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20958 PR tree-optimization/81354
20959 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
20960 Insert on edges rather than explicitly creating landing pads.
20961 (analyze_candidates_and_replace): Commit edge inserts.
20963 2017-08-08 Richard Biener <rguenther@suse.de>
20965 PR middle-end/81719
20966 * tree-ssa-loop-niter.c: Include tree-dfa.h.
20967 (expand_simple_operations): Also look through ADDR_EXPRs with
20968 MEM_REF bases treating them as POINTER_PLUS_EXPR.
20970 2017-08-08 Richard Biener <rguenther@suse.de>
20972 PR tree-optimization/81723
20973 * tree-vect-slp.c (struct bst_traits): New hash traits.
20974 (bst_fail): New global.
20975 (vect_build_slp_tree_2): New worker, split out from ...
20976 (vect_build_slp_tree): ... this now wrapping it with using
20977 bst_fail set to cache SLP tree build fails. Properly handle
20979 (vect_analyze_slp_instance): Allocate and free bst_fail.
20981 2017-08-08 Martin Liska <mliska@suse.cz>
20984 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
20985 LABEL_DECLs that can be from a different function.
20987 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20989 PR tree-optimization/81744
20990 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
20991 loop's number of iterations.
20993 2017-08-08 Martin Liska <mliska@suse.cz>
20995 * asan.c: Include header files.
20996 * attribs.c (build_decl_attribute_variant): New function moved
20998 (build_type_attribute_qual_variant): Likewise.
20999 (cmp_attrib_identifiers): Likewise.
21000 (simple_cst_list_equal): Likewise.
21001 (omp_declare_simd_clauses_equal): Likewise.
21002 (attribute_value_equal): Likewise.
21003 (comp_type_attributes): Likewise.
21004 (build_type_attribute_variant): Likewise.
21005 (lookup_ident_attribute): Likewise.
21006 (remove_attribute): Likewise.
21007 (merge_attributes): Likewise.
21008 (merge_type_attributes): Likewise.
21009 (merge_decl_attributes): Likewise.
21010 (merge_dllimport_decl_attributes): Likewise.
21011 (handle_dll_attribute): Likewise.
21012 (attribute_list_equal): Likewise.
21013 (attribute_list_contained): Likewise.
21014 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21015 (lookup_attribute_by_prefix): Likewise.
21016 * bb-reorder.c: Include header files.
21017 * builtins.c: Likewise.
21018 * calls.c: Likewise.
21019 * cfgexpand.c: Likewise.
21020 * cgraph.c: Likewise.
21021 * cgraphunit.c: Likewise.
21022 * convert.c: Likewise.
21023 * dwarf2out.c: Likewise.
21024 * final.c: Likewise.
21025 * fold-const.c: Likewise.
21026 * function.c: Likewise.
21027 * gimple-expr.c: Likewise.
21028 * gimple-fold.c: Likewise.
21029 * gimple-pretty-print.c: Likewise.
21030 * gimple.c: Likewise.
21031 * gimplify.c: Likewise.
21032 * hsa-common.c: Likewise.
21033 * hsa-gen.c: Likewise.
21034 * internal-fn.c: Likewise.
21035 * ipa-chkp.c: Likewise.
21036 * ipa-cp.c: Likewise.
21037 * ipa-devirt.c: Likewise.
21038 * ipa-fnsummary.c: Likewise.
21039 * ipa-inline.c: Likewise.
21040 * ipa-visibility.c: Likewise.
21042 * lto-cgraph.c: Likewise.
21043 * omp-expand.c: Likewise.
21044 * omp-general.c: Likewise.
21045 * omp-low.c: Likewise.
21046 * omp-offload.c: Likewise.
21047 * omp-simd-clone.c: Likewise.
21048 * opts-global.c: Likewise.
21049 * passes.c: Likewise.
21050 * predict.c: Likewise.
21051 * sancov.c: Likewise.
21052 * sanopt.c: Likewise.
21053 * symtab.c: Likewise.
21054 * toplev.c: Likewise.
21055 * trans-mem.c: Likewise.
21056 * tree-chkp.c: Likewise.
21057 * tree-eh.c: Likewise.
21058 * tree-into-ssa.c: Likewise.
21059 * tree-object-size.c: Likewise.
21060 * tree-parloops.c: Likewise.
21061 * tree-profile.c: Likewise.
21062 * tree-ssa-ccp.c: Likewise.
21063 * tree-ssa-live.c: Likewise.
21064 * tree-ssa-loop.c: Likewise.
21065 * tree-ssa-sccvn.c: Likewise.
21066 * tree-ssa-structalias.c: Likewise.
21067 * tree-ssa.c: Likewise.
21068 * tree-streamer-in.c: Likewise.
21069 * tree-vectorizer.c: Likewise.
21070 * tree-vrp.c: Likewise.
21071 * tsan.c: Likewise.
21072 * ubsan.c: Likewise.
21073 * varasm.c: Likewise.
21074 * varpool.c: Likewise.
21075 * tree.c: Remove functions moved to attribs.[ch].
21076 * tree.h: Likewise.
21077 * config/aarch64/aarch64.c: Add attrs.h header file.
21078 * config/alpha/alpha.c: Likewise.
21079 * config/arc/arc.c: Likewise.
21080 * config/arm/arm.c: Likewise.
21081 * config/avr/avr.c: Likewise.
21082 * config/bfin/bfin.c: Likewise.
21083 * config/c6x/c6x.c: Likewise.
21084 * config/cr16/cr16.c: Likewise.
21085 * config/cris/cris.c: Likewise.
21086 * config/darwin.c: Likewise.
21087 * config/epiphany/epiphany.c: Likewise.
21088 * config/fr30/fr30.c: Likewise.
21089 * config/frv/frv.c: Likewise.
21090 * config/ft32/ft32.c: Likewise.
21091 * config/h8300/h8300.c: Likewise.
21092 * config/i386/winnt.c: Likewise.
21093 * config/ia64/ia64.c: Likewise.
21094 * config/iq2000/iq2000.c: Likewise.
21095 * config/lm32/lm32.c: Likewise.
21096 * config/m32c/m32c.c: Likewise.
21097 * config/m32r/m32r.c: Likewise.
21098 * config/m68k/m68k.c: Likewise.
21099 * config/mcore/mcore.c: Likewise.
21100 * config/microblaze/microblaze.c: Likewise.
21101 * config/mips/mips.c: Likewise.
21102 * config/mmix/mmix.c: Likewise.
21103 * config/mn10300/mn10300.c: Likewise.
21104 * config/moxie/moxie.c: Likewise.
21105 * config/msp430/msp430.c: Likewise.
21106 * config/nds32/nds32-isr.c: Likewise.
21107 * config/nds32/nds32.c: Likewise.
21108 * config/nios2/nios2.c: Likewise.
21109 * config/nvptx/nvptx.c: Likewise.
21110 * config/pa/pa.c: Likewise.
21111 * config/pdp11/pdp11.c: Likewise.
21112 * config/powerpcspe/powerpcspe.c: Likewise.
21113 * config/riscv/riscv.c: Likewise.
21114 * config/rl78/rl78.c: Likewise.
21115 * config/rx/rx.c: Likewise.
21116 * config/s390/s390.c: Likewise.
21117 * config/sh/sh.c: Likewise.
21118 * config/sol2.c: Likewise.
21119 * config/sparc/sparc.c: Likewise.
21120 * config/spu/spu.c: Likewise.
21121 * config/stormy16/stormy16.c: Likewise.
21122 * config/tilegx/tilegx.c: Likewise.
21123 * config/tilepro/tilepro.c: Likewise.
21124 * config/v850/v850.c: Likewise.
21125 * config/vax/vax.c: Likewise.
21126 * config/visium/visium.c: Likewise.
21127 * config/xtensa/xtensa.c: Likewise.
21129 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21132 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
21133 constraints since the -mupper-regs-* switches have been
21135 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
21136 into a vector from a double word element that was extracted from
21137 another vector, and eliminate extra XXPERMDI instructions.
21138 (vsx_concat_<mode>_2): Likewise.
21139 (vsx_concat_<mode>_3): Likewise.
21140 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
21141 concat to allow optimizing inserts from previous extracts.
21143 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
21145 * config/i386/i386.c (ix86_stack_protect_guard): Generate
21146 memory reference to a SSP offset in TLS address space.
21147 (ix86_print_operand) <case '@'>: Remove.
21148 (ix86_print_operand_punct_valid_p): Remove '@' code.
21149 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
21150 UNSPEC_SP_TLS_TEST.
21151 (stack_tls_protect_set_<mode>): Remove.
21152 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
21153 (stack_tls_protect_test_<mode>): Remove.
21154 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
21156 2017-08-07 Olivier Hainque <hainque@adacore.com>
21159 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
21161 2017-08-07 Douglas Rupp <rupp@adacore.com>
21163 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
21164 variable was referenced as multidir in command.
21166 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21169 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
21172 2017-08-07 Martin Liska <mliska@suse.cz>
21174 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
21175 * config/rl78/rl78.c: Add include of attribs.h.
21176 * config/sh/sh.c: Likewise.
21177 * config/v850/v850.c: Likewise.
21179 2017-08-07 Tom de Vries <tom@codesourcery.com>
21181 PR middle-end/78266
21182 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
21184 2017-08-07 Martin Liska <mliska@suse.cz>
21186 * config/mips/mips.c: Include attribs.h.
21188 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
21191 * doc/invoke.texi: Document change in behvaior for -Ofast for
21194 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
21196 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
21198 (aarch64_pop_regs): Likewise.
21199 (aarch64_gen_load_pair): Likewise.
21200 (aarch64_save_callee_saves): Likewise.
21201 (aarch64_restore_callee_saves): Likewise.
21203 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21205 * config/i386/i386.c: Revert the last change.
21207 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21210 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21212 (ix86_finalize_stack_frame_flags): This. Also clear
21213 frame_pointer_needed if -fno-omit-frame-pointer is used without
21215 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21216 with ix86_finalize_stack_frame_flags.
21217 (ix86_expand_epilogue): Likewise.
21218 (ix86_expand_split_stack_prologue): Likewise.
21220 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21223 * config/i386/i386.c (get_builtin_code_for_version): Set priority
21224 to P_AES for Westmere.
21226 2017-08-07 Jonathan Yong <10walls@gmail.com>
21228 * config/i386/mingw.opt (fset-stack-executable): Removed.
21229 * config/i386/cygming.opt (fset-stack-executable): Moved
21231 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
21233 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
21235 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
21237 2017-08-07 Marek Polacek <polacek@redhat.com>
21239 PR middle-end/81737
21240 * fold-const.c (fold_indirect_ref_1): Check type_domain.
21242 2017-08-07 Martin Liska <mliska@suse.cz>
21244 * attribs.h (canonicalize_attr_name): New function.
21245 (cmp_attribs): Move from c-format.c and adjusted.
21246 (is_attribute_p): Moved from tree.h.
21247 * tree-inline.c: Add new includes.
21248 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
21249 (private_is_attribute_p): Remove.
21250 (private_lookup_attribute): Likewise.
21251 (private_lookup_attribute_by_prefix): Simplify.
21252 (remove_attribute): Use is_attribute_p.
21253 * tree.h: Remove removed declarations.
21255 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21257 PR middle-end/81698
21258 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
21259 instead of computing it in the function. Formatting fix.
21260 (expand_case): Don't rely on default_edge being the first edge,
21261 clear it if removing it, pass default_edge to
21262 emit_case_dispatch_table.
21263 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
21266 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
21268 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
21269 insn in the function, emit NOP after the insn.
21271 2017-08-06 Tom de Vries <tom@codesourcery.com>
21273 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
21276 2017-08-06 Tom de Vries <tom@codesourcery.com>
21278 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
21281 2017-08-04 Yury Gribov <tetra2005@gmail.com>
21283 PR tree-optimization/57371
21284 * match.pd: New pattern.
21286 2017-08-04 Marek Polacek <polacek@redhat.com>
21288 PR middle-end/81695
21289 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
21290 perform the computation in offset_int.
21292 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21294 PR tree-optimization/81136
21295 * tree-vectorizer.h: Include tree-hash-traits.h.
21296 (vec_base_alignments): New typedef.
21297 (vec_info): Add a base_alignments field.
21298 (vect_record_base_alignments): Declare.
21299 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
21301 (DR_IS_CONDITIONAL_IN_STMT): New macro.
21302 (create_data_ref): Add an is_conditional_in_stmt argument.
21303 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
21304 the is_conditional_in_stmt field.
21305 (data_ref_loc): Add an is_conditional_in_stmt field.
21306 (get_references_in_stmt): Set the is_conditional_in_stmt field.
21307 (find_data_references_in_stmt): Update call to create_data_ref.
21308 (graphite_find_data_references_in_stmt): Likewise.
21309 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
21310 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
21311 (vect_record_base_alignment): New function.
21312 (vect_record_base_alignments): Likewise.
21313 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
21314 for nested statements even if we fail to compute a misalignment.
21315 Use pooled base alignments for unconditional references.
21316 (vect_find_same_alignment_drs): Compare base addresses instead
21318 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
21319 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
21321 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21323 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
21324 Add an explicit name for the enum. Use auto_vec for slp_instances
21325 and grouped_stores.
21326 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
21328 (_bb_vec_info): Add a constructor and destructor.
21329 (vinfo_for_stmt): Return NULL for uids of -1 as well.
21330 (destroy_loop_vec_info): Delete.
21331 (vect_destroy_datarefs): Likewise.
21332 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
21333 (vec_info::vec_info): New function.
21334 (vec_info::~vec_info): Likewise.
21335 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
21336 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
21337 destroy_loop_vec_info.
21338 * tree-vect-loop.c (new_loop_vec_info): Replace with...
21339 (_loop_vec_info::_loop_vec_info): ...this.
21340 (destroy_loop_vec_info): Replace with...
21341 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
21342 the stmt_vec_infos. Leave handling of vec_info information to its
21343 destructor. Remove explicit vector releases.
21344 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
21345 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
21346 * tree-vect-slp.c (new_bb_vec_info): Replace with...
21347 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
21348 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
21349 (destroy_bb_vec_info): Replace with...
21350 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
21351 information to its destructor.
21352 (vect_slp_analyze_bb_1): Use new and delete instead of
21353 new_bb_vec_info and destroy_bb_vec_info.
21354 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
21357 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21359 * tree-data-ref.h (subscript): Add access_fn field.
21360 (data_dependence_relation): Add could_be_independent_p.
21361 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
21362 (same_access_functions): Move to tree-data-ref.c.
21363 * tree-data-ref.c (ref_contains_union_access_p): New function.
21364 (access_fn_component_p): Likewise.
21365 (access_fn_components_comparable_p): Likewise.
21366 (dr_analyze_indices): Add a reference to access_fn_component_p.
21367 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
21369 (constant_access_functions): Likewise.
21370 (add_other_self_distances): Likewise.
21371 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
21372 (initialize_data_dependence_relation): Use XCNEW and remove
21373 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
21374 of access functions that have the same type. Allow the
21375 subsequence to end with different bases in some circumstances.
21376 Record the chosen access functions in SUB_ACCESS_FN.
21377 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
21378 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
21379 (subscript_dependence_tester_1): Likewise dra and drb.
21380 (build_classic_dist_vector): Update calls accordingly.
21381 (subscript_dependence_tester): Likewise.
21382 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
21383 DDR_COULD_BE_INDEPENDENT_P.
21384 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
21385 comp_alias_ddrs instead of may_alias_ddrs.
21386 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
21388 (vect_analyze_data_ref_dependence): Use it if
21389 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
21390 distance vectors if that fails.
21391 (dependence_distance_ge_vf): New function.
21392 (vect_prune_runtime_alias_test_list): Use it. Don't clear
21393 LOOP_VINFO_MAY_ALIAS_DDRS.
21395 2017-08-04 Richard Biener <rguenther@suse.de>
21397 PR middle-end/81705
21398 * fold-const.c (fold_binary_loc): Properly restrict
21399 minus_var0 && minus_var1 case when associating undefined overflow
21402 2017-08-04 Tom de Vries <tom@codesourcery.com>
21404 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
21406 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21408 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21409 Don't start diagnostic messages with a capital letter.
21410 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21412 (rs6000_invalid_builtin): Likewise.
21413 (rs6000_trampoline_init): Likewise.
21415 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21418 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
21419 after setting changeable df flags.
21421 2017-08-03 Richard Biener <rguenther@suse.de>
21423 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
21424 up if the use is in USE - X.
21426 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
21428 * toplev.c (dumpfile.h): New include.
21429 (internal_error_reentered): New static function. Use it...
21430 (internal_error_function): ...here to handle reentered internal_error.
21432 2017-08-03 Richard Biener <rguenther@suse.de>
21434 PR middle-end/81148
21435 * fold-const.c (split_tree): Add minus_var and minus_con
21436 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
21437 here but always use minus_*.
21438 (associate_trees): Assert we never associate with MINUS_EXPR
21439 and NULL first operand. Do not recurse for PLUS_EXPR operands
21440 when associating as MINUS_EXPR either.
21441 (fold_binary_loc): Track minus_var and minus_con.
21443 2017-08-03 Tom de Vries <tom@codesourcery.com>
21446 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
21447 ACCEL_COMPILER, apply finish_options on
21448 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
21450 2017-08-03 Tom de Vries <tom@codesourcery.com>
21453 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
21454 function_entry_patch_area_size > 0.
21456 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21459 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
21460 instead of 10??LU, perform unit multiplication in wide_int,
21461 don't change alloc_object_size_limit if the limit is larger
21464 PR tree-optimization/81655
21465 PR tree-optimization/81588
21466 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
21467 the case when ranges[i].low and high are 1 for unsigned type with
21470 PR middle-end/81052
21471 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
21472 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
21474 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21476 * tree-vrp.h: Add include guard.
21478 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
21481 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
21482 (ud2): New insn pattern.
21483 * config/i386/i386.c (ix86_expand_epilogue):
21484 For naked functions, generate ud2 instead of trap insn.
21486 2017-08-02 Marek Polacek <polacek@redhat.com>
21489 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
21491 2017-08-02 Tom de Vries <tom@codesourcery.com>
21492 Cesar Philippidis <cesar@codesourcery.com>
21494 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
21495 Add missing edge probabilities.
21497 2017-08-02 Tamar Christina <tamar.christina@arm.com>
21499 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
21500 Correct endianness.
21502 2017-08-02 Jakub Jelinek <jakub@redhat.com>
21504 PR middle-end/79499
21505 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
21506 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
21507 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
21508 prologue_seq sequences - if any.
21510 2017-08-02 Richard Biener <rguenther@suse.de>
21512 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
21513 via vectors if supported, integer extracts via punning if supported
21514 or otherwise vector extracts.
21516 2017-08-02 Richard Biener <rguenther@suse.de>
21518 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
21520 (bitmap_insert_into_set): ... this.
21522 2017-08-02 Richard Biener <rguenther@suse.de>
21524 PR tree-optimization/81633
21526 2015-08-17 Alan Hayward <alan.hayward@arm.com>
21528 PR tree-optimization/71752
21529 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
21531 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
21533 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
21534 (machine_function::call_ms2sysv_pad_out): Remove field.
21535 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
21536 (ix86_compute_frame_layout): Likewise.
21538 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
21541 * config/i386/i386.c (ix86_set_func_type): Disallow naked
21542 attribute with interrupt attribute.
21544 2017-08-01 Andrew Pinski <apinski@cavium.com>
21546 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
21547 BIT_INSERT_EXPR's operand 1
21548 to see if the types precision matches.
21550 2017-08-01 Martin Liska <mliska@suse.cz>
21552 PR middle-end/70140
21553 * builtins.c (expand_builtin_memcpy_args): Remove.
21554 (expand_builtin_memcpy): Call newly added function
21555 expand_builtin_memory_copy_args.
21556 (expand_builtin_memcpy_with_bounds): Likewise.
21557 (expand_builtin_mempcpy): Remove last argument.
21558 (expand_builtin_mempcpy_with_bounds): Likewise.
21559 (expand_builtin_memory_copy_args): New function created from
21560 expand_builtin_mempcpy_args with small modifications.
21561 (expand_builtin_mempcpy_args): Remove.
21562 (expand_builtin_stpcpy): Remove unused argument.
21563 (expand_builtin): Likewise.
21564 (expand_builtin_with_bounds): Likewise.
21566 2017-08-01 Martin Liska <mliska@suse.cz>
21569 Make mempcpy more optimal (PR middle-end/70140).
21571 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21574 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
21575 __builtin_vec_cmpne verify both arguments are compatible vectors
21576 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
21577 verify arg1_type is a pointer or array type. For __builtin_vec_st,
21578 move computation of aligned to after checking the argument types.
21582 * config/rs6000/vsx.md (vextract_fp_from_shorth,
21583 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
21586 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
21587 Doug Rupp <rupp@adacore.com>
21588 Olivier Hainque <hainque@adacore.com>
21590 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
21591 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
21593 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
21594 for TARGET_OS_CPP_BUILTIN.
21595 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
21596 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
21598 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
21599 passing required abi options to the assembler for EABI configurations.
21600 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
21601 of .text.hot and .text.unlikely sections for kernel modules when
21602 using ARM style exceptions.
21603 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
21604 options. Add EXTRA_CC1_SPEC.
21605 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
21607 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
21609 (ARM_TARGET2_DWARF_FORMAT): Define.
21610 * config/arm/t-vxworks: Adjust multilib control to removal of the
21611 Diab command line options.
21613 2017-08-01 Martin Liska <mliska@suse.cz>
21615 PR gcov-profile/81561
21616 * gcov.c (unblock): Make unblocking safe as we need to preserve
21617 index correspondence of blocks and block_lists.
21619 2017-08-01 Richard Biener <rguenther@suse.de>
21621 PR tree-optimization/81181
21622 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21623 (compute_antic): ... end of iteration here.
21625 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21627 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21628 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21629 (ftree-slp-vectorize): Likewise.
21630 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21631 can no longer be set independent of flag_tree_loop_vectorize.
21632 * omp-general.c (emp_max_vf): Likewise.
21633 * opts.c (enable_fdo_optimizations): Remove references to
21634 flag_tree_vectorize, these are now implicit.
21635 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21636 and leave it for the options machinery.
21638 2017-08-01 Martin Liska <mliska@suse.cz>
21640 PR middle-end/70140
21641 * builtins.c (expand_builtin_memcpy_args): Remove.
21642 (expand_builtin_memcpy): Call newly added function
21643 expand_builtin_memory_copy_args.
21644 (expand_builtin_memcpy_with_bounds): Likewise.
21645 (expand_builtin_mempcpy): Remove last argument.
21646 (expand_builtin_mempcpy_with_bounds): Likewise.
21647 (expand_builtin_memory_copy_args): New function created from
21648 expand_builtin_mempcpy_args with small modifications.
21649 (expand_builtin_mempcpy_args): Remove.
21650 (expand_builtin_stpcpy): Remove unused argument.
21651 (expand_builtin): Likewise.
21652 (expand_builtin_with_bounds): Likewise.
21654 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21657 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21658 print "ds:" only for immediates in generic address space.
21660 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21663 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21664 (ix86_function_ok_for_sibcall): Return false for naked functions.
21666 2017-08-01 Richard Biener <rguenther@suse.de>
21668 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21669 (compute_antic): Seed worklist with exit block predecessors.
21670 * cfganal.c (dfs_find_deadend): For a cycle return the source
21671 of the edge closing it.
21673 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21675 * config/aarch64/aarch64.c
21676 (aarch64_can_const_movi_rtx_p): Move 0 check.
21678 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21680 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21681 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21683 * match.pd: Ditto in address comparison pattern.
21685 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21687 PR tree-optimization/81627
21688 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21689 closed ssa form for store-store chain.
21691 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21693 PR tree-optimization/81620
21694 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21695 for store-store chain.
21697 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21699 PR tree-optimization/81588
21700 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21701 ranges[i].in_p, invert comparison code ccode. For >/>=,
21702 swap rhs1 and rhs2 and comparison code unconditionally,
21703 for </<= don't do that. Don't swap rhs1/rhs2 again if
21704 ranges[i].in_p, instead invert comparison code ccode if
21705 opcode or oe->rank is BIT_IOR_EXPR.
21708 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21709 a direct optab to conversion optab.
21710 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21711 with GET_MODE_INNER as last argument instead of optab_handler.
21712 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21713 vector extraction if possible and optab is available.
21714 * expr.c (store_constructor): Use convert_optab_handler instead
21715 of optab_handler. Use vector initialization from smaller
21716 vectors if possible and optab is available.
21717 * tree-vect-stmts.c (vectorizable_load): Likewise.
21718 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21719 now have two modes.
21720 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21721 of vec_init from half-sized vectors with the same element mode.
21722 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21723 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21724 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21725 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21726 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21727 after mode in gen_vec_extract* calls.
21728 (vec_extract<mode>): Renamed to ...
21729 (vec_extract<mode><ssescalarmodelower>): ... this.
21730 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21731 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21732 element mode after mode in gen_vec_init* calls.
21733 (VEC_INIT_HALF_MODE): New mode iterator.
21734 (vec_init<mode>): Renamed to ...
21735 (vec_init<mode><ssescalarmodelower>): ... this.
21736 (vec_init<mode><ssehalfvecmodelower>): New expander.
21737 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21738 (vec_extractv2sfsf): ... this.
21739 (vec_initv2sf): Renamed to ...
21740 (vec_initv2sfsf): ... this.
21741 (vec_extractv2si): Renamed to ...
21742 (vec_extractv2sisi): ... this.
21743 (vec_initv2si): Renamed to ...
21744 (vec_initv2sisi): ... this.
21745 (vec_extractv4hi): Renamed to ...
21746 (vec_extractv4hihi): ... this.
21747 (vec_initv4hi): Renamed to ...
21748 (vec_initv4hihi): ... this.
21749 (vec_extractv8qi): Renamed to ...
21750 (vec_extractv8qiqi): ... this.
21751 (vec_initv8qi): Renamed to ...
21752 (vec_initv8qiqi): ... this.
21753 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21754 (vec_init<mode>): Renamed to ...
21755 (vec_init<mode><VEC_base_l>): ... this.
21756 (vec_extract<mode>): Renamed to ...
21757 (vec_extract<mode><VEC_base_l>): ... this.
21758 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21759 (vec_initv2sfsf): ... this.
21760 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21761 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21762 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21763 element mode after mode in gen_vec_init* calls.
21764 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21765 (vec_init<mode><Vel>): ... this.
21766 (vec_extract<mode>): Renamed to ...
21767 (vec_extract<mode><Vel>): ... this.
21768 * config/aarch64/iterators.md (Vel): New mode attribute.
21769 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21770 Add element mode after mode in gen_vec_extract* calls.
21771 * config/s390/vector.md (non_vec_l): New mode attribute.
21772 (vec_extract<mode>): Renamed to ...
21773 (vec_extract<mode><non_vec_l>): ... this.
21774 (vec_init<mode>): Renamed to ...
21775 (vec_init<mode><non_vec_l>): ... this.
21776 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21777 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21779 * config/arm/iterators.md (V_elem_l): New mode attribute.
21780 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21781 (vec_extract<mode><V_elem_l>): ... this.
21782 (vec_extractv2di): Renamed to ...
21783 (vec_extractv2didi): ... this.
21784 (vec_init<mode>): Renamed to ...
21785 (vec_init<mode><V_elem_l>): ... this.
21786 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21787 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21788 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21789 Add element mode after gen_vec_extract* calls.
21790 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21791 (vec_init<mode><unitmode>): ... this.
21792 (vec_extract<mode>): Renamed to ...
21793 (vec_extract<mode><unitmode>): ... this.
21794 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21795 (vec_init<mode><unitmode>): ... this.
21796 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21797 (vec_initv2sfsf): ... this.
21798 (vec_extractv2sf): Renamed to ...
21799 (vec_extractv2sfsf): ... this.
21800 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21801 Add element mode after gen_vec_extract* calls.
21802 * config/mips/mips.md (unitmode): New mode iterator.
21803 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21804 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21805 * config/spu/spu.md (inner_l): New mode attribute.
21806 (vec_init<mode>): Renamed to ...
21807 (vec_init<mode><inner_l>): ... this.
21808 (vec_extract<mode>): Renamed to ...
21809 (vec_extract<mode><inner_l>): ... this.
21810 * config/sparc/sparc.md (veltmode): New mode iterator.
21811 (vec_init<VMALL:mode>): Renamed to ...
21812 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21813 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21814 (vec_initv2sisi): ... this.
21815 (vec_initv2sf): Renamed to ...
21816 (vec_initv2sfsf): ... this.
21817 (vec_extractv2sf): Renamed to ...
21818 (vec_extractv2sfsf): ... this.
21819 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21820 (vec_init<mode>): Renamed to ...
21821 (vec_init<mode><VEC_base_l>): ... this.
21822 (vec_extract<mode>): Renamed to ...
21823 (vec_extract<mode><VEC_base_l>): ... this.
21824 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21825 (vec_initv2sfsf): ... this.
21826 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21827 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21828 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21829 gen_vec_init* calls.
21831 2017-08-01 Richard Biener <rguenther@suse.de>
21833 PR tree-optimization/81297
21834 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21835 TREE_OVERFLOW from INTEGER_CSTs.
21837 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21839 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21841 2017-07-31 Carl Love <cel@us.ibm.com>
21843 * config/rs6000/rs6000-c: Add support for built-in functions
21844 vector signed char vec_xl_be (signed long long, signed char *);
21845 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21846 vector signed int vec_xl_be (signed long long, signed int *);
21847 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21848 vector signed long long vec_xl_be (signed long long, signed long long *);
21849 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21850 vector signed short vec_xl_be (signed long long, signed short *);
21851 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21852 vector double vec_xl_be (signed long long, double *);
21853 vector float vec_xl_be (signed long long, float *);
21854 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21855 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21856 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21858 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21859 (altivec_expand_builtin): Add switch statement to call
21860 altivec_expand_xl_be for each builtin.
21861 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21862 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21863 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21864 __builtin_vsx_le_be_v16qi.
21865 * doc/extend.texi: Update the built-in documentation file for the
21866 new built-in functions.
21868 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21871 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
21873 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21875 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21877 * config.gcc: Add z14.
21878 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
21879 CPU model numbers for z13s and z14.
21880 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
21882 * config/s390/s390-opts.h (enum processor_type): Rename
21883 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21884 * config/s390/s390.c (processor_table): Add field for CPU name to
21885 be passed to Binutils.
21886 (s390_asm_output_machine_for_arch): Use the new field in
21887 processor_table for Binutils.
21888 (s390_expand_builtin): Replace arch12 with z14.
21889 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21890 (s390_get_sched_attrmask): Likewise.
21891 (s390_get_unit_mask): Likewise.
21892 * config/s390/s390.opt: Add z14 to processor_type enum.
21894 2017-07-31 Martin Jambor <mjambor@suse.cz>
21897 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
21898 regardless of optimization level.
21900 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
21901 Martin Liska <mliska@suse.cz>
21903 * predict.def: Remove old comment and adjust probability.
21904 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
21905 PREDICT statements.
21907 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21910 * config/i386/i386.c (ix86_function_naked): New function.
21911 (ix86_can_use_return_insn_p): Return false for naked functions.
21912 (ix86_expand_prologue): Skip prologue for naked functions.
21913 (ix86_expand_epilogue): Skip epilogue for naked functions
21914 and emit trap instruction.
21915 (ix86_warn_func_return): New function.
21916 (ix86_attribute_table): Add "naked" attribute specification.
21917 (TARGET_WARN_FUNC_RETURN): Define.
21918 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
21920 2017-07-31 Martin Liska <mliska@suse.cz>
21922 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
21923 (dump_gimple_bb_header): Always dump BB info.
21924 (pp_cfg_jump): Do not append info about BB when dumping a jump.
21926 2017-07-31 Martin Liska <mliska@suse.cz>
21929 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
21930 also with current_function_decl non-null equality.
21932 2017-07-31 Jakub Jelinek <jakub@redhat.com>
21935 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
21936 change type to the element type, instead add eltype variable and
21937 use it where we are interested in the element type.
21939 PR tree-optimization/81603
21940 * ipa-polymorphic-call.c
21941 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
21942 offset arithmetic in offset_int, bail out if the resulting bit offset
21943 doesn't fit into shwi.
21945 2017-07-31 Martin Liska <mliska@suse.cz>
21947 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
21948 (gimplify_save_expr): Fix comment.
21950 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
21953 * config/i386/i386.c (ix86_function_arg): Update arguments for
21955 (ix86_compute_frame_layout): Set the initial stack offset to
21956 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
21957 INCOMING_FRAME_SP_OFFSET.
21958 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
21959 stack before exception handler returns.
21960 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
21961 the 'ERROR_CODE' for exception handler.
21963 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
21965 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
21966 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
21967 (ASM_OUTPUT_REG_POP): Ditto.
21968 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
21969 instead of asm_fprintf to output pure string.
21971 2017-07-29 Jakub Jelinek <jakub@redhat.com>
21973 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
21974 to imported_module_or_decl hook.
21975 (debug_nothing_tree_tree_tree_bool): Remove.
21976 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
21977 * debug.c (do_nothing_debug_hooks): Use
21978 debug_nothing_tree_tree_tree_bool_bool instead of
21979 debug_nothing_tree_tree_tree_bool.
21980 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21981 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
21982 * sdbout.c (sdb_debug_hooks): Likewise.
21983 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
21984 (gen_namespace_die): Add DW_AT_export_symbols attribute if
21986 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
21987 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
21988 attribute, don't add anything.
21990 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21992 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
21993 (fold_build2_stat_loc): Likewise.
21994 (fold_build3_stat_loc): Likewise.
21995 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
21996 (fold_build1_loc): Remove macro.
21997 (fold_build2_loc): Likewise.
21998 (fold_build3_loc): Likewise.
22000 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22002 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
22003 (gimple_build_debug_bind_source_stat): Likewise.
22004 * gimple.h (gimple_build_debug_bind): Remove macro.
22005 (gimple_build_debug_bind_source): Likewise.
22007 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22009 * bitmap.c (bitmap_alloc): Adjust.
22010 (bitmap_gc_alloc): Likewise.
22011 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22013 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22015 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22016 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22017 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22018 (bitmap_gc_alloc_stat): Likewise.
22019 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22021 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22023 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22024 * rtl.h (shallow_copy_rtx): Remove macro.
22026 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22028 * emit-rtl.c (gen_raw_REG): Adjust.
22029 * gengenrtl.c (gendef): Likewise.
22030 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22031 * rtl.h (rtx_alloc): Remove macro.
22033 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22035 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22036 (build_tree_list_stat): Likewise.
22037 * tree.h (build_tree_list): Remove macro.
22038 (build_tree_list_vec): Likewise.
22040 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22042 * tree.c (make_vector_stat): Remove _stat from name.
22043 (build_vector_stat): Likewise.
22044 * tree.h (make_vector_stat): Remove macro.
22045 (build_vector_stat): Likewise.
22047 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22049 * tree.h (build_var_debug_value): Remove prototype.
22051 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22053 * tree.c (tree_cons_stat): Remove _stat from name.
22054 * tree.h (tree_cons): Remove macro.
22056 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22058 * tree.c (build_vl_exp_stat): Remove _stat from name.
22059 * tree.h (build_vl_exp): Remove macro.
22061 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22063 * tree.c (build_decl_stat): Remove _stat from name.
22064 * tree.h (build_decl): Remove macro.
22066 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22068 * gimple.c (gimple_build_with_ops_stat): Adjust.
22069 (gimple_alloc_stat): Remove _stat from name.
22070 * gimple.h (gimple_alloc): Remove macro.
22072 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22074 * tree.c (make_tree_vec_stat): Remove _stat from name.
22075 (grow_tree_vec_stat): Likewise.
22076 * tree.h (make_tree_vec_stat): Adjust prototype.
22077 (grow_tree_vec_stat): Likewise.
22078 (make_tree_vec): Remove macro.
22079 (grow_tree_vec): Likewise.
22081 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22083 * fold-const.c (fold_build1_stat_loc): Adjust.
22084 (fold_build2_stat_loc): Likewise.
22085 (fold_build3_stat_loc): Likewise.
22086 * tree.c (build0_stat): Remove _stat from name.
22087 (build1_stat): Likewise.
22088 (build2_stat): Likewise.
22089 (build3_stat): Likewise.
22090 (build4_stat): Likewise.
22091 (build5_stat): Likewise.
22092 * tree.h (build1_loc): Remove macro, and rename _stat function
22094 (build2_loc): Likewise.
22095 (build3_loc): Likewise.
22096 (build4_loc): Likewise.
22097 (build5_loc): Likewise.
22099 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22101 * tree.c (make_int_cst_stat): Remove _stat from name.
22102 * tree.h (make_int_cst_stat): Adjust prototype.
22103 (make_int_cst): Remove macro.
22105 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22107 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22108 * tree.h (make_tree_binfo_stat): Adjust prototype.
22109 (make_tree_binfo): Remove.
22111 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22113 * tree.c (copy_node_stat): Rename to copy_node.
22114 (build_distinct_type_copy): Adjust.
22115 * tree.h (copy_node_stat): Adjust prototype.
22116 (copy_node): Remove macro.
22118 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22120 * tree.c (make_node_stat): rename to make_node.
22121 (build_tree_list_stat): Adjust.
22122 (build0_stat): Likewise.
22123 (build2_stat): Likewise.
22124 (build3_stat): Likewise.
22125 (build4_stat): Likewise.
22126 (build5_stat): Likewise.
22127 (build_decl_stat): Likewise.
22128 * tree.h (make_node_stat): Adjust prototype.
22129 (make_node): remove macro.
22131 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
22133 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
22134 (PPC_FEATURE2_SCV): Likewise.
22135 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
22137 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22139 * config/aarch64/aarch64.c
22140 (aarch64_internal_mov_immediate): Add new special pattern.
22141 * config/aarch64/aarch64.md (*movdi_aarch64):
22142 Add reg/32bit const mov case.
22144 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22145 Richard Sandiford <richard.sandiford@linaro.org>
22147 * config/aarch64/aarch64.md (mov<mode>): Generalize.
22148 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
22149 Add integer and movi cases.
22150 (movi-split-hf-df-sf split, fp16): New.
22151 (enabled): Added TARGET_FP_F16INST.
22152 * config/aarch64/iterators.md (GPF_HF): New.
22153 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
22155 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22157 * config/aarch64/aarch64.c
22158 (aarch64_simd_container_mode): Add prototype.
22159 (aarch64_expand_mov_immediate): Add HI support.
22160 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
22161 (aarch64_can_const_movi_rtx_p): New.
22162 (aarch64_preferred_reload_class):
22163 Remove restrictions of using FP registers for certain SIMD operations.
22164 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
22165 (aarch64_valid_floating_const): Add integer move validation.
22166 (aarch64_simd_imm_scalar_p): Remove.
22167 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
22168 (aarch64_legitimate_constant_p): Expand list of supported cases.
22169 * config/aarch64/aarch64-protos.h
22170 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
22171 (aarch64_reinterpret_float_as_int): New.
22172 (aarch64_simd_imm_scalar_p): Remove.
22173 * config/aarch64/constraints.md (Uvi): New.
22174 (Dd): Split into Ds and new Dd.
22175 * config/aarch64/aarch64.md (*movsi_aarch64):
22177 (*movdi_aarch64): Add SIMD mov case.
22179 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22181 * tree-predcom.c: (struct chain): Handle store-store chain in which
22182 stores for elimination only store loop invariant values.
22183 (execute_pred_commoning_chain): Ditto.
22184 (prepare_initializers_chain_store_elim): Ditto.
22185 (prepare_finalizers): Ditto.
22186 (is_inv_store_elimination_chain): New function.
22187 (initialize_root_vars_store_elim_1): New function.
22189 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22191 * tree-predcom.c: Revise general description of the pass.
22192 (enum chain_type): New enum type for store elimination.
22193 (struct chain): New field supporting store elimination.
22194 (struct component): Ditto.
22195 (dump_chain): Dump store-stores chain.
22196 (release_chain): Release resources.
22197 (split_data_refs_to_components): Compute and create component
22198 contains only stores for elimination.
22199 (get_chain_last_ref_at): New function.
22200 (make_invariant_chain): Initialization.
22201 (make_rooted_chain): Specify chain type in parameter and record it.
22202 (add_looparound_copies): Skip for store-stores chain.
22203 (determine_roots_comp): Compute type of chain and pass it to
22205 (initialize_root_vars_store_elim_2): New function.
22206 (finalize_eliminated_stores): New function.
22207 (remove_stmt): Handle store for elimination.
22208 (execute_pred_commoning_chain): Execute predictive commoning on
22209 store-store chains.
22210 (determine_unroll_factor): Skip unroll for store-stores chain.
22211 (prepare_initializers_chain_store_elim): New function.
22212 (prepare_initializers_chain): Hanlde store-store chain.
22213 (prepare_finalizers_chain, prepare_finalizers): New function.
22214 (tree_predictive_commoning_loop): Return integer value indicating
22215 if loop is unrolled or lcssa form is corrupted.
22216 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
22218 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22220 * tree-predcom.c (initialize_root): Delete.
22221 (execute_pred_commoning_chain): Initialize root vars and replace
22222 reference of non-combined chain directly, rather than call above
22225 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22227 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
22228 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
22230 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22232 * tree-predcom.c (struct chain): New field init_seq.
22233 (release_chain): Release init_seq.
22234 (prepare_initializers_chain): Record intialization stmts in above
22236 (insert_init_seqs): New function.
22237 (tree_predictive_commoning_loop): Call insert_init_seqs.
22239 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22241 * tree-predcom.c (determine_roots_comp): Skip trivial components.
22243 2017-07-28 Richard Biener <rguenther@suse.de>
22245 * match.pd: Remove superfluous :c.
22246 * genmatch.c (simplify::id): Add member.
22247 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
22249 (current_id): New global.
22250 (dt_node::parent): Move from ...
22251 (dt_operand::parent): ... here. Add for_id member.
22252 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
22253 (decision_tree::find_node): Relax order requirement when
22254 merging DT_TRUE nodes to ones inbetween the current simplify
22255 and the one we try to merge with. Add diagnostic whenever
22256 we need to enforce pattern order by not merging.
22257 (decision_tree::insert): Set current_id.
22258 (decision_tree::print_node): Dump parent node and for_id.
22259 (parser::last_id): Add member.
22260 (parser::push_simplify): Assign unique id.
22261 (parser::parser): Initialize last_id.
22263 2017-07-28 Martin Liska <mliska@suse.cz>
22266 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
22267 gimple_build_debug_bind.
22269 2017-07-28 Richard Biener <rguenther@suse.de>
22271 PR tree-optimization/81502
22272 * match.pd: Add pattern combining BIT_INSERT_EXPR with
22274 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
22276 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
22277 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
22278 for BIT_FIELD_REF args.
22279 * fold-const.c (make_bit_field_ref): Likewise.
22280 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
22282 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22285 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
22286 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
22287 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
22288 Or it into SANITIZER_UNDEFINED.
22289 * ubsan.c: Include gimple-fold.h and varasm.h.
22290 (ubsan_expand_ptr_ifn): New function.
22291 (instrument_pointer_overflow): New function.
22292 (maybe_instrument_pointer_overflow): New function.
22293 (instrument_object_size): Formatting fix.
22294 (pass_ubsan::execute): Call instrument_pointer_overflow
22295 and maybe_instrument_pointer_overflow.
22296 * internal-fn.c (expand_UBSAN_PTR): New function.
22297 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
22298 * sanitizer.def (__ubsan_handle_pointer_overflow,
22299 __ubsan_handle_pointer_overflow_abort): New builtins.
22300 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
22301 * internal-fn.def (UBSAN_PTR): New internal function.
22302 * opts.c (sanitizer_opts): Add pointer-overflow.
22303 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
22304 * fold-const.c (build_range_check): Compute pointer range check in
22305 integral type if pointer arithmetics would be needed. Formatting
22308 2017-07-28 Martin Liska <mliska@suse.cz>
22311 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
22312 parameters that are of a variable-length.
22314 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
22316 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
22318 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
22319 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
22320 (CRT_CALL_STATIC_FUNCTION): Likewise.
22321 (ASM_DEFAULT_SPEC): New define.
22322 (ASM_SPEC32): Likewise.
22323 (ASM_SPEC64): Likewise.
22324 (ASM_SPEC_COMMON): Likewise.
22325 (ASM_SPEC): Likewise.
22326 (INVALID_64BIT): Likewise.
22327 (LINK_OS_DEFAULT_SPEC): Likewise.
22328 (LINK_OS_SPEC32): Likewise.
22329 (LINK_OS_SPEC64): Likewise.
22330 (POWERPC_LINUX): Likewise.
22331 (PTRDIFF_TYPE): Likewise.
22332 (RESTORE_FP_PREFIX): Likewise.
22333 (RESTORE_FP_SUFFIX): Likewise.
22334 (SAVE_FP_PREFIX): Likewise.
22335 (SAVE_FP_SUFFIX): Likewise.
22336 (SIZE_TYPE): Likewise.
22337 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
22338 (TARGET_64BIT): Likewise.
22339 (TARGET_64BIT): Likewise.
22340 (TARGET_AIX): Likewise.
22341 (WCHAR_TYPE_SIZE): Likewise.
22342 (WCHAR_TYPE): Undefine.
22343 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
22344 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
22345 (CPP_OS_RTEMS_SPEC): Delete.
22346 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
22347 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
22349 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
22351 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22353 PR tree-optimization/81578
22354 * tree-parloops.c (build_new_reduction): Bail out if
22355 reduction_code isn't one of the standard OpenMP reductions.
22356 Move the details printing after that decision.
22358 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22360 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
22361 related to reload_in_progress.
22362 (splat_input_operand): Likewise.
22363 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
22365 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
22367 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
22368 (TARGET_INSTANTIATE_DECLS): Likewise.
22369 (legitimate_indexed_address_p): Delete reload_in_progress code.
22370 (rs6000_debug_legitimate_address_p): Likewise.
22371 (rs6000_eliminate_indexed_memrefs): Likewise.
22372 (rs6000_emit_le_vsx_store): Likewise.
22373 (rs6000_emit_move_si_sf_subreg): Likewise.
22374 (rs6000_emit_move): Likewise.
22375 (register_to_reg_type): Likewise.
22376 (rs6000_pre_atomic_barrier): Likewise.
22377 (rs6000_machopic_legitimize_pic_address): Likewise.
22378 (rs6000_allocate_stack_temp): Likewise.
22379 (rs6000_address_for_fpconvert): Likewise.
22380 (rs6000_address_for_altivec): Likewise.
22381 (rs6000_secondary_memory_needed_rtx): Delete function.
22382 (rs6000_check_sdmode): Likewise.
22383 (rs6000_alloc_sdmode_stack_slot): Likewise.
22384 (rs6000_instantiate_decls): Likewise.
22385 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
22386 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
22387 Delete reload_in_progress.
22388 (*vec_reload_and_plus_<mptrsize>): Likewise.
22389 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
22390 (vsx_div_v2di): Likewise.
22391 (vsx_udiv_v2di): Likewise.
22393 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22395 * config/rs6000/rs6000.opt (mlra): Replace with stub.
22396 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
22397 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
22398 (rs6000_debug_reg_global): Delete print of LRA status.
22399 (rs6000_option_override_internal): Delete dead LRA related code.
22400 (rs6000_lra_p): Delete function.
22401 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
22403 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
22405 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
22406 * config/riscv/rtems.h: New file.
22408 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22409 Sudakshina Das <sudi.das@arm.com>
22411 * config/aarch64/aarch64.md
22412 (define_split for and<mode>3nr_compare): Move
22413 non aarch64_logical_operand to a register.
22414 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
22415 register immediate operand to a register.
22416 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
22418 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22420 PR middle-end/81564
22421 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
22423 2017-07-27 Richard Biener <rguenther@suse.de>
22425 PR tree-optimization/81573
22426 PR tree-optimization/81494
22427 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
22428 multi defuse cycle case.
22430 2017-07-27 Richard Biener <rguenther@suse.de>
22432 PR tree-optimization/81571
22433 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
22436 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
22438 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
22439 earlier and only if MASK_FPU is set. Adjust formatting.
22441 2017-07-27 Martin Liska <mliska@suse.cz>
22443 * opt-functions.awk: Add validation of value of Init.
22444 * optc-gen.awk: Pass new argument.
22446 2017-07-27 Martin Liska <mliska@suse.cz>
22448 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
22449 Fix wrong condition.
22451 2017-07-27 Martin Liska <mliska@suse.cz>
22453 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
22454 BBs and edges seen by autoFDO.
22456 2017-07-27 Richard Biener <rguenther@suse.de>
22458 PR tree-optimization/81502
22459 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
22460 with incompatible but same sized type.
22461 (execute_update_addresses_taken): Likewise.
22463 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
22465 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
22466 flag_tree_loop_vectorize rather than flag_tree_vectorize.
22468 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22471 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
22472 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
22473 Change s_operand to memory_operand.
22475 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
22477 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
22478 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
22479 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
22480 Emit instructions rather than returning an expression. Handle TFmode
22481 and KFmode by casting to TImode.
22482 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
22483 (rs6000_emit_le_vsx_store): Likewise.
22484 * config/rs6000/vsx.md (VSX_TI): New iterator.
22485 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
22486 (*vsx_le_undo_permute_<mode>): Likewise.
22487 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
22488 emit the split sequence.
22489 (*vsx_le_perm_store_<mode>): Likewise.
22491 2017-07-27 Jakub Jelinek <jakub@redhat.com>
22493 PR tree-optimization/81555
22494 PR tree-optimization/81556
22495 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
22496 if true, force CHANGED for the recursive invocation.
22497 (reassociate_bb): Remember original length of ops array, pass
22498 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
22500 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
22501 attributes for noipa attribute. For naked attribute use
22502 lookup_attribute first before lookup_attribute_spec.
22503 * final.c (rest_of_handle_final): Disable IPA RA for functions with
22505 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
22506 for functions with noipa attribute.
22507 (cgraph_externally_visible_p): Return true for functions with noipa
22509 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
22510 for functions with noipa attribute.
22511 * doc/extend.texi: Document noipa function attribute.
22512 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
22513 also for functions with noipa attribute.
22514 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
22516 2017-07-26 Andrew Pinski <apinski@cavium.com>
22518 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
22519 vec_unalign_load_cost and vec_unalign_store_cost.
22521 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
22523 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
22524 -mvsx-small-integer option.
22525 (ISA_3_0_MASKS_IEEE): Likewise.
22526 (OTHER_VSX_VECTOR_MASKS): Likewise.
22527 (POWERPC_MASKS): Likewise.
22528 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
22529 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
22530 code, only testing for DImode being allowed in non-VSX floating
22532 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
22533 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
22535 (rs6000_option_override_internal): Delete -mvsx-small-integer.
22536 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
22537 TARGET_P8_VECTOR test.
22538 (rs6000_secondary_reload_simple_move): Likewise.
22539 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
22540 since TARGET_P9_VECTOR was already tested.
22541 (rs6000_opt_masks): Remove -mvsx-small-integer.
22542 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
22543 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22545 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
22546 test for TARGET_VEXTRACTUB was used, and that uses
22548 (p9 extract splitter): Likewise.
22549 (vsx_extract_<mode>_di_p9): Likewise.
22550 (vsx_extract_<mode>_store_p9): Likewise.
22551 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
22552 for TARGET_P9_VECTOR was used. Delete code that is now dead with
22553 the elimination of TARGET_VSX_SMALL_INTEGER.
22554 (vsx_extract_<mode>_p8): Likewise.
22555 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
22556 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
22557 (vsx_set_<mode>_p9): Likewise.
22558 (vsx_set_v4sf_p9): Likewise.
22559 (vsx_set_v4sf_p9_zero): Likewise.
22560 (vsx_insert_extract_v4sf_p9): Likewise.
22561 (vsx_insert_extract_v4sf_p9_2): Likewise.
22562 * config/rs6000/rs6000.md (sign extend splitter): Change
22563 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
22564 (floatsi<mode>2_lfiwax_mem): Likewise.
22565 (floatunssi<mode>2_lfiwzx_mem): Likewise.
22566 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
22567 since a test for TARGET_P9_VECTOR was used.
22568 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22569 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
22570 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22571 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22572 TARGET_P8_VECTOR test.
22573 (fix_trunc<mode>si2_stfiwx): Likewise.
22574 (fix_trunc<mode>si2_internal): Likewise.
22575 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
22576 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22578 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22579 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22580 TARGET_P8_VECTOR test.
22581 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22582 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
22583 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22585 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22586 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
22587 since a test for TARGET_P9_VECTOR was used.
22588 (splitter for loading small constants): Likewise.
22590 2017-07-26 Andrew Pinski <apinski@cavium.com>
22592 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
22595 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
22598 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
22599 (fp_valid_at): Likewise.
22601 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22603 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
22604 (qdf24xx_addrcost_table): Likewise.
22605 (cortexa57_tunings): Update to use generic_branch_cost.
22606 (cortexa72_tunings): Likewise.
22607 (cortexa73_tunings): Likewise.
22608 (qdf24xx_tunings): Likewise.
22610 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22612 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
22613 (thunderx2t99_branch_cost): Likewise.
22614 (cortexa35_tunings): Update to use generic_branch_cost.
22615 (cortexa53_tunings): Likewise.
22616 (cortexa57_tunings): Likewise.
22617 (cortexa72_tunings): Likewise.
22618 (cortexa73_tunings): Likewise.
22619 (thunderx2t99_tunings): Likewise.
22621 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22623 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22624 (sparc_option_override): Honour MASK_FSMULD.
22625 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22626 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22627 * config/sparc/sparc.opt (mfsmuld): New option.
22628 * doc/invoke.texi (mfsmuld): Document option.
22630 2017-07-26 Marek Polacek <polacek@redhat.com>
22632 PR middle-end/70992
22633 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22635 2017-07-26 Richard Biener <rguenther@suse.de>
22637 * gimple-match-head.c (do_valueize): Return OP if valueize
22639 (get_def): New helper to get at the def stmt of a SSA name
22640 if valueize allows.
22641 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22642 do_valueize to get at the def stmt.
22643 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22645 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22647 PR middle-end/46932
22648 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22650 2017-07-26 Martin Liska <mliska@suse.cz>
22653 * function.c (expand_function_start): Make expansion of
22654 nonlocal_goto_save_area after parm_birth_insn.
22656 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22658 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22659 from all CPU target flags enable members.
22661 2017-07-26 Richard Biener <rguenther@suse.de>
22663 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22664 (decision_tree::gen): Make 'type' const.
22665 (write_predicate): Likewise.
22667 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22669 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22670 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22671 (rs6000_option_override_internal): Likewise.
22672 (rs6000_expand_vector_set): Likewise.
22673 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22674 (TARGET_UPPER_REGS_SF): Likewise.
22675 (TARGET_UPPER_REGS_DI): Likewise.
22676 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22677 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22678 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22679 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22680 (Splitters for DI constants in Altivec registers): Likewise.
22681 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22682 (vsx_set_v4sf_p9): Likewise.
22683 (vsx_set_v4sf_p9_zero): Likewise.
22684 (vsx_insert_extract_v4sf_p9): Likewise.
22685 (vsx_insert_extract_v4sf_p9_2): Likewise.
22687 2017-07-25 Carl Love <cel@us.ibm.com>
22689 * doc/extend.texi: Update the built-in documentation file for the
22690 existing built-in functions
22691 vector signed char vec_cnttz (vector signed char);
22692 vector unsigned char vec_cnttz (vector unsigned char);
22693 vector signed short vec_cnttz (vector signed short);
22694 vector unsigned short vec_cnttz (vector unsigned short);
22695 vector signed int vec_cnttz (vector signed int);
22696 vector unsigned int vec_cnttz (vector unsigned int);
22697 vector signed long long vec_cnttz (vector signed long long);
22698 vector unsigned long long vec_cnttz (vector unsigned long long);
22700 2017-07-25 Andrew Pinski <apinski@cavium.com>
22702 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22703 accesses where the use is for the first operand of a BIT_INSERT.
22705 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22708 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22709 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22711 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22713 * config/i386/gstabs.h: Delete.
22714 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22716 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22718 * config/i386/i386.c (ix86_decompose_address): Do not check for
22719 register RTX when looking at index_reg or base_reg.
22720 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22722 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22724 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22725 to update EH info here.
22727 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22729 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22731 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22733 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22735 2017-07-25 Torsten Duwe <duwe@suse.de>
22737 * common.opt: Introduce -fpatchable-function-entry
22738 command line option, and its variables function_entry_patch_area_size
22739 and function_entry_patch_area_start.
22740 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22741 including a two-value parser.
22742 * target.def (print_patchable_function_entry): New target hook.
22743 * targhooks.h (default_print_patchable_function_entry): New function.
22744 * targhooks.c (default_print_patchable_function_entry): Likewise.
22745 * toplev.c (process_options): Switch off IPA-RA if
22746 patchable function entries are being generated.
22747 * varasm.c (assemble_start_function): Look at the
22748 patchable-function-entry command line switch and current
22749 function attributes and maybe generate NOP instructions by
22750 calling the print_patchable_function_entry hook.
22751 * doc/extend.texi: Document patchable_function_entry attribute.
22752 * doc/invoke.texi: Document -fpatchable_function_entry
22753 command line option.
22754 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22756 * doc/tm.texi: Re-generate.
22758 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22761 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22762 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22764 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22766 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22768 2017-07-25 Richard Biener <rguenther@suse.de>
22770 PR tree-optimization/81455
22771 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22772 not walk in cycles when looking for guards.
22774 2017-07-25 Richard Biener <rguenther@suse.de>
22776 PR tree-optimization/81529
22777 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22778 when optimizing backedge uses.
22780 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22782 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22784 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22785 to dl_section_ref. On AIX, append an expression to subtract
22786 the size of the section length to dl_section_ref.
22788 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22790 * configure.ac: If any of the config.* scripts fail, exit 1.
22791 * configure: Regenerate.
22793 2017-07-25 Richard Biener <rguenther@suse.de>
22795 PR middle-end/81546
22796 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22797 of immediate uses, be more verbose on errors.
22799 2017-07-25 Richard Biener <rguenther@suse.de>
22801 PR tree-optimization/81510
22802 * tree-vect-loop.c (vect_is_simple_reduction): When the
22803 reduction stmt is not inside the loop bail out.
22805 2017-07-25 Richard Biener <rguenther@suse.de>
22807 PR tree-optimization/81303
22808 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22809 profitability check against LOOP_VINFO_NITERSM1.
22811 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22813 * domwalk.c (cmp_bb_postorder): Simplify.
22814 (sort_bbs_postorder): New function. Use it...
22815 (dom_walker::walk): ...here to optimize common cases.
22817 2017-07-25 Martin Liska <mliska@suse.cz>
22820 * ipa-visibility.c (function_and_variable_visibility): Make the
22821 redirection just on target that supports aliasing.
22822 Fix GNU coding style.
22824 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22827 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22829 * config/arm/rtems.h: Likewise.
22830 * config/bfin/rtems.h: Likewise.
22831 * config/i386/rtemself.h: Likewise.
22832 * config/lm32/rtems.h: Likewise.
22833 * config/m32c/rtems.h: Likewise.
22834 * config/m68k/rtemself.h: Likewise.
22835 * config/microblaze/rtems.h: Likewise.
22836 * config/mips/rtems.h: Likewise.
22837 * config/moxie/rtems.h: Likewise.
22838 * config/nios2/rtems.h: Likewise.
22839 * config/powerpcspe/rtems.h: Likewise.
22840 * config/rs6000/rtems.h: Likewise.
22841 * config/rtems.h: Likewise.
22842 * config/sh/rtems.h: Likewise.
22843 * config/sh/rtemself.h: Likewise.
22844 * config/sparc/rtemself.h: Likewise.
22846 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22849 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22850 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22851 * tree-ssa-structalias.c (alias_get_name): Same.
22853 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22856 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22857 instructions if no du chain is found.
22859 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22861 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22863 2017-07-25 Richard Biener <rguenther@suse.de>
22865 PR middle-end/81505
22866 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
22869 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22871 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
22872 upper-regs options.
22873 (ISA_2_7_MASKS_SERVER): Likewise.
22874 (ISA_3_0_MASKS_IEEE): Likewise.
22875 (OTHER_P8_VECTOR_MASKS): Likewise.
22876 (OTHER_VSX_VECTOR_MASKS): Likewise.
22877 (POWERPC_MASKS): Likewise.
22878 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
22879 duplicate list of options.
22880 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
22881 explicit -mupper-regs options.
22882 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
22883 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
22884 alias for -mupper-regs-df.
22885 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
22886 (rs6000_init_hard_regno_mode_ok): Likewise.
22887 (rs6000_option_override_internal): Likewise.
22888 (rs6000_opt_masks): Likewise.
22889 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
22890 options in terms of whether -mvsx or -mpower8-vector was used.
22891 (TARGET_UPPER_REGS_DI): Likewise.
22892 (TARGET_UPPER_REGS_SF): Likewise.
22893 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
22894 -mupper-regs-* options.
22896 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22898 * passes.c (emergency_dump_function): Print some empty lines and a
22899 header before the RTL dump.
22901 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22903 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
22905 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
22908 * config/aarch64/aarch64.c (aarch64_classify_symbol):
22909 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
22911 2017-07-24 Carl Love <cel@us.ibm.com>
22913 * config/rs6000/rs6000-c.c: Add support for built-in functions
22914 vector float vec_extract_fp32_from_shorth (vector unsigned short);
22915 vector float vec_extract_fp32_from_shortl (vector unsigned short);
22916 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
22917 vec_extract_fp_from_shortl): Add defines for the two builtins.
22918 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
22919 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
22921 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
22922 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
22923 * doc/extend.texi: Update the built-in documentation file for the
22924 new built-in function.
22926 2017-07-24 Jakub Jelinek <jakub@redhat.com>
22929 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
22931 * doc/generic.texi: Likewise.
22932 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22933 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22935 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
22937 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
22938 (aarch64_mls_elt_merge<mode>): Likewise.
22940 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
22942 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22943 having __cxa_atexit.
22945 2017-07-23 Michael Collison <michael.collison@arm.com>
22947 * config/arm/arm.c (arm_option_override): Deprecate
22948 use of -mstructure-size-boundary.
22949 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
22950 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
22952 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22955 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22956 Reduce cost estimate for direct moves.
22958 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
22961 * config/i386/i386.c (ix86_option_override_internal): Disable
22962 BMI, BMI2 and TBM instructions for -m16.
22964 2017-07-21 Carl Love <cel@us.ibm.com>
22966 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22967 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22968 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22969 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22970 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22971 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22972 VMULOSW): New enum "unspec" values.
22973 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22974 altivec_vmulosw): New patterns.
22975 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22976 VMULOSW): Add definitions.
22978 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
22980 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
22981 (qdf24xx): Likewise.
22982 * config/aarch64/aarch64-options-extensions.def (rdma); New.
22983 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
22984 (AARCH64_FL_V8_1): Renumber.
22985 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
22986 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
22987 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
22988 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
22989 rdma to feature modifiers list.
22991 2017-07-21 Yury Gribov <tetra2005@gmail.com>
22993 PR middle-end/56727
22994 * ipa-visibility (function_and_variable_visibility): Convert
22995 recursive PLT call to direct call if appropriate.
22997 2017-07-21 Andrew Pinski <apinski@cavium.com>
22999 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
23000 operand 1 to see if the types precision matches.
23001 * fold-const.c (operand_equal_p): Likewise.
23003 2017-07-21 Richard Biener <rguenther@suse.de>
23005 PR tree-optimization/81303
23006 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23007 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23008 (vect_peeling_hash_get_lowest_cost): Adjust.
23009 (vect_enhance_data_refs_alignment): Likewise. Use
23010 vect_get_peeling_costs_all_drs to compute the penalty for no
23011 peeling to match up costs.
23013 2017-07-21 Richard Biener <rguenther@suse.de>
23015 PR tree-optimization/81500
23016 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23017 we didn't identify a reduction path.
23019 2017-07-21 Tom de Vries <tom@codesourcery.com>
23020 Cesar Philippidis <cesar@codesourcery.com>
23022 PR gcov-profile/81442
23023 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23026 2017-07-21 Tom de Vries <tom@codesourcery.com>
23029 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23031 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23032 nvptx_override_options_after_change.
23034 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23036 * dwarf2out.c (output_file_names): Avoid double testing for
23037 dwarf_version >= 5.
23039 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23041 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
23043 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23045 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23047 (try_crossjump_to_edge): Do not punt on partitioned functions.
23049 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23051 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23052 Put all BBs reachable only via paths crossing cold region to cold
23054 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23056 2016-07-21 Richard Biener <rguenther@suse.de>
23058 PR tree-optimization/81303
23059 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23060 into account prologue and epilogue iterations when raising
23061 min_profitable_iters to sth at least covering one vector iteration.
23063 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23065 * config/arm/arm.c (arm_test_cpu_arch_dat):
23068 2017-07-20 Nathan Sidwell <nathan@acm.org>
23070 Remove TYPE_METHODS.
23071 * tree.h (TYPE_METHODS): Delete.
23072 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23073 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23074 (dbxout_type_methods): Scan TYPE_FIELDS.
23075 (dbxout_type): Don't check TYPE_METHODS here.
23076 * function.c (use_register_for_decl): Always ignore register for
23077 class types when not optimizing.
23078 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23079 * tree.c (free_lang_data_in_type): Stitch out member functions and
23080 templates from TYPE_FIELDS.
23081 (build_distinct_type_copy, verify_type_variant,
23082 verify_type): Member fns are on TYPE_FIELDS.
23083 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23084 * tree-pretty-print.c (dump_generic_node): Likewise.
23086 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23089 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23090 V2TImode and V4TImode.
23091 (ix86_expand_vector_extract): Likewise.
23092 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23093 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23094 (ssescalarmode): Handle V4TImode and V2TImode.
23095 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23096 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23097 (VEXTRACTI128_MODE): New mode iterator.
23098 (splitter for *vec_extractv?ti first element): New.
23099 (VEC_INIT_MODE): New mode iterator.
23100 (vec_init<mode>): Consolidate 3 expanders into one using
23101 VEC_INIT_MODE mode iterator.
23103 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23105 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23106 non_spilled_static_chain_regno_p.
23108 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23110 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
23112 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
23114 * bb-reorder.c (connect_traces): Allow copying of blocks within
23117 2017-07-20 Richard Biener <rguenther@suse.de>
23119 * gimple.h (gimple_phi_result): Add gphi * overload.
23120 (gimple_phi_result_ptr): Likewise.
23121 (gimple_phi_arg): Likewise. Adjust index assert to only
23122 allow actual argument accesses rather than all slots available
23124 (gimple_phi_arg_def): Add gphi * overload.
23125 * tree-phinodes.c (make_phi_node): Initialize only actual
23127 (resize_phi_node): Clear memory not covered by old node,
23128 do not initialize excess argument slots.
23129 (reserve_phi_args_for_new_edge): Initialize new argument slot
23132 2017-07-20 Bin Cheng <bin.cheng@arm.com>
23134 PR tree-optimization/81388
23136 2016-07-21 Bin Cheng <bin.cheng@arm.com>
23138 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
23139 by removing computation of may_be_zero.
23141 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23142 Tom de Vries <tom@codesourcery.com>
23144 PR middle-end/81030
23145 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
23146 when gimple level profile disagrees with what RTL expander did.
23148 2017-07-20 Richard Biener <rguenther@suse.de>
23150 PR tree-optimization/61171
23151 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
23152 (vect_analyze_stmt): Add slp instance parameter.
23153 (vectorizable_reduction): Likewise.
23154 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
23155 (vect_is_simple_reduction): Deal with chains not detected
23156 as SLP reduction chain, specifically not properly associated
23157 chains containing a mix of plus/minus.
23158 (get_reduction_op): Remove.
23159 (get_initial_defs_for_reduction): Simplify, pass in whether
23160 this is a reduction chain, pass in the SLP node for the PHIs.
23161 (vect_create_epilog_for_reduction): Get the SLP instance as
23163 (vectorizable_reduction): Get the SLP instance as arg.
23164 During analysis remember the SLP node with the PHIs in the
23165 instance. Simplify getting at the vectorized reduction PHIs.
23166 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
23167 through SLP instance.
23168 (vect_slp_analyze_operations): Likewise.
23169 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
23170 (vect_transform_stmt): Likewise.
23172 2017-07-20 Tom de Vries <tom@codesourcery.com>
23174 PR tree-optimization/81489
23175 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
23176 read of phi arg location to before loop that modifies phi.
23178 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23180 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
23183 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23185 PR middle-end/81331
23186 * except.c (execute): Fix ordering issue.
23188 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23190 PR rtl-optimization/81423
23191 * combine.c (make_compound_operation_int): Don't try to optimize
23192 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
23194 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23196 PR rtl-optimization/81423
23197 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
23198 with a constant that is -1 in the truncated to mode.
23200 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23202 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
23203 (determine_unlikely_bbs): ... here.
23204 * predict.h (propagate_unlikely_bbs_forward): Declare.
23205 * cfgexpand.c (pass_expand::execute): Use it.
23206 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
23208 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
23209 propagate_unlikely_bbs_forward.
23211 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23213 PR middle-end/81331
23214 * except.c (maybe_add_nop_after_section_switch): New function.
23217 2017-07-19 Tom de Vries <tom@codesourcery.com>
23219 * gimple.h (gimple_phi_set_arg): Make assert more strict.
23221 2017-07-19 Tom de Vries <tom@codesourcery.com>
23223 * gimple.h (gimple_phi_arg): Make assert more strict.
23225 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
23227 * config.gcc (powerpc*-*-*): Add mmintrin.h.
23228 * config/rs6000/mmintrin.h: New file.
23229 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
23231 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23233 PR tree-optimization/81346
23234 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
23236 2017-07-19 Tom de Vries <tom@codesourcery.com>
23238 * config/nvptx/nvptx.md (VECIM): Add V2DI.
23240 2017-07-19 Tom de Vries <tom@codesourcery.com>
23242 * config/nvptx/nvptx-modes.def: Add V2DImode.
23243 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
23244 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
23245 (nvptx_output_mov_insn): Handle lack of mov.b128.
23246 (nvptx_print_operand): Handle 'H' and 'L' codes.
23247 (nvptx_vector_mode_supported): Allow V2DImode.
23248 (nvptx_preferred_simd_mode): New function.
23249 (nvptx_data_alignment): New function.
23250 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
23251 nvptx_preferred_simd_mode.
23252 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
23254 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
23256 2017-07-19 Tom de Vries <tom@codesourcery.com>
23258 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
23259 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
23260 (nvptx_vector_mode_supported): New function. Allow V2SImode.
23261 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
23262 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
23263 (mov<VECIM>_insn): New define_insn.
23264 (define_expand "mov<VECIM>): New define_expand.
23266 2017-07-19 Tom de Vries <tom@codesourcery.com>
23268 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
23270 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23272 PR tree-optimization/81346
23273 * fold-const.h (fold_div_compare, range_check_type): Declare.
23274 * fold-const.c (range_check_type): New function.
23275 (build_range_check): Use range_check_type.
23276 (fold_div_compare): No longer static, rewritten into
23277 a match.pd helper function.
23278 (fold_comparison): Don't call fold_div_compare here.
23279 * match.pd (X / C1 op C2): New optimization using fold_div_compare
23280 as helper function.
23282 2017-07-19 Nathan Sidwell <nathan@acm.org>
23284 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
23285 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
23286 * tree.c (find_decls_types_r, verify_type): Use
23287 TYPE_{MIN,MAX}_VALUE_RAW.
23288 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
23289 (hash_tree): Likewise.
23290 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
23292 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
23295 2017-07-18 Tom de Vries <tom@codesourcery.com>
23297 PR middle-end/81464
23298 * omp-expand.c (expand_omp_for_static_chunk): Handle
23299 equal-argument loop exit phi.
23301 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
23304 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
23305 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
23306 operand 2 predicate.
23307 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
23308 operand 2 predicate.
23309 (ror,rol -> rorx splitters): Use const_int_operand as
23310 operand 2 predicate.
23312 2017-06-18 Richard Biener <rguenther@suse.de>
23314 PR tree-optimization/81410
23315 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
23316 the gap in the ! slp_perm SLP case after each group.
23318 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23320 PR middle-end/81463
23321 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
23324 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23326 PR middle-end/81462
23327 * predict.c (set_even_probabilities): Cleanup; do not affect
23328 probabilities that are already known.
23329 (combine_predictions_for_bb): Call even when count is set.
23331 2017-07-18 Nathan Sidwell <nathan@acm.org>
23333 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
23336 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23339 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
23340 optimization for loop niter analysis.
23342 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
23345 * config/avr/avr.c (avr_optimize_casesi): Don't use
23346 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
23348 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
23350 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
23351 body_cost_vec from _vect_peel_extended_info.
23352 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
23353 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
23356 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23358 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
23360 2017-07-18 Richard Biener <rguenther@suse.de>
23362 PR tree-optimization/80620
23363 PR tree-optimization/81403
23364 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
23365 info when re-using a VN table entry.
23367 2017-07-18 Richard Biener <rguenther@suse.de>
23369 PR tree-optimization/81418
23370 * tree-vect-loop.c (vectorizable_reduction): Properly compute
23371 vectype_in. Verify that with lane-reducing reduction operations
23372 we have a single def-use cycle.
23374 2017-07-17 Carl Love <cel@us.ibm.com>
23376 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
23378 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23379 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23380 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23381 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23382 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23383 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23384 VMULOSW): New enum "unspec" values.
23385 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23386 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23387 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23388 altivec_vmulosw): New patterns.
23389 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23390 VMULOSW): Add definitions.
23392 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
23394 * config/alpha/alpha.c: Include predict.h.
23396 2017-07-17 Yury Gribov <tetra2005@gmail.com>
23398 * tree-vrp.c (compare_assert_loc): Fix comparison function
23399 to return predictable results.
23401 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23403 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
23405 (subdi3): Likewise.
23406 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
23407 * doc/invoke.texi (mexpand-adddi): Update text.
23409 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23411 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
23412 that also clobbers the CC register. The old expand code is moved
23414 (*arc_clzsi2): ... here.
23415 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
23416 the CC register. The old expand code is moved to ...
23417 (arc_ctzsi2): ... here.
23419 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23421 * config/arc/arc.opt (mindexed-loads): Use initial value
23422 TARGET_INDEXED_LOADS_DEFAULT.
23423 (mauto-modify-reg): Use initial value
23424 TARGET_AUTO_MODIFY_REG_DEFAULT.
23425 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23426 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23427 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23428 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23430 2017-07-17 Martin Liska <mliska@suse.cz>
23433 * opts.c (finish_options): Do not allow -fgnu-tm
23434 w/ -fsanitize={kernel-,}address. Say sorry.
23436 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23439 * tree-loop-distribution.c (classify_partition): Only assert on
23440 numer of iterations.
23441 (merge_dep_scc_partitions): Delete prameter. Update function call.
23442 (distribute_loop): Remove code handling loop with unknown niters.
23443 (pass_loop_distribution::execute): Skip loop with unknown niters.
23445 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23448 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
23449 function sort_partitions_by_post_order.
23451 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23453 PR tree-optimization/81374
23454 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
23455 the max index of basic blocks, rather than number of basic blocks.
23457 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23459 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
23461 (arc_legitimate_pic_operand_p): Likewise.
23462 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
23464 (arc_needs_pcl_p): Likewise.
23465 (arc_legitimate_pc_offset_p): Likewise.
23466 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
23467 function is also used in constrains.md.
23468 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
23469 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
23470 PLUS. Only return true/false in known cases, otherwise assert.
23471 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
23472 is already called in arc_legitimate_constant_p.
23473 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
23475 (LEGITIMATE_PIC_OPERAND_P): Use
23476 arc_raw_symbolic_reference_mentioned_p function.
23477 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
23482 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23483 Andrew Burgess <andrew.burgess@embecosm.com>
23485 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
23486 (arc_return_address_register): New function.
23487 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
23488 (arc_handle_fndecl_attribute): Add naked attribute.
23489 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
23490 (TARGET_WARN_FUNC_RETURN): Likewise.
23491 (arc_allocate_stack_slots_for_args): New function.
23492 (arc_warn_func_return): Likewise.
23493 (machine_function): Change type fn_type.
23494 (arc_compute_function_type): Consider new naked function type,
23495 change function return type.
23496 (arc_must_save_register): Adapt to handle new
23497 arc_compute_function_type's return type.
23498 (arc_expand_prologue): Likewise.
23499 (arc_expand_epilogue): Likewise.
23500 (arc_return_address_regs): Delete.
23501 (arc_return_address_register): New function.
23502 (arc_epilogue_uses): Use above function.
23503 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
23504 (arc_function_type): Change encoding, add naked type.
23505 (ARC_INTERRUPT_P): Change to handle the new encoding.
23506 (ARC_FAST_INTERRUPT_P): Likewise.
23507 (ARC_NORMAL_P): Define.
23508 (ARC_NAKED_P): Likewise.
23509 (arc_compute_function_type): Delete prototype.
23510 * config/arc/arc.md (in_ret_delay_slot): Use
23511 arc_return_address_register function.
23512 (simple_return): Likewise.
23513 (p_return_i): Likewise.
23515 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23517 PR tree-optimization/81428
23518 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
23519 can't be built for those types.
23521 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23523 Remove stuff dead since r239246.
23525 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
23526 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
23527 (avr_inform_devices): Remove dead stuff.
23529 2017-07-17 Tamar Christina <tamar.christina@arm.com>
23531 * config/arm/arm_neon.h: Fix softp typo.
23533 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23535 PR tree-optimization/81365
23536 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
23537 aggregate moves onto bb predecessor edges, make sure there are no
23538 loads that could alias the lhs in between the start of bb and the
23541 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23544 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
23545 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
23546 [LSHIFTRT, outer_code = TRUNCATE]: Same.
23548 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23550 PR tree-optimization/81396
23551 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
23552 (init_symbolic_number): Initialize it to 1.
23553 (perform_symbolic_merge): Add n_ops from both operands into the new
23555 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
23556 without base_addr as useless if they need more than one operation.
23557 (bswap_replace): Handle !bswap case for NULL base_addr.
23559 2017-07-17 Tom de Vries <tom@codesourcery.com>
23562 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
23565 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
23567 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
23568 conditional builtin define __FIX_LEON3FT_B2BST.
23570 2017-07-17 Daniel Cederman <cederman@gaisler.com>
23572 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
23573 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
23576 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23578 PR rtl-optimization/81424
23579 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
23580 to remove potential trapping from operands if -fnon-call-exceptions.
23582 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23584 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
23585 profile_proability for scalling.
23586 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
23588 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23590 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
23592 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23594 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
23595 fixpoint arithmetics.
23597 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23599 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
23600 fixpoint arithmetics.
23602 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23604 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
23605 fixpoint arithmetics.
23607 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23609 * profile-count.h (profile_probability::from_reg_br_prob_note,
23610 profile_probability::to_reg_br_prob_note): New functions.
23611 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
23612 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
23613 * predict.c (probability_reliable_p): Update.
23614 (edge_probability_reliable_p): Update.
23615 (br_prob_note_reliable_p): Update.
23616 (invert_br_probabilities): Update.
23617 (add_reg_br_prob_note): New function.
23618 (combine_predictions_for_insn): Update.
23619 * asan.c (asan_clear_shadow): Update.
23620 * cfgbuild.c (compute_outgoing_frequencies): Update.
23621 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23622 (update_br_prob_note): Update.
23623 (rtl_verify_edges): Update.
23624 (purge_dead_edges): Update.
23625 (fixup_reorder_chain): Update.
23626 * emit-rtl.c (try_split): Update.
23627 * ifcvt.c (cond_exec_process_insns): Update.
23628 (cond_exec_process_if_block): Update.
23629 (dead_or_predicable): Update.
23630 * internal-fn.c (expand_addsub_overflow): Update.
23631 (expand_neg_overflow): Update.
23632 (expand_mul_overflow): Update.
23633 * loop-doloop.c (doloop_modify): Update.
23634 * loop-unroll.c (compare_and_jump_seq): Update.
23635 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23636 * predict.h: Update.
23637 * reorg.c (mostly_true_jump): Update.
23639 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23640 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23641 * config/arc/arc.c: (emit_unlikely_jump): Update.
23642 * config/arm/arm.c: (emit_unlikely_jump): Update.
23643 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23644 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23645 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23646 (ix86_print_operand): Update.
23647 (ix86_split_fp_branch): Update.
23648 (predict_jump): Update.
23649 * config/ia64/ia64.c (ia64_print_operand): Update.
23650 * config/mmix/mmix.c (mmix_print_operand): Update.
23651 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23652 (rs6000_expand_split_stack_prologue): Update.
23653 * config/rs6000/rs6000.c: Update.
23654 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23655 (s390_expand_vec_movstr): Update.
23656 (s390_expand_cs_tdsi): Update.
23657 (s390_expand_split_stack_prologue): Update.
23658 * config/sh/sh.c (sh_print_operand): Update.
23659 (expand_cbranchsi4): Update.
23660 (expand_cbranchdi4): Update.
23661 * config/sparc/sparc.c (output_v9branch): Update.
23662 * config/spu/spu.c (get_branch_target): Update.
23663 (ea_load_store_inline): Update.
23664 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23665 * config/tilepro/tilepro.c: Update.
23667 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23669 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23670 (gimplify_save_expr): Likewise.
23672 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23674 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23676 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23678 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23680 * timevar.def (TV_IPA_FNSUMMARY): Define.
23682 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23684 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23685 to back store errata sensitive sequence from being generated.
23686 (sqrtdf2_fix): Likewise.
23688 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23690 * tree-ssa-threadupdate.c (compute_path_counts,
23691 update_joiner_offpath_counts): Use profile_probability.
23693 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23696 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23698 * config/arm/arm-c.c (arm_cpu_builtins): Define
23699 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23701 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23703 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23704 array entries to represent __ieee128 versions of the
23705 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23706 scalar_extract_sig, and scalar_insert_exp built-in functions.
23707 (altivec_resolve_overloaded_builtin): Add special case handling
23708 for the __builtin_scalar_insert_exp function, as represented by
23709 the P9V_BUILTIN_VEC_VSIEDP constant.
23710 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23711 exponent support for __ieee128 argument.
23712 (VSESQP): Add scalar extract signature support for __ieee128
23714 (VSTDCNQP): Add scalar test negative support for __ieee128
23716 (VSIEQP): Add scalar insert exponent support for __int128 argument
23717 with __ieee128 result.
23718 (VSIEQPF): Add scalar insert exponent support for __ieee128
23719 argument with __ieee128 result.
23720 (VSTDCQP): Add scalar test data class support for __ieee128
23722 (VSTDCNQP): Add overload support for scalar test negative with
23723 __ieee128 argument.
23724 (VSTDCQP): Add overload support for scalar test data class
23725 __ieee128 argument.
23726 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23727 UNSPEC_VSX_SXSIGDP.
23728 (UNSPEC_VSX_SIEXPQP): New constant.
23729 (xsxexpqp): New insn for VSX scalar extract exponent quad
23731 (xsxsigqp): New insn for VSX scalar extract significand quad
23733 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23734 precision with floating point argument.
23735 (xststdcqp): New expand for VSX scalar test data class quad
23737 (xststdcnegqp): New expand for VSX scalar test negative quad
23739 (xststdcqp): New insn to match expansions for VSX scalar test data
23740 class quad precision and VSX scalar test negative quad precision.
23741 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23742 special case operand checking to enforce that second operand of
23743 VSX scalar test data class with quad precision argument is a 7-bit
23745 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23746 prototypes and descriptions of __ieee128 versions of
23747 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23748 scalar_test_data_class, and scalar_test_neg built-in functions.
23750 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23752 PR tree-optimization/81162
23753 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23754 replace a negate with an add.
23756 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23758 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23760 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23762 * config/arm/arm-c.c (arm_cpu_builtins): Define
23763 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23765 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23767 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23768 (armv8-r): Set ARM Cortex-R52 as default CPU.
23769 * config/arm/arm-tables.opt: Regenerate.
23770 * config/arm/arm-tune.md: Regenerate.
23771 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23773 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23774 extension for -mcpu=cortex-r52.
23776 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23778 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23779 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23780 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23781 (fp-armv8): Define it as FP_ARMv8 only.
23782 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23783 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23785 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23786 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23787 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23788 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23789 than TARGET_FPU_ARMV8.
23790 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23791 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23792 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23794 * config/arm/neon.md (neon_vrint): Likewise.
23795 (neon_vcvt): Likewise.
23796 (neon_<fmaxmin_op><mode>): Likewise.
23797 (<fmaxmin><mode>3): Likewise.
23798 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23799 * config/arm/predicates.md (arm_cond_move_operator): Check against
23800 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23802 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23804 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23805 to top of function.
23807 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23809 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23810 loop in comment with memset.
23812 2017-07-14 Martin Liska <mliska@suse.cz>
23814 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23815 * dwarf2out.c (is_java): Remove the function.
23816 (output_pubname): Remove usage of the function.
23817 (lower_bound_default): Remove usage of DW_LANG_Java.
23818 (gen_compile_unit_die): Likewise.
23819 * gcc.c: Remove compiler defaults for .java and .zip files.
23820 * gimple-expr.c (remove_suffix): Change as there's no longer
23821 extension than 4-letter one.
23822 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23823 (gimplify_save_expr): Likewise.
23824 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23825 as it's possible even for other languages than Java.
23826 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23827 * lto-opts.c (lto_write_options): Remove reference to Java.
23828 * opts.c (strip_off_ending): Update file extension handling.
23829 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23830 * tree-eh.c (lower_resx): Likewise.
23831 * tree.c (free_lang_data_in_type): Remove dead code.
23832 (find_decls_types_r): Likewise.
23833 (build_common_builtin_nodes): Remove Java from a comment.
23834 (verify_type): Remove dead code.
23835 * varasm.c (assemble_external): Remove Java from a comment.
23837 2017-07-14 Martin Liska <mliska@suse.cz>
23839 * opts.c (finish_options): Add quotes.
23840 (common_handle_option): Likewise.
23842 2017-07-14 Martin Liska <mliska@suse.cz>
23844 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23845 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23846 Remove N_SO_PASCAL.
23847 * dwarf2out.c (lower_bound_default): Do not handle
23849 (gen_compile_unit_die): Likewise.
23850 * gcc.c: Remove default extension binding for GNU Pascal.
23851 * stmt.c: Remove Pascal language from a comment.
23852 * xcoffout.c: Likewise.
23854 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23857 * diagnostic-show-locus.c (fixit_cmp): New function.
23858 (layout::layout): Sort m_fixit_hints.
23859 (column_range::column_range): Assert that the values are valid.
23860 (struct char_span): New struct.
23861 (correction::overwrite): New method.
23862 (struct source_line): New struct.
23863 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23864 calls in terms of classes source_line and char_span, and
23865 correction::overwrite.
23866 (selftest::test_overlapped_fixit_printing_2): New function.
23867 (selftest::diagnostic_show_locus_c_tests): Call it.
23869 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
23871 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
23872 early if there is no lhs.
23874 2017-07-13 Martin Liska <mliska@suse.cz>
23876 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
23877 (gen_reference_type_die): Likewise.
23878 * stor-layout.c: Remove Pascal-related comment.
23880 2017-07-13 Martin Liska <mliska@suse.cz>
23882 * opts.c (finish_options): Add quotes to error messages.
23883 (parse_sanitizer_options): Likewise.
23885 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
23887 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
23889 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
23891 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
23893 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23895 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
23897 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
23899 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23902 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
23903 provides the hardware capability bits, define the macro
23904 __BUILTIN_CPU_SUPPORTS__.
23905 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
23906 if GLIBC does not provide the hardware capability bits. Add a
23907 gcc_unreachable call if the built-in cpu function is neither
23908 __builtin_cpu_is nor __builtin_cpu_supports.
23909 (rs6000_get_function_versions_dispatcher): Change the warning
23910 that an old GLIBC is used which does not export the capability
23911 bits to be an error.
23912 * doc/extend.texi (target_clones attribute): Document the
23913 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
23914 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
23915 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
23916 the macros defined by GCC if the newer GLIBC is available.
23918 2017-07-12 Jeff Law <law@redhat.com>
23920 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
23921 remaining includes slightly.
23922 * config/riscv/riscv-builtins.c: Include profile-count.h.
23924 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23927 * config/avr/avr.c (avr_set_current_function): In diagnostic
23928 messages: Quote keywords and (parts of) identifiers.
23929 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
23932 2017-07-12 Carl Love <cel@us.ibm.com>
23934 * config/rs6000/rs6000-c.c: Add support for built-in functions
23935 vector bool char vec_revb (vector bool char);
23936 vector bool short vec_revb (vector short char);
23937 vector bool int vec_revb (vector bool int);
23938 vector bool long long vec_revb (vector bool long long);
23939 * doc/extend.texi: Update the built-in documentation file for the
23940 new built-in functions.
23942 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23944 * config/s390/s390.md: Remove movcc splitter.
23946 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23948 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
23949 load/store on condition.
23951 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23954 * config/avr/avr.c (avr_encode_section_info)
23955 [progmem && !TREE_READONLY]: Error if progmem object needs
23958 2017-07-11 Michael Collison <michael.collison@arm.com>
23960 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
23963 2017-07-11 Carl Love <cel@us.ibm.com>
23965 * config/rs6000/rs6000-c.c: Add support for builtins
23966 vector unsigned int vec_parity_lsbb (vector signed int);
23967 vector unsigned int vec_parity_lsbb (vector unsigned int);
23968 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
23969 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
23970 vector unsigned long long vec_parity_lsbb (vector signed long long);
23971 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
23972 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
23973 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
23974 * doc/extend.texi: Update the built-in documentation file for the
23975 new built-in functions.
23977 2017-07-11 David Malcolm <dmalcolm@redhat.com>
23979 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
23980 (layout::m_primary_loc): New field.
23981 (layout::layout): Initialize new field. Move location filtering
23982 logic from here to...
23983 (layout::maybe_add_location_range): ...this new method. Add
23984 support for filtering to just the lines already specified by other
23986 (layout::will_show_line_p): New method.
23987 (gcc_rich_location::add_location_if_nearby): New method.
23988 (selftest::test_add_location_if_nearby): New test function.
23989 (selftest::diagnostic_show_locus_c_tests): Call it.
23990 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
23993 2017-07-11 Tom de Vries <tom@codesourcery.com>
23995 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
23996 (bb_first_real_insn): New function.
23997 (nvptx_single): Add extra initialization of broadcasted condition
24000 2017-07-11 Nathan Sidwell <nathan@acm.org>
24002 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
24004 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24006 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24007 Binutils doc as TEXI will mess them up.
24008 * doc/invoke.texi (AVR Options): Same here.
24010 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24012 * config/sparc/sparc.opt (mfix-ut700): New option.
24013 (mfix-gr712rc): Likewise.
24014 (sparc_fix_b2bst): New variable.
24015 * doc/invoke.texi (SPARC options): Document them.
24016 (ARM options): Fix warnings.
24017 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24018 instructions to prevent sequences that can trigger the store-store
24019 errata for certain LEON3FT processors.
24020 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24021 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24022 * config/sparc/sparc.md (fix_b2bst): New attribute.
24023 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24025 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24028 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24030 (*rsqrtsf2_sse): Ditto.
24032 (div<mode>3): Macroize insn from divdf3 and divsf3
24033 using MODEF mode iterator.
24035 2017-07-10 Martin Sebor <msebor@redhat.com>
24037 PR tree-optimization/80397
24038 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24039 instead of testing for equality to INTEGER_TYPE.
24041 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
24043 * config.gcc: Remove uclibc from arc target spec.
24045 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24047 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24049 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24052 * lto-wrapper.c (remove_option): New function.
24053 (merge_and_complain): Merge PIC/PIE options more realistically.
24055 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24057 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24061 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24062 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24063 * config.in: Regenerate.
24064 * configure: Regenerate.
24065 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24066 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24067 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24068 (TARGET_GASISR_PROLOGUES): ...target mask.
24069 * common/config/avr/avr-common.c
24070 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24071 Set -mgas-isr-prologues.
24072 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24073 INSERT_PASS_BEFORE for it.
24074 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24075 * config/avr/avr.c (avr_option_override)
24076 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24077 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24078 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24079 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24080 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24082 (make_avr_pass_pre_proep): New function.
24083 (emit_push_sfr) <treg>: Add argument to function and use it
24084 instead of TMP_REG.
24085 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24086 and set machine->gasisr.yes.
24087 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24088 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24089 __gcc_isr.n_pushed to .L__stack_usage.
24090 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24091 (avr_asm_final_postscan_insn): ...this new static function.
24092 * config/avr/avr.h (machine_function)
24093 <is_no_gccisr, use_L__stack_usage>: New fields.
24094 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24095 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24096 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24097 (gasisr, *gasisr): New expander and insn.
24098 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24099 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24100 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24102 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24104 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24107 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24109 Move jump-tables out of .text again.
24112 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
24113 (ASM_OUTPUT_ADDR_VEC): New function.
24114 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
24115 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
24116 INSN_ADDRESSes as asm comment.
24117 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
24118 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
24119 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
24120 * config/avr/avr.md (*tablejump): Adjust comment.
24121 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
24122 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
24124 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
24125 (avr_output_addr_vec): New proto.
24126 (avr_log_t) <insn_addresses>: New field.
24128 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
24131 * config/i386/i386.c (ix86_function_arg_advance): Set
24132 outgoing_args_on_stack to true if there are outgoing arguments
24134 (ix86_function_arg): Likewise.
24135 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
24136 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
24137 * config/i386/i386.h (machine_function): Add
24138 outgoing_args_on_stack.
24140 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
24142 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
24143 supporting pthreds.
24144 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
24146 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
24148 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
24149 (REAL_H): Remove $(MACHMODE_H).
24150 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
24152 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
24153 $(MACHMODE_H) and double-int.h.
24154 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
24156 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
24159 2017-07-07 Andrew Pinski <apinski@cavium.com>
24161 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
24162 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
24164 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
24166 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
24167 Add warning if GCC was not configured to link against a GLIBC that
24168 exports the hardware capability bits.
24169 (make_resolver_func): Make resolver function private and not a
24170 COMDAT function. Create the name with clone_function_name instead
24171 of make_unique_name.
24174 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
24175 correct operand in doing the split.
24177 2017-07-07 Carl Love <cel@us.ibm.com>
24179 * config/rs6000/rs6000-c: Add support for built-in function
24180 vector unsigned short vec_pack_to_short_fp32 (vector float,
24182 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
24183 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
24184 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
24185 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
24186 (convert_4f32_8i16): Add define_expand.
24187 * doc/extend.texi: Update the built-in documentation file for the
24188 new built-in function.
24190 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24192 * config/sparc/m8.md: New file.
24193 * config/sparc/sparc.md: Include m8.md.
24195 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24197 * config/sparc/sparc.opt: New option -mvis4b.
24198 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
24199 (sparc_option_override): Handle VIS4B.
24200 (enum sparc_builtins): Define
24201 SPARC_BUILTIN_DICTUNPACK{8,16,32},
24202 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
24203 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
24204 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
24205 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
24206 (check_constant_argument): New function.
24207 (sparc_vis_init_builtins): Define builtins
24208 __builtin_vis_dictunpack{8,16,32},
24209 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
24210 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
24211 __builtin_vis_fpcmpde{8,16,32}shl and
24212 __builtin_vis_fpcmpur{8,16,32}shl.
24213 (sparc_expand_builtin): Check that the constant operands to
24214 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
24215 constant and in range.
24216 * config/sparc/sparc-c.c (sparc_target_macros): Handle
24218 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
24219 (SPARC_IMM5_P): Likewise.
24220 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
24221 (enabled): Handle vis4b.
24222 (UNSPEC_DICTUNPACK): New unspec.
24223 (UNSPEC_FPCMPSHL): Likewise.
24224 (UNSPEC_FPUCMPSHL): Likewise.
24225 (UNSPEC_FPCMPDESHL): Likewise.
24226 (UNSPEC_FPCMPURSHL): Likewise.
24227 (cpu_feature): New CPU feature `vis4b'.
24228 (dictunpack{8,16,32}): New insns.
24229 (FPCSMODE): New mode iterator.
24230 (fpcscond): New code iterator.
24231 (fpcsucond): Likewise.
24232 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
24233 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
24234 (fpcmpde{8,16,32}{si,di}shl): Likewise.
24235 (fpcmpur{8,16,32}{si,di}shl): Likewise.
24236 * config/sparc/constraints.md: Define constraints `q' for unsigned
24237 2-bit integer constants and `t' for unsigned 5-bit integer
24239 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
24241 (imm5_operand_dictunpack16): Likewise.
24242 (imm5_operand_dictunpack32): Likewise.
24243 (imm2_operand): Likewise.
24244 * doc/invoke.texi (SPARC Options): Document -mvis4b.
24245 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
24246 ditunpack* and fpcmp*shl builtins.
24248 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24250 * config.gcc: Handle m8 in --with-{cpu,tune} options.
24251 * config.in: Add HAVE_AS_SPARC6 define.
24252 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
24254 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
24256 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
24257 (CPP_CPU_SPEC): Handle m8.
24258 (ASM_CPU_SPEC): Likewise.
24259 * config/sparc/sparc-opts.h (enum processor_type): Add
24261 * config/sparc/sparc.c (m8_costs): New struct.
24262 (sparc_option_override): Handle TARGET_CPU_m8.
24263 (sparc32_initialize_trampoline): Likewise.
24264 (sparc64_initialize_trampoline): Likewise.
24265 (sparc_issue_rate): Likewise.
24266 (sparc_register_move_cost): Likewise.
24267 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
24268 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
24269 (ASM_CPU64_DEFAULT_SPEC): Likewise.
24270 (CPP_CPU_SPEC): Handle M8.
24271 (ASM_CPU_SPEC): Likewise.
24272 (AS_M8_FLAG): Define.
24273 * config/sparc/sparc.md: Add m8 to the cpu attribute.
24274 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
24275 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
24277 * configure: Regenerate.
24278 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
24281 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24283 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
24285 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
24286 ("*movdi_insn_sp32"): Do not set v3pipe.
24287 ("*movsi_insn"): Likewise.
24288 ("*movdi_insn_sp64"): Likewise.
24289 ("*movsf_insn"): Likewise.
24290 ("*movdf_insn_sp32"): Likewise.
24291 ("*movdf_insn_sp64"): Likewise.
24292 ("*zero_extendsidi2_insn_sp64"): Likewise.
24293 ("*sign_extendsidi2_insn"): Likewise.
24294 ("*mov<VM32:mode>_insn"): Likewise.
24295 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24296 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24297 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24298 ("<vlop:code><VL:mode>3"): Likewise.
24299 ("*not_<vlop:code><VL:mode>3"): Likewise.
24300 ("*nand<VL:mode>_vis"): Likewise.
24301 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
24302 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
24303 ("one_cmpl<VL:mode>2"): Likewise.
24304 ("faligndata<VM64:mode>_vis"): Likewise.
24305 ("alignaddrsi_vis"): Likewise.
24306 ("alignaddrdi_vis"): Likweise.
24307 ("alignaddrlsi_vis"): Likewise.
24308 ("alignaddrldi_vis"): Likewise.
24309 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24310 ("bmaskdi_vis"): Likewise.
24311 ("bmasksi_vis"): Likewise.
24312 ("bshuffle<VM64:mode>_vis"): Likewise.
24313 ("cmask8<P:mode>_vis"): Likewise.
24314 ("cmask16<P:mode>_vis"): Likewise.
24315 ("cmask32<P:mode>_vis"): Likewise.
24316 ("pdistn<P:mode>_vis"): Likewise.
24317 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24319 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24321 * config/sparc/sparc.md ("subtype"): New insn attribute.
24322 ("*wrgsr_sp64"): Set insn subtype.
24323 ("*rdgsr_sp64"): Likewise.
24324 ("alignaddrsi_vis"): Likewise.
24325 ("alignaddrdi_vis"): Likewise.
24326 ("alignaddrlsi_vis"): Likewise.
24327 ("alignaddrldi_vis"): Likewise.
24328 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24329 ("fexpand_vis"): Likewise.
24330 ("fpmerge_vis"): Likewise.
24331 ("faligndata<VM64:mode>_vis"): Likewise.
24332 ("bshuffle<VM64:mode>_vis"): Likewise.
24333 ("cmask8<P:mode>_vis"): Likewise.
24334 ("cmask16<P:mode>_vis"): Likewise.
24335 ("cmask32<P:mode>_vis"): Likewise.
24336 ("fchksm16_vis"): Likewise.
24337 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
24338 ("fmean16_vis"): Likewise.
24339 ("fp<plusminus_insn>64_vis"): Likewise.
24340 ("<plusminus_insn>v8qi3"): Likewise.
24341 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24342 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
24343 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
24344 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
24345 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
24346 ("*movqi_insn"): Likewise.
24347 ("*movhi_insn"): Likewise.
24348 ("*movsi_insn"): Likewise.
24349 ("movsi_pic_gotdata_op"): Likewise.
24350 ("*movdi_insn_sp32"): Likewise.
24351 ("*movdi_insn_sp64"): Likewise.
24352 ("movdi_pic_gotdata_op"): Likewise.
24353 ("*movsf_insn"): Likewise.
24354 ("*movdf_insn_sp32"): Likewise.
24355 ("*movdf_insn_sp64"): Likewise.
24356 ("*zero_extendhisi2_insn"): Likewise.
24357 ("*zero_extendqihi2_insn"): Likewise.
24358 ("*zero_extendqisi2_insn"): Likewise.
24359 ("*zero_extendqidi2_insn"): Likewise.
24360 ("*zero_extendhidi2_insn"): Likewise.
24361 ("*zero_extendsidi2_insn_sp64"): Likewise.
24362 ("ldfsr"): Likewise.
24363 ("prefetch_64"): Likewise.
24364 ("prefetch_32"): Likewise.
24365 ("tie_ld32"): Likewise.
24366 ("tie_ld64"): Likewise.
24367 ("*tldo_ldub_sp32"): Likewise.
24368 ("*tldo_ldub1_sp32"): Likewise.
24369 ("*tldo_ldub2_sp32"): Likewise.
24370 ("*tldo_ldub_sp64"): Likewise.
24371 ("*tldo_ldub1_sp64"): Likewise.
24372 ("*tldo_ldub2_sp64"): Likewise.
24373 ("*tldo_ldub3_sp64"): Likewise.
24374 ("*tldo_lduh_sp32"): Likewise.
24375 ("*tldo_lduh1_sp32"): Likewise.
24376 ("*tldo_lduh_sp64"): Likewise.
24377 ("*tldo_lduh1_sp64"): Likewise.
24378 ("*tldo_lduh2_sp64"): Likewise.
24379 ("*tldo_lduw_sp32"): Likewise.
24380 ("*tldo_lduw_sp64"): Likewise.
24381 ("*tldo_lduw1_sp64"): Likewise.
24382 ("*tldo_ldx_sp64"): Likewise.
24383 ("*mov<VM32:mode>_insn"): Likewise.
24384 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24385 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24387 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24389 * config/sparc/sparc.md ("type"): New insn type viscmp.
24390 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
24392 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
24393 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
24394 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24395 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
24397 ("n7_vis_logical_11cycle"): Likewise.
24398 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
24399 * config/sparc/niagara2.md ("niag3_vis": Likewise.
24400 * config/sparc/niagara.md ("niag_vis"): Likewise.
24401 * config/sparc/ultra3.md ("us3_fga"): Likewise.
24402 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
24404 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24406 * config/sparc/sparc.md: New instruction type `bmask'.
24407 (bmaskdi_vis): Use the `bmask' type.
24408 (bmasksi_vis): Likewise.
24409 * config/sparc/ultra3.md (us3_array): Likewise.
24410 * config/sparc/niagara7.md (n7_array): Likewise.
24411 * config/sparc/niagara4.md (n4_array): Likewise.
24412 * config/sparc/niagara2.md (niag2_vis): Likewise.
24413 (niag3_vis): Likewise.
24414 * config/sparc/niagara.md (niag_vis): Likewise.
24416 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24418 * ipa-comdats.c: Remove optimize check from gate.
24419 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
24420 for functions not optimized.
24421 (ipa_fn_summary_read): Skip optimize check.
24422 (ipa_fn_summary_write): Likewise.
24423 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
24425 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
24427 (can_inline_edge_p): Check flag_pcc_struct_return for match.
24428 (check_callers): Give up on caller which is not optimized.
24429 (inline_small_functions): Likewise.
24430 (ipa_inline): Do not give up when not optimizing.
24431 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
24432 away unoptimizes cdtors.
24433 (whole_program_function_and_variable_visibility): Do
24434 ipa_discover_readonly_nonaddressable_vars in LTO mode.
24435 * ipa.c (process_references): Do not check optimize.
24436 (symbol_table::remove_unreachable_nodes): Update optimize check.
24437 (set_writeonly_bit): Update optimize check.
24438 (pass_ipa_cdtor_merge::gate): Do not check optimize.
24439 (pass_ipa_single_use::gate): Remove.
24441 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24443 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
24444 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
24445 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
24446 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
24447 permute_load, permute_store, adjust_extract, adjust_splat,
24448 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
24449 replace_swap_with_copy, dump_swap_insn_table,
24450 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
24451 recombine_lvx_pattern, recombine_stvx_pattern,
24452 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
24453 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
24454 to file rs6000-p8swap.c.
24455 * config/rs6000/rs6000-p8swap.c: New file.
24456 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
24457 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
24458 and rs6000*-*-* targets.
24460 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24462 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
24464 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24466 * lto-wrapper.c (merge_and_complain): Do not merge
24467 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
24468 fsigned_zeros, ftrapping_math, fwrapv.
24469 (append_compiler_options): Do not track these options.
24470 (append_linker_options): Likewie
24472 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24474 * cgraphunit.c (cgraph_node::finalize_function): When
24475 !flag_toplevel_reorde set no_reorder flag.
24476 (varpool_node::finalize_decl): Likewise.
24477 (symbol_table::compile): Drop no toplevel reorder path.
24479 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24481 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
24482 edges; zero probability is not better than uninitialized.
24484 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24486 * asan.h (asan_sanitize_allocas_p): Declare.
24487 * asan.c (asan_sanitize_allocas_p): New function.
24488 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
24489 (handle_builtin_alloca): Likewise.
24490 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
24491 if !asan_sanitize_allocas_p.
24492 * params.def (asan-instrument-allocas): Add new option.
24493 * params.h (ASAN_PROTECT_ALLOCAS): Define.
24494 * opts.c (common_handle_option): Disable allocas sanitization for
24497 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24499 * asan.c: Include gimple-fold.h.
24500 (get_last_alloca_addr): New function.
24501 (handle_builtin_stackrestore): Likewise.
24502 (handle_builtin_alloca): Likewise.
24503 (asan_emit_allocas_unpoison): Likewise.
24504 (get_mem_refs_of_builtin_call): Add new parameter, remove const
24505 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
24506 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
24507 (instrument_builtin_call): Pass gimple iterator to
24508 get_mem_refs_of_builtin_call.
24509 (last_alloca_addr): New global.
24510 * asan.h (asan_emit_allocas_unpoison): Declare.
24511 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
24512 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
24513 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
24514 if function calls alloca.
24515 * gimple-fold.c (replace_call_with_value): Remove static keyword.
24516 * gimple-fold.h (replace_call_with_value): Declare.
24517 * internal-fn.c: Include asan.h.
24518 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
24519 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
24521 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24523 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
24524 (C_SELFTEST_FLAGS): New.
24525 (CPP_SELFTEST_FLAGS): New.
24526 (SELFTEST_DEPS): New, from deps of s-selftest.
24527 (C_SELFTEST_DEPS): New, from deps of s-selftest.
24528 (CPP_SELFTEST_DEPS): New.
24529 (selftest): Add dependency on s-selftest-c++.
24530 (s-selftest): Rename to...
24531 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
24532 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
24533 than SELFTEST_FLAGS.
24534 (selftest-gdb): Rename to...
24535 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
24537 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
24538 (selftest-valgrind): Rename to...
24539 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
24541 (selftest-valgrind): Reintroduce as an alias for
24542 selftest-c-valgrind.
24543 (s-selftest-c++): New.
24544 (selftest-c++-gdb): New.
24545 (selftest-c++-valgrind): New.
24547 2017-07-06 Olivier Hainque <hainque@adacore.com>
24549 * gcc.c (process_command): When deciding if undefined variables
24550 should be ignored when processing specs, accept "gcc -v" as well.
24552 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24554 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
24555 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
24557 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24559 * config/arm/arm-cpus.in (armv8-r): Add new entry.
24560 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
24561 * config/arm/arm-tables.opt: Regenerate.
24562 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
24564 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
24566 2017-07-06 Carl Love <cel@us.ibm.com>
24568 * ChangeLog: Clean up from mid air collision
24570 2017-07-06 Carl Love <cel@us.ibm.com>
24572 * config/rs6000/rs6000-c.c: Add support for built-in functions
24573 vector signed int vec_subc (vector signed int, vector signed int);
24574 vector signed __int128 vec_subc (vector signed __int128,
24575 vector signed __int128);
24576 vector unsigned __int128 vec_subc (vector unsigned __int128,
24577 vector unsigned __int128);
24578 vector signed int vec_sube (vector signed int, vector signed int,
24579 vector signed int);
24580 vector unsigned int vec_sube (vector unsigned int,
24581 vector unsigned int,
24582 vector unsigned int);
24583 vector signed __int128 vec_sube (vector signed __int128,
24584 vector signed __int128,
24585 vector signed__int128);
24586 vector unsigned __int128 vec_sube (vector unsigned __int128,
24587 vector unsigned __int128,
24588 vector unsigned __int128);
24589 vector signed int vec_subec (vector signed int, vector signed int,
24590 vector signed int);
24591 vector unsigned int vec_subec (vector unsigned int,
24592 vector unsigned int,
24593 vector unsigned int);
24594 vector signed __int128 vec_subec (vector signed __int128,
24595 vector signed __int128,
24596 vector signed__int128);
24597 vector unsigned __int128 vec_subec (vector unsigned __int128,
24598 vector unsigned __int128,
24599 vector unsigned __int128);
24600 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
24601 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
24602 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
24603 BU_ALTIVEC_OVERLOAD_X definitions.
24604 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
24605 * doc/extend.texi: Update the built-in documentation file for the new
24606 built-in functions.
24608 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24611 * diagnostic-show-locus.c (layout::layout): Use start and finish
24612 spelling location for the start and finish of each range.
24613 * genmatch.c (linemap_client_expand_location_to_spelling_point):
24614 Add unused aspect param.
24615 * input.c (expand_location_1): Add "aspect" param, and use it
24616 to access the correct part of the location.
24617 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
24619 (expand_location_to_spelling_point): Likewise.
24620 (linemap_client_expand_location_to_spelling_point): Add "aspect"
24621 param, and pass it to expand_location_1.
24623 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24625 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24626 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24627 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24628 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24629 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24630 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24631 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24632 _mm_maskz_getmant_ss): New intrinsics.
24633 (__builtin_ia32_getexpss128_mask): Changed to ...
24634 __builtin_ia32_getexpss128_round ... this.
24635 (__builtin_ia32_getexpsd128_mask): Changed to ...
24636 __builtin_ia32_getexpsd128_round ... this.
24637 * config/i386/i386-builtin-types.def
24638 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24639 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24640 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24641 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24642 __builtin_ia32_getmantss_mask_round): New builtins.
24643 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24644 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24645 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24646 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24647 * config/i386/sse.md
24648 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24649 avx512f_sgetexp<mode><mask_scalar_name>
24650 <round_saeonly_scalar_name> ... this.
24651 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24652 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24653 vgetexp<ssescalarmodesuffix>
24654 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24655 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24656 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24657 avx512f_vgetmant<mode><mask_scalar_name>
24658 <round_saeonly_scalar_name> ... this.
24659 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24660 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24661 vgetmant<ssescalarmodesuffix>
24662 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24663 %0<mask_scalar_operand4>, %1, %2
24664 <round_saeonly_scalar_mask_op4>, %3} ... this.
24665 * config/i386/subst.md (mask_scalar_operand4,
24666 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24667 round_saeonly_scalar_nimm_predicate): New subst attributes.
24669 2017-07-06 Julia Koval <julia.koval@intel.com>
24671 * config/i386/i386.c (ix86_erase_embedded_rounding):
24672 Remove code for old rounding pattern.
24674 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24676 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24678 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24680 * doc/sourcebuild.texi (Test Directives, Variants of
24681 dg-require-support): Add documentation for dg-require-stack-check.
24683 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24685 * config/i386/subst.md (mask_scalar, round_scalar,
24686 round_saeonly_scalar): New meta-templates.
24687 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24688 round_scalar_mask_operand3, round_scalar_mask_op3,
24689 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24690 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24691 round_saeonly_scalar_constraint,
24692 round_saeonly_scalar_prefix): New subst attribute.
24693 * config/i386/sse.md
24694 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24695 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24696 <round_scalar_name> ... this.
24697 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24698 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24699 <round_scalar_name> ... this.
24700 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24701 <sse>_vm<code><mode>3<mask_scalar_name>
24702 <round_saeonly_scalar_name> ... this.
24703 (v<plusminus_mnemonic><ssescalarmodesuffix>
24704 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24705 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24706 v<plusminus_mnemonic><ssescalarmodesuffix>
24707 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24708 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24709 (v<multdiv_mnemonic><ssescalarmodesuffix>
24710 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24711 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24712 v<multdiv_mnemonic><ssescalarmodesuffix>
24713 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24714 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24715 (v<maxmin_float><ssescalarmodesuffix>
24716 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24717 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24718 v<maxmin_float><ssescalarmodesuffix>
24719 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24720 %0<mask_scalar_operand3>, %1, %<iptr>2
24721 <round_saeonly_scalar_mask_op3>} ... this.
24723 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24725 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24726 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24728 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24729 Alan Hayward <alan.hayward@arm.com>
24730 David Sherwood <david.sherwood@arm.com>
24732 * combine.c (simplify_if_then_else): Remove "enum" before
24734 * compare-elim.c (can_eliminate_compare): Likewise.
24735 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24737 (aarch64_lookup_simd_builtin_type): Likewise.
24738 (aarch64_simd_builtin_type): Likewise.
24739 (aarch64_init_simd_builtin_types): Likewise.
24740 (aarch64_simd_expand_args): Likewise.
24741 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24743 (aarch64_reverse_mask): Likewise.
24744 (aarch64_simd_emit_reg_reg_move): Likewise.
24745 (aarch64_gen_adjusted_ldpstp): Likewise.
24746 (aarch64_ccmp_mode_to_code): Likewise.
24747 (aarch64_operands_ok_for_ldpstp): Likewise.
24748 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24749 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24751 (aarch64_min_divisions_for_recip_mul): Likewise.
24752 (aarch64_reassociation_width): Likewise.
24753 (aarch64_get_condition_code_1): Likewise.
24754 (aarch64_simd_emit_reg_reg_move): Likewise.
24755 (aarch64_simd_attr_length_rglist): Likewise.
24756 (aarch64_reverse_mask): Likewise.
24757 (aarch64_operands_ok_for_ldpstp): Likewise.
24758 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24759 (aarch64_gen_adjusted_ldpstp): Likewise.
24760 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24762 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24763 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24764 (arm_lookup_simd_builtin_type): Likewise.
24765 (arm_simd_builtin_type): Likewise.
24766 (arm_init_simd_builtin_types): Likewise.
24767 (arm_expand_builtin_args): Likewise.
24768 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24769 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24770 (ft32_setup_incoming_varargs): Likewise.
24771 (ft32_function_arg): Likewise.
24772 (ft32_function_arg_advance): Likewise.
24773 (ft32_pass_by_reference): Likewise.
24774 (ft32_arg_partial_bytes): Likewise.
24775 (ft32_valid_pointer_mode): Likewise.
24776 (ft32_addr_space_pointer_mode): Likewise.
24777 (ft32_addr_space_legitimate_address_p): Likewise.
24778 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24780 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24781 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24782 (iamcu_alignment): Likewise.
24783 (canonicalize_vector_int_perm): Likewise.
24784 (ix86_noce_conversion_profitable_p): Likewise.
24785 (ix86_mpx_bound_mode): Likewise.
24786 (ix86_operands_ok_for_move_multiple): Likewise.
24787 * config/microblaze/microblaze-protos.h
24788 (microblaze_expand_conditional_branch_reg): Likewise.
24789 * config/microblaze/microblaze.c
24790 (microblaze_expand_conditional_branch_reg): Likewise.
24791 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24793 (rs6000_reassociation_width): Likewise.
24794 (rs6000_invalid_binary_op): Likewise.
24795 (fusion_p9_p): Likewise.
24796 (emit_fusion_p9_load): Likewise.
24797 (emit_fusion_p9_store): Likewise.
24798 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24800 (riscv_hard_regno_mode_ok_p): Likewise.
24801 (riscv_address_insns): Likewise.
24802 (riscv_split_symbol): Likewise.
24803 (riscv_legitimize_move): Likewise.
24804 (riscv_function_value): Likewise.
24805 (riscv_hard_regno_nregs): Likewise.
24806 (riscv_expand_builtin): Likewise.
24807 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24808 (riscv_build_integer): Likewise.
24809 (riscv_split_integer): Likewise.
24810 (riscv_legitimate_constant_p): Likewise.
24811 (riscv_cannot_force_const_mem): Likewise.
24812 (riscv_regno_mode_ok_for_base_p): Likewise.
24813 (riscv_valid_base_register_p): Likewise.
24814 (riscv_valid_offset_p): Likewise.
24815 (riscv_valid_lo_sum_p): Likewise.
24816 (riscv_classify_address): Likewise.
24817 (riscv_legitimate_address_p): Likewise.
24818 (riscv_address_insns): Likewise.
24819 (riscv_load_store_insns): Likewise.
24820 (riscv_force_binary): Likewise.
24821 (riscv_split_symbol): Likewise.
24822 (riscv_force_address): Likewise.
24823 (riscv_legitimize_address): Likewise.
24824 (riscv_move_integer): Likewise.
24825 (riscv_legitimize_const_move): Likewise.
24826 (riscv_legitimize_move): Likewise.
24827 (riscv_address_cost): Likewise.
24828 (riscv_subword): Likewise.
24829 (riscv_output_move): Likewise.
24830 (riscv_canonicalize_int_order_test): Likewise.
24831 (riscv_emit_int_order_test): Likewise.
24832 (riscv_function_arg_boundary): Likewise.
24833 (riscv_pass_mode_in_fpr_p): Likewise.
24834 (riscv_pass_fpr_single): Likewise.
24835 (riscv_pass_fpr_pair): Likewise.
24836 (riscv_get_arg_info): Likewise.
24837 (riscv_function_arg): Likewise.
24838 (riscv_function_arg_advance): Likewise.
24839 (riscv_arg_partial_bytes): Likewise.
24840 (riscv_function_value): Likewise.
24841 (riscv_pass_by_reference): Likewise.
24842 (riscv_setup_incoming_varargs): Likewise.
24843 (riscv_print_operand): Likewise.
24844 (riscv_elf_select_rtx_section): Likewise.
24845 (riscv_save_restore_reg): Likewise.
24846 (riscv_for_each_saved_reg): Likewise.
24847 (riscv_register_move_cost): Likewise.
24848 (riscv_hard_regno_mode_ok_p): Likewise.
24849 (riscv_hard_regno_nregs): Likewise.
24850 (riscv_class_max_nregs): Likewise.
24851 (riscv_memory_move_cost): Likewise.
24852 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24853 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24854 (rl78_addr_space_address_mode): Likewise.
24855 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24857 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24858 (rs6000_reassociation_width): Likewise.
24859 (rs6000_invalid_binary_op): Likewise.
24860 (fusion_p9_p): Likewise.
24861 (emit_fusion_p9_load): Likewise.
24862 (emit_fusion_p9_store): Likewise.
24863 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24864 (ok_for_simple_move_operands): Likewise.
24865 (ok_for_simple_move_strict_operands): Likewise.
24866 (ok_for_simple_arith_logic_operands): Likewise.
24867 (visium_legitimize_reload_address): Likewise.
24868 (visium_select_cc_mode): Likewise.
24869 (output_cbranch): Likewise.
24870 (visium_split_double_move): Likewise.
24871 (visium_expand_copysign): Likewise.
24872 (visium_expand_int_cstore): Likewise.
24873 (visium_expand_fp_cstore): Likewise.
24874 * config/visium/visium.c (visium_pass_by_reference): Likewise.
24875 (visium_function_arg): Likewise.
24876 (visium_function_arg_advance): Likewise.
24877 (visium_libcall_value): Likewise.
24878 (visium_setup_incoming_varargs): Likewise.
24879 (visium_legitimate_constant_p): Likewise.
24880 (visium_legitimate_address_p): Likewise.
24881 (visium_legitimize_address): Likewise.
24882 (visium_secondary_reload): Likewise.
24883 (visium_register_move_cost): Likewise.
24884 (visium_memory_move_cost): Likewise.
24885 (prepare_move_operands): Likewise.
24886 (ok_for_simple_move_operands): Likewise.
24887 (ok_for_simple_move_strict_operands): Likewise.
24888 (ok_for_simple_arith_logic_operands): Likewise.
24889 (visium_function_value_1): Likewise.
24890 (rtx_ok_for_offset_p): Likewise.
24891 (visium_legitimize_reload_address): Likewise.
24892 (visium_split_double_move): Likewise.
24893 (visium_expand_copysign): Likewise.
24894 (visium_expand_int_cstore): Likewise.
24895 (visium_expand_fp_cstore): Likewise.
24896 (visium_split_cstore): Likewise.
24897 (visium_select_cc_mode): Likewise.
24898 (visium_split_cbranch): Likewise.
24899 (output_cbranch): Likewise.
24900 (visium_print_operand_address): Likewise.
24901 * expmed.c (flip_storage_order): Likewise.
24902 * expmed.h (emit_cstore): Likewise.
24903 (flip_storage_order): Likewise.
24904 * genrecog.c (validate_pattern): Likewise.
24905 * hsa-gen.c (gen_hsa_addr): Likewise.
24906 * internal-fn.c (expand_arith_overflow): Likewise.
24907 * ira-color.c (allocno_copy_cost_saving): Likewise.
24908 * lra-assigns.c (find_hard_regno_for_1): Likewise.
24909 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
24910 (process_invariant_for_inheritance): Likewise.
24911 * lra-eliminations.c (move_plus_up): Likewise.
24912 * omp-low.c (lower_oacc_reductions): Likewise.
24913 * simplify-rtx.c (simplify_subreg): Likewise.
24914 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
24915 (TARGET_CHKP_BOUND_MODE): Likewise..
24916 * targhooks.c (default_chkp_bound_mode): Likewise.
24917 (default_setup_incoming_vararg_bounds): Likewise.
24918 * targhooks.h (default_chkp_bound_mode): Likewise.
24919 (default_setup_incoming_vararg_bounds): Likewise.
24920 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
24921 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
24922 (have_whole_vector_shift): Likewise.
24923 * tree-vect-stmts.c (vectorizable_load): Likewise.
24924 * doc/tm.texi: Regenerate.
24926 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24928 Graceful degrade if Binutils PR21472 is not available.
24931 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
24932 .rodata in flash test fails.
24933 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
24934 * confgure: Regenerate.
24935 * config.in: Regenerate.
24936 * config/avr/avr.c (avr_asm_named_section)
24937 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
24938 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
24939 (avr_asm_init_sections): Same.
24941 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24943 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
24944 (fma<VH:mode>4_intrinsic): Likewise.
24945 (*fmsub<VCVTF:mode>4): Likewise.
24946 (*fmsub<VH:mode>4_intrinsic): Likewise.
24948 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24951 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
24952 Don't depend on "optimize > 0".
24953 (out_movhi_r_mr, out_movqi_mr_r): Same.
24954 (out_movhi_mr_r, out_movqi_r_mr): Same.
24955 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
24956 io_address_operand on "optimize > 0".
24958 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24960 * tree-loop-distribution.c: Add general explanantion on the pass.
24961 (generate_loops_for_partition): Mark distributed loop.
24962 (pg_add_dependence_edges): New parameter. Handle alias data
24963 dependence specially and record it in the parameter if asked.
24964 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
24965 (init_partition_graph_vertices, add_partition_graph_edge): New.
24966 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
24967 (free_partition_graph_vdata, build_partition_graph): New.
24968 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
24969 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
24970 (data_ref_segment_size, latch_dominated_by_data_ref): New.
24971 (compute_alias_check_pairs, version_loop_by_alias_check): New.
24972 (version_for_distribution_p, finalize_partitions): New.
24973 (distribute_loop): Handle alias data dependence specially. Factor
24974 out loop fusion code as functions and call these functions.
24976 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24978 * tree-loop-distribution.c (classify_partition): New parameter and
24979 better handle reduction statement.
24980 (rdg_build_partitions): Revise comment.
24981 (distribute_loop): Compute statements in all partitions and pass it
24982 to classify_partition.
24984 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24986 * tree-loop-distribution.c (enum partition_type): New.
24987 (struct partition): New field type.
24988 (partition_merge_into): Add parameter. Update partition type.
24989 (data_dep_in_cycle_p, update_type_for_merge): New functions.
24990 (build_rdg_partition_for_vertex): Compute partition type.
24991 (rdg_build_partitions): Dump partition type.
24992 (distribute_loop): Update calls to partition_merge_into.
24994 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24996 * tree-loop-distribution.c (struct ddr_hasher): New.
24997 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
24999 (classify_partition): Call get_data_dependence.
25000 (pg_add_dependence_edges): Ditto.
25001 (distribute_loop): Release data dependence hash table.
25003 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25005 * tree-loop-distribution.c (ref_base_address): Delete.
25006 (similar_memory_accesses): Rename ...
25007 (share_memory_accesses): ... to this. Check if partitions access
25008 the same memory reference.
25009 (distribute_loop): Call share_memory_accesses.
25011 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25013 * tree-loop-distribution.c (struct partition): New field recording
25014 its data reference.
25015 (partition_alloc, partition_free): Init and release data refs.
25016 (partition_merge_into): Merge data refs.
25017 (build_rdg_partition_for_vertex): Collect data refs for partition.
25018 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25020 (distribute_loop): Remve data refs from vertice data of partition
25023 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25025 * tree-loop-distribution.c (params.h): Include header file.
25026 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25027 (datarefs_vec): New global var.
25028 (create_rdg_vertices): Use datarefs_vec directly.
25029 (free_rdg): Don't free data references.
25030 (build_rdg): Update use. Don't free data references.
25031 (distribute_loop): Compute global variable for data references.
25032 Bail out if there are too many data references.
25034 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25036 * tree-loop-distribution.c (loop_nest): New global var.
25037 (build_rdg): Use loop directly, rather than loop nest.
25038 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25040 (distribute_loop): Compute global variable loop nest. Update use.
25042 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25044 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
25045 (partition_merge_into): New parameter. Dump reason for fusion.
25046 (distribute_loop): Update use of partition_merge_into.
25048 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25050 * tree-loop-distribution.c (bb_top_order_index): New.
25051 (bb_top_order_index_size, bb_top_order_cmp): New.
25052 (stmts_from_loop): Use topological order.
25053 (pass_loop_distribution::execute): Compute and release topological
25054 order for basic blocks.
25056 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25058 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25061 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25063 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25064 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25065 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25066 * internal-fn.def (LOOP_DIST_ALIAS): New.
25067 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25068 (fold_loop_internal_call): ... this.
25069 (vect_loop_dist_alias_call): New function.
25070 (set_uid_loop_bbs): Call fold_loop_internal_call.
25071 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25074 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25077 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25078 Require dead FLAGS_REG at the beginning of a peephole.
25080 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25083 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25084 arguments in the call to __builtin_ia32_sbb_u32.
25085 (_subborrow_u64): Swap _X and _Y arguments in the call to
25086 __builtin_ia32_sbb_u64.
25088 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25091 * tree-vrp.c (compare_assert_loc): Turn into a function template
25092 with stable template parameter. Only test if a->e is NULL,
25093 !a->e == !b->e has been verified already. Use e == NULL or
25094 e != NULL instead of e or ! e tests. If stable is true, don't use
25095 iterative_hash_expr, on the other side allow a or b or both NULL
25096 and sort the NULLs last.
25097 (process_assert_insertions): Sort using compare_assert_loc<false>
25098 instead of compare_assert_loc, later sort using
25099 compare_assert_loc<true> before calling process_assert_insertions_for
25100 in a loop. Use break instead of continue once seen NULL pointer.
25102 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25104 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25105 Cortex-R7 and Cortex-R8 processors.
25107 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25109 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25110 uninitialized while src is not.
25112 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
25114 * common/config/arm/arm-common.c: Adjust include path for
25116 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
25117 (arm-cpu.h): Create in build directory. Adjust dependency rules.
25118 (arm-cpu-data.h): Likewise.
25119 (arm-cpu-cdata.h): Likewise.
25120 * config/arm/arm-cpu.h: Delete.
25121 * config/arm/arm-cpu-cdata.h: Delete.
25122 * config/arm/arm-cpu-data.h: Delete.
25124 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
25126 * config/arm/arm-cpus.in (cortex-a55): New.
25127 (cortex-a75): Likewise.
25128 (cortex-a75.cortex-a55): Likewise.
25129 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
25131 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
25132 * config/arm/arm-cpu-cdata.h: Regenerate.
25133 * config/arm/arm-cpu-data.h: Regenerate.
25134 * config/arm/arm-cpu.h: Regenerate.
25135 * config/arm/arm-tables.opt: Regenerate.
25136 * config/arm/arm-tune.md: Regenerate.
25138 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25140 * haifa-sched.c (sched_create_recovery_edges): Update profile.
25142 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25144 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
25147 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
25149 PR tree-optimization/81292
25150 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
25151 full_string_p, also call adjust_related_strinfos if the adjustment
25152 is simple, otherwise invalidate related strinfos.
25154 2017-07-04 Martin Liska <mliska@suse.cz>
25157 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
25158 newly created variable as DECL_IGNORED_P.
25160 2017-07-04 Martin Liska <mliska@suse.cz>
25163 * ipa-inline.c (inline_small_functions):
25164 Use xstrdup_for_dump.
25166 2017-07-04 Tom de Vries <tom@codesourcery.com>
25168 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
25170 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25173 * config/darwin.c (darwin_function_switched_text_sections):
25176 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
25178 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
25180 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25182 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
25184 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25186 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
25187 min_profitable_iters, and th as inclusive lower bounds.
25188 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
25189 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
25190 for min_profitable_iters and min_profitable_estimate.
25191 (vect_transform_loop): Treat th as an inclusive lower bound.
25192 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25194 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25197 * config/darwin.c (darwin_function_switched_text_sections):
25198 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
25199 in two pieces, and suppress the use of buf.
25201 2017-07-03 Nathan Sidwell <nathan@acm.org>
25203 * hash-table.h (hash_table_mod1): Fix indentation.
25205 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25207 PR middle-end/81290
25208 * predict.c (force_edge_cold): Be more careful about propagation
25210 * profile-count.h (profile_probability::guessed,
25211 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
25213 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
25215 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
25217 * doc/invoke.texi (rcpc architecture extension): Document it.
25219 2017-07-03 Richard Biener <rguenther@suse.de>
25221 PR tree-optimization/60510
25222 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
25223 the scalar reduction PHI and use it.
25224 (vectorizable_reduction): Properly guard the single_defuse_cycle
25225 path for non-SLP reduction chains where we cannot use it.
25226 Rework reduc_def/index and vector type deduction. Rework
25227 vector operand gathering during reduction op code-gen.
25228 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
25229 chains dissolve the chain and leave it to non-SLP reduction
25232 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25234 * tree-data-ref.h (dr_alignment): Declare.
25235 * tree-data-ref.c (dr_alignment): New function.
25236 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
25237 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
25239 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
25241 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25243 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
25244 and base_misalignment fields.
25245 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
25246 * tree-data-ref.c: Include builtins.h.
25247 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
25248 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
25249 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
25250 * tree-vect-data-refs.c: Include tree-cfg.h.
25251 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
25252 fields instead of calculating an alignment here.
25253 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
25254 innermost_loop_behavior fields.
25256 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25258 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
25260 (DR_STEP_ALIGNMENT): New macro.
25261 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
25262 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
25263 (create_data_ref): Print it.
25264 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
25265 to tell whether the step preserves vector (mis)alignment.
25266 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25267 Move the check for an integer step and generalise to all INTEGER_CST.
25268 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
25269 Print the outer step alignment.
25271 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25273 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
25274 with offset_alignment.
25275 (DR_ALIGNED_TO): Delete.
25276 (DR_OFFSET_ALIGNMENT): New macro.
25277 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
25278 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
25279 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
25280 (create_data_ref): Likewise.
25281 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25282 (vect_analyze_data_refs): Likewise.
25283 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
25284 creating dummy innermost behavior.
25286 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25288 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
25289 with a "innermost_loop_behavior *" and refeence tree.
25290 * tree-data-ref.c (dr_analyze_innermost): Likewise.
25291 (create_data_ref): Update call accordingly.
25292 * tree-predcom.c (find_looparound_phi): Likewise.
25294 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25296 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
25297 fields with dr_wrt_vec_loop.
25298 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
25299 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
25300 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
25301 (vect_dr_behavior): New function.
25302 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25303 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
25304 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
25305 track whether the step preserves the misalignment.
25306 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25307 Use vect_dr_behavior.
25308 (vect_setup_realignment): Update call accordingly.
25309 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
25310 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
25311 call to vect_create_addr_base_for_vector_ref.
25312 (vect_create_cond_for_align_checks): Likewise.
25313 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
25314 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
25315 (vect_recog_mask_conversion_pattern): Likewise.
25316 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
25317 (new_stmt_vec_info): Remove redundant zeroing.
25319 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25321 * common/config/arm/arm-common.c (arm_be8_option): New function.
25322 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
25323 (ISA_ARMv6): Add isa_bit_be8.
25324 * config/arm/arm.h (arm_be8_option): Add prototype.
25325 (BE8_SPEC_FUNCTION): New define.
25326 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
25327 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
25328 (mlittle-endian): Similarly.
25329 (mbe8, mbe32): New options.
25330 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
25331 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
25333 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25335 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
25337 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25339 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
25340 (cleanup_tree_cfg_bb): Use it.
25341 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
25343 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
25345 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25348 * loop-doloop.c (add_test): Update profile.
25350 2017-07-03 Martin Liska <mliska@suse.cz>
25353 * sanopt.c (rewrite_usage_of_param): New function.
25354 (sanitize_rewrite_addressable_params): Likewise.
25355 (pass_sanopt::execute): Call rewrite_usage_of_param.
25357 2017-07-03 Richard Biener <rguenther@suse.de>
25359 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
25360 back to using VIEW_CONVERT_EXPR.
25362 2017-07-03 Martin Liska <mliska@suse.cz>
25365 * doc/extend.texi: Document when a resolver function is
25366 generated for target_clones.
25368 2017-07-03 Martin Liska <mliska@suse.cz>
25370 * asan.c (asan_emit_stack_protection): Unpoison just red zones
25371 and shadow memory of auto variables which are subject of
25372 use-after-scope sanitization.
25373 (asan_expand_mark_ifn): Add do set only when is_poison.
25375 2016-07-03 Richard Biener <rguenther@suse.de>
25377 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
25379 (vect_force_simple_reduction): Record reduction def -> phi mapping.
25380 (vectorizable_reduction): Perform reduction PHI creation when
25381 visiting a reduction PHI and adjust and simplify code generation
25382 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
25383 (vect_transform_loop): Visit reduction PHIs.
25384 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
25385 defs into the SLP tree.
25386 (vect_build_slp_tree): Reduction defs terminate the recursion.
25387 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
25389 (vect_get_vec_defs_for_stmt_copy): Export.
25390 (vect_get_vec_defs): Likewise.
25391 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
25393 (vect_get_vec_defs_for_stmt_copy): Declare.
25394 (vect_get_vec_defs): Likewise.
25396 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25398 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
25399 parameter with a "loop" parameter and use it instead of the
25400 loop containing DR_STMT. Don't check simple_iv when doing
25401 BB analysis. Describe the two analysis modes in the comment.
25403 2017-07-03 Tom de Vries <tom@codesourcery.com>
25405 PR tree-optimization/69468
25406 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
25407 (find_same_succ_bb): Handle ignore_edge_flags.
25409 2017-07-03 Tom de Vries <tom@codesourcery.com>
25411 PR tree-optimization/81192
25412 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
25414 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
25416 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
25418 2017-07-03 Tom de Vries <tom@codesourcery.com>
25420 PR tree-optimization/81192
25421 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
25422 BB_SAME_SUCC (bb) == NULL.
25424 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25426 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
25429 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25431 * dumpfile.c: Include profile-count.h
25432 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
25434 (insert_cond_bb): Update profile.
25435 * tree-cfg.h (insert_cond_bb): Update prototype.
25436 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
25437 * tree-dump.c: Do not include tree-cfg.
25439 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25441 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
25443 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25445 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
25448 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25450 * tree-complex.c (expand_complex_div_wide): update profile.
25452 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25453 Alan Hayward <alan.hayward@arm.com>
25454 David Sherwood <david.sherwood@arm.com>
25456 * Makefile.in (MACHMODE_H): Remove insn-modes.h
25457 (CORETYPES_H): New define.
25458 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
25459 (insn-modes-inline.h, s-modes-inline-h): New rules.
25460 (generated_files): Add insn-modes-inline.h.
25461 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
25462 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
25463 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
25464 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
25465 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
25466 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
25467 (build/gencodes.o, build/genconditions.o): Likewise.
25468 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
25469 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
25470 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
25471 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
25472 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
25473 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
25474 * coretypes.h: Include everything up to real.h for generators.
25475 Include insn-modes.h first. Include wide-int-print.h after
25476 wide-int.h. Include insn-modes-inline.h and then machmode.h.
25477 * machmode.h: Don't include insn-modes.h here.
25478 * function-tests.c: Remove includes of signop.h, machmode.h,
25479 double-int.h and wide-int.h.
25481 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
25483 * optc-save-gen.awk: Likewise.
25484 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
25485 * godump.c: Remove include of wide-int-print.h.
25486 * pretty-print.h: Likewise.
25487 * wide-int-print.cc: Likewise.
25488 * wide-int.cc: Likewise.
25489 * hash-map-tests.c: Remove include of signop.h.
25490 * hash-set-tests.c: Likewise.
25491 * rtl-tests.c: Likewise.
25492 * mkconfig.sh: Remove include of machmode.h.
25493 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
25495 (emit_insn_modes_inline_h): ...this new function. Emit the code
25496 into an insn-modes-inline.h header file, adding appropriate
25497 include guards and end comments.
25498 (emit_insn_modes_c_header): Remove include of machmode.h.
25499 (emit_min_insn_modes_c_header): Include coretypes.h rather than
25501 (main): Handle -i flag and call emit_insn_modes_inline_h when
25504 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25506 * tree-ssa-strlen.c (strinfo): Rename the length field to
25507 nonzero_chars. Add a full_string_p field.
25508 (compare_nonzero_chars, zero_length_string_p): New functions.
25509 (get_addr_stridx): Add an offset_out parameter.
25510 Use compare_nonzero_chars.
25511 (get_stridx): Update accordingly. Use compare_nonzero_chars.
25512 (new_strinfo): Update after above changes to strinfo.
25513 (set_endptr_and_length): Set full_string_p.
25514 (get_string_length): Update after above changes to strinfo.
25515 (unshare_strinfo): Update call to new_strinfo.
25516 (maybe_invalidate): Likewise.
25517 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
25518 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
25519 as a uhwi instead of an shwi. Update after above changes to
25520 strinfo and new_strinfo.
25521 (zero_length_string): Assert that chainsi contains full strings.
25522 Use zero_length_string_p. Update call to new_strinfo.
25523 (adjust_related_strinfos): Update after above changes to strinfo.
25524 Copy full_string_p from origsi.
25525 (adjust_last_stmt): Use zero_length_string_p.
25526 (handle_builtin_strlen): Update after above changes to strinfo and
25527 new_strinfo. Install the lhs as the string length if the previous
25528 entry didn't describe a full string.
25529 (handle_builtin_strchr): Update after above changes to strinfo
25531 (handle_builtin_strcpy): Likewise.
25532 (handle_builtin_strcat): Likewise.
25533 (handle_builtin_malloc): Likewise.
25534 (handle_pointer_plus): Likewise.
25535 (handle_builtin_memcpy): Likewise. Track nonzero characters
25536 that aren't necessarily followed by a nul terminator.
25537 (handle_char_store): Likewise.
25539 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25541 PR tree-optimization/80769
25542 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
25543 for malloc and calloc. Document the new invariant that all related
25544 strinfos have delayed lengths or none do.
25545 (verify_related_strinfos): Move earlier in file.
25546 (set_endptr_and_length): New function, split out from...
25547 (get_string_length): ...here. Also set the lengths of related
25549 (zero_length_string): Assert that chainsi has known (rather than
25551 (adjust_related_strinfos): Likewise.
25553 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25555 PR tree-optimization/81136
25556 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
25557 assert that two references with the same misalignment have the same
25558 compile-time misalignment if those compile-time misalignments
25561 2017-07-01 Andi Kleen <ak@linux.intel.com>
25563 * print-tree.c (print_node): Print all attributes.
25565 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25567 * cfg.c (scale_bbs_frequencies): New function.
25568 * cfg.h (scale_bbs_frequencies): Declare it.
25569 * cfgloopanal.c (single_likely_exit): Cleanup.
25570 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
25572 (scale_loop_profile): Likewise.
25573 (loop_version): Likewise.
25574 (create_empty_loop_on_edge): Update.
25575 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
25576 scale_loop_frequencies, scale_loop_profile, loopify,
25577 loop_version): Update prototypes.
25578 * modulo-sched.c (sms_schedule): Update.
25579 * predict.c (unlikely_executed_edge_p): Also check probability.
25580 (probably_never_executed_edge_p): Fix typo.
25581 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25582 * tree-parloops.c (gen_parallel_loop): Update.
25583 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
25584 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25585 * tree-ssa-loop-split.c (split_loop): Update.
25586 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25587 * tree-vect-loop-manip.c (vect_do_peeling): Update.
25588 (vect_loop_versioning): Update.
25589 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25591 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25593 * trans-mem.c (split_bb_make_tm_edge): Update profile.
25595 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25597 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
25598 to keep profile consistent.
25600 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25602 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
25603 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
25604 * profile-count.h (max_safe_multiplier): Make unsigned.
25605 (profile_count::guessed_zero): New.
25607 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25609 * bb-reorder.c (fix_up_crossing_landing_pad,
25610 fix_crossing_conditional_branches): Use make_single_succ_edge
25611 to keep profile consistent.
25613 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25615 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
25618 2017-07-01 Jakub Jelinek <jakub@redhat.com>
25621 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
25622 the right scopes, make sure cond_jump isn't preserved between multiple
25623 iterations. Search for fallthru edge whenever there are 3+ edges and
25624 use find_fallthru_edge for it.
25626 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25628 Patch by Alexander Monakov <amonakov@ispras.ru>
25629 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25630 probabilities consistently.
25632 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25634 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25635 emit_cmp_and_jump_insns.
25637 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25640 * tree-inline.c (expand_call_inline): Combine profile statuses.
25642 2017-06-30 Andrew Pinski <apinski@cavium.com>
25644 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25645 fold_stmt returned true.
25647 2017-06-30 Nathan Sidwell <nathan@acm.org>
25649 * ggc.h (empty_string): Delete.
25650 * cfgexpand.c (expand_asm_stmt): Use plain "".
25651 * optabs.c (expand_asm_memory_barrier): Likewise.
25652 * stringpool.c (empty_string): Delete.
25653 (digit_vector, digit_string): Delete.
25654 (ggc_alloc_string): Use plain "", don't optimize single digit
25655 strings. Use ggc_alloc_atomic.
25657 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25659 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25660 comparison set and one other set, use the cost of the non-comparison
25663 2017-06-30 Nathan Sidwell <nathan@acm.org>
25665 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25668 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25670 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25671 loops. Remove now unneeded calls to gimple_switch_set_label() that
25672 just set removed labels to NULL_TREE.
25674 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25676 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25677 (set_range_info): ...here. Only call set_range_info_raw if domain
25679 (set_nonzero_bits): Call set_range_info_raw.
25680 * tree-ssanames.h (set_range_info_raw): New.
25682 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25685 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25686 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25687 of nonimmediate_operand and <store_mask_constraint> instead of m
25688 for the input operand. For V8FI iterator, always split if input
25689 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25690 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25691 <store_mask_predicate> instead of register_operand and
25692 <store_mask_constraint> instead of v for the input operand. Make
25693 sure both operands aren't MEMs for if not <mask_applied>.
25695 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25697 * lto-wrapper.c (copy_file) Close both file descriptors before
25700 2017-06-30 Martin Liska <mliska@suse.cz>
25703 * multiple_target.c (create_dispatcher_calls): Make ifunc
25704 also for function that don't have calls or are not referenced.
25706 2017-06-30 Richard Biener <rguenther@suse.de>
25708 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25709 analyze the first scalar stmt. Move vector type computation
25710 for the BB case here from ...
25711 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25712 live operation processing in the SLP case properly.
25714 2017-06-30 Richard Biener <rguenther@suse.de>
25716 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25718 2017-06-30 Martin Liska <mliska@suse.cz>
25721 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25722 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25724 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25726 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25729 2017-06-30 Richard Biener <rguenther@suse.de>
25731 PR tree-optimization/81249
25732 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25733 condition reduction result to original scalar type.
25735 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25737 * profile-count.h (enum profile_quality): Fix typos and whitespace
25740 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25742 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25743 type for branch probabilities.
25745 2017-06-29 Julian Brown <julian@codesourcery.com>
25746 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25748 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25749 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25750 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25751 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25753 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25755 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25756 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25757 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25758 CC usage from generic code to here.
25759 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25760 CC usage into the target macros.
25762 2017-06-29 Maya Rashish <coypu@sdf.org>
25764 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25767 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25769 * arm/arm-builtins.c: Include profile-count.h
25770 * except.c (sjlj_emit_function_enter): Use
25771 profile_probability::unlikely.
25773 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25775 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25776 and tocrel_offset be pointer args rather than implicitly using
25778 (legitimate_constant_pool_address_p, rs6000_emit_move,
25779 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25780 tocrel_offset and use in toc_relative_expr_p call.
25781 (print_operand, print_operand_address): Use static tocrel_base_oac
25782 and tocrel_offset_oac.
25783 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25786 2017-06-29 Maya Rashish <coypu@sdf.org>
25788 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25790 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25792 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25793 objects, take into account only the alignment of 'op0' and 'mode1' if
25796 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25798 * ccmp.c (ccmp_tree_comparison_p): New function.
25799 (ccmp_candidate_p): Update to use above function.
25800 (get_compare_parts): New function.
25801 (expand_ccmp_next): Update to use new functions.
25802 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25804 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25805 take mode as argument.
25806 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25807 * expr.c (expand_expr_real_1): Pass mode as argument.
25809 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25811 * combine.c (combine_instructions): Print insns to dump_file, together
25814 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25816 * asan.c (asan_emit_stack_protection): Update.
25817 (create_cond_insert_point): Update.
25818 * auto-profile.c (afdo_propagate_circuit): Update.
25819 * basic-block.h (struct edge_def): Turn probability to
25820 profile_probability.
25821 (EDGE_FREQUENCY): Update.
25822 * bb-reorder.c (find_traces_1_round): Update.
25823 (better_edge_p): Update.
25824 (sanitize_hot_paths): Update.
25825 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25826 (make_single_succ_edge): Update.
25827 (check_bb_profile): Update.
25828 (dump_edge_info): Update.
25829 (update_bb_profile_for_threading): Update.
25830 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25831 probabilitycount to 0.
25832 * cfgbuild.c (compute_outgoing_frequencies): Update.
25833 * cfgcleanup.c (try_forward_edges): Update.
25834 (outgoing_edges_match): Update.
25835 (try_crossjump_to_edge): Update.
25836 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25837 (expand_gimple_tailcall): Update.
25838 (construct_init_block): Use make_single_succ_edge.
25839 (construct_exit_block): Use make_single_succ_edge.
25840 * cfghooks.c (verify_flow_info): Update.
25841 (redirect_edge_succ_nodup): Update.
25842 (split_edge): Update.
25843 (account_profile_record): Update.
25844 * cfgloopanal.c (single_likely_exit): Update.
25845 * cfgloopmanip.c (scale_loop_profile): Update.
25846 (set_zero_probability): Remove.
25847 (duplicate_loop_to_header_edge): Update.
25848 * cfgloopmanip.h (loop_version): Update prototype.
25849 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25850 (force_nonfallthru_and_redirect): Update.
25851 (update_br_prob_note): Update.
25852 (rtl_verify_edges): Update.
25853 (purge_dead_edges): Update.
25854 (rtl_lv_add_condition_to_bb): Update.
25855 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25856 * cgraphunit.c (init_lowered_empty_function): Update.
25857 (cgraph_node::expand_thunk): Update.
25858 * cilk-common.c: Include profile-count.h
25859 * dojump.c (inv): Remove.
25860 (jumpifnot): Update.
25861 (jumpifnot_1): Update.
25862 (do_jump_1): Update.
25864 (do_jump_by_parts_greater_rtx): Update.
25865 (do_compare_rtx_and_jump): Update.
25866 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25867 do_jump_1. do_compare_rtx_and_jump): Update prototype.
25868 * dwarf2cfi.c: Include profile-count.h
25869 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
25870 (sjlj_emit_dispatch_table): Likewise.
25871 * explow.c: Include profile-count.h
25872 * expmed.c (emit_store_flag_force): Update.
25873 (do_cmp_and_jump): Update.
25874 * expr.c (compare_by_pieces_d::generate): Update.
25875 (compare_by_pieces_d::finish_mode): Update.
25876 (emit_block_move_via_loop): Update.
25877 (store_expr_with_bounds): Update.
25878 (store_constructor): Update.
25879 (expand_expr_real_2): Update.
25880 (expand_expr_real_1): Update.
25881 * expr.h (try_casesi, try_tablejump): Update prototypes.
25882 * gimple-pretty-print.c (dump_probability): Update.
25883 (dump_profile): New.
25884 (dump_gimple_label): Update.
25885 (dump_gimple_bb_header): Update.
25886 * graph.c (draw_cfg_node_succ_edges): Update.
25887 * hsa-gen.c (convert_switch_statements): Update.
25888 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
25889 (find_if_case_1): Update.
25890 (find_if_case_2): Update.
25891 * internal-fn.c (expand_arith_overflow_result_store): Update.
25892 (expand_addsub_overflow): Update.
25893 (expand_neg_overflow): Update.
25894 (expand_mul_overflow): Update.
25895 (expand_vector_ubsan_overflow): Update.
25896 * ipa-cp.c (good_cloning_opportunity_p): Update.
25897 * ipa-split.c (split_function): Use make_single_succ_edge.
25898 * ipa-utils.c (ipa_merge_profiles): Update.
25899 * loop-doloop.c (add_test): Update.
25900 (doloop_modify): Update.
25901 * loop-unroll.c (compare_and_jump_seq): Update.
25902 (unroll_loop_runtime_iterations): Update.
25903 * lra-constraints.c (lra_inheritance): Update.
25904 * lto-streamer-in.c (input_cfg): Update.
25905 * lto-streamer-out.c (output_cfg): Update.
25906 * mcf.c (adjust_cfg_counts): Update.
25907 * modulo-sched.c (sms_schedule): Update.
25908 * omp-expand.c (expand_omp_for_init_counts): Update.
25909 (extract_omp_for_update_vars): Update.
25910 (expand_omp_ordered_sink): Update.
25911 (expand_omp_for_ordered_loops): Update.
25912 (expand_omp_for_generic): Update.
25913 (expand_omp_for_static_nochunk): Update.
25914 (expand_omp_for_static_chunk): Update.
25915 (expand_cilk_for): Update.
25916 (expand_omp_simd): Update.
25917 (expand_omp_taskloop_for_outer): Update.
25918 (expand_omp_taskloop_for_inner): Update.
25919 * omp-simd-clone.c (simd_clone_adjust): Update.
25920 * optabs.c (expand_doubleword_shift): Update.
25921 (expand_abs): Update.
25922 (emit_cmp_and_jump_insn_1): Update.
25923 (expand_compare_and_swap_loop): Update.
25924 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
25925 * predict.c (predictable_edge_p): Update.
25926 (edge_probability_reliable_p): Update.
25927 (set_even_probabilities): Update.
25928 (combine_predictions_for_insn): Update.
25929 (combine_predictions_for_bb): Update.
25930 (propagate_freq): Update.
25931 (estimate_bb_frequencies): Update.
25932 (force_edge_cold): Update.
25933 * profile-count.c (profile_count::dump): Add missing space into dump.
25934 (profile_count::debug): Add newline.
25935 (profile_count::differs_from_p): Explicitly convert to unsigned.
25936 (profile_count::stream_in): Update.
25937 (profile_probability::dump): New member function.
25938 (profile_probability::debug): New member function.
25939 (profile_probability::differs_from_p): New member function.
25940 (profile_probability::differs_lot_from_p): New member function.
25941 (profile_probability::stream_in): New member function.
25942 (profile_probability::stream_out): New member function.
25943 * profile-count.h (profile_count_quality): Rename to ...
25944 (profile_quality): ... this one.
25945 (profile_probability): New.
25946 (profile_count): Update.
25947 * profile.c (compute_branch_probabilities): Update.
25948 * recog.c (peep2_attempt): Update.
25949 * sched-ebb.c (schedule_ebbs): Update.
25950 * sched-rgn.c (find_single_block_region): Update.
25951 (compute_dom_prob_ps): Update.
25952 (schedule_region): Update.
25953 * sel-sched-ir.c (compute_succs_info): Update.
25954 * stmt.c (struct case_node): Update.
25955 (do_jump_if_equal): Update.
25956 (get_outgoing_edge_probs): Update.
25957 (conditional_probability): Update.
25958 (emit_case_dispatch_table): Update.
25959 (expand_case): Update.
25960 (expand_sjlj_dispatch_table): Update.
25961 (emit_case_nodes): Update.
25962 * targhooks.c: Update.
25963 * tracer.c (better_p): Update.
25964 (find_best_successor): Update.
25965 * trans-mem.c (expand_transaction): Update.
25966 * tree-call-cdce.c: Update.
25967 * tree-cfg.c (gimple_split_edge): Upate.
25968 (move_sese_region_to_fn): Upate.
25969 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
25970 * tree-eh.c (lower_resx): Upate.
25971 (cleanup_empty_eh_move_lp): Upate.
25972 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25973 * tree-inline.c (copy_edges_for_bb): Update.
25974 (copy_cfg_body): Update.
25975 * tree-parloops.c (gen_parallel_loop): Update.
25976 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
25977 (gimple_gen_time_profiler): Update.
25978 * tree-ssa-dce.c (remove_dead_stmt): Update.
25979 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25980 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
25981 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
25982 (unloop_loops): Update.
25983 (try_peel_loop): Update.
25984 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25985 * tree-ssa-loop-split.c (connect_loops): Update.
25986 (split_loop): Update.
25987 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25988 (hoist_guard): Update.
25989 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
25990 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
25991 (value_replacement): Update.
25992 * tree-ssa-reassoc.c (branch_fixup): Update.
25993 * tree-ssa-tail-merge.c (replace_block_by): Update.
25994 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
25995 (create_edge_and_update_destination_phis): Update.
25996 (compute_path_counts): Update.
25997 (recompute_probabilities): Update.
25998 (update_joiner_offpath_counts): Update.
25999 (freqs_to_counts_path): Update.
26000 (duplicate_thread_path): Update.
26001 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
26002 (struct switch_conv_info): Update.
26003 (gen_inbound_check): Update.
26004 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26005 (vect_do_peeling): Update.
26006 (vect_loop_versioning): Update.
26007 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26008 (optimize_mask_stores): Update.
26009 * ubsan.c (ubsan_expand_null_ifn): Update.
26010 * value-prof.c (gimple_divmod_fixed_value): Update.
26011 (gimple_divmod_fixed_value_transform): Update.
26012 (gimple_mod_pow2): Update.
26013 (gimple_mod_pow2_value_transform): Update.
26014 (gimple_mod_subtract): Update.
26015 (gimple_mod_subtract_transform): Update.
26016 (gimple_ic): Update.
26017 (gimple_stringop_fixed_value): Update.
26018 (gimple_stringops_transform): Update.
26019 * value-prof.h: Update.
26021 2017-06-29 Carl Love <cel@us.ibm.com>
26023 * config/rs6000/rs6000-c.c: Add support for built-in functions
26024 vector signed int vec_signed (vector float);
26025 vector signed long long vec_signed (vector double);
26026 vector signed int vec_signed2 (vector double, vector double);
26027 vector signed int vec_signede (vector double);
26028 vector signed int vec_signedo (vector double);
26029 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26030 instruction generator.
26031 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26032 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26033 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26035 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26036 vunsignede_v2df): Add define_expands.
26037 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26038 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26039 VEC_UNSIGNEDO): Add definitions.
26040 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26041 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26042 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
26043 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
26044 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
26045 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26046 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26047 vec_unsignede, vec_unsignedo): Add builtin defines.
26048 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26050 * doc/extend.texi: Update the built-in documentation file for the
26051 new built-in functions.
26053 2017-06-29 Richard Biener <rguenther@suse.de>
26055 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26056 reduction chains to LOOP_VINFO_REDUCTIONS.
26057 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26058 SLP reductions after processing reduction chains.
26060 2017-06-29 Nathan Sidwell <nathan@acm.org>
26062 * builtins.c (fold_builtin_FUNCTION): Use
26063 lang_hooks.decl_printable_name.
26065 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26067 PR middle-end/81194
26068 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26069 with only one label.
26070 * stmt.c (expand_case): Assert NCASES is greater than one.
26072 2017-06-29 Richard Biener <rguenther@suse.de>
26074 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26076 (group_case_labels): Likewise.
26077 (find_taken_edge): Push sanity checking on val to workers...
26078 (find_taken_edge_cond_expr): ... here
26079 (find_taken_edge_switch_expr): ... and here, handle cases
26080 with just a default label.
26081 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26082 (group_case_labels): Likewise.
26083 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26084 group_case_labels does anything cleanup the CFG again.
26086 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26088 PR tree-optimization/81196
26089 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26090 exit condition comparing two IVs.
26092 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26094 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26095 profile to the dummy entry at the end of the list of architectures.
26096 * config/arm/arm-cpu-cdata.h: Regenerated.
26098 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26099 Michael Collison <michael.collison@arm.com>
26102 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26104 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26105 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26106 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26107 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26108 of shift when the shift amount is masked with constant equal to
26109 the size of the mode.
26110 * config/aarch64/predicates.md (subreg_lowpart_operator): New
26113 2017-06-29 Martin Liska <mliska@suse.cz>
26115 * config/i386/i386.opt: Change range from [1,5] to [0,5].
26117 2017-06-29 Yury Gribov <tetra2005@gmail.com>
26120 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
26123 (edge_growth_cache_entry::edge_growth_cache_entry): New
26125 (reset_edge_growth_cache): Update to use constructor.
26127 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26129 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
26130 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26131 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26133 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
26135 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
26136 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
26138 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
26140 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
26141 (*-linux-uclibc*): Add t-uclibc tmake_file.
26142 * config/t-musl: New.
26143 * config/t-uclibc: New.
26145 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
26147 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
26149 (gen_comm_data): Emit architectural setting of arch_prof.
26150 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
26152 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
26153 (armv8-m.base, armv8-m.main): Likewise.
26154 * arm-protos.h (arm_build_target): Add profile field.
26155 (arch_option): Likewise.
26156 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
26158 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
26159 arm_active_target.profile.
26161 2017-06-28 Richard Biener <rguenther@suse.de>
26163 PR middle-end/81227
26164 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
26165 TYPE_OVERFLOW_WRAPS.
26166 * match.pd (negate_expr_p): Likewise.
26167 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
26168 fold_build2, not fold_binary.
26170 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26172 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
26173 Convert memory address to Pmode.
26174 (aarch64_print_operand): Assert MEM operands are always Pmode.
26176 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26179 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
26180 Remove redundant if.
26181 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
26182 * config/arm/aarch-common-protos.h
26183 (aarch_forward_to_shift_is_not_shifted_re): Remove.
26184 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
26186 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
26189 * multiple_target.c (create_dispatcher_calls): Set the default
26190 clone to be static, not public.
26192 2017-06-28 Richard Biener <rguenther@suse.de>
26194 * tree-vect-loop.c (vectorizable_reduction): Move special
26195 cond reduction IV var creation ...
26196 (vect_create_epilog_for_reduction): ... here. Remove induction_index
26197 parameter. Use STMT_VINFO_VECTYPE.
26198 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
26201 2017-06-28 Martin Liska <mliska@suse.cz>
26204 * ipa-visibility.c (non_local_p): Handle visibility.
26206 2017-06-28 Martin Liska <mliska@suse.cz>
26209 * common.opt: Add IntegerRange to various options.
26210 * opt-functions.awk (integer_range_info): New function.
26211 * optc-gen.awk: Add integer_range_info to cl_options struct.
26212 * opts-common.c (decode_cmdline_option): Handle
26213 CL_ERR_INT_RANGE_ARG.
26214 (cmdline_handle_error): Likewise.
26215 * opts.c (print_filtered_help): Show valid interval in
26216 when --help is provided.
26217 * opts.h (struct cl_option): Add range_min and range_max fields.
26218 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
26220 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
26222 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
26223 (x * C EQ/NE y * C): New transformation.
26225 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
26227 * genmultilib (combination_space): Accept '+' in option names.
26229 2017-06-28 Martin Liska <mliska@suse.cz>
26232 * asan.c (instrument_derefs): Bail out inner references
26233 that are hard register variables.
26235 2017-06-28 Jakub Jelinek <jakub@redhat.com>
26238 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
26239 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
26241 2017-06-28 Richard Biener <rguenther@suse.de>
26243 * tree-vectorizer.h (vect_get_vec_defs): Remove.
26244 (vect_get_slp_defs): Adjust.
26245 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
26247 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
26249 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
26250 get_initial_defs_for_reduction instead of vect_get_vec_defs.
26251 (vectorizable_reduction): Adjust.
26252 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
26254 (vect_get_slp_defs): Likewise.
26255 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
26256 (vectorizable_bswap): Adjust.
26257 (vectorizable_call): Likewise.
26258 (vectorizable_conversion): Likewise.
26259 (vectorizable_assignment): Likewise.
26260 (vectorizable_shift): Likewise.
26261 (vectorizable_operation): Likewise.
26262 (vectorizable_store): Likewise.
26263 (vectorizable_condition): Likewise.
26264 (vectorizable_comparison): Likewise.
26266 2017-06-28 Michael Collison <michael.collison@arm.com>
26269 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
26271 (arm_gen_movmemqi): Removed unused variable 'i'.
26272 Convert 'for' loop into 'while' loop.
26273 (arm_expand_prologue): Remove last unnecessary set of insn.
26274 (thumb_pop): Remove unused variable 'pushed_words'.
26275 (thumb_exit): Remove last unnecessary set of regs_to_pop.
26277 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26279 * config/s390/predicates.md: Use s390_rel_address_ok_p.
26280 * config/s390/s390-protos.h: Add prototype of
26281 s390_rel_address_ok_p.
26282 * config/s390/s390.c (s390_got_symbol): New function.
26283 (s390_rel_address_ok_p): New function.
26284 (legitimize_pic_address): Use s390_rel_address_ok_p.
26285 (s390_load_got): Use s390_got_symbol.
26286 (s390_option_override): Issue error if
26287 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
26288 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
26290 * config/s390/s390.opt: New option mpic-data-is-text-relative.
26292 2017-06-27 Andrew Pinski <apinski@cavium.com>
26294 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
26295 (X * copysign (1.0, X)): New pattern.
26296 (X * copysign (1.0, -X)): New pattern.
26297 (copysign (-1.0, CST)): New pattern.
26299 2017-06-27 Joseph Myers <joseph@codesourcery.com>
26301 * genmultilib (combination_space): Remove variable.
26302 Validate reuse rules against regular expression for any sequence
26303 of multilib options in any order.
26305 2017-06-27 Michael Collison <michael.collison@arm.com>
26307 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
26308 call aarch64_split_simd_combine.
26309 * (aarch64_combine_internal<mode>): Delete pattern.
26310 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
26311 Allow register and subreg operands.
26313 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26315 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
26316 specific need, just fallback on defaults.
26317 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
26319 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26320 Olivier Hainque <hainque@adacore.com>
26322 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
26324 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
26325 targets. Pick a default if no particular attempt applied.
26326 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
26329 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26331 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
26332 (x86_64-wrs-vxworks7): Likewise.
26334 2017-06-27 Marek Polacek <polacek@redhat.com>
26337 * ubsan.c (instrument_null): Check get_base_address's result for null.
26339 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26341 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
26343 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26345 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
26346 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
26347 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
26348 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
26349 New function types.
26350 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
26351 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
26352 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
26353 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
26354 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
26355 BUILT_IN_FEUPDATEENV): New builtins.
26356 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
26357 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
26358 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
26359 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
26361 (builtin_structptr_types): Adjust size.
26362 * tree.c (builtin_structptr_types): Add four entries.
26364 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26365 Olivier Hainque <hainque@adacore.com>
26367 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
26368 (TLS_SYM): New local macro, forcing reference to __tls__ on
26369 link command lines for VxWorks 7 RTPs, triggering initialization
26371 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
26372 OS features TLS support, true for RTPs on VxWorks 7.
26373 * config/vxworks.c (vxworks_override_options): Setup emutls
26376 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26378 * predict.c (test_prediction_value_range): Use -1U instead of -1
26379 to avoid narrowing conversion warning.
26380 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
26381 to avoid narrowing conversion warning.
26382 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
26384 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
26386 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26388 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
26389 64bit configurations.
26390 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
26391 (SIZE_TYPE): Likewise.
26392 * config/vxworks.c (vxworks_emutls_var_fields): Use
26393 long_unsigned_type_node instead of unsigned_type_node as the offset
26394 field type, which is "pointer" mode in emutls.c.
26396 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26399 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
26401 PR middle-end/81207
26402 * gimple-fold.c (replace_call_with_call_and_fold): Handle
26403 gimple_vuse copying separately from gimple_vdef copying.
26405 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26407 * value-prof.c (free_hist): Remove call to memset and the enclosing if
26410 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26411 Olivier Hainque <hainque@adacore.com>
26413 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
26414 for all vxworks7 targets.
26415 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
26416 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
26417 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
26418 variations for VX6/VX7 and 32/64bits later on in ...
26419 (VXWORKS_LIB_SPEC): Leverage new macros.
26420 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
26421 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
26423 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26425 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
26426 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
26428 2017-06-26 Carl Love <cel@us.ibm.com>
26430 * config/rs6000/rs6000-c.c: Add support for built-in functions
26431 vector bool char vec_reve (vector bool char);
26432 vector signed char vec_reve (vector signed char);
26433 vector unsigned char vec_reve (vector unsigned char);
26434 vector bool int vec_reve (vector bool int);
26435 vector signed int vec_reve (vector signed int);
26436 vector unsigned int vec_reve (vector unsigned int);
26437 vector bool long long vec_reve (vector bool long long);
26438 vector signed long long vec_reve (vector signed long long);
26439 vector unsigned long long vec_reve (vector unsigned long long);
26440 vector bool short vec_reve (vector bool short);
26441 vector signed short vec_reve (vector signed short);
26442 vector double vec_reve (vector double);
26443 vector float vec_reve (vector float);
26444 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
26445 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
26446 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
26447 (altivec_vreve): New pattern.
26448 * config/rs6000/altivec.h (vec_reve): New define.
26449 * doc/extend.texi (vec_rev): Update the built-in documentation file
26450 for the new built-in functions.
26452 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26454 PR tree-optimization/71815
26455 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
26457 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
26459 (slsr_process_phi): Likewise.
26460 (replace_uncond_cands_and_profitable_phis): Don't replace a
26461 multiply candidate with a stride of 1 (copy or cast).
26462 (phi_incr_cost): Call uses_consumed_by_stmt rather than
26464 (lowest_cost_path): Likewise.
26465 (total_savings): Likewise.
26467 2017-06-26 Richard Biener <rguenther@suse.de>
26470 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
26471 Use def_builtin_pure for all gather builtins.
26473 2017-06-26 Richard Biener <rguenther@suse.de>
26475 PR tree-optimization/81203
26476 * tree-tailcall.c (find_tail_calls): Do not move stmts into
26477 non-dominating BBs.
26479 2017-06-26 Marek Polacek <polacek@redhat.com>
26482 * doc/invoke.texi: Document -Wmultistatement-macros.
26484 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
26486 * doc/sourcebuild.texi (ARM-specific attributes): Document new
26487 arm_neon_ok_no_float_abi effective target.
26489 2017-06-26 Richard Biener <rguenther@suse.de>
26491 PR tree-optimization/80928
26492 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
26493 (copy_bbs): Set BB_DUPLICATED flag early.
26494 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
26496 (execute_on_shrinking_pred): Likewise.
26497 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
26498 BB_DUPLICATED blocks.
26499 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
26500 iterate over all PHIs considering removal of *gsi.
26502 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
26504 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
26507 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26509 * config/rs6000/rs6000-string.c: (expand_block_clear,
26510 do_load_for_compare, select_block_compare_mode,
26511 compute_current_alignment, expand_block_compare,
26512 expand_strncmp_align_check, expand_strn_compare,
26513 expand_block_move, rs6000_output_load_multiple)
26514 Move functions related to string/block move/compare
26515 to a separate file.
26516 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
26517 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
26518 for this function which is now used in two files.
26519 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
26520 * config.gcc: Add rs6000-string.o to extra_objs for
26521 targets powerpc*-*-* and rs6000*-*-*.
26523 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
26526 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
26527 32-bit, since indexed is not valid for DImode.
26528 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
26529 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
26530 (define_peephole2 for Altivec d-form load): Add 32-bit support.
26531 (define_peephole2 for Altivec d-form store): Likewise.
26534 * multiple_target.c (create_dispatcher_calls): Only create the
26535 dispatcher call if the function is the default clone of a
26536 versioned function.
26538 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
26540 PR middle-end/80902
26541 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
26542 a call, force the call to not be a tail call.
26544 2017-06-23 Jeff Law <law@redhat.com>
26546 * doc/contrib.texi: Add entry for Steven Pemberton's work on
26549 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
26551 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
26552 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
26553 handling for early expansion of vector shifts (sl,sr,sra,rl).
26554 (builtin_function_type): Add vector shift right instructions
26555 to the unsigned argument list.
26557 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
26559 rtl-optimizatoin/79286
26560 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
26561 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
26562 trap. PIC register plus a const unspec without offset can never trap.
26564 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
26566 * tree.h (builtin_structptr_type): New type.
26567 (builtin_structptr_types): Declare new array.
26568 * tree.c (builtin_structptr_types): New array.
26569 (free_lang_data, build_common_tree_nodes): Use it.
26571 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
26574 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
26577 2017-06-22 Matt Turner <mattst88@gmail.com>
26579 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
26580 Lake models to skylake case. Assume skylake for unknown
26581 models with clflushopt.
26583 2017-06-22 Jeff Law <law@redhat.com>
26585 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
26586 frame sizes that do not satisfy aarch64_uimm12_shift.
26588 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
26590 * profile-count.h (apply_probability,
26591 apply_scale, probability_in): Fix checks for zero.
26593 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26595 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
26596 * doc/cppdiropts.texi (-I @var{dir}): Document it.
26598 2016-06-22 Richard Biener <rguenther@suse.de>
26600 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26601 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
26602 REDUC_MAX_EXPR support.
26603 (vectorizable_reduction): Likewise.
26604 (vect_create_epilog_for_reduction): Likewise.
26606 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
26608 * match.pd (A / (1 << B) -> A >> B): New.
26609 * generic-match-head.c: Include optabs-tree.h.
26610 * gimple-match-head.c: Likewise.
26611 * optabs-tree.h (target_supports_op_p): New.
26612 * optabs-tree.c (target_supports_op_p): New.
26614 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26616 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
26617 $gcc_cv_ld --help output.
26618 (gcc_cv_ld_demangle): Likewise.
26619 (gcc_cv_ld_eh_frame_hdr): Likewise.
26620 (gcc_cv_ld_pie): Likewise.
26621 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
26622 (gcc_cv_ld_buildid): Likewise.
26623 (gcc_cv_ld_sysroot): Likewise.
26624 (ld_bndplt_support): Likewise.
26625 (ld_pushpopstate_support): Likewise.
26626 * configure: Regenerate.
26627 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26629 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26632 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26633 operands indexes to avoid gap between operands and match_dups.
26635 2017-06-21 Andrew Pinski <apinski@cavium.com>
26637 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26638 Increment Arith_shift and Arith_shift_reg by 1.
26639 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26641 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26642 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26643 (aarch64_strip_extend): Add new argument and test for it.
26644 (aarch64_cheap_mult_shift_p): New function.
26645 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26646 add a cost if it is true.
26647 Update calls to aarch64_strip_extend.
26648 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26650 2017-06-21 Andrew Pinski <apinski@cavium.com>
26652 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26654 (thunderxt88): Likewise.
26655 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26656 (thunderx_prefetch_tune): New variable.
26657 (thunderx2t99_prefetch_tune): Update for the correct values.
26658 (thunderxt88_tunings): New variable.
26659 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26660 generic_prefetch_tune.
26661 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26663 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26665 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26666 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26667 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26668 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26669 (aarch64_atomic_cas<mode>, GPI): Likewise.
26671 2017-06-21 Martin Liska <mliska@suse.cz>
26673 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26674 statements on cold and hot labels.
26675 * predict.c (tree_estimate_probability_bb): Remove the
26676 prediction from this place.
26678 2017-06-21 Martin Liska <mliska@suse.cz>
26680 PR tree-optimization/79489
26681 * gimplify.c (maybe_add_early_return_predict_stmt): New
26683 (gimplify_return_expr): Call the function.
26684 * predict.c (tree_estimate_probability_bb): Remove handling
26686 * predict.def: Update comment about early return predictor.
26687 * gimple-predict.h (is_gimple_predict): New function.
26688 * predict.def: Change default value of early return to 66.
26689 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26691 * passes.def: Put pass_strip_predict_hints to the beginning of
26694 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26696 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26697 FUNCTION_DECL declarations.
26698 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26700 (dwaf2out_decl): Likewise.
26701 * godump.c (go_early_global_decl): Skip call to the real debug hook
26702 for FUNCTION_DECL declarations.
26703 * passes.c (rest_of_decl_compilation): Skip call to the
26704 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26705 -fdump-go-spec is passed.
26707 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26709 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26710 Reorder for compactness.
26711 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26712 (def_builtin_pure, def_builtin_pure2): New functions.
26713 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26715 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26717 * match.pd (nop_convert): New predicate.
26718 ((A +- CST1) +- CST2): Allow some NOP conversions.
26720 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26723 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26724 with ctors/dtors if GOVD_SHARED is set.
26726 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26728 * config/aarch64/aarch64.md (movti_aarch64):
26729 Emit mov rather than orr.
26730 (movtf_aarch64): Likewise.
26731 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26732 Emit mov rather than orr.
26734 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26736 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26737 Swap alternatives, make integer dup more expensive.
26739 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26741 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26742 Return true for non-tls symbols.
26744 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26746 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26747 (cortex-a75): Likewise.
26748 (cortex-a75.cortex-a55): Likewise.
26749 * config/aarch64/aarch64-tune.md: Regenerate.
26750 * doc/invoke.texi (-mtune): Document new values for -mtune.
26752 2017-06-21 Tom de Vries <tom@codesourcery.com>
26754 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26755 stack_size feature.
26756 (Effective-Target Keywords, Other attributes): Suggest using
26757 dg-add-options stack_size feature to get stack limit in stack_size
26758 effective target documentation.
26760 2017-06-21 Julian Brown <julian@codesourcery.com>
26761 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26763 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26764 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26765 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26767 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26768 attribute type list for neon_multiply.
26769 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26770 attribute type list for neon_multiply.
26771 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26772 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26773 attribute type list for neon_multiply.
26774 * config/arm/types.md (crypto_pmull): Add.
26775 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26776 attribute type list.
26778 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26780 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26783 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26785 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26786 to make sure not to dereference a NULL cost_classes_ptr pointer.
26788 2017-06-20 Carl Love <cel@us.ibm.com>
26790 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26791 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26792 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26793 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26794 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26795 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26796 VMULOSW): New enum "unspec" values.
26797 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26798 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26799 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26800 altivec_vmulosw): New patterns.
26801 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26802 VMULOSW): Add definitions.
26804 2017-06-20 Julia Koval <julia.koval@intel.com>
26806 * config/i386/i386.c: Fix rounding expand for new pattern.
26807 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26809 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26811 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26812 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26814 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26816 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26819 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26821 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26822 architecture, then by implementer ID.
26823 * config/aarch64/aarch64-tune.md: Regenerate.
26825 2017-06-20 Richard Biener <rguenther@suse.de>
26827 PR middle-end/81097
26828 * fold-const.c (split_tree): Fold to type before negating.
26830 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26832 * diagnostic-show-locus.c
26833 (selftest::test_fixit_deletion_affecting_newline): New function.
26834 (selftest::diagnostic_show_locus_c_tests): Call it.
26836 2017-06-20 Andreas Schwab <schwab@suse.de>
26839 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26842 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26844 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26845 __ARM_FEATURE_COPROC according to support.
26847 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26849 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26850 Rewritten to avoid overflow for > 32-bit pointers.
26853 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26854 by removing enum keyword.
26855 (ubsan_type_descriptor): Likewise. Formatting fix.
26858 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26859 splitter): Require TARGET_SSE2 in the condition.
26861 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26864 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26865 for doing vector set of SFmode on ISA 3.0.
26866 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26867 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
26869 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
26870 SFmode value into a V4SF variable that was extracted from another
26871 V4SF variable without converting the element to double precision
26872 and back to single precision vector format.
26873 (vsx_insert_extract_v4sf_p9_2): Likewise.
26875 2017-06-19 Jakub Jelinek <jakub@redhat.com>
26877 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
26878 in UWHI to avoid undefined overflow.
26881 * ubsan.h (enum ubsan_encode_value_phase): New.
26882 (ubsan_encode_value): Change second argument to
26883 enum ubsan_encode_value_phase with default value of
26884 UBSAN_ENCODE_VALUE_GENERIC.
26885 * ubsan.c (ubsan_encode_value): Change second argument to
26886 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
26887 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
26888 create_tmp_var_raw instead of create_tmp_var and use a
26890 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
26891 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
26892 ubsan_encode_value callers.
26895 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
26896 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
26897 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
26899 2017-06-19 Richard Biener <rguenther@suse.de>
26901 PR middle-end/81118
26902 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
26903 estimates if we changed anything.
26905 2017-06-19 Richard Biener <rguenther@suse.de>
26907 PR tree-optimization/80887
26908 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
26909 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
26910 simplified lookups, then reset mprts_hook.
26911 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
26913 (try_to_simplify): Likewise.
26915 2017-06-19 Martin Liska <mliska@suse.cz>
26918 * gimplify.c (gimplify_switch_expr):
26919 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
26921 2017-06-19 Martin Liska <mliska@suse.cz>
26923 * doc/install.texi: Document that PGO runs in 4 stages.
26925 2017-06-19 Martin Liska <mliska@suse.cz>
26928 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
26929 to dispatcher function name.
26930 * multiple_target.c (replace_function_decl): New function.
26931 (create_dispatcher_calls): Redirect both edges and references.
26933 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
26935 * profile-count.c (profile_count::dump): Dump quality.
26936 (profile_count::differs_from_p): Update for unsigned val.
26937 * profile-count.h (profile_count_quality): New enum.
26938 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
26940 2017-06-19 Richard Biener <rguenther@suse.de>
26942 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
26943 struct function as arg.
26944 (estimate_numbers_of_iterations): Export overload with loop arg.
26945 (free_numbers_of_iterations_estimates_loop): Use an overload of
26946 free_numbers_of_iterations_estimates instead.
26947 * tree-cfg.c (remove_bb): Adjust.
26948 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
26949 * tree-parloops.c (gen_parallel_loop): Likewise.
26950 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
26952 (tree_unroll_loops_completely): Likewise.
26953 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
26954 Use an overload instead and export.
26955 (estimated_loop_iterations): Adjust.
26956 (max_loop_iterations): Likewise.
26957 (likely_max_loop_iterations): Likewise.
26958 (estimate_numbers_of_iterations): Take struct function as arg
26960 (loop_exits_before_overflow): Adjust.
26961 (free_numbers_of_iterations_estimates_loop): Use an overload.
26962 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
26963 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
26965 2017-06-19 Richard Biener <rguenther@suse.de>
26968 * ipa-prop.c (find_constructor_constant_at_offset): Handle
26969 RANGE_EXPR conservatively.
26971 2017-06-16 Carl Love <cel@us.ibm.com>
26973 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26974 definitions for vec_float, vec_float2, vec_floato,
26975 vec_floate built-ins.
26976 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
26977 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
26979 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
26980 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
26981 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
26982 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
26983 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
26984 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
26985 vec_floato): Add builtin defines.
26986 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
26987 Update the built-in documentation file for the new built-in
26990 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26992 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
26993 (mthumb): Mark as the negative of -marm.
26995 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26997 * doc/invoke.texi (ARM Options, -mcpu): Document supported
26999 (ARM Options, -mtune): Document that this accepts the same
27000 extension options as -mcpu.
27001 (ARM Options, -mfpu): Document addition of -mfpu=auto.
27003 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27005 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27006 permitted extensions.
27008 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27010 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27011 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27012 (armv8-m.main): Add option +nodsp.
27013 * config/arm/arm-cpu-cdata.h: Regenerated.
27015 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27017 * config/arm/t-fuchsia: New file.
27018 * config.gcc (arm*-*-fuchsia*): Use it.
27020 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27022 * config/arm/t-symbian: Rewrite for new option infrastructure.
27024 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27026 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27027 (MULTILIB_REQUIRED): Likewise.
27029 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27031 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27032 (MULTILIB_RESUE): Likewise.
27033 (MULTILIB_MATCHES): Likewise.
27034 (MULTLIB_REQUIRED): Likewise.
27036 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27038 * config/arm/t-rtems: Rewrite for new option framework.
27040 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27042 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
27043 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
27044 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
27045 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27046 * config/arm/t-multilib: ... here.
27047 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27048 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27049 armv7-a and armv8*-a when A-profile libraries have not been built.
27050 * config/arm/t-rmprofile: Rewrite.
27052 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27054 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27055 with a backslash. Remove the backslash after substituting unescaped
27057 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27059 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27061 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27062 the driver to pass through the default setting of -mfloat-abi.
27063 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27065 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27066 (all_feat_combs): New rule.
27067 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27069 * config/arm/t-aprofile: Rewrite.
27071 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27073 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27074 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27075 fpu is not specified by the user/command-line.
27076 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27077 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27078 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27079 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27080 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27081 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27083 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27085 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27086 * config/arm/t-arm-elf: Rewritten.
27088 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27090 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27091 have some floating-point instructions.
27092 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27093 (TARGET_MAYBE_HARD_FLOAT): New macro.
27094 * config/arm/arm-builtins.c (arm_init_builtins): Use
27095 TARGET_MAYBE_HARD_FLOAT.
27096 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27098 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27100 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27101 (configargs.h): Include it.
27102 (arm_print_hint_for_fpu_option): New function.
27103 (arm_parse_fpu_option): New function.
27104 (candidate_extension): New class.
27105 (arm_canon_for_multilib): New function.
27106 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27107 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27108 (ARCH_CANONICAL_SPECS): New macro.
27109 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
27111 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27113 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
27114 are set after handling multilib fragments. Set target_cpu_default2
27117 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27119 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
27121 (arm*-*-*): Set target_cpu_default2 to a quoted string.
27122 * config/arm/parsecpu.awk (check_cpu): Validate any extension
27124 (check_arch): Likewise.
27125 * config/arm/arm.c (arm_configure_build_target): Handle
27126 TARGET_CPU_DEFAULT being a string constant. Scan any feature
27127 options in the default.
27129 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27131 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
27132 when an option is an alias of another.
27133 * config/arm/parsecpu.awk (optalias): New parser token.
27134 (gen_comm_data): Mark non-alias options as such. Emit entries
27135 for extension aliases.
27136 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
27137 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
27138 (armv6kz, armv6zk, armv6t2): Likewise.
27139 (armv7): Make vfpv3-d16 an alias.
27140 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
27142 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
27143 Sort in canonical order.
27144 (armv8-a): Sort in canonical order.
27145 (armv8.1-a, armv8.2-a): Likewise.
27146 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
27148 (cortex-a9): Sort in canonical order.
27149 * config/arm/arm.c (selftests.h): Include it.
27150 (arm_test_cpu_arch_data): New function.
27151 (arm_run_self_tests): New function.
27152 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
27153 (targetm): Move declaration to the end of the file.
27154 * arm-cpu-cdata.h: Regenerated.
27156 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27158 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
27159 call to target_mode_check describing the type of option passed.
27160 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
27161 (arm_target_thumb_only): Use arm_parse_arch_option_name or
27162 arm_parse_cpu_option_name to match parameters against list of
27164 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
27165 arm_arch_core_flags data structure.
27166 * config/arm/arm-cpu_cdata.h: Regenerated.
27168 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27170 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
27172 (arm_print_hint_for_cpu_option): Likewise.
27173 (arm_print_hint_for_arch_option): Likewise.
27174 (arm_parse_cpu_option_name): Likewise.
27175 (arm_parse_arch_option_name): Likewise.
27176 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
27177 of entries in the all_fpus list.
27178 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
27179 (arm_parse_cpu_option_name): Declare.
27180 (arm_parse_arch_option_name): Declare.
27181 (arm_parse_option_features): Declare.
27182 (arm_intialize_isa): Declare.
27183 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
27185 (gen_comm_data): ... here. Make definitions non-static.
27186 * config/arm/arm-cpu-data.h: Regenerated.
27187 * config/arm/arm-cpu-cdata.h: Regenerated.
27189 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27191 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
27192 (cpu_arch_extension): New structure.
27193 (cpu_arch_option, arch_option, cpu_option): New structures.
27194 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
27195 architecture types.
27196 (gen_data): Generate new format data tables.
27197 * config/arm/arm.c (cpu_tune): New structure.
27198 (cpu_option, processors): Delete.
27199 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
27200 (arm_print_hint_for_cpu_option): ... this and ...
27201 (arm_print_hint_for_arch_option): ... this.
27202 (arm_parse_arch_cpu_name): Delete. Replace with ...
27203 (arm_parse_cpu_option_name): ... this and ...
27204 (arm_parse_arch_option_name): ... this.
27205 (arm_unrecognized_feature): Change type of target parameter to
27207 (arm_parse_arch_cpu_features): Delete. Replace with ...
27208 (arm_parse_option_features): ... this.
27209 (arm_configure_build_target): Rework to use new configuration data
27211 (arm_print_tune_info): Rework for new configuration data tables.
27212 * config/arm/arm-cpu-data.h: Regenerated.
27213 * config/arm/arm-cpu.h: Regenerated.
27215 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27217 * Makefile.in (OBJS): Move sbitmap.o from here ...
27218 (OBJS-libcommon): ... to here.
27220 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27222 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
27223 (ISA_ALL_CRYPTO): New macro.
27224 (ISA_ALL_SIMD): New macro
27225 (ISA_ALL_FP): New macro.
27226 * config/arm/arm.c (fpu_bitlist): Update initializer.
27227 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
27229 (arm9e): Add fpu. Add option for nofp
27230 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
27231 (arm926ej-s, arm1026ej-s): Likewise.
27232 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
27233 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
27234 neon-fp16, neon-vfpv4, nofp and nosimd.
27235 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
27236 (cortex-a8): Add fpu. Add option for nofp.
27237 (cortex-a9): Add fpu. Add options for nosimd and nofp.
27238 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
27239 (cortex-r4f): Add fpu.
27240 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
27241 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
27243 (cortex-r8): Likewise.
27244 (cortex-m4): Add fpu. Add option for nofp.
27245 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
27246 (cortex-a17.cortex-a7): Likewise.
27247 (cortex-a32): Add fpu. Add options for crypto and nofp.
27248 (cortex-a35, cortex-a53): Likewise.
27249 (cortex-a57): Add fpu. Add option for crypto.
27250 (cortex-a72, cortex-a73): Likewise.
27251 (exynos-m1): Likewise.
27252 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
27253 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
27254 (cortex-m33): Add fpu. Add option for nofp.
27255 * config/arm/arm-cpu-cdata.h: Regenerated
27256 * config/arm/arm-cpu-data.h: Regenerated.
27258 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27260 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
27261 (armv5te, armv5tej): Likewise.
27262 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
27263 (armv7): Add options fp and vfpv3-d16.
27264 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
27265 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
27267 (armv7ve): Likewise.
27268 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
27269 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
27270 (armv8-a): Add nocrypto option.
27271 (armv8.1-a, armv8.2-a): Likewise.
27272 (armv8-m.main): add options fp, fp.dp and nofp.
27274 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27276 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
27278 (armv8-a+crc): Delete.
27279 (armv8.1-a): Add options simd, crypto and nofp.
27280 (armv8.2-a): Add options fp16, simd, crypto and nofp.
27281 (armv8.2-a+fp16): Delete.
27282 (armv8-m.main): Add option dsp.
27283 (armv8-m.main+dsp): Delete.
27284 (cortex-a8): Add fpu. Add nofp option.
27285 (cortex-a9): Add fpu. Add nofp and nosimd options.
27286 * config/arm/parsecpu.awk (gen_data): Generate option tables and
27287 link to main cpu and architecture data structures.
27288 (gen_comm_data): Only put isa attributes from the main architecture
27290 (option): New statement for architecture and CPU entries.
27291 * arm.c (struct cpu_option): New structure.
27292 (struct processors): Add entry for options.
27293 (arm_unrecognized_feature): New function.
27294 (arm_parse_arch_cpu_name): Ignore any characters after the first
27296 (arm_parse_arch_cpu_feature): New function.
27297 (arm_configure_build_target): Separate out any CPU and architecture
27298 features and parse separately. Don't error out if -mfpu=auto is
27299 used with only an architecture string.
27300 (arm_print_asm_arch_directives): New function.
27301 (arm_file_start): Call it.
27302 * config/arm/arm-cpu-cdata.h: Regenerated.
27303 * config/arm/arm-cpu-data.h: Likewise.
27304 * config/arm/arm-tables.opt: Likewise.
27306 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27308 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
27309 assembler when it is not -mfpu=auto.
27311 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27313 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
27314 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
27315 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
27316 (ASM_CPU_SPEC): Rewrite.
27317 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
27318 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
27319 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
27320 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
27321 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
27322 copied string is NUL-terminated. Also strip any characters prefixed
27324 (arm_rewrite_selected_arch): New function.
27325 (arm_rewrite_march): New function.
27327 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27329 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
27330 (x_arm_cpu_string, x_arm_tune_string): Likewise.
27331 (march, mcpu, mtune): Convert to string-based options.
27332 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
27333 (arm_parse_arch_cpu_name): New function.
27334 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
27335 identify selected architecture or CPU.
27336 (arm_option_save): New function.
27337 (TARGET_OPTION_SAVE): Redefine.
27338 (arm_option_restore): Restore string options.
27339 (arm_option_print): Print string options.
27341 2017-06-16 Martin Sebor <msebor@redhat.com>
27343 PR tree-optimization/80933
27344 PR tree-optimization/80934
27345 * builtins.c (fold_builtin_3): Do not handle bcmp here.
27346 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
27347 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
27348 (gimple_fold_builtin): Call them.
27350 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27352 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
27353 as unlikely; update profile.
27355 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27357 * predict.c (force_edge_cold): Handle declaring edges impossible
27360 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27362 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
27364 (try_unroll_loop_completely): Fix reporting.
27366 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27368 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
27370 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
27373 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
27374 if given a non-constant argument for an intrinsic which requires a
27377 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27379 * profile.c (compare_freqs): New function.
27380 (branch_prob): Sort edge list.
27381 (find_spanning_tree): Assume that the list is priority sorted.
27383 2017-06-16 Richard Biener <rguenther@suse.de>
27385 PR tree-optimization/81090
27386 * passes.def (pass_record_bounds): Remove.
27387 * tree-pass.h (make_pass_record_bounds): Likewise.
27388 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
27389 make_pass_record_bounds): Likewise.
27390 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
27391 not free niter estimates at the beginning but at the end.
27392 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
27394 2017-06-16 Richard Biener <rguenther@suse.de>
27396 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
27397 initializer to workaround ICE in host GCC 4.8.
27399 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27401 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
27403 (clone_inlined_nodes): Update.
27405 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27407 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
27408 prefetch settings, and enable prefetching by default at -O3.
27410 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27412 * config/aarch64/aarch64.c (aarch64_override_options_internal):
27413 Set flag_prefetch_loop_arrays according to tuning data.
27415 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27417 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
27418 New tune structure.
27419 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
27420 [Unrelated to main purpose of the patch] Place the pointer field last
27421 to enable type checking errors when tune structure are wrongly merged.
27422 * config/aarch64/aarch64.c (generic_prefetch_tune,)
27423 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
27424 (thunderx2t99_prefetch_tune): New tune constants.
27425 (tune_params *_tunings): Update all tunings (no functional change).
27426 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
27427 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
27428 from tunings structures.
27430 2017-06-16 Jakub Jelinek <jakub@redhat.com>
27433 * ubsan.c (instrument_null): Add T argument, use it instead
27434 of computing it based on IS_LHS.
27435 (instrument_object_size): Likewise.
27436 (pass_ubsan::execute): Adjust instrument_null and
27437 instrument_object_size callers to pass gimple_get_lhs or
27438 gimple_assign_rhs1 result to it. Use instrument_null instead of
27439 calling get_base_address and instrument_mem_ref. Handle
27440 aggregate call arguments for object-size sanitization.
27442 2017-06-16 Yury Gribov <tetra2005@gmail.com>
27444 PR tree-optimization/81089
27445 * tree-vrp.c (is_masked_range_test): Validate operands of
27448 2017-06-15 Martin Sebor <msebor@redhat.com>
27451 * dumpfile.c (dump_register): Avoid calling memset to initialize
27452 a class with a default ctor.
27453 * gcc.c (struct compiler): Remove const qualification.
27454 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
27455 * hash-table.h: Ditto.
27456 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
27458 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
27459 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
27461 * params.h (struct param_info): Make struct members non-const.
27462 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
27463 with default initialization.
27464 * vec.h (vec_copy_construct, vec_default_construct): New helper
27466 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
27467 with vec_copy_construct.
27468 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
27469 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
27470 * doc/invoke.texi (-Wclass-memaccess): Document.
27472 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27474 * emit-rtl.h (is_leaf): Update comment about local
27475 register allocator.
27477 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27480 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
27481 for a variable to have a section before checking if the section has a
27483 Set section to.persistent if persistent attribute is set.
27484 Warn if .persistent attribute is used on an automatic variable.
27486 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
27488 PR rtl-optimization/80474
27489 * reorg.c (update_block): Do not ignore instructions in a delay slot.
27491 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
27493 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
27496 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
27498 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
27499 (casesi): Emit bounds checking as RTL.
27500 (casesi_internal_mips16_<mode>): Remove bounds checking.
27502 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
27504 * config/xtensa/xtensa.c (xtensa_option_override): Append
27505 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
27506 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
27507 xtensa_doloop_hooks): Define unconditionally.
27508 (xtensa_reorg_loops): Only call reorg_loops in the presence of
27510 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
27511 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
27512 for it in xtensa_option_override.
27513 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
27514 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
27516 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
27518 * doc/cppopts.texi: Document '-' special value to -MF.
27520 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
27522 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
27523 (cortex_a53_fconst): Likewise.
27524 (cortex_a53_fpmul): Likewise.
27525 (cortex_a53_f_load_64): Likewise.
27526 (cortex_a53_f_load_many): Likewise.
27527 (cortex_a53_advsimd_alu): Likewise.
27528 (cortex_a53_advsimd_alu_q): Likewise.
27529 (cortex_a53_advsimd_mul): Likewise.
27530 (cortex_a53_advsimd_mul_q): Likewise.
27531 (fpmac bypass): Add new bypass for fpmac-fpmac case.
27532 Add missing fmul, r2f_cvt and fconst cases.
27534 2017-06-14 Richard Biener <rguenther@suse.de>
27536 PR middle-end/81088
27537 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
27539 (fold_binary_loc): When associating do not treat pre-existing
27540 TREE_OVERFLOW on literal constants as a reason to allow
27541 TREE_OVERFLOW on associated literal constants.
27543 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27545 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
27546 (MASK_FEATURES): New macro.
27547 * config/sparc/sparc.c (sparc_option_override): Remove the special
27548 handling of -mfpu and generalize it to all MASK_FEATURES switches.
27550 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27552 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
27553 a division of 0 if non-call exceptions are enabled.
27555 2017-06-14 Andrew Pinski <apinski@cavium.com>
27556 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27559 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
27560 Improve vector initialization code gen for only variable case.
27562 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27564 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
27566 2017-06-14 Richard Biener <rguenther@suse.de>
27568 PR tree-optimization/81083
27569 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
27572 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27574 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
27575 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
27576 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
27577 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
27578 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
27579 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
27581 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27583 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
27584 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
27586 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27588 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
27590 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27592 * config/rs6000/t-rtems: Don't handle SPE.
27594 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27596 * config/rs6000/t-linux: Don't handle SPE.
27598 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27600 * config/rs6000/eabispe.h: Delete file.
27602 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27604 * config/rs6000/t-spe: Delete file.
27606 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27608 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
27609 (rs6000_legitimate_offset_address_p): Return false for anything in
27610 V2SImode or V2SFmode.
27612 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27614 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
27615 except V2SF and V2SI. Rearrange the vector modes, and add comments.
27616 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
27618 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
27619 (rs6000_legitimate_offset_address_p): Ditto.
27620 (rs6000_emit_move): Ditto.
27621 (rs6000_init_builtins): Remove V4HI_type_node.
27623 2017-06-13 Martin Liska <mliska@suse.cz>
27626 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27627 (gate_asan): Likewise.
27628 * asan.h (asan_no_sanitize_address_p): Remove the function.
27629 (sanitize_flags_p): New function.
27630 * builtins.def: Fix coding style.
27631 * common.opt: Use renamed enum value.
27632 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27633 * doc/extend.texi: Document no_sanitize attribute.
27634 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27635 to SANITIZE_UNDEFINED_NONDEFAULT.
27636 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27637 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27638 Use sanitize_flags_p.
27639 * gimplify.c (gimplify_function_tree): Likewise.
27640 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27641 * opts.c (parse_no_sanitize_attribute): New function.
27642 (common_handle_option): Use renamed enum value.
27643 * opts.h (parse_no_sanitize_attribute): Declare.
27644 * tree.c (sanitize_flags_p): New function.
27645 * tree.h: Declared here.
27646 * tsan.c: Use sanitize_flags_p.
27647 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27648 (instrument_mem_ref): Likewise.
27649 (instrument_bool_enum_load): Likewise.
27650 (do_ubsan_in_current_function): Remove the function.
27651 (pass_ubsan::execute): Use sanitize_flags_p.
27652 * ubsan.h: Remove do_ubsan_in_current_function
27653 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27654 (dump_function_to_file): Use it here.
27656 2017-06-13 Martin Jambor <mjambor@suse.cz>
27658 PR tree-optimization/80803
27659 PR tree-optimization/81063
27660 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27661 (propagate_subaccesses_across_link): Enqueue subtree whenever
27662 necessary instead of relying on the caller.
27664 2017-06-13 Martin Jambor <mjambor@suse.cz>
27666 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27667 that have a first_link.
27668 (sort_and_splice_var_accesses): Do not check first_link before
27670 (subtree_mark_written_and_enqueue): Likewise.
27671 (propagate_all_subaccesses): Likewise and do not stop at first
27672 parent with a first_link.
27674 2017-06-13 Martin Jambor <mjambor@suse.cz>
27676 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27679 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27681 * match.pd: New pattern.
27683 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27685 * tree-vrp.c (is_masked_range_test): New function.
27686 (register_edge_assert_for): Determine ranges for
27689 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27691 PR tree-optimization/67328
27692 * fold-const.c (maskable_range_p): New function.
27693 (build_range_check): Generate bittests if possible.
27695 2017-06-13 Martin Liska <mliska@suse.cz>
27697 * gimple-pretty-print.c (dump_probability): Add new argument.
27698 (dump_edge_probability): Dump both probability and count.
27699 (dump_gimple_label): Likewise.
27700 (dump_gimple_bb_header): Likewise.
27702 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27705 * config/avr/avr-devices.c: Fix indentation.
27706 * config/avr/gen-avr-mmcu-specs.c: Dito.
27708 2017-06-13 Richard Biener <rguenther@suse.de>
27710 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27711 instead get vector type from stmt_info.
27712 (vectorizable_reduction): Adjust. Remove dead code.
27714 2017-06-13 Richard Biener <rguenther@suse.de>
27716 PR middle-end/81065
27717 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27718 case of C * (x * C2 + C3).
27719 (fold_addr_of_array_ref_difference): Properly fold index difference.
27721 2017-06-12 David S. Miller <davem@davemloft.net>
27724 * config/sparc/sparc.md (return expander): Emit frame blockage if
27725 function uses alloca.
27727 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27729 * combine.c (make_field_assignment): Check len rather than the mode
27730 precision when calling force_to_mode.
27732 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27734 Support multilibs and devices that see flash in RAM address range.
27737 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27738 (avr_mcu_t) <flash_pm_offset>: New field.
27739 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27740 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27741 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27742 (AVR_TINY_PM_OFFSET): Remove macro.
27743 * config/avr/avr.opt (-mshort-calls): New option.
27744 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27745 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27746 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27747 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27748 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27749 instead of avr_arch->have_jmp_call.
27750 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27751 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27752 avr_arch->flash_pm_offset to define.
27753 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27754 new field flash_pm_offset. Add entry for avrxmega3.
27755 (avr_texinfo): Add entry for avrxmega3.
27756 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27757 attiny212, attiny214,
27758 attiny412, attiny414, attiny416, attiny417,
27759 attiny814, attiny816, attiny817,
27760 attiny1614, attiny1616, attiny1617,
27761 attiny3214, attiny3216, attiny3217.
27762 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27763 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27764 (avr_print_operand_address) [AVR_TINY]: Same.
27765 (avr_asm_init_sections) <readonly_data_section>: Only patch
27766 callback if avr_arch->flash_pm_offset = 0.
27767 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27768 for rodata if avr_arch->flash_pm_offset != 0.
27769 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27770 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27771 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27772 (m_options): Append opt_rcall.
27773 (m_dirnames): Append dir_rcall.
27774 * config/avr/t-multilib: Regenerate.
27776 * configure.ac [target=avr]: Check whether avrxmega3 default
27777 linker description file works as needed.
27778 * configure: Regenerate.
27779 * doc/avr-mmcu.texi: Regenerate.
27780 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27781 <__AVR_ARCH__>: Document avrxmega3 and 103.
27782 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27783 <__AVR_SHORT_CALLS__>: Document it.
27784 <__AVR_PM_BASE_ADDRESS__>: Document it.
27785 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27786 (AVR Variable Attributes) <progmem>: Document this is
27787 not needed for avrxmega3.
27788 (AVR Named Address Spaces) <__flash>: Dito.
27790 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27792 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27794 2017-06-12 Doug Rupp <rupp@adacore.com>
27796 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27797 Append vxworks-stdint.h to the tm_file list.
27798 * config/vxworks-stdint.h: New file.
27800 2017-06-12 Martin Liska <mliska@suse.cz>
27802 PR tree-optimization/81041
27803 * tree-profile.c (gimple_gen_ic_func_profiler):
27804 Create an extra BB in profile-generate
27805 (gimple_gen_time_profiler): Likewise.
27807 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27809 PR tree-optimization/81003
27810 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27811 (update_range_test): Use it instead of force_gimple_operand_gsi.
27813 2017-06-12 Richard Biener <rguenther@suse.de>
27815 PR tree-optimization/81053
27816 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27817 with backedge value not defined in loop. Simplify def stmt
27820 2017-06-11 Tom de Vries <tom@codesourcery.com>
27823 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27825 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27826 nvptx_cannot_force_const_mem.
27828 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27830 * opts.c (finish_options): Move test for flag_split_stack after
27831 it has been initialized.
27833 2017-06-11 Jason Merrill <jason@redhat.com>
27835 * tree.h (id_equal): New.
27836 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27837 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27838 instead of strcmp of IDENTIFIER_POINTER.
27840 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27842 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27843 (mark_all_inlined_calls_cdtor): Fix formating.
27844 (inline_transform): Rescale profile before inlining.
27846 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27848 * cgraph.h (cgraph_edge::clone): Update prototype.
27849 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27850 (cgraph_node::create_clone): Update.
27851 (cgraph_node::create_version_clone): Update.
27852 * tree-inline.c (copy_bb): Update.
27853 (expand_call_inline): Update.
27855 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27857 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27858 factored out from ...
27859 (rs6000_emit_prologue): ... here.
27861 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27863 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27864 factored out from ...
27865 (rs6000_emit_prologue): ... here.
27867 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27869 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
27871 (handle_missing_profiles): Fix computation of tp_first_run.
27872 (counts_to_freqs): Do not touch freqs when count is 0.
27874 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27876 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
27879 2017-06-10 Tom de Vries <tom@codesourcery.com>
27881 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27882 attributes): Document signal effective target.
27884 2017-06-10 Tom de Vries <tom@codesourcery.com>
27886 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27887 Document effective target stack_size.
27889 2017-06-09 David Malcolm <dmalcolm@redhat.com>
27891 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
27892 to the edit_context if they can be auto-applied.
27894 2017-06-9 Ian Lance Taylor <iant@golang.org>
27896 * opts.c (finish_options): If -fsplit-stack, disable implicit
27897 -forder-blocks-and-partition.
27898 * doc/invoke.texi (Optimize Options): Document that when using
27899 -fsplit-stack -forder-blocks-and-partition is not implicitly
27902 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27904 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
27905 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
27906 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
27907 * builtins.def (abort, trap, unreachable): Declare cold.
27908 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
27909 * tree-core.h (ECF_COLD): New.
27910 * tree.c (set_call_expr_flags): Handle ECF_COLD.
27911 (build_common_builtin_nodes): Mark unreachable and abort as cold.
27913 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27915 * predict.c (unlikely_executed_stmt_p): Cleanup.
27917 2017-06-09 Richard Biener <rguenther@suse.de>
27919 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
27920 model if the ref is always written to.
27922 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27924 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
27926 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27928 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
27931 2017-06-09 Tom de Vries <tom@codesourcery.com>
27934 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
27935 "target cannot support label values" when encountering LABEL_REF.
27937 2017-06-09 Martin Liska <mliska@suse.cz>
27939 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
27940 (gimple_gen_ic_func_profiler): Emit direct comparison
27941 of __gcov_indirect_call_callee with NULL.
27942 (gimple_gen_time_profiler): Change probability from
27943 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
27945 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27947 * profile.c (edge_gcov_counts): Turn to pointer.
27948 (compute_branch_probabilities, compute_branch_probabilities): Update.
27949 (branch_prob): Do not clear edge_gcov_count.
27950 * profile.h (edge_gcov_counts): Turn to pointer.
27951 (edge_gcov_count): Update.
27953 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27955 * gimple.h (gimple_check_failed): Mark cold.
27957 2017-06-09 Richard Biener <rguenther@suse.de>
27959 PR tree-optimization/66623
27960 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
27961 refactor check_reduction into two parts, properly computing
27962 whether we have to check reduction validity for outer loop
27965 2017-06-09 Richard Biener <rguenther@suse.de>
27967 PR tree-optimization/79483
27968 * graphite-scop-detection.c (order): New global.
27969 (get_order): Compute bb to order mapping that satisfies code
27970 generation constraints.
27971 (cmp_pbbs): New helper.
27972 (build_scops): Start domwalk at entry block, sort generated
27975 2017-06-09 Richard Biener <rguenther@suse.de>
27977 PR middle-end/81007
27978 * ipa-polymorphic-call.c
27979 (ipa_polymorphic_call_context::restrict_to_inner_class):
27980 Skip FIELD_DECLs with error_mark_node type.
27981 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
27984 2017-06-09 Martin Liska <mliska@suse.cz>
27986 * predict.c (struct branch_predictor): New struct.
27987 (test_prediction_value_range): New test.
27988 (predict_c_tests): New function.
27989 * selftest-run-tests.c (selftest::run_tests): Run the function.
27990 * selftest.h: Declare new tests.
27992 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
27995 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
27996 gen_add3_insn did not fail.
27997 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
27998 r0, construct that number in a temporary reg and add that reg to r0.
27999 If asked to put the result in r0 as well, fail.
28001 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
28003 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28004 for early expansion of vec_eqv.
28006 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28008 PR middle-end/81005
28009 * ubsan.c (instrument_null): Avoid pointless code temporary.
28010 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28012 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28016 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28019 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28022 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28024 * predict.c (unlikely_executed_edge_p): Move ahead.
28025 (probably_never_executed_edge_p): Use it.
28027 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28029 PR middle-end/79988
28030 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28031 gimple_call_builtin_p call.
28033 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28035 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28036 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28037 rtl_check_failed_type2, rtl_check_failed_code1,
28038 rtl_check_failed_code2, rtl_check_failed_code_mode,
28039 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28040 rtvec_check_failed_bounds, rtl_check_failed_flag,
28041 _fatal_insn_not_found, _fatal_insn): Likewise.
28042 * tree.h (tree_contains_struct_check_failed,
28043 tree_check_failed, tree_not_check_failed,
28044 tree_class_check_failed, tree_range_check_failed,
28045 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28046 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28047 tree_operand_check_failed, omp_clause_check_failed,
28048 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28051 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28053 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28054 flag_branch_probabilities.
28055 * ipa-inline.c (edge_badness): Likewise.
28056 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28057 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28058 * predict.c (maybe_hot_frequency_p): Likewise.
28059 (probably_never_executed): Likewise.
28060 * sched-ebb.c (schedule_ebbs): Likewise.
28061 * sched-rgn.c (find_single_block_region): Likewise.
28062 * tracer.c (tail_duplicate): Likewise.
28064 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28066 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28067 longer requires x_flag_profile_use.
28069 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28071 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28072 instead of flag_reorder_blocks_and_partition.
28073 * dbxout.c (dbxout_function_end): Likewise.
28074 * dwarf2out.c (gen_subprogram_die): Likewise.
28075 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28076 * hw-doloop.c (reorg_loops): Likewise.
28077 * varasm.c (assemble_start_function,
28078 assemble_end_function): Likewise.
28079 (decide_function_section): Do not check for
28080 flag_reorder_blocks_and_partition.
28082 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28084 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28086 (chkp_get_hard_register_fake_addr_expr): Ditto.
28087 (chkp_build_addr_expr): Add check for hard reg case.
28088 (chkp_parse_array_and_component_ref): Ditto.
28089 (chkp_find_bounds_1): Ditto.
28090 (chkp_process_stmt): Don't generate bounds store for
28093 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28095 * predict.c (maybe_hot_bb_p): Do not check profile status.
28096 (maybe_hot_edge_p): Likewise.
28097 (probably_never_executed): Check for zero counts even if profile
28099 (unlikely_executed_edge_p): New function.
28100 (unlikely_executed_stmt_p): New function.
28101 (unlikely_executed_bb_p): New function.
28102 (set_even_probabilities): Use unlikely predicates.
28103 (combine_predictions_for_bb): Likewise.
28104 (predict_paths_for_bb): Likewise.
28105 (predict_paths_leading_to_edge): Likewise.
28106 (determine_unlikely_bbs): New function.
28107 (estimate_bb_frequencies): Use it.
28108 (compute_function_frequency): Use zero counts even if profile is
28110 * profile-count.h: Fix typo.
28112 2017-08-08 Julia Koval <julia.koval@intel.com>
28114 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
28115 _mm512_mask_cvtsepi16_storeu_epi8,
28116 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
28117 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
28118 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
28119 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
28120 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
28121 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
28122 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
28123 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
28124 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
28125 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
28126 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
28127 __builtin_ia32_pmovuswb256mem_mask,
28128 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
28129 __builtin_ia32_pmovwb512mem_mask): New builtins.
28131 2017-08-08 Julia Koval <julia.koval@intel.com>
28133 PR target/73350,80862
28134 * config/i386/subst.md (round): Fix round pattern.
28135 * config/i386/i386.c (ix86_erase_embedded_rounding):
28136 Fix erasing rounding for the fixed pattern.
28138 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28140 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
28142 2017-06-08 Martin Liska <mliska@suse.cz>
28144 PR gcov-profile/80911
28145 * gcov.c (block_info::block_info): New constructor.
28147 2017-06-07 Carl Love <cel@us.ibm.com>
28149 * config/rs6000/rs6000-c: The return type of the following
28150 built-in functions was implemented as int not long long. Fix sign
28151 of return value for the unsigned version of vec_mulo and vec_mule.
28152 vector unsigned long long vec_bperm (vector unsigned long long,
28153 vector unsigned char)
28154 vector signed long long vec_mule (vector signed int,
28156 vector unsigned long long vec_mule (vector unsigned int,
28157 vector unsigned int)
28158 vector signed long long vec_mulo (vector signed int,
28160 vector unsigned long long vec_mulo (vector unsigned int,
28161 vector unsigned int)
28162 * doc/extend.texi: Fix the documentation for the built-in
28165 2017-06-07 Carl Love <cel@us.ibm.com>
28168 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
28171 2017-06-07 Carl Love <cel@us.ibm.com>
28173 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
28174 support, Generate doublehv for signed int/float for BE case only.
28176 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
28178 * doc/invoke.texi (mcx16): Rewrite.
28180 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28182 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
28183 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
28184 *mov<mode>_softfloat, and an anonymous splitter): Use
28185 nonimmediate_operand instead of rs6000_nonimmediate_operand.
28187 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28189 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
28191 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28192 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
28193 (rs6000_debug_reg_global): Adjust.
28194 (rs6000_init_hard_regno_mode_ok): Adjust.
28195 (rs6000_dbx_register_number): Adjust.
28196 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
28197 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28198 Remove SPE_ACC and SPEFSCR.
28199 (REG_ALLOC_ORDER): Ditto.
28200 (FRAME_POINTER_REGNUM): Change to 111.
28201 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
28202 (REG_CLASS_NAMES): Ditto.
28203 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
28204 (REGISTER_NAMES): Ditto.
28205 (ADDITIONAL_REG_NAMES): Ditto.
28206 (rs6000_reg_names): Ditto.
28207 * config/rs6000/rs6000.md: Renumber some register number
28210 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28212 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
28214 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28215 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
28217 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
28218 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28219 Delete the SPE high registers.
28220 (REG_ALLOC_ORDER): Ditto.
28221 (enum reg_class): Remove SPE_HIGH_REGS.
28222 (REG_CLASS_NAMES): Ditto.
28223 (REG_CLASS_CONTENTS): Delete the SPE high registers.
28224 (REGISTER_NAMES): Ditto.
28225 (rs6000_reg_names): Ditto.
28226 * doc/tm.texi.in: Remove SPE as example.
28227 * doc/tm.texi: Regenerate.
28229 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28231 * config/rs6000/8540.md (ppc8540_brinc): Delete.
28232 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
28233 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
28234 * config/rs6000/rs6000.md (type): Remove "brinc".
28236 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28238 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
28239 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
28240 * config/rs6000/linuxspe.h: Delete file.
28241 * config/rs6000/rs6000.md: Don't include spe.md.
28242 * config/rs6000/spe.h: Delete file.
28243 * config/rs6000/spe.md: Delete file.
28244 * config/rs6000/t-rs6000: Remove spe.md.
28246 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28248 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
28249 (reg_or_none500mem_operand): Delete.
28250 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
28251 instead of reg_or_none500mem_operand.
28253 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28255 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28256 handling of SPE flags.
28257 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
28259 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28261 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
28263 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
28264 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
28265 paired_divv2sf3): Similar.
28266 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
28267 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
28268 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
28270 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
28271 Rename the paired_* instruction patterns.
28272 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
28274 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
28275 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
28276 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
28277 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
28278 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
28279 PAIRED_VECTOR_MODE.
28280 (struct machine_function): Delete field spe_insn_chain_scanned_p.
28281 (spe_func_has_64bit_regs_p): Delete.
28282 (spe_expand_predicate_builtin): Delete.
28283 (spe_expand_evsel_builtin): Delete.
28284 (TARGET_DWARF_REGISTER_SPAN): Do not define.
28285 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
28286 (invalid_e500_subreg): Delete.
28287 (rs6000_legitimize_address): Always force_reg op2 as well, for
28288 paired single memory accesses.
28289 (rs6000_member_type_forces_blk): Delete.
28290 (rs6000_spe_function_arg): Delete.
28291 (rs6000_expand_unop_builtin): Delete SPE handling.
28292 (rs6000_expand_binop_builtin): Ditto.
28293 (spe_expand_stv_builtin): Delete.
28294 (bdesc_2arg_spe): Delete.
28295 (spe_expand_builtin): Delete.
28296 (spe_expand_predicate_builtin): Delete.
28297 (spe_expand_evsel_builtin): Delete.
28298 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
28299 (spe_init_builtins): Delete.
28300 (spe_func_has_64bit_regs_p): Delete.
28301 (savres_routine_name): Delete "info" parameter. Adjust callers.
28302 (rs6000_emit_stack_reset): Ditto.
28303 (rs6000_dwarf_register_span): Delete.
28304 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
28305 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
28306 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
28308 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
28310 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
28311 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
28312 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
28313 mulv2sf3, divv2sf3): Delete expanders.
28315 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28317 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
28319 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28321 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
28322 * config/rs6000/rs6000.c: Ditto.
28324 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28326 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
28327 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
28328 comparison_operator.
28330 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28332 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
28333 * config/rs6000/rs6000.opt: Ditto.
28334 * config/rs6000/t-rtems: Ditto.
28336 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28338 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
28339 TARGET_E500_SINGLE by 0, simplify.
28340 * config/rs6000/rs6000.c: Ditto.
28341 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
28342 (spe_build_register_parallel): Delete.
28343 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
28344 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
28345 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
28346 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
28347 (E500_CONVERT): Delete.
28348 * config/rs6000/spe.md: Remove many patterns and all define_constants.
28350 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28352 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
28353 * config/rs6000/dfp.md: Ditto.
28354 (negdd2, *negdd2_fpr): Merge.
28355 (absdd2, *absdd2_fpr): Merge.
28356 (negtd2, *negtd2_fpr): Merge.
28357 (abstd2, *abstd2_fpr): Merge.
28358 * config/rs6000/e500.h: Delete file.
28359 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
28360 TARGET_FPRS by 1 and simplify.
28361 * config/rs6000/rs6000-c.c: Ditto.
28362 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
28363 TARGET_DF_SPE by 0.
28364 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
28366 * config/rs6000/rs6000.md: Ditto.
28367 (floatdidf2, *floatdidf2_fpr): Merge.
28368 (move_from_CR_gt_bit): Delete.
28369 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
28370 (E500_CR_IOR_COMPARE): Delete.
28371 (All patterns that require !TARGET_FPRS): Delete.
28372 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
28374 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28376 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
28378 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28380 * graphds.c (add_edge): Intitialize edge's attached data.
28381 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
28382 pointer parameter. Call pointed function on each edge during
28383 graph traversing. Skip traversing the edge when the function
28385 (graphds_dfs, graphds_scc): Ditto.
28386 (for_each_edge): New parameter. Pass the new parameter to callback
28388 * graphds.h (skip_edge_callback): New function pointer type.
28389 (graphds_dfs, graphds_scc): New function pointer parameter.
28390 (graphds_edge_callback, for_each_edge): New parameter.
28392 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28394 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
28395 out code checking if runtime alias check is possible to below ...
28396 Call the new function.
28397 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
28398 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
28400 2017-06-07 Marek Polacek <polacek@redhat.com>
28403 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
28404 TYPE_OVERFLOW_WRAPS check.
28406 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28408 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
28409 if versioning is required.
28410 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
28411 peeling with the check for versioning.
28413 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28415 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
28416 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
28417 Set true to new parameter if new ssa variable is defined.
28418 (vect_gen_vector_loop_niters): Refactor. Set range information
28419 for the new vector loop bound variable.
28420 (vect_do_peeling): Ditto.
28422 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28424 * tree-affine.c (ssa.h): Include header file.
28425 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
28426 has wrapping overflow behavior.
28428 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28430 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
28432 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28434 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
28435 (tree_to_aff_combination): ... here.
28437 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28439 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
28440 reg_pressure model function.
28441 (ivopts_global_cost_for_size): Delete.
28442 (determine_set_costs, iv_ca_recount_cost): Call new model function
28443 ivopts_estimate_reg_pressure.
28445 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28447 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
28448 expensive than udiv. Remove floating point cases from mod.
28450 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28452 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
28453 Increase idiv cost.
28455 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28457 * config/aarch64/aarch64.md
28458 (copysignsf3): Fix mask generation.
28460 2017-06-07 Jakub Jelinek <jakub@redhat.com>
28462 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
28464 (class dump_manager): Add register_dumps method.
28465 * dumpfile.c: Include langhooks.h.
28466 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
28467 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
28468 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
28469 (dump_manager::dump_register): Start with 512 entries instead of 32.
28470 (dump_manager::register_dumps): New method.
28471 * toplev.c (general_init): Instead of invoking register_dumps
28472 langhook, invoke register_dumps method on the dump manager.
28473 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
28476 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
28478 * doc/md.texi: Clarify the restrictions on a define_insn condition.
28479 Say that # requires an associated define_split to exist, and that
28480 the define_split must be suitable for use after register allocation.
28482 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28484 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
28485 (compute_outgoing_frequencies): Also initialize zero counts.
28486 (find_many_sub_basic_blocks): Do not produce uninitialized profile
28487 around loops; preserve more of profile when nothing changes.
28489 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
28491 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
28493 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
28494 * config/arm/arm-cpu-cdata.h: Regenerate.
28495 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
28496 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
28497 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
28498 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
28499 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
28501 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
28503 * config/arm/t-rmprofile: Likewise.
28504 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
28506 2017-06-06 David S. Miller <davem@davemloft.net>
28509 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
28510 blockage if function uses alloca.
28512 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28514 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
28515 New "uid" fields to hold pretty-print IDs of group and ref.
28516 Memory references are now identified as <group_id>:<ref_id>
28517 instead of using [random] addresses.
28518 (dump_mem_details): Simplify, no functional change.
28519 (dump_mem_ref): Simplify and make output more concise.
28520 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
28521 (find_or_create_group): Initialize group uid.
28522 (record_ref): Initialize ref uid. Improve debug output.
28523 (prune_group_by_reuse, should_issue_prefetch_p,)
28524 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
28525 (mark_nontemporal_store, determine_loop_nest_reuse):
28526 Improve debug output.
28528 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28530 * dbgcnt.def (prefetch): New debug counter.
28531 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
28532 (schedule_prefetches): Stop issueing prefetches if debug counter
28535 2017-06-06 Tom de Vries <tom@codesourcery.com>
28537 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
28538 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
28540 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28542 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
28543 Use aarch64_reg_or_zero predicate for operand 4.
28544 (aarch64_compare_and_swap<mode> define_insn_and_split):
28545 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
28546 (aarch64_store_exclusive<mode>): Likewise for operand 2.
28548 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
28550 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
28551 (arm_compute_save_core_reg_mask): This.
28552 (thumb1_compute_save_reg_mask): Rename into ...
28553 (thumb1_compute_save_core_reg_mask): This.
28554 (arm_compute_save_reg0_reg12_mask): Adapt comment.
28555 (arm_compute_frame_layout): Likewise.
28557 2017-06-06 Richard Biener <rguenther@suse.de>
28559 PR tree-optimization/80974
28560 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
28561 keep or clear leaders SSA info.
28563 2017-06-06 Tom de Vries <tom@codesourcery.com>
28565 * config/nvptx/nvptx.c (split_mode_p): New function.
28566 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
28568 2017-06-06 Tom de Vries <tom@codesourcery.com>
28570 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
28572 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28575 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
28578 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28580 * shrink-wrap.c (handle_simple_exit): Update profile.
28581 (try_shrink_wrapping): Upate profile.
28583 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28585 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
28586 (tree_guess_outgoing_edge_probabilities): New.
28587 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
28588 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
28590 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28592 * ipa-split.c (split_function): Initialize return bb profile.
28594 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28596 * profile.c (compute_branch_probabilities): Also initialize
28597 EXIT_BLOCK profile.
28599 2017-06-06 Richard Biener <rguenther@suse.de>
28601 PR tree-optimization/80928
28602 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
28603 (vect_analyze_loop_operations): Properly guard analysis for
28605 (vect_transform_loop): Likewise.
28606 (vect_analyze_loop_2): Also reset SLP type on PHIs.
28607 (vect_model_induction_cost): Do not cost for pure SLP.
28608 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
28609 of induction in inner loop vectorization.
28610 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
28611 (vect_get_and_check_slp_defs): Handle vect_induction_def.
28612 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
28614 (vect_analyze_slp_cost_1): Cost induction.
28615 (vect_detect_hybrid_slp_stmts): Handle PHIs.
28616 (vect_get_slp_vect_defs): Likewise.
28617 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
28618 (vect_transform_stmt): Handle SLP reductions.
28619 * tree-vectorizer.h (vectorizable_induction): Adjust.
28621 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28623 * config/rs6000/rs6000.c (make_resolver_func): Update
28624 init_lowered_empty_function call.
28626 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28628 * doc/invoke.texi: Document the -fprofile-abs-path option.
28629 * common.opt (fprofile-abs-path): New option.
28630 * gcov-io.h (gcov_write_filename): Declare.
28631 * gcov-io.c (gcov_write_filename): New function.
28632 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28633 * profile.c (output_location): Likewise.
28635 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28637 * shring-wrap.c: Revert accidental commit.
28639 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28641 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28643 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28645 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28647 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28648 profile in callgraph edge.
28649 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28650 (apply_scale): Likewise.
28651 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28652 Also scale profile when inlining function with zero profile.
28653 (initialize_cfun): Update exit block profile even when it is zero.
28654 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28655 when profile is read.
28657 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28659 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28660 (CLONE_*): New constants to define the processors we can generate
28661 code for with the target_clone attribute.
28662 (rs6000_clone_map): New array to identify which clone processors
28663 the current program is running on.
28664 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28665 target_clone attribute.
28666 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28667 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28668 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28669 (cpu_expand_builtin): Add support for target_clone attribute.
28670 (rs6000_valid_attribute_p): Allow "default" attribute.
28671 (get_decl_name): New debug function to simplify printing the
28672 current function name in debugging statements.
28673 (rs6000_clone_priority): New functions to support the target_clone
28674 attribute, and be able to generate code to switch between ISA 2.05
28675 through ISA 3.0 (power6 through power9).
28676 (rs6000_compare_version_priority): Likewise.
28677 (rs6000_get_function_versions_dispatcher): Likewise.
28678 (make_resolver_func): Likewise.
28679 (add_condition_to_bb): Likewise.
28680 (dispatch_function_versions): Likewise.
28681 (rs6000_generate_version_dispatcher_body): Likewise.
28682 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28683 (fusion_gpr_load_p): Fix a spacing issue.
28684 * doc/extend.texi (Common Function Attributes): Document that the
28685 PowerPC supports the target_clone attribute.
28687 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28689 * config/arm/arm.h: explain F symbol found in description of ARM
28690 register allocation in its legend.
28692 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28694 * config/mips/frame-header-opt.c: Include profile-count.h.
28695 * config/riscv/riscv.c: Include profile-count.h
28697 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28699 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28701 (sm_set_flag_if_changed): Add bbs field.
28702 (execute_sm_if_changed_flag_set): Pass BBS.
28703 (execute_sm): Update.
28705 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28707 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28710 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28712 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28713 (peephole2): New peephole2 to emit the above.
28714 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28716 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28718 * config/aarch64/aarch64.c (define_peephole2 above
28719 *sub_<shift>_<mode>): New peephole.
28721 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28723 * config/i386/i386.c (make_resolver_func): Update.
28724 * Makefile.in: Add profile-count.h and profile-count.o
28725 * auto-profile.c (afdo_indirect_call): Update to new API.
28726 (afdo_set_bb_count): Update.
28727 (afdo_propagate_edge): Update.
28728 (afdo_propagate_circuit): Update.
28729 (afdo_calculate_branch_prob): Update.
28730 (afdo_annotate_cfg): Update.
28731 * basic-block.h: Include profile-count.h
28732 (struct edge_def): Turn count to profile_count.
28733 (struct basic_block_def): Likewie.
28734 (REG_BR_PROB_BASE): Move to profile-count.h
28735 (RDIV): Move to profile-count.h
28736 * bb-reorder.c (max_entry_count): Turn to profile_count.
28737 (find_traces): Update.
28738 (rotate_loop):Update.
28739 (connect_traces):Update.
28740 (sanitize_hot_paths):Update.
28741 * bt-load.c (migrate_btr_defs): Update.
28742 * cfg.c (RDIV): Remove.
28743 (init_flow): Use alloc_block.
28744 (alloc_block): Uninitialize count.
28745 (unchecked_make_edge): Uninitialize count.
28746 (check_bb_profile): Update.
28747 (dump_edge_info): Update.
28748 (dump_bb_info): Update.
28749 (update_bb_profile_for_threading): Update.
28750 (scale_bbs_frequencies_int): Update.
28751 (scale_bbs_frequencies_gcov_type): Update.
28752 (scale_bbs_frequencies_profile_count): New.
28753 * cfg.h (update_bb_profile_for_threading): Update.
28754 (scale_bbs_frequencies_profile_count): Declare.
28755 * cfgbuild.c (compute_outgoing_frequencies): Update.
28756 (find_many_sub_basic_blocks): Update.
28757 * cfgcleanup.c (try_forward_edges): Update.
28758 (try_crossjump_to_edge): Update.
28759 * cfgexpand.c (expand_gimple_tailcall): Update.
28760 (construct_exit_block): Update.
28761 * cfghooks.c (verify_flow_info): Update.
28762 (dump_bb_for_graph): Update.
28763 (split_edge): Update.
28764 (make_forwarder_block): Update.
28765 (duplicate_block): Update.
28766 (account_profile_record): Update.
28767 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28768 (get_estimated_loop_iterations): Update.
28769 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28770 (single_likely_exit): Update.
28771 * cfgloopmanip.c (scale_loop_profile): Update.
28773 (set_zero_probability): Update.
28774 (lv_adjust_loop_entry_edge): Update.
28775 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28776 (purge_dead_edges): Update.
28777 (rtl_account_profile_record): Update.
28778 * cgraph.c (cgraph_node::create): Uninitialize count.
28779 (symbol_table::create_edge): Uninitialize count.
28780 (cgraph_update_edges_for_call_stmt_node): Update.
28781 (cgraph_edge::dump_edge_flags): Update.
28782 (cgraph_node::dump): Update.
28783 (cgraph_edge::maybe_hot_p): Update.
28784 * cgraph.h: Include profile-count.h
28785 (create_clone), create_edge, create_indirect_edge): Update.
28786 (cgraph_node): Turn count to profile_count.
28787 (cgraph_edge0: Likewise.
28788 (make_speculative, clone): Update.
28789 (create_edge): Update.
28790 (init_lowered_empty_function): Update.
28791 * cgraphclones.c (cgraph_edge::clone): Update.
28792 (duplicate_thunk_for_node): Update.
28793 (cgraph_node::create_clone): Update.
28794 * cgraphunit.c (cgraph_node::analyze): Update.
28795 (cgraph_node::expand_thunk): Update.
28796 * final.c (dump_basic_block_info): Update.
28797 * gimple-streamer-in.c (input_bb): Update.
28798 * gimple-streamer-out.c (output_bb): Update.
28799 * graphite.c (print_global_statistics): Update.
28800 (print_graphite_scop_statistics): Update.
28801 * hsa-brig.c: Include basic-block.h.
28802 * hsa-dump.c: Include basic-block.h.
28803 * hsa-gen.c (T sum_slice): Update.
28804 (convert_switch_statements):Update.
28805 * hsa-regalloc.c: Include basic-block.h.
28806 * ipa-chkp.c (chkp_produce_thunks): Update.
28807 * ipa-cp.c (struct caller_statistics): Update.
28808 (init_caller_stats): Update.
28809 (gather_caller_stats): Update.
28810 (ipcp_cloning_candidate_p): Update.
28811 (good_cloning_opportunity_p): Update.
28812 (get_info_about_necessary_edges): Update.
28813 (dump_profile_updates): Update.
28814 (update_profiling_info): Update.
28815 (update_specialized_profile): Update.
28816 (perhaps_add_new_callers): Update.
28817 (decide_about_value): Update.
28818 (ipa_cp_c_finalize): Update.
28819 * ipa-devirt.c (struct odr_type_warn_count): Update.
28820 (struct decl_warn_count): Update.
28821 (struct final_warning_record): Update.
28822 (possible_polymorphic_call_targets): Update.
28823 (ipa_devirt): Update.
28824 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28825 * ipa-icf.c (sem_function::merge): Update.
28826 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28827 * ipa-inline.c (compute_uninlined_call_time): Update.
28828 (compute_inlined_call_time): Update.
28829 (want_inline_small_function_p): Update.
28830 (want_inline_self_recursive_call_p): Update.
28831 (edge_badness): Update.
28832 (lookup_recursive_calls): Update.
28833 (recursive_inlining): Update.
28834 (inline_small_functions): Update.
28835 (dump_overall_stats): Update.
28836 (dump_inline_stats): Update.
28837 * ipa-profile.c (ipa_profile_generate_summary): Update.
28838 (ipa_propagate_frequency): Update.
28839 (ipa_profile): Update.
28840 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28841 * ipa-utils.c (ipa_merge_profiles): Update.
28842 * loop-doloop.c (doloop_modify): Update.
28843 * loop-unroll.c (report_unroll): Update.
28844 (unroll_loop_runtime_iterations): Update.
28845 * lto-cgraph.c (lto_output_edge): Update.
28846 (lto_output_node): Update.
28847 (input_node): Update.
28848 (input_edge): Update.
28849 (merge_profile_summaries): Update.
28850 * lto-streamer-in.c (input_cfg): Update.
28851 * lto-streamer-out.c (output_cfg): Update.
28852 * mcf.c (create_fixup_graph): Update.
28853 (adjust_cfg_counts): Update.
28854 (sum_edge_counts): Update.
28855 * modulo-sched.c (sms_schedule): Update.
28856 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28857 * predict.c (maybe_hot_count_p): Update.
28858 (probably_never_executed): Update.
28859 (dump_prediction): Update.
28860 (combine_predictions_for_bb): Update.
28861 (propagate_freq): Update.
28862 (handle_missing_profiles): Update.
28863 (counts_to_freqs): Update.
28864 (rebuild_frequencies): Update.
28865 (force_edge_cold): Update.
28866 * predict.h: Include profile-count.h
28867 (maybe_hot_count_p, counts_to_freqs): UPdate.
28868 * print-rtl-function.c: Do not include cfg.h
28869 * print-rtl.c: Include basic-block.h
28870 * profile-count.c: New file.
28871 * profile-count.h: New file.
28872 * profile.c (is_edge_inconsistent): Update.
28873 (correct_negative_edge_counts): Update.
28874 (is_inconsistent): Update.
28875 (set_bb_counts): Update.
28876 (read_profile_edge_counts): Update.
28877 (compute_frequency_overlap): Update.
28878 (compute_branch_probabilities): Update; Initialize and deinitialize
28880 (branch_prob): Update.
28881 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
28882 (edge_gcov_count): New.
28883 (bb_gcov_count): New.
28884 * shrink-wrap.c (try_shrink_wrapping): Update.
28885 * tracer.c (better_p): Update.
28886 * trans-mem.c (expand_transaction): Update.
28887 (ipa_tm_insert_irr_call): Update.
28888 (ipa_tm_insert_gettmclone_call): Update.
28889 * tree-call-cdce.c: Update.
28890 * tree-cfg.c (gimple_duplicate_sese_region): Update.
28891 (gimple_duplicate_sese_tail): Update.
28892 (gimple_account_profile_record): Update.
28893 (execute_fixup_cfg): Update.
28894 * tree-inline.c (copy_bb): Update.
28895 (copy_edges_for_bb): Update.
28896 (initialize_cfun): Update.
28897 (freqs_to_counts): Update.
28898 (copy_cfg_body): Update.
28899 (expand_call_inline): Update.
28900 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
28901 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
28902 (try_unroll_loop_completely): Update.
28903 (try_peel_loop): Update.
28904 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
28905 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
28906 * tree-ssa-loop-split.c (connect_loops): Update.
28907 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
28908 * tree-ssa-reassoc.c (branch_fixup): Update.
28909 * tree-ssa-tail-merge.c (replace_block_by): Update.
28910 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
28911 (compute_path_counts): Update.
28912 (update_profile): Update.
28913 (recompute_probabilities): Update.
28914 (update_joiner_offpath_counts): Update.
28915 (estimated_freqs_path): Update.
28916 (freqs_to_counts_path): Update.
28917 (clear_counts_path): Update.
28918 (ssa_fix_duplicate_block_edges): Update.
28919 (duplicate_thread_path): Update.
28920 * tree-switch-conversion.c (case_bit_test_cmp): Update.
28921 (struct switch_conv_info): Update.
28922 * tree-tailcall.c (decrease_profile): Update.
28923 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
28924 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
28925 * value-prof.c (check_counter): Update.
28926 (gimple_divmod_fixed_value): Update.
28927 (gimple_mod_pow2): Update.
28928 (gimple_mod_subtract): Update.
28929 (gimple_ic_transform): Update.
28930 (gimple_stringop_fixed_value): Update.
28931 * value-prof.h (gimple_ic): Update.
28933 2017-06-02 Carl Love <cel@us.ibm.com>
28935 * config/rs6000/rs6000-c: Add support for built-in functions
28936 vector double vec_doublee (vector signed int);
28937 vector double vec_doublee (vector unsigned int);
28938 vector double vec_doublee (vector float);
28939 vector double vec_doubleh (vector signed int);
28940 vector double vec_doubleh (vector unsigned int);
28941 vector double vec_doubleh (vector float);
28942 vector double vec_doublel (vector signed int);
28943 vector double vec_doublel (vector unsigned int);
28944 vector double vec_doublel (vector float);
28945 vector double vec_doubleo (vector signed int);
28946 vector double vec_doubleo (vector unsigned int);
28947 vector double vec_doubleo (vector float);.
28948 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
28949 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
28951 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
28952 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
28953 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
28955 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
28956 vec_doublel, vec_doubleh.
28957 * doc/extend.texi: Update the built-in documentation file for the
28958 new built-in functions.
28960 2017-06-02 David Malcolm <dmalcolm@redhat.com>
28963 * ipa-inline-analysis.c (free_growth_caches): Set
28964 edge_removal_hook_holder to NULL after removing it.
28966 2017-06-02 Sudakshina Das <sudi.das@arm.com>
28968 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
28969 comparision with zero.
28971 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28972 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28973 for early expansion of vec_min and vec_max builtins.
28974 (builtin_function_type): Add min/max unsigned variants to those
28975 identified as having unsigned arguments.
28977 2017-06-02 Olivier Hainque <hainque@adacore.com>
28979 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
28981 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28983 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
28984 Use VALL_F16 iterator rather than VALL.
28986 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28988 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
28989 Emit CBNZ inside loop when doing a strong exchange and comparing
28990 against zero. Generate the CC flags after the loop.
28992 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
28994 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
28995 (dl_section_ref): New.
28996 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
28997 On AIX, append an expression to subtract the size of the
28998 section length to dl_section_ref.
29000 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29002 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29003 for early expansion of vector absolute builtins.
29005 2017-06-02 Richard Biener <rguenther@suse.de>
29007 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29008 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29010 2017-06-02 Richard Biener <rguenther@suse.de>
29012 PR tree-optimization/80948
29013 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29014 stmt order as well.
29016 2017-06-02 Richard Biener <rguenther@suse.de>
29018 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29020 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29021 for inductions as relevant.
29023 2017-06-02 Richard Biener <rguenther@suse.de>
29025 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29026 (vectorizable_induction): ... this. Remove dead code.
29028 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29030 * builtins. (expand_builtin_alloca): Remove second parameter and
29031 infer its value from the first parameter instead.
29032 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29034 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29036 PR rtl-optimization/80903
29037 * loop-doloop.c (add_test): Unshare sequence.
29039 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29041 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
29043 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
29045 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29047 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29048 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29049 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29050 (xlogue_layout::get_stub_rtx): Make static.
29051 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29052 (xlogue_layout::compute_stub_managed_regs): Rename to...
29053 (xlogue_layout::count_stub_managed_regs): ...this.
29054 (xlogue_layout::is_stub_managed_reg): New function.
29055 (xlogue_layout::m_stub_names): Rename to...
29056 (xlogue_layout::s_stub_names): ...this, make static.
29057 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29058 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29059 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29060 xlogue_layout::s_stub_names): Instantiate statics.
29061 (stub_managed_regs): Remove.
29062 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29063 (disable_call_ms2sysv_xlogues): Rename to...
29064 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29065 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29067 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29068 change after reload_completed.
29069 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29071 (ix86_expand_prologue): Likewise.
29072 (ix86_expand_epilogue): Likewise.
29073 (ix86_expand_split_stack_prologue): Likewise.
29074 (ix86_compute_frame_layout): Remove frame parameter ...
29075 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29076 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29078 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29079 (ix86_frame): Move from here ...
29080 * config/i386/i386.h (ix86_frame): ... to here.
29081 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29082 complete ix86_frame data structure instead. Remove some_ld_name.
29084 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29086 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29087 symbols that hold a DECL_VALUE_EXPR.
29089 2017-06-01 Martin Jambor <mjambor@suse.cz>
29091 PR tree-optimization/80898
29092 * tree-sra.c (process_subtree_disqualification): Removed.
29093 (disqualify_candidate): Do not acll
29094 process_subtree_disqualification.
29095 (subtree_mark_written_and_enqueue): New function.
29096 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29097 RHS has been disqualified and re-queue LHS if necessary. Apart
29098 from that, ignore disqualified RHS.
29100 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29102 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29103 address fetch for z10 or later.
29105 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29107 * config/arc/arc.md (tst_movb): Add guard when splitting.
29109 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29111 * config/arc/arc.c (arc_can_eliminate): Test against
29112 arc_frame_pointer_needed.
29114 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29116 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
29117 to prevent store reordering.
29118 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
29119 (type): Add block type.
29120 (stack_tie): Define special instruction to be used in
29123 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29125 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
29126 constraint. It is not valid for the pattern.
29127 (noncommutative_binary_comparison): Likewise.
29129 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29131 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
29134 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29136 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
29137 be used by the reg-alloc.
29139 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29141 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
29142 reg-alloc when having mul64 or mul32x16 instructions.
29143 (mulsidi3): Likewise.
29144 (umulsidi3): Likewise.
29145 (mulsi32x16): New pattern.
29146 (mulsi64): Likewise.
29147 (mulsidi64): Likewise.
29148 (umulsidi64): Likewise.
29149 (MUL32x16_REG): Define.
29150 (mul64_600): Use MUL32x16_REG.
29151 (mac64_600): Likewise.
29152 (umul64_600): Likewise.
29153 (umac64_600): Likewise.
29155 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29157 * config/arc/arc.md (mulsi3_700): Make it commutative.
29159 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
29161 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
29163 (*sign_extendsidi2_insn): Likewise for movstosw.
29165 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29167 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
29168 the type of the input discriminant value. Convert the
29169 discriminant value of signedness vary.
29171 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
29173 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
29174 Add to -Wall section.
29176 2017-06-01 Richard Biener <rguenther@suse.de>
29178 PR middle-end/66313
29179 * fold-const.c (fold_plusminus_mult_expr): If the factored
29180 factor may be zero use a wrapping type for the inner operation.
29181 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
29182 and handle moved defs.
29183 (process_assignment): Properly guard the unary op case. Return a
29184 tri-state indicating that moving the stmt before the call may allow
29185 to continue. Pass through to_move.
29186 (find_tail_calls): Handle moving unrelated defs before
29189 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
29192 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
29193 splitter result in the canonical way.
29195 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
29197 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
29198 also for 32bit target. Update insn attributes.
29199 (zero-extendsidi2 splitter): Allow all registers for operand 1.
29201 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
29203 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
29204 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
29205 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
29206 (_mm_maskz_min_ss): New intrinsics.
29208 2017-05-31 Martin Liska <mliska@suse.cz>
29210 * tree-vect-loop.c (vect_create_epilog_for_reduction):
29211 Change comment style to one we normally use.
29212 (vectorizable_reduction): Likewise.
29213 (vectorizable_induction): Likewise.
29214 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
29215 (vectorizable_call): Likewise.
29216 (vectorizable_simd_clone_call): Likewise.
29217 (vectorizable_conversion): Likewise.
29218 (vectorizable_assignment): Likewise.
29219 (vectorizable_shift): Likewise.
29220 (vectorizable_operation): Likewise.
29221 (vectorizable_store): Likewise.
29222 (vectorizable_load): Likewise.
29223 * tree-vectorizer.h: Likewise.
29225 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
29227 * passes.c (emergency_dump_function): New.
29228 * tree-pass.h (emergency_dump_function): Declare.
29229 * plugin.c (plugins_internal_error_function): Remove.
29230 * plugin.h (plugins_internal_error_function): Remove declaration.
29231 * toplev.c (internal_error_function): New static function. Use it...
29232 (general_init): ...here.
29234 2017-05-31 Graham Markall <graham.markall@embecosm.com>
29236 * config/arc/arc.c (arc_print_operand): Handle constant operands.
29237 (arc_rtx_costs): Add costs for new patterns.
29238 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
29239 * config/arc/predicates.md: Add _1_2_3_operand predicate.
29241 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29243 * tree-ssa-strlen.c (get_next_strinfo): New function.
29244 (get_stridx_plus_constant): Use it.
29245 (zero_length_string): Likewise.
29246 (adjust_related_strinfos): Likewise.
29247 (adjust_last_stmt): Likewise.
29249 2017-05-31 Richard Biener <rguenther@suse.de>
29252 * config/i386/i386.c (ix86_expand_builtin): Remove assert
29253 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
29255 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29257 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
29258 loop_vinfo argument and use of dependence distance vectors.
29259 Check instead whether the two references differ only in their
29260 initial value and assume that they have the same alignment if the
29261 difference is a multiple of the vector alignment.
29262 (vect_analyze_data_refs_alignment): Update call accordingly.
29264 2017-05-31 Martin Liska <mliska@suse.cz>
29267 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
29269 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29271 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
29272 (create_intersect_range_checks): Move from ...
29273 * tree-data-ref.c (create_intersect_range_checks_index)
29274 (create_intersect_range_checks): ... to here.
29275 (create_runtime_alias_checks): New function factored from ...
29276 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
29277 here. Call above function.
29278 * tree-data-ref.h (create_runtime_alias_checks): New function.
29280 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29282 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
29283 segment length for dr_b and compute it in wide_int.
29285 2017-05-31 Richard Biener <rguenther@suse.de>
29287 PR tree-optimization/80906
29288 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
29289 and pass through iv_map.
29290 (copy_bb_and_scalar_dependences): Adjust.
29291 (translate_pending_phi_nodes): Likewise.
29292 (copy_loop_close_phi_args): Handle code-generating IVs instead
29295 2017-05-30 David Malcolm <dmalcolm@redhat.com>
29297 * diagnostic-color.c (color_dict): Add "type-diff".
29298 (parse_gcc_colors): Update comment.
29299 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
29300 -fdiagnostics-show-template-tree and -fno-elide-type.
29301 (GCC_COLORS): Add type-diff to example.
29303 (-fdiagnostics-show-template-tree): New.
29304 (-fno-elide-type): New.
29305 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
29306 the pp_format_decoder callback. Call any m_format_postprocessor's
29308 (pretty_printer::pretty_printer): Initialize
29309 m_format_postprocessor.
29310 (pretty_printer::~pretty_printer): Delete any
29311 m_format_postprocessor.
29312 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
29313 (class format_postprocessor): New class.
29314 (struct pretty_printer::format_decoder): Document the new parameters.
29315 (struct pretty_printer::m_format_postprocessor): New field.
29316 * tree-diagnostic.c (default_tree_printer): Update for new
29317 bool and const char ** params.
29318 * tree-diagnostic.h (default_tree_printer): Likewise.
29320 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
29322 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
29323 (lwa_operand): Delete rs6000_gen_cell_microcode test.
29324 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29325 rs6000_gen_cell_microcode code.
29326 (rs6000_final_prescan_insn): Delete.
29327 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
29328 "warn-cell-microcode" entries.
29329 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
29330 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
29331 throughout. Change cc_reg_not_micro_cr0_operand to
29332 cc_reg_not_cr0_operand throughout.
29333 (*extendhi<mode>2_noload): Delete.
29334 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
29335 (mwarn-cell-microcode): Delete.
29336 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
29337 -mgen-cell-microcode and -mwarn-cell-microcode.
29339 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
29342 * config/i386/constraints.md (Yd): New constraint.
29344 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
29345 and (?Yd, r) alternatives. Update insn attributes.
29346 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
29347 and (?*Yd, r) alternatives. Update insn attributes.
29348 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
29350 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
29352 * gimplify.c (gimplify_modify_expr): Don't create a
29353 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
29356 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
29358 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
29360 2017-05-30 Richard Biener <rguenther@suse.de>
29362 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
29363 and reduc_def fields.
29364 (STMT_VINFO_REDUC_TYPE): New define.
29365 (STMT_VINFO_REDUC_DEF): Likewise.
29366 (vect_force_simple_reduction): Adjust prototype.
29367 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
29368 (vect_is_simple_reduction): Remove check_reduction argument.
29369 (vect_force_simple_reduction): Adjust and set
29370 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29371 (vectorizable_reduction): Do not re-do reduction analysis
29372 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29373 * tree-parloops.c (gather_scalar_reductions): Adjust.
29375 2017-05-30 Richard Biener <rguenther@suse.de>
29377 PR middle-end/80901
29378 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
29381 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
29383 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
29384 Introduce unknown_misalignment parameter and remove vf.
29385 (vect_peeling_hash_get_lowest_cost):
29386 Pass unknown_misalignment parameter.
29387 (vect_enhance_data_refs_alignment):
29388 Fix unsupportable data ref treatment.
29390 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29392 * tree-vect-data-refs.c (vect_get_data_access_cost):
29393 Workaround for SLP handling.
29394 (vect_enhance_data_refs_alignment):
29395 Compute costs for doing no peeling at all, compare to the best
29396 peeling costs so far and avoid peeling if cheaper.
29398 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29400 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
29401 Return peeling info and set costs to zero for unlimited cost
29403 (vect_enhance_data_refs_alignment): Also inspect all datarefs
29404 with unknown misalignment. Compute and costs for unknown
29405 misalignment, compare them to the costs for known misalignment
29406 and choose the cheapest for peeling.
29408 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29410 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
29411 (vect_get_peeling_costs_all_drs): Create function.
29412 (vect_peeling_hash_get_lowest_cost):
29413 Use vect_get_peeling_costs_all_drs.
29414 (vect_peeling_supportable): Create function.
29415 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
29417 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29419 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
29420 DR_HAS_NEGATIVE_STEP.
29421 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
29422 (vect_enhance_data_refs_alignment): Use.
29423 (vect_duplicate_ssa_name_ptr_info): Use.
29424 * tree-vectorizer.h (dr_misalignment): Use.
29425 (known_alignment_for_access_p): Use.
29427 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
29430 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
29432 (has_section_name): New function.
29434 2017-05-30 Martin Liska <mliska@suse.cz>
29437 * auto-profile.c (get_function_decl_from_block): Fix
29440 2017-05-30 Richard Biener <rguenther@suse.de>
29442 PR middle-end/80876
29443 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
29445 2017-05-30 Martin Liska <mliska@suse.cz>
29447 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
29448 * dumpfile.h (struct dump_file_info): Remove ctors.
29450 2017-05-30 Martin Liska <mliska@suse.cz>
29452 * predict.def: Fix GNU coding style.
29454 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29456 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29457 Mark 'to' argument with ATTRIBUTE_UNUSED.
29459 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29461 * config/xtensa/xtensa.c (xtensa_emit_call): Use
29462 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
29463 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
29466 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
29468 * doc/install.texi (Options specification): Restore entry of
29469 --enable-sjlj-exceptions.
29471 2017-05-27 Michael Eager <eager@eagercon.com>
29474 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
29476 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
29478 * config/microblaze/microblaze.h
29479 (FIXED_REGISTERS): Update in macro.
29480 (CALL_USED_REGISTERS): Update in macro.
29482 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
29484 * doc/install.texi: Add links to macOS binary distributions.
29486 2017-05-27 Jakub Jelinek <jakub@redhat.com>
29490 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29492 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29494 2017-05-26 Martin Liska <mliska@suse.cz>
29496 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
29498 2017-05-26 Martin Liska <mliska@suse.cz>
29500 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
29501 always leading ';; '.
29502 (dump_bb_info): Likewise.
29503 (brief_dump_cfg): Likewise.
29504 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
29505 * dumpfile.c: Remove usage of TDF_VERBOSE.
29506 * dumpfile.h (enum dump_kind): Likewise.
29507 (dump_gimple_bb_header): Do not use TDF_COMMENT.
29508 * print-tree.c (debug_verbose): Remove.
29509 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
29510 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
29511 * tree-diagnostic.c (default_tree_printer): Replace
29512 TDF_DIAGNOSTIC with TDF_SLIM.
29514 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29516 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
29517 in parameter loop, rather than loop_vinfo.
29518 (create_intersect_range_checks): Ditto.
29519 (vect_create_cond_for_alias_checks): Update call to above functions.
29521 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29523 PR tree-optimization/80815
29524 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
29525 for merging runtime alias checks. Handle negative DR_STEPs.
29527 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29529 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
29531 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
29532 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
29533 out code pruning runtime alias checks.
29534 * tree-data-ref.c (prune_runtime_alias_test_list): New function
29535 factored out from above.
29536 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29538 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29540 (prune_runtime_alias_test_list): New decalaration.
29542 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29544 * tree-vect-data-refs.c (compare_tree): Rename and move ...
29545 * tree-data-ref.c (data_ref_compare_tree): ... to here.
29546 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
29547 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
29548 (operator==, comp_dr_with_seg_len_pair): Ditto.
29549 (vect_prune_runtime_alias_test_list): Ditto.
29551 2017-05-26 Martin Liska <mliska@suse.cz>
29554 * params.def: Bound partial-inlining-entry-probability param.
29556 2017-05-26 Marek Polacek <polacek@redhat.com>
29559 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
29562 2017-05-26 Richard Biener <rguenther@suse.de>
29564 PR tree-optimization/80842
29565 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
29568 2017-05-26 Richard Biener <rguenther@suse.de>
29570 PR tree-optimization/80844
29571 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
29573 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
29575 * doc/md.texi (Machine Constraints): Update x86 family
29576 machine constraints section to match 'config/i386/constraints.md'.
29578 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
29580 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
29582 2017-05-25 Nathan Sidwell <nathan@acm.org>
29584 * doc/invoke.texi (--enable-languages): Update documentation.
29586 2017-05-25 Martin Liska <mliska@suse.cz>
29588 * dumpfile.c: Add TDF_FOLDING.
29589 * dumpfile.h (enum dump_kind): Likewise.
29590 * genmatch.c (dt_simplify::gen_1): Use it.
29592 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29594 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
29596 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29598 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29599 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
29601 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29603 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
29604 * match.pd (X == C): Rewrite it here.
29605 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
29606 with_certain_nonzero_bits2): New predicates.
29607 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
29609 2017-05-24 Nathan Sidwell <nathan@acm.org>
29611 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
29614 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
29617 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29619 * config/powerpcspe: New port. Files are copied from the rs6000
29620 port, with "rs6000" in filenames replaced by "powerpcspe".
29622 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29624 PR rtl-optimization/80754
29625 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29627 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29629 * config/sparc/sparc.md (length): Return the correct value for -mflat
29630 sibcalls to match output_sibcall.
29632 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29636 * config/rs6000/rs6000.c (struct machine_function): Add new field
29638 (rs6000_get_separate_components): Init that field, use it.
29639 (rs6000_components_for_bb): Use the field.
29641 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29643 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29645 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29647 PR middle-end/80823
29648 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29650 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29653 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29654 address against address_operand predicate.
29655 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29657 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29659 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29660 parameters passed indirectly.
29662 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29664 * config/i386/i386.md (*movdi_internal): Remove SSE4
29665 alternative 18 (?r, *v). Update insn attributes.
29666 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29667 Update insn attributes.
29668 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29669 Update insn attributes.
29670 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29671 alternative 1 (r, v). Remove isa attribute.
29672 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29673 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29674 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29676 2017-05-23 Tom de Vries <tom@codesourcery.com>
29678 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29681 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29683 * cgraphunit.c (symbol_table::process_new_functions): Update.
29684 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29685 (inline_generate_summary): Rename to ...
29686 (ipa_fn_summary_generate): ... this one.
29687 (inline_read_summary): Rename to ...
29688 (ipa_fn_summary_read): ... this one.
29689 (inline_write_summary): Rename to ...
29690 (ipa_fn_summary_write): ... this one.
29691 (inline_free_summary): Rename to ...
29692 (ipa_free_fn_summary): ... this one.
29693 (pass_data_local_fn_summary, pass_local_fn_summary,
29694 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29695 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29696 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29697 make_pass_ipa_fn_summary): New.
29698 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29699 inline_write_summary, inline_free_summary): Remove.
29700 (ipa_free_fn_summary) : New.
29701 * ipa-inline.c (ipa_inline): Update.
29702 (pass_ipa_inline): Do not generate summaries.
29703 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29705 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29706 and add pass_ipa_fn_summary.
29707 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29709 (make_pass_inline_parameters): Remove.
29711 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29713 * omp-low.c (struct omp_context): Remove "default_kind" member.
29716 * omp-offload.c (execute_oacc_device_lower): Remove the
29717 parallelism dimensions function attributes for unparallelized
29718 OpenACC kernels constructs.
29720 2017-05-23 Martin Liska <mliska@suse.cz>
29722 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29724 (cgraph_edge::make_speculative): Likewise.
29725 (cgraph_edge::resolve_speculation): Likewise.
29726 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29727 (cgraph_node::dump): Likewise.
29728 * cgraph.h: Likewise.
29729 * cgraphunit.c (analyze_functions): Likewise.
29730 (symbol_table::compile): Likewise.
29731 * ipa-cp.c (print_all_lattices): Likewise.
29732 (determine_versionability): Likewise.
29733 (initialize_node_lattices): Likewise.
29734 (ipcp_verify_propagated_values): Likewise.
29735 (estimate_local_effects): Likewise.
29736 (update_profiling_info): Likewise.
29737 (create_specialized_node): Likewise.
29738 (perhaps_add_new_callers): Likewise.
29739 (decide_about_value): Likewise.
29740 (decide_whether_version_node): Likewise.
29741 (identify_dead_nodes): Likewise.
29742 (ipcp_store_bits_results): Likewise.
29743 * ipa-devirt.c (dump_targets): Likewise.
29744 (ipa_devirt): Likewise.
29745 * ipa-icf.c (sem_item::dump): Likewise.
29746 (sem_function::equals): Likewise.
29747 (sem_variable::equals): Likewise.
29748 (sem_item_optimizer::read_section): Likewise.
29749 (sem_item_optimizer::execute): Likewise.
29750 (congruence_class::dump): Likewise.
29751 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29752 (dump_inline_summary): Likewise.
29753 (estimate_node_size_and_time): Likewise.
29754 (inline_analyze_function): Likewise.
29755 * ipa-inline-transform.c (inline_call): Likewise.
29756 * ipa-inline.c (report_inline_failed_reason): Likewise.
29757 (want_early_inline_function_p): Likewise.
29758 (edge_badness): Likewise.
29759 (update_edge_key): Likewise.
29760 (inline_small_functions): Likewise.
29761 * ipa-profile.c (ipa_profile): Likewise.
29762 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29763 (ipa_make_edge_direct_to_target): Likewise.
29764 (remove_described_reference): Likewise.
29765 (ipa_impossible_devirt_target): Likewise.
29766 (propagate_controlled_uses): Likewise.
29767 (ipa_print_node_params): Likewise.
29768 (ipcp_transform_function): Likewise.
29769 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29770 (propagate_pure_const): Likewise.
29771 * ipa-reference.c (generate_summary): Likewise.
29772 (read_write_all_from_decl): Likewise.
29773 (propagate): Likewise.
29774 (ipa_reference_read_optimization_summary): Likewise.
29775 * ipa-utils.c (ipa_merge_profiles): Likewise.
29776 * ipa.c (walk_polymorphic_call_targets): Likewise.
29777 (symbol_table::remove_unreachable_nodes): Likewise.
29778 (ipa_single_use): Likewise.
29779 * passes.c (execute_todo): Likewise.
29780 * predict.c (drop_profile): Likewise.
29781 * symtab.c (symtab_node::get_dump_name): New function.
29782 (symtab_node::dump_name): Likewise.
29783 (symtab_node::dump_asm_name): Likewise.
29784 (symtab_node::dump_references): Likewise.
29785 (symtab_node::dump_referring): Likewise.
29786 (symtab_node::dump_base): Likewise.
29787 (symtab_node::debug_symtab): Likewise.
29788 * tree-sra.c (convert_callers_for_node): Likewise.
29789 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29790 * value-prof.c (init_node_map): Likewise.
29792 2017-05-23 Martin Liska <mliska@suse.cz>
29794 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29795 and symtab_node::debug_symtab to symbol_table::debug.
29796 * cgraphunit.c (analyze_functions): Use the renamed function.
29797 (symbol_table::compile): Likewise.
29798 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29799 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29800 * passes.c (execute_todo): Likewise.
29801 * symtab.c (symbol_table::dump): New function.
29802 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29804 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29806 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29807 that nonconst implies exec.
29809 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29811 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29812 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29813 (inline_edge_summary_vec): Turn into ...
29814 (ipa_call_summaries): ... this one.
29815 (redirect_to_unreachable, edge_set_predicate,
29816 evaluate_properties_for_edge, inline_summary_alloc,
29817 reset_ipa_call_summary, reset_inline_summary,
29818 inline_summary_t::duplicate): Update.
29819 (inline_edge_duplication_hook): Turn to ...
29820 (ipa_call_summary_t::duplicate): ... this one.
29821 (inline_edge_removal_hook): Turn to ...
29822 (ipa_call_summary_t::remove): ... this one.
29823 (dump_inline_edge_summary): Turn to ...
29824 (dump_ipa_call_summary): ... this one.
29825 (estimate_function_body_sizes): Update.
29826 (inline_update_callee_summaries): Update.
29827 (remap_edge_change_prob): Update.
29828 (remap_edge_summaries): Update.
29829 (inline_merge_summary): Update.
29830 (do_estimate_edge_time): Update.
29831 (inline_generate_summary): Update.
29832 (inline_read_section): Update.
29833 (inline_read_summary): Update.
29834 (inline_free_summary): Update.
29835 * ipa-inline.c (can_inline_edge_p): Update.
29836 (compute_inlined_call_time): Update.
29837 (want_inline_small_function_p): Update.
29838 (edge_badness): Update.
29839 (early_inliner): Update.
29840 * ipa-inline.h (inline_edge_summary): Turn to ...
29841 (ipa_call_summary): ... this one.
29842 (ipa_call_summary_t): New class.
29843 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29844 (ipa_call_summaries): New.
29845 (inline_edge_summary): Remove.
29846 (estimate_edge_growth): Update.
29847 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29848 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29849 * ipa-split.c (execute_split_functions): Update.
29850 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29852 2017-05-23 Tom de Vries <tom@codesourcery.com>
29854 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29855 attributes): Document rdrand effective target.
29857 2017-05-23 Tom de Vries <tom@codesourcery.com>
29859 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29860 attributes): Sort alphabetically.
29862 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29864 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29866 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29869 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
29870 V2DF/V2DI splat into two separate patterns, one that handles
29871 registers, and the other that only handles memory. Drop support
29872 for splatting from a GPR on ISA 2.07 and then splitting the
29873 splat into direct move and splat.
29874 (vsx_splat_<mode>_reg): Likewise.
29875 (vsx_splat_<mode>_mem): Likewise.
29877 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
29879 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
29881 2017-05-22 Jakub Jelinek <jakub@redhat.com>
29883 PR middle-end/80809
29884 * omp-low.c (finish_taskreg_remap): New function.
29885 (finish_taskreg_scan): If unit size of ctx->record_type
29886 is non-constant, unshare the size expression and replace
29887 decls in it with possible outer var refs.
29889 PR middle-end/80809
29890 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
29891 GOVD_SHARED rather than GOVD_PRIVATE with it.
29892 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
29893 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
29895 PR middle-end/80853
29896 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
29897 as last argument to build_outer_var_ref for pointer bases of array
29898 section reductions.
29900 2017-05-19 Martin Sebor <msebor@redhat.com>
29902 * print-tree.c (print_node): Print DECL_READ_P flag.
29904 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29906 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
29907 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
29908 * cgraph.c: Likewise.
29909 * cgraphunit.c: Likewise.
29910 * gengtype.c: Likewise.
29911 * ipa-cp.c: Likewise.
29912 * ipa-devirt.c: Likewise.
29913 * ipa-icf.c: Likewise.
29914 * ipa-predicate.c: Likewise.
29915 * ipa-profile.c: Likewise.
29916 * ipa-prop.c: Likewise.
29917 * ipa-split.c: Likewise.
29919 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
29920 edge_predicate_pool, dump_inline_hints,
29921 inline_summary::account_size_time, redirect_to_unreachable,
29922 edge_set_predicate, set_hint_predicate,
29923 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
29924 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
29925 inline_summary_t::remove, remap_hint_predicate_after_duplication,
29926 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
29927 ipa_call_summary_t::remove, initialize_growth_caches,
29928 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
29929 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
29930 mark_modified, unmodified_parm_1, unmodified_parm,
29931 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
29932 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
29933 compute_bb_predicates, will_be_nonconstant_expr_predicate,
29934 will_be_nonconstant_predicate, record_modified_bb_info,
29935 get_minimal_bb, record_modified, param_change_prob,
29936 phi_result_unknown_predicate, predicate_for_phi_result,
29937 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
29938 estimate_function_body_sizes, compute_inline_parameters,
29939 compute_inline_parameters_for_curren, pass_data_inline_parameters,
29940 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
29941 inline_update_callee_summaries, remap_edge_change_prob,
29942 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
29943 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
29944 inline_analyze_function, inline_summary_t::insert,
29945 inline_generate_summary, read_ipa_call_summary, inline_read_section,
29946 inline_read_summary, write_ipa_call_summary, inline_write_summary,
29947 inline_free_summary): Move to ipa-fnsummary.h
29948 (predicate_t): Remove.
29949 * ipa-fnsummary.c: New file.
29950 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
29951 (enum inline_hints_vals, inline_hints, agg_position_info,
29952 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
29953 inline_summaries, ipa_call_summary, ipa_call_summary_t,
29954 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
29955 dump_inline_summary, dump_inline_hints, inline_generate_summary,
29956 inline_read_summary, inline_write_summary, inline_free_summary,
29957 inline_analyze_function, initialize_inline_failed,
29958 inline_merge_summary, inline_update_overall_summary,
29959 compute_inline_parameters): Move to ipa-fnsummary.h
29960 * ipa-fnsummary.h: New file.
29961 * ipa-inline-transform.h: Include ipa-inline.h.
29962 * ipa-inline.c: LIkewise.
29964 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29966 * ipa-inline.c (edge_badness): Use inlined_time instead of
29967 inline_summaries->get.
29969 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29971 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
29973 2017-05-22 Nathan Sidwell <nathan@acm.org>
29975 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
29976 (fdump-lang): Document 'raw' option.
29977 * dumpfile.h (TDI_tu): Delete.
29978 * dumpfile.c (dump_files): Remove translation-unit.
29979 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
29981 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
29983 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
29984 command option from $(AWK) call.
29985 * config/avr/genmultilib.awk: Simplify and rewrite so that it
29986 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
29987 [FORMAT]: Remove handling of variable.
29988 * config/avr/t-multilib: Regenerate.
29990 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29992 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
29994 (dump_inline_summary): Do not print self_time.
29995 (estimate_function_body_sizes): Do not set self_time.
29996 (compute_inline_parameters): Likewise.
29997 (inline_read_section, inline_write_summary): Do not stream self_time.
29998 * ipa-inline.h (inline_summary): Drop self_time.
30000 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30002 * ipa-inline-analysis.c (account_size_time): Rename to ...
30003 (inline_summary::account_size_time): ... this one.
30004 (reset_ipa_call_summary): Turn to ...
30005 (ipa_call_summary::reset): ... this one.
30006 (reset_inline_summary): Turn to ...
30007 (inline_summary::reset): ... this one.
30008 (inline_summary_t::remove): Update.
30009 (inline_summary_t::duplicate): Update.
30010 (ipa_call_summary_t::remove): Update.
30011 (dump_inline_summary): Update.
30012 (estimate_function_body_sizes): Update.
30013 (compute_inline_parameters): Update.
30014 (estimate_node_size_and_time): Update.
30015 (inline_merge_summary): Update.
30016 (inline_update_overall_summary): Update.
30017 (inline_read_section): Update.
30018 (inline_write_summary): Update.
30019 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30020 add account_size_time and reset member functions.
30021 (ipa_call_summary): Add reset function.
30022 * ipa-predicate.h (predicate::operator &): Constify.
30024 2017-05-22 Richard Biener <rguenther@suse.de>
30026 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30028 2017-05-19 Jason Merrill <jason@redhat.com>
30030 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30032 2017-05-19 Marek Polacek <polacek@redhat.com>
30035 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30036 TYPE_OVERFLOW_WRAPS checks.
30038 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30040 * tree-core.h (enum omp_clause_default_kind): Add
30041 "OMP_CLAUSE_DEFAULT_PRESENT".
30042 * tree-pretty-print.c (dump_omp_clause): Handle it.
30043 * gimplify.c (enum gimplify_omp_var_data): Add
30044 "GOVD_MAP_FORCE_PRESENT".
30045 (gimplify_adjust_omp_clauses_1): Map it to
30046 "GOMP_MAP_FORCE_PRESENT".
30047 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30049 * gimplify.c (oacc_default_clause): Clarify.
30051 2017-05-19 Nathan Sidwell <nathan@acm.org>
30053 LANG_HOOK_REGISTER_DUMPS
30054 * toplev.c (general_init): Call register dump lang hook.
30055 * doc/invoke.texi: Document -fdump-lang option family.
30056 * dumpfile.c (dump_files): Remove class dump here.
30057 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30058 * dumpfile.h (tree_dump_index): Remove TDI_class.
30059 * langhooks-def.h (lhd_register_dumps): Declare.
30060 (LANG_HOOKS_REGISTER_DUMPS): Define.
30061 (LANG_HOOKS_INITIALIZER): Add it.
30062 * langhooks.c (lhd_register_dumps): Define.
30063 * langhooks.h (struct lang_hooks): Add register_dumps.
30065 2017-05-19 Nathan Sidwell <nathan@acm.org>
30067 * context.h (context::set_passes): New.
30068 * context.c (context::context): Do not create pass manager.
30069 * toplev.c (general_init): Create pass manager here.
30071 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30073 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30074 use this splitter if two add or or instructions would also work for
30075 the constant we want to generate.
30077 2017-05-19 Richard Biener <rguenther@suse.de>
30080 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30081 predicate evaluation.
30083 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30085 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30087 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30090 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30092 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30093 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30094 (inline_edge_summary_vec): Turn into ...
30095 (ipa_call_summaries): ... this one.
30096 (redirect_to_unreachable, edge_set_predicate,
30097 evaluate_properties_for_edge, inline_summary_alloc,
30098 reset_ipa_call_summary, reset_inline_summary,
30099 inline_summary_t::duplicate): Update.
30100 (inline_edge_duplication_hook): Turn to ...
30101 (ipa_call_summary_t::duplicate): ... this one.
30102 (inline_edge_removal_hook): Turn to ...
30103 (ipa_call_summary_t::remove): ... this one.
30104 (dump_inline_edge_summary): Turn to ...
30105 (dump_ipa_call_summary): ... this one.
30106 (estimate_function_body_sizes): Update.
30107 (inline_update_callee_summaries): Update.
30108 (remap_edge_change_prob): Update.
30109 (remap_edge_summaries): Update.
30110 (inline_merge_summary): Update.
30111 (do_estimate_edge_time): Update.
30112 (inline_generate_summary): Update.
30113 (inline_read_section): Update.
30114 (inline_read_summary): Update.
30115 (inline_free_summary): Update.
30116 * ipa-inline.c (can_inline_edge_p): Update.
30117 (compute_inlined_call_time): Update.
30118 (want_inline_small_function_p): Update.
30119 (edge_badness): Update.
30120 (early_inliner): Update.
30121 * ipa-inline.h (inline_edge_summary): Turn to ...
30122 (ipa_call_summary): ... this one.
30123 (ipa_call_summary_t): New class.
30124 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30125 (ipa_call_summaries): New.
30126 (inline_edge_summary): Remove.
30127 (estimate_edge_growth): Update.
30128 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30129 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30130 * ipa-split.c (execute_split_functions): Update.
30131 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30133 2017-05-19 Richard Biener <rguenther@suse.de>
30135 PR middle-end/80764
30136 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
30138 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
30140 * config/rs6000/rs6000.c (struct machine_function): Add field
30141 fpr_is_wrapped_separately.
30142 (rs6000_get_separate_components): Use 64 components. Handle the
30143 new FPR components.
30144 (rs6000_components_for_bb): Handle the FPR components.
30145 (rs6000_emit_prologue_components): Handle the FPR components.
30146 (rs6000_emit_epilogue_components): Handle the FPR components.
30147 (rs6000_set_handled_components): Handle the FPR components.
30148 (rs6000_emit_prologue): Don't output prologue code for those FPRs
30149 that are already separately shrink-wrapped.
30150 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
30151 that are already separately shrink-wrapped.
30153 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30156 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
30159 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
30160 (define_peephole2 for Altivec d-form load): Add peepholes to catch
30161 cases where the register allocator uses a move and an offsettable
30162 memory operation to/from a FPR register on ISA 2.06/2.07.
30163 (define_peephole2 for Altivec d-form store): Likewise.
30165 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
30168 * config/i386/mmx.md (*mov<mode>_internal): Enable
30169 alternatives 11, 12, 13 and 14 also for 32bit targets.
30170 Remove alternatives 15, 16, 17 and 18.
30171 * config/i386/sse.md (vec_concatv2di): Change
30172 alternative (!x, *y) to (x, ?!*Yn).
30174 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
30176 * dumpfile.h (enum dump_kind): Remove stray comma.
30178 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30180 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
30181 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
30182 predicate::num_conditions
30183 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
30184 (CHANGED): turn into predicate::changed.
30185 (agg_position_info): Move to ipa-predicate.h
30186 (add_condition, predicate::add_clause, predicate::operator &=,
30187 predicate::or_with, predicate::evaluate, predicate::probability,
30188 dump_condition, dump_clause, predicate::dump,
30189 predicate::remap_after_duplication, predicate::remap_after_inlining,
30190 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
30191 (evaluate_conditions_for_known_args): Update.
30192 (set_cond_stmt_execution_predicate): Update.
30193 * ipa-inline.h: Include ipa-predicate.h
30194 (condition, inline_param_summary, conditions, agg_position_info,
30195 predicate): Move to ipa-predicate.h
30196 * ipa-predicate.c: New file.
30197 * ipa-predicate.h: New file.
30199 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
30201 * final.c (leaf_function_p): Check we are not in a sequence.
30203 2017-05-18 Martin Liska <mliska@suse.cz>
30205 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
30206 * dumpfile.c (dump_register): Use new enum dump_kind.
30207 (get_dump_file_name): Likewise.
30208 (dump_enable_all): Likewise.
30209 (dump_switch_p_1): Likewise.
30210 (enable_rtl_dump_file): Remove usage of TDF_RTL.
30211 * dumpfile.h (enum dump_kind): New enum type.
30212 (struct dump_file_info): Create constructor and
30213 format fields and comments.
30214 * passes.c (pass_manager::register_one_dump_file):
30216 * statistics.c (statistics_early_init): Likewise.
30217 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
30218 TDF_TREE with TDF_SLIM.
30219 (gather_memory_references_ref): Likewise.
30221 2017-05-18 Martin Liska <mliska@suse.cz>
30223 * vec.h (struct vnull): Use it.
30225 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30227 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
30228 (true_predicate, false_predicate, true_predicate_p,
30229 false_predicate_p): Remove.
30230 (single_cond_predicate, not_inlined_predicate): Turn to member function
30232 (add_condition): Update.
30233 (add_clause): Turn to...
30234 (predicate::add_clause): ... this one; update; allow passing NULL
30236 (and_predicates): Turn to ...
30237 (predicate::operator &=): ... this one.
30238 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
30239 (or_predicates): Turn to ...
30240 (predicate::or_with): ... this one.
30241 (evaluate_predicate): Turn to ...
30242 (predicate::evaluate): ... this one.
30243 (predicate_probability): Turn to ...
30244 (predicate::probability): ... this one.
30245 (dump_condition): Update.
30246 (dump_predicate): Turn to ...
30247 (predicate::dump): ... this one.
30248 (account_size_time): Update.
30249 (edge_set_predicate): Update.
30250 (set_hint_predicate): UPdate.
30251 (evaluate_conditions_for_known_args): Update.
30252 (evaluate_properties_for_edge): Update.
30253 (remap_predicate_after_duplication): Turn to...
30254 (predicate::remap_after_duplication): ... this one.
30255 (remap_hint_predicate_after_duplication): Update.
30256 (inline_summary_t::duplicate): UPdate.
30257 (dump_inline_edge_summary): Update.
30258 (dump_inline_summary): Update.
30259 (set_cond_stmt_execution_predicate): Update.
30260 (set_switch_stmt_execution_predicate): Update.
30261 (compute_bb_predicates): Update.
30262 (will_be_nonconstant_expr_predicate): Update.
30263 (will_be_nonconstant_predicate): Update.
30264 (phi_result_unknown_predicate): Update.
30265 (predicate_for_phi_result): Update.
30266 (array_index_predicate): Update.
30267 (estimate_function_body_sizes): Update.
30268 (estimate_node_size_and_time): Update.
30269 (estimate_ipcp_clone_size_and_time): Update.
30270 (remap_predicate): Rename to ...
30271 (predicate::remap_after_inlining): ... this one.
30272 (remap_hint_predicate): Update.
30273 (inline_merge_summary): Update.
30274 (inline_update_overall_summary): Update.
30275 (estimate_size_after_inlining): Update.
30276 (read_predicate): Rename to ...
30277 (predicate::stream_in): ... this one.
30278 (read_inline_edge_summary): Update.
30279 (write_predicate): Rename to ...
30280 (predicate::stream_out): ... this one.
30281 (write_inline_edge_summary): Update.
30282 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
30283 (clause_t): Turn to uint32_t
30284 (predicate): Turn to class; implement constructor and operators
30286 (size_time_entry): Update.
30287 (inline_summary): Update.
30288 (inline_edge_summary): Update.
30290 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
30292 * fold-const.c (fold_binary_loc): Move transformation...
30293 * match.pd (C - X CMP X): ... here.
30295 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
30297 * config/sparc/sparc.c (sparc_option_override): Set function
30298 alignment for -mcpu=niagara7 to 64 to match the I$ line.
30299 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
30301 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
30303 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
30305 2017-05-18 Marek Polacek <polacek@redhat.com>
30308 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
30309 (pass_ubsan::execute): Call gimple_assign_single_p instead of
30310 gimple_assign_load_p.
30312 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
30314 PR middle-end/80692
30315 * real.c (do_compare): Give decimal_do_compare preference over
30316 comparing just the signs.
30318 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
30320 * doc/md.texi (Canonicalization of Instructions): Describe the
30321 canonical form of instructions that inherently set a condition
30324 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
30326 PR middle-end/80775
30327 * tree-cfg.c: Move deletion of unreachable case statements to after
30328 the merging of consecutive case labels.
30330 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
30332 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
30333 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
30334 restoring of callee-saved registers.
30336 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
30338 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
30339 * config/visium/visium.c (single_set_and_flags): Likewise.
30340 * config/visium/visium.md (Substitutions): Likewise.
30342 2017-05-17 Martin Liska <mliska@suse.cz>
30344 * cfg.c: Introduce dump_flags_t type and
30345 use it instead of int type.
30347 * cfghooks.c: Likewise.
30348 * cfghooks.h (struct cfg_hooks): Likewise.
30349 * cfgrtl.c: Likewise.
30350 * cfgrtl.h: Likewise.
30351 * cgraph.c (cgraph_node::get_body): Likewise.
30352 * coretypes.h: Likewise.
30353 * domwalk.c: Likewise.
30354 * domwalk.h: Likewise.
30355 * dumpfile.c (struct dump_option_value_info): Likewise.
30356 (dump_enable_all): Likewise.
30357 (dump_switch_p_1): Likewise.
30358 (opt_info_switch_p): Likewise.
30359 * dumpfile.h (enum tree_dump_index): Likewise.
30360 (struct dump_file_info): Likewise.
30361 * genemit.c: Likewise.
30362 * generic-match-head.c: Likewise.
30363 * gengtype.c (open_base_files): Likewise.
30364 * gimple-pretty-print.c: Likewise.
30365 * gimple-pretty-print.h: Likewise.
30366 * graph.c (print_graph_cfg): Likewise.
30367 * graphite-scop-detection.c (dot_all_sese): Likewise.
30368 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
30369 * loop-unroll.c (report_unroll): Likewise.
30370 * passes.c (pass_manager::register_one_dump_file): Likewise.
30371 * print-tree.c: Likewise.
30372 * statistics.c: Likewise.
30373 * tree-cfg.c: Likewise.
30374 * tree-cfg.h: Likewise.
30375 * tree-dfa.c: Likewise.
30376 * tree-dfa.h: Likewise.
30377 * tree-dump.c (dump_function): Likewise.
30378 * tree-dump.h (struct dump_info): Likewise.
30379 * tree-pretty-print.c: Likewise.
30380 * tree-pretty-print.h: Likewise.
30381 * tree-ssa-live.c: Likewise.
30382 * tree-ssa-live.h: Likewise.
30383 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
30384 * tree-vect-loop.c: Likewise.
30385 * tree-vect-slp.c: Likewise.
30387 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
30388 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30390 PR tree-optimization/80457
30391 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
30392 of all arguments to a statement as scalar_to_vec operations.
30393 (vectorizable_call): Adjust call to vect_model_simple_cost for
30395 (vectorizable_conversion): Likewise.
30396 (vectorizable_assignment): Likewise.
30397 (vectorizable_shift): Likewise.
30398 (vectorizable_operation): Likewise.
30399 (vectorizable_comparison): Likewise.
30400 (vect_is_simple_cond): Record the def types for operands.
30401 (vectorizable_condition): Likewise, call vect_model_simple_cost.
30402 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
30403 for statement argument count.
30405 2017-05-16 Carl Love <cel@us.ibm.com>
30407 * config/rs6000/rs6000-c: Add support for built-in functions
30408 vector unsigned long long vec_bperm (vector unsigned long long,
30409 vector unsigned char)
30410 vector signed long long vec_mule (vector signed int,
30412 vector unsigned long long vec_mule (vector unsigned int,
30413 vector unsigned int)
30414 vector signed long long vec_mulo (vector signed int,
30416 vector unsigned long long vec_mulo (vector unsigned int,
30417 vector unsigned int)
30418 vector signed char vec_sldw (vector signed char,
30419 vector signed char,
30421 vector unsigned char vec_sldw (vector unsigned char,
30422 vector unsigned char,
30424 vector signed short vec_sldw (vector signed short,
30425 vector signed short,
30427 vector unsigned short vec_sldw (vector unsigned short,
30428 vector unsigned short,
30430 vector signed int vec_sldw (vector signed int,
30433 vector unsigned int vec_sldw (vector unsigned int,
30434 vector unsigned int,
30436 vector signed long long vec_sldw (vector signed long long,
30437 vector signed long long,
30439 vector unsigned long long vec_sldw (vector unsigned long long,
30440 vector unsigned long long,
30442 * config/rs6000/rs6000-c: Add support for built-in functions
30443 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
30444 * config/rs6000/altivec.h: Add defintion for vec_sldw.
30445 * doc/extend.texi: Update the built-in documentation for the
30446 new built-in functions.
30448 2017-05-16 Marek Polacek <polacek@redhat.com>
30452 * tree.c (save_expr): Don't fold the expression.
30454 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
30456 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
30457 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
30458 and (?*y,m). Update insn attributes.
30460 2017-05-16 Martin Liska <mliska@suse.cz>
30462 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
30463 flags argument of print_gimple_stmt, print_gimple_expr,
30464 print_generic_stmt and print_generic_expr.
30465 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
30466 * coretypes.h: Likewise.
30467 * except.c (dump_eh_tree): Likewise.
30468 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
30469 * gimple-pretty-print.h: Likewise.
30470 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
30471 (backprop::push_to_worklist): Likewise.
30472 (backprop::pop_from_worklist): Likewise.
30473 (backprop::process_use): Likewise.
30474 (backprop::intersect_uses): Likewise.
30475 (note_replacement): Likewise.
30476 * gimple-ssa-store-merging.c
30477 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
30478 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
30479 (pass_store_merging::execute): Likewise.
30480 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
30481 (ssa_base_cand_dump_callback): Likewise.
30482 (dump_incr_vec): Likewise.
30483 (replace_refs): Likewise.
30484 (replace_mult_candidate): Likewise.
30485 (create_add_on_incoming_edge): Likewise.
30486 (create_phi_basis): Likewise.
30487 (insert_initializers): Likewise.
30488 (all_phi_incrs_profitable): Likewise.
30489 (introduce_cast_before_cand): Likewise.
30490 (replace_one_candidate): Likewise.
30491 * gimplify.c (gimplify_expr): Likewise.
30492 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
30493 (set_rename): Likewise.
30494 (rename_uses): Likewise.
30495 (copy_loop_phi_nodes): Likewise.
30496 (add_close_phis_to_merge_points): Likewise.
30497 (copy_loop_close_phi_args): Likewise.
30498 (copy_cond_phi_args): Likewise.
30499 (graphite_copy_stmts_from_block): Likewise.
30500 (translate_pending_phi_nodes): Likewise.
30501 * graphite-poly.c (print_pdr): Likewise.
30502 (dump_gbb_cases): Likewise.
30503 (dump_gbb_conditions): Likewise.
30504 (print_scop_params): Likewise.
30505 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
30506 (build_cross_bb_scalars_use): Likewise.
30507 (gather_bbs::before_dom_children): Likewise.
30508 * hsa-dump.c (dump_hsa_immed): Likewise.
30509 * ipa-cp.c (print_ipcp_constant_value): Likewise.
30510 (get_replacement_map): Likewise.
30511 * ipa-inline-analysis.c (dump_condition): Likewise.
30512 (estimate_function_body_sizes): Likewise.
30513 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
30514 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
30515 * ipa-prop.c (ipa_dump_param): Likewise.
30516 (ipa_print_node_jump_functions_for_edge): Likewise.
30517 (ipa_modify_call_arguments): Likewise.
30518 (ipa_modify_expr): Likewise.
30519 (ipa_dump_param_adjustments): Likewise.
30520 (ipa_dump_agg_replacement_values): Likewise.
30521 (ipcp_modif_dom_walker::before_dom_children): Likewise.
30522 * ipa-pure-const.c (check_stmt): Likewise.
30523 (pass_nothrow::execute): Likewise.
30524 * ipa-split.c (execute_split_functions): Likewise.
30525 * omp-offload.c (dump_oacc_loop_part): Likewise.
30526 (dump_oacc_loop): Likewise.
30527 * trans-mem.c (tm_log_emit): Likewise.
30528 (tm_memopt_accumulate_memops): Likewise.
30529 (dump_tm_memopt_set): Likewise.
30530 (dump_tm_memopt_transform): Likewise.
30531 * tree-cfg.c (gimple_verify_flow_info): Likewise.
30532 (print_loop): Likewise.
30533 * tree-chkp-opt.c (chkp_print_addr): Likewise.
30534 (chkp_gather_checks_info): Likewise.
30535 (chkp_get_check_result): Likewise.
30536 (chkp_remove_check_if_pass): Likewise.
30537 (chkp_use_outer_bounds_if_possible): Likewise.
30538 (chkp_reduce_bounds_lifetime): Likewise.
30539 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
30540 (chkp_mark_completed_bounds): Likewise.
30541 (chkp_register_incomplete_bounds): Likewise.
30542 (chkp_mark_invalid_bounds): Likewise.
30543 (chkp_maybe_copy_and_register_bounds): Likewise.
30544 (chkp_build_returned_bound): Likewise.
30545 (chkp_get_bound_for_parm): Likewise.
30546 (chkp_build_bndldx): Likewise.
30547 (chkp_get_bounds_by_definition): Likewise.
30548 (chkp_generate_extern_var_bounds): Likewise.
30549 (chkp_get_bounds_for_decl_addr): Likewise.
30550 * tree-chrec.c (chrec_apply): Likewise.
30551 * tree-data-ref.c (dump_data_reference): Likewise.
30552 (dump_subscript): Likewise.
30553 (dump_data_dependence_relation): Likewise.
30554 (analyze_overlapping_iterations): Likewise.
30555 * tree-inline.c (expand_call_inline): Likewise.
30556 (tree_function_versioning): Likewise.
30557 * tree-into-ssa.c (dump_defs_stack): Likewise.
30558 (dump_currdefs): Likewise.
30559 (dump_names_replaced_by): Likewise.
30560 (dump_update_ssa): Likewise.
30561 (update_ssa): Likewise.
30562 * tree-object-size.c (pass_object_sizes::execute): Likewise.
30563 * tree-parloops.c (build_new_reduction): Likewise.
30564 (try_create_reduction_list): Likewise.
30565 (ref_conflicts_with_region): Likewise.
30566 (oacc_entry_exit_ok_1): Likewise.
30567 (oacc_entry_exit_single_gang): Likewise.
30568 * tree-pretty-print.h: Likewise.
30569 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
30570 (get_scalar_evolution): Likewise.
30571 (add_to_evolution): Likewise.
30572 (get_loop_exit_condition): Likewise.
30573 (analyze_evolution_in_loop): Likewise.
30574 (analyze_initial_condition): Likewise.
30575 (analyze_scalar_evolution): Likewise.
30576 (instantiate_scev): Likewise.
30577 (number_of_latch_executions): Likewise.
30578 (gather_chrec_stats): Likewise.
30579 (final_value_replacement_loop): Likewise.
30580 (scev_const_prop): Likewise.
30581 * tree-sra.c (dump_access): Likewise.
30582 (disqualify_candidate): Likewise.
30583 (create_access): Likewise.
30584 (reject): Likewise.
30585 (maybe_add_sra_candidate): Likewise.
30586 (create_access_replacement): Likewise.
30587 (analyze_access_subtree): Likewise.
30588 (analyze_all_variable_accesses): Likewise.
30589 (sra_modify_assign): Likewise.
30590 (initialize_constant_pool_replacements): Likewise.
30591 (find_param_candidates): Likewise.
30592 (decide_one_param_reduction): Likewise.
30593 (replace_removed_params_ssa_names): Likewise.
30594 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
30595 * tree-ssa-copy.c (dump_copy_of): Likewise.
30596 (copy_prop_visit_cond_stmt): Likewise.
30597 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
30598 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
30599 (record_equivalences_from_stmt): Likewise.
30600 * tree-ssa-dse.c (compute_trims): Likewise.
30601 (delete_dead_call): Likewise.
30602 (delete_dead_assignment): Likewise.
30603 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
30604 (forward_propagate_into_cond): Likewise.
30605 (pass_forwprop::execute): Likewise.
30606 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
30607 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
30609 (move_computations_worker): Likewise.
30610 (execute_sm): Likewise.
30611 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
30612 (remove_exits_and_undefined_stmts): Likewise.
30613 (remove_redundant_iv_tests): Likewise.
30614 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
30615 (adjust_iv_update_pos): Likewise.
30616 * tree-ssa-math-opts.c (bswap_replace): Likewise.
30617 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
30618 (value_replacement): Likewise.
30619 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
30620 * tree-ssa-pre.c (print_pre_expr): Likewise.
30621 (get_representative_for): Likewise.
30622 (create_expression_by_pieces): Likewise.
30623 (insert_into_preds_of_block): Likewise.
30624 (eliminate_insert): Likewise.
30625 (eliminate_dom_walker::before_dom_children): Likewise.
30626 (eliminate): Likewise.
30627 (remove_dead_inserted_code): Likewise.
30628 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30629 * tree-ssa-reassoc.c (get_rank): Likewise.
30630 (eliminate_duplicate_pair): Likewise.
30631 (eliminate_plus_minus_pair): Likewise.
30632 (eliminate_not_pairs): Likewise.
30633 (undistribute_ops_list): Likewise.
30634 (eliminate_redundant_comparison): Likewise.
30635 (update_range_test): Likewise.
30636 (optimize_range_tests_var_bound): Likewise.
30637 (optimize_vec_cond_expr): Likewise.
30638 (rewrite_expr_tree): Likewise.
30639 (rewrite_expr_tree_parallel): Likewise.
30640 (linearize_expr): Likewise.
30641 (break_up_subtract): Likewise.
30642 (linearize_expr_tree): Likewise.
30643 (attempt_builtin_powi): Likewise.
30644 (attempt_builtin_copysign): Likewise.
30645 (transform_stmt_to_copy): Likewise.
30646 (transform_stmt_to_multiply): Likewise.
30647 (dump_ops_vector): Likewise.
30648 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30649 (print_scc): Likewise.
30650 (set_ssa_val_to): Likewise.
30651 (visit_reference_op_store): Likewise.
30652 (visit_use): Likewise.
30653 (sccvn_dom_walker::before_dom_children): Likewise.
30654 (run_scc_vn): Likewise.
30655 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30657 (expr_hash_elt::print): Likewise.
30658 (const_and_copies::pop_to_marker): Likewise.
30659 (const_and_copies::record_const_or_copy_raw): Likewise.
30660 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30661 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30662 (dump_predicates): Likewise.
30663 (find_uninit_use): Likewise.
30664 (warn_uninitialized_phi): Likewise.
30665 (pass_late_warn_uninitialized::execute): Likewise.
30666 * tree-ssa.c (verify_vssa): Likewise.
30667 (verify_ssa): Likewise.
30668 (maybe_optimize_var): Likewise.
30669 * tree-vrp.c (dump_value_range): Likewise.
30670 (dump_all_value_ranges): Likewise.
30671 (dump_asserts_for): Likewise.
30672 (register_edge_assert_for_2): Likewise.
30673 (vrp_visit_cond_stmt): Likewise.
30674 (vrp_visit_switch_stmt): Likewise.
30675 (vrp_visit_stmt): Likewise.
30676 (vrp_visit_phi_node): Likewise.
30677 (simplify_cond_using_ranges_1): Likewise.
30678 (fold_predicate_in): Likewise.
30679 (evrp_dom_walker::before_dom_children): Likewise.
30680 (evrp_dom_walker::push_value_range): Likewise.
30681 (evrp_dom_walker::pop_value_range): Likewise.
30682 (execute_early_vrp): Likewise.
30684 2017-05-16 Richard Biener <rguenther@suse.de>
30686 * dwarf2out.c (loc_list_from_tree_1): Do not create
30687 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30689 2017-05-16 Richard Biener <rguenther@suse.de>
30691 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30693 (note_variable_value_in_expr): If we resolved the decl ref
30694 do not push to the stack.
30696 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30698 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30699 operations in fast-math mode.
30700 (vaddq_f16): Likewise.
30701 (vmul_f16): Likewise.
30702 (vmulq_f16): Likewise.
30703 (vsub_f16): Likewise.
30704 (vsubq_f16): Likewise.
30705 * config/arm/neon.md (add<mode>3): New.
30707 (fma:<VH:mode>3): New. Also remove outdated comment.
30710 2017-05-16 Martin Liska <mliska@suse.cz>
30714 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30715 (odr_types_equivalent_p): Likewise.
30717 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30719 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30721 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30724 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30725 non-interunit SSE move alternatives with '?'.
30726 (zero-extendsidi peephole2): New peephole to skip intermediate
30727 general register in SSE zero-extend sequence.
30729 2017-05-15 Jeff Law <law@redhat.com>
30731 * reorg.c (relax_delay_slots): Create a new variable to hold
30732 the temporary target rather than clobbering TARGET_LABEL.
30734 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30735 missing argument to extract_bit_field call.
30736 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30738 2017-05-15 Martin Liska <mliska@suse.cz>
30741 * gcc.c (process_command): Do not allow empty argument of -o option.
30743 2017-05-15 Renlin Li <renlin.li@arm.com>
30745 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30746 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30747 * config/aarch64/constraints.md (Usf): Add long call check.
30748 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30749 (call_value): Likewise.
30750 (sibcall): Likewise.
30751 (sibcall_value): Likewise.
30753 (call_value_insn): New.
30754 (sibcall_insn): Update rtx pattern.
30755 (sibcall_value_insn): Likewise.
30756 (call_internal): Remove.
30757 (call_value_internal): Likewise.
30758 (sibcall_internal): Likewise.
30759 (sibcall_value_internal): Likewise.
30760 (call_reg): Likewise.
30761 (call_symbol): Likewise.
30762 (call_value_reg): Likewise.
30763 (call_value_symbol): Likewise.
30765 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30768 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30770 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30772 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30773 compatible with CCGOCmode and with CCZmode.
30775 2017-05-14 Martin Sebor <msebor@redhat.com>
30777 PR middle-end/77671
30778 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30779 (gimple_fold_builtin_snprintf): Same.
30780 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30781 (gimple_fold_builtin_snprintf): Same.
30782 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30783 of character types.
30784 (is_call_safe): New function.
30785 (try_substitute_return_value): Call it.
30786 (try_simplify_call): New function.
30787 (pass_sprintf_length::handle_gimple_call): Call it.
30789 2017-05-14 Martin Sebor <msebor@redhat.com>
30791 PR middle-end/80669
30792 * builtins.c (expand_builtin_stpncpy): Simplify.
30794 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30796 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30797 * config/i386/i386.h
30798 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30799 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30800 (struct machine_function): Add new members call_ms2sysv,
30801 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30802 (struct machine_frame_state): New fields sp_realigned and
30803 sp_realigned_offset.
30804 * config/i386/i386.c
30805 (enum xlogue_stub): New enum.
30806 (enum xlogue_stub_sets): New enum.
30807 (class xlogue_layout): New class.
30808 (struct ix86_frame): New fields stack_realign_allocate_offset,
30809 stack_realign_offset and outlined_save_offset. Modify comments to
30810 detail stack layout when using out-of-line stubs.
30811 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30812 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30813 -mcall-ms2sysv-xlogues.
30814 (stub_managed_regs): New static variable.
30815 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30816 registers managed by out-of-line stub.
30817 (disable_call_ms2sysv_xlogues): New function.
30818 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30819 m->call_ms2sysv when appropriate and compute frame layout for
30821 (sp_valid_at, fp_valid_at): New inline functions.
30822 (choose_basereg): New function.
30823 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30825 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30826 Use align parameter of choose_baseaddr to generated aligned SSE movs
30828 (pro_epilogue_adjust_stack): Modify to track
30829 machine_frame_state::sp_realigned.
30830 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30831 (ix86_nsaved_sseregs): Likewise.
30832 (ix86_emit_save_regs): Likewise.
30833 (ix86_emit_save_regs_using_mov): Likewise.
30834 (ix86_emit_save_sse_regs_using_mov): Likewise.
30835 (get_scratch_register_on_entry): Likewise.
30836 (gen_frame_set): New function.
30837 (gen_frame_load): Likewise.
30838 (gen_frame_store): Likewise.
30839 (emit_outlined_ms2sysv_save): Likewise.
30840 (emit_outlined_ms2sysv_restore): Likewise.
30841 (ix86_expand_prologue): Modify stack re-alignment code and call
30842 emit_outlined_ms2sysv_save when appropriate.
30843 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30844 parameter rtx_insn *insn, which allows the function to be used to only
30845 generate the notes.
30846 (ix86_expand_epilogue): Modify validity checks of frame and stack
30847 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30848 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30849 * config/i386/predicates.md
30850 (save_multiple): New predicate.
30851 (restore_multiple): Likewise.
30852 * config/i386/sse.md
30853 (save_multiple<mode>): New pattern.
30854 (save_multiple_realign<mode>): Likewise.
30855 (restore_multiple<mode>): Likewise.
30856 (restore_multiple_and_return<mode>): Likewise.
30857 (restore_multiple_leave_return<mode>): Likewise.
30858 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30860 2017-05-14 Julia Koval <julia.koval@intel.com>
30862 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30863 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30864 (__builtin_ia32_xsetbv): New builtins.
30865 * config/i386/i386.c (ix86_expand_special_args_builtin):
30867 (ix86_expand_builtin): Special expand for new intrinsics.
30868 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
30869 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
30870 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
30872 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30874 * cfganal.c (inverted_post_order_compute): Change argument type
30876 * cfganal.h (inverted_post_order_compute): Adjust prototype.
30877 * df-core.c (rest_of_handle_df_initialize): Adjust.
30878 (rest_of_handle_df_finish): Likewise.
30879 (df_analyze_1): Likewise.
30880 (df_analyze): Likewise.
30881 (loop_inverted_post_order_compute): Change argument to be a vec *.
30882 (df_analyze_loop): Adjust.
30883 (df_get_n_blocks): Likewise.
30884 (df_get_postorder): Likewise.
30885 * df.h (struct df_d): Change field to be a vec.
30886 * lcm.c (compute_laterin): Adjust.
30887 (compute_available): Likewise.
30888 * lra-lives.c (lra_create_live_ranges_1): Likewise.
30889 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
30890 * tree-ssa-pre.c (compute_antic): Likewise.
30892 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30894 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
30895 (depth_first_search::depth_first_search): Change structure init
30896 function to this constructor.
30897 (depth_first_search::add_bb): Rename function to this member.
30898 (depth_first_search::execute): Likewise.
30899 (flow_dfs_compute_reverse_finish): Adjust.
30901 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30903 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
30904 (longest_simple_path): Likewise.
30905 * shrink-wrap.c (spread_components): Likewise.
30906 (disqualify_problematic_components): Likewise.
30907 (emit_common_heads_for_components): Likewise.
30908 (emit_common_tails_for_components): Likewise.
30909 (insert_prologue_epilogue_for_components): Likewise.
30911 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30913 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
30916 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30918 * df-core.c (df_set_blocks): Start using auto_bitmap.
30919 (df_compact_blocks): Likewise.
30920 * df-problems.c (df_rd_confluence_n): Likewise.
30921 * df-scan.c (df_insn_rescan_all): Likewise.
30922 (df_process_deferred_rescans): Likewise.
30923 (df_update_entry_block_defs): Likewise.
30924 (df_update_exit_block_uses): Likewise.
30925 (df_entry_block_bitmap_verify): Likewise.
30926 (df_exit_block_bitmap_verify): Likewise.
30927 (df_scan_verify): Likewise.
30928 * lra-constraints.c (lra_constraints): Likewise.
30929 (undo_optional_reloads): Likewise.
30930 (lra_undo_inheritance): Likewise.
30931 * lra-remat.c (calculate_gen_cands): Likewise.
30932 (do_remat): Likewise.
30933 * lra-spills.c (assign_spill_hard_regs): Likewise.
30934 (spill_pseudos): Likewise.
30935 * tree-ssa-pre.c (bitmap_set_and): Likewise.
30936 (bitmap_set_subtract_values): Likewise.
30938 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30940 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
30941 management with auto_bitmap.
30942 (fix_inter_tick): Likewise.
30943 (fix_recovery_deps): Likewise.
30944 * ira.c (add_store_equivs): Likewise.
30945 (find_moveable_pseudos): Likewise.
30946 (split_live_ranges_for_shrink_wrap): Likewise.
30947 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
30948 (rtx_reuse_manager::seen_def_p): Likewise.
30949 (rtx_reuse_manager::set_seen_def): Likewise.
30950 * print-rtl.h (class rtx_reuse_manager): Likewise.
30952 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30954 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
30956 (migrate_btr_def): Likewise.
30957 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
30958 * df-core.c (loop_post_order_compute): Likewise.
30959 (loop_inverted_post_order_compute): Likewise.
30960 * hsa-common.h: Likewise.
30961 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
30962 * init-regs.c (initialize_uninitialized_regs): Likewise.
30963 * ipa-inline.c (resolve_noninline_speculation): Likewise.
30964 (inline_small_functions): Likewise.
30965 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
30966 * ira.c (combine_and_move_insns): Likewise.
30967 (build_insn_chain): Likewise.
30968 * loop-invariant.c (find_invariants): Likewise.
30969 * lower-subreg.c (propagate_pseudo_copies): Likewise.
30970 * predict.c (tree_predict_by_opcode): Likewise.
30971 (predict_paths_leading_to): Likewise.
30972 (predict_paths_leading_to_edge): Likewise.
30973 (estimate_loops_at_level): Likewise.
30974 (estimate_loops): Likewise.
30975 * shrink-wrap.c (try_shrink_wrapping): Likewise.
30976 (spread_components): Likewise.
30977 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
30978 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
30979 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
30980 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
30981 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
30982 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
30983 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
30984 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
30985 (mark_threaded_blocks): Likewise.
30986 (thread_through_all_blocks): Likewise.
30987 * tree-ssa.c (verify_ssa): Likewise.
30988 (execute_update_addresses_taken): Likewise.
30989 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
30991 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30993 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
30995 (post_order_compute): Likewise.
30996 (inverted_post_order_compute): Likewise.
30997 (pre_and_rev_post_order_compute_fn): Likewise.
30999 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31001 * genrecog.c (int_set::int_set): Explicitly construct our
31002 auto_vec base class.
31003 * vec.h (auto_vec::auto_vec): New constructor.
31005 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31007 * bitmap.h (class auto_bitmap): New constructor taking
31008 bitmap_obstack * argument.
31010 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31012 * bitmap.h (class auto_bitmap): Change type of m_bits to
31013 bitmap_head, and adjust ctor / dtor and member operators.
31015 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31017 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31018 when returned register mode doesn't match original mode.
31020 2017-05-12 Jeff Law <law@redhat.com>
31021 Jakub Jelinek <jakub@redhat.com>
31023 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31024 we look for cc setter after the compare-elim changes.
31025 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31026 within the vector to match what compare-elim now expects.
31027 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31028 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31030 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31031 after the compare-elim changes.
31032 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31033 the vector to match what compare-elim now expects.
31034 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31035 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31036 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31037 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31038 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31040 * config/visium/visium.c (single_set_and_flags): Fix where
31041 we look for cc setter after the compare-elim changes.
31042 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
31043 with the vector to match what compare-elim now expects.
31044 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
31045 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31046 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31047 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31048 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31049 (neg<mode>2_insn_set_overflow): Likewise.
31051 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31053 PR middle-end/79794
31054 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31055 maybe_expand_insn call, set ops[0].target. If still set after call,
31056 set alt_rtl. Add extra arg to recursive calls.
31057 (extract_bit_field): Add alt_rtl argument. Pass to
31059 * expmed.h (extract_bit_field): Fix prototype.
31060 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31061 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31062 to extract_bit_field_calls.
31063 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31064 Pass alt_rtl to extract_bit_field calls.
31065 * calls.c (store_unaligned_arguments_into_psuedos)
31066 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31067 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31069 * optabs.h (struct expand_operand): Add target bitfield.
31071 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31073 * compare-elim.c (try_eliminate_compare): Canonicalize
31074 operation with embedded compare to
31075 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31076 (set (reg) (operation)].
31078 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31080 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31083 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31084 cost of adding a carry flag for ADC instruction.
31085 [case MINUS]: Ignore the cost of subtracting a carry flag
31086 for SBB instruction.
31088 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31090 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31092 * config/rs6000/bmiintrin.h: New file.
31093 * config/rs6000/bmi2intrin.h: New file.
31094 * config/rs6000/x86intrin.h: New file.
31096 2017-05-12 Jeff Law <law@redhat.com>
31098 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31101 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31103 PR middle-end/80707
31104 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31106 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31108 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31109 early expansion of vector divide builtins.
31110 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
31111 builtins identified as having unsigned arguments.
31113 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31115 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
31116 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
31117 expansion of vector logical operations (and, andc, or, xor,
31120 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31122 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
31123 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
31125 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31127 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31128 early GIMPLE expansion of vector multiplies.
31130 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31132 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
31133 TARGET_HAVE_MOVT conditional.
31134 (movt splitter): Likewise.
31136 2017-05-12 Richard Biener <rguenther@suse.de>
31138 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
31139 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31140 Fold all stmts not inplace.
31142 2017-05-12 Richard Biener <rguenther@suse.de>
31144 PR tree-optimization/80713
31145 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
31146 inserted_exprs bit for not removed stmts.
31148 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
31150 PR middle-end/69921
31151 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
31152 parallelized" attribute for parallelized OpenACC kernels.
31153 * omp-offload.c (execute_oacc_device_lower): Use it.
31155 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
31156 Set "oacc kernels" attribute.
31157 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
31158 parameter. Adjust all users.
31159 (oacc_fn_attrib_kernels_p): Remove function.
31160 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
31161 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
31162 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
31163 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
31164 assert "oacc kernels" attribute is set.
31166 2017-05-11 Carl Love <cel@us.ibm.com>
31168 * config/rs6000/rs6000-c: Add support for built-in functions
31169 vector unsigned char vec_popcnt (vector signed char)
31170 vector unsigned char vec_popcnt (vector unsigned char)
31171 vector unsigned short vec_popcnt (vector signed short)
31172 vector unsigned short vec_popcnt (vector unsigned short)
31173 vector unsigned int vec_popcnt (vector signed int)
31174 vector unsigned int vec_popcnt (vector unsigned int)
31175 vector unsigned long long vec_popcnt (vector signed long long)
31176 vector unsigned long long vec_popcnt (vector unsigned long long)
31177 vector signed long long vec_slo (vector signed long long,
31178 vector signed char)
31179 vector signed long long vec_slo (vector signed long long,
31180 vector unsigned char)
31181 vector unsigned long long vec_slo (vector unsigned long long,
31182 vector signed char)
31183 vector unsigned long long vec_slo (vector unsigned long long,
31184 vector unsigned char)
31185 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
31186 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
31187 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
31188 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
31189 * doc/extend.texi: Update the built-in documentation file for the
31190 new built-in functions.
31192 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
31194 * attribs.h (sorted_attr_string): Move machine independent
31195 functions for target clone support from the i386 port to common
31196 code. Rename ix86_function_versions to common_function_versions.
31197 Rename make_name to make_unique_name.
31198 (common_function_versions): Likewise.
31199 (make_unique_name): Likewise.
31200 (make_dispatcher_decl): Likewise.
31201 (is_function_default_version): Likewise.
31202 * attribs.c (attr_strcmp): Likewise.
31203 (sorted_attr_string): Likewise.
31204 (common_function_versions): Likewise.
31205 (make_unique_name): Likewise.
31206 (make_dispatcher_decl): Likewise.
31207 (is_function_default_version): Likewise.
31208 * config/i386/i386.c (attr_strcmp): Likewise.
31209 (sorted_attr_string): Likewise.
31210 (ix86_function_versions): Likewise.
31211 (make_name): Likewise.
31212 (make_dispatcher_decl): Likewise.
31213 (is_function_default_version): Likewise.
31214 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
31216 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31219 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
31220 Account for direct move costs for vec_construct of integer
31223 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
31226 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
31227 (UNSPEC_STX_ATOMIC): Ditto.
31228 (loaddi_via_sse): New insn.
31229 (storedi_via_sse): Ditto.
31230 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
31231 Update corresponding peephole2 patterns.
31232 (atomic_storedi_fpu): Ditto.
31234 2017-05-11 Julia Koval <julia.koval@intel.com>
31236 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
31237 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
31239 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
31240 (__builtin_ia32_rsqrt14ss_mask): New builtins.
31241 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
31243 2017-05-11 Nathan Sidwell <nathan@acm.org>
31245 * graphite-poly.c: Include dumpfile.h.
31247 * dumpfle.h (dump_function): Declare here ...
31248 * tree-dump.h (dump_function): ... not here.
31249 * dumpfile.c: #include tree-cfg.h.
31250 (dump_function): Move here from ...
31251 * tree-dump.c (dump_function): ... here.
31252 * gimplify.c: #include splay-tree.h, not tree-dump.h.
31253 * graphite-poly.c: Don't include tree-dump.h.
31254 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
31255 * print-tree.c: Likewise.
31256 * stor-layout.c: Likewise.
31257 * tree-nested.c: Likewise.
31259 * dumpfile.c (dump_start): Use TDF_FLAGS.
31260 (dump_enable_all): Fix TDF_KIND check thinko.
31262 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
31264 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31265 array entries to represent two legal parameterizations of the
31266 overloaded __builtin_cmpb function, as represented by the
31267 P6_OV_BUILTIN_CMPB constant.
31268 (altivec_resolve_overloaded_builtin): Add special case handling
31269 for the __builtin_cmpb function, as represented by the
31270 P6_OV_BUILTIN_CMPB constant.
31271 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
31272 (BU_P6_64BIT_2): New macro.
31273 (BU_P6_OVERLOAD_2): New macro
31274 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
31275 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
31276 (CMPB): Add overload support to represent both 32-bit and 64-bit
31277 compare-bytes function.
31278 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
31279 support for TARGET_CMPB.
31280 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
31281 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31282 documentation of the __builtin_cmpb overloaded built-in function.
31284 2017-05-11 Richard Biener <rguenther@suse.de>
31286 PR tree-optimization/80705
31287 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
31288 bases are not vectorizable.
31290 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31292 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
31293 when counting register pressure.
31295 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31297 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
31298 (iv_ca_more_deps): Renamed to ...
31299 (iv_ca_compare_deps): ... this.
31300 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
31302 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31304 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
31306 (determine_group_iv_costs): ... here.
31307 (find_inv_vars_cb): Record inv var if it's not recorded before.
31309 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31311 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
31312 (get_shiftadd_cost): Ditto.
31314 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31316 * tree-ssa-address.c: Include header file.
31317 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
31319 (add_to_parts): Refactor.
31320 (addr_to_parts): New parameter. Update use of move_hint_to_base.
31321 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
31324 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31326 PR tree-optimization/53090
31327 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
31329 (extract_cond_operands): Detect condition with IV on both sides
31330 and return COMP_IV_EXPR_2.
31331 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
31332 (rewrite_use_compare): Simplify by removing call to function
31333 extract_cond_operands.
31335 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31337 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
31338 (extract_cond_operands): Detect condition comparing against non-
31339 invariant bound and return appropriate enum value.
31340 (find_interesting_uses_cond): Update use of extract_cond_operands.
31341 Handle its return value accordingly.
31342 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
31344 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31346 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
31347 nonlinear iv_use computation in loop invariant sensitive way.
31349 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31351 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
31352 (find_iv_candidates): Call relate_compare_use_with_all_cands.
31354 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31356 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
31357 (dump_cand): Support iv_cand.inv_exprs.
31358 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
31360 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
31363 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31365 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
31367 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
31368 as local function. Include necessary header files.
31369 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
31371 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31373 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
31375 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31377 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
31378 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
31379 RSHIFT_EXPR and BIT_NOT_EXPR.
31381 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31383 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
31384 (adjust_setup_cost): New parameter supporting round up adjustment.
31385 (struct address_cost_data): Delete.
31386 (force_expr_to_var_cost): Don't bound cost with spill_cost.
31387 (split_address_cost, ptr_difference_cost): Delete.
31388 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
31389 (struct ainc_cost_data): New struct.
31390 (get_address_cost_ainc): New function.
31391 (get_address_cost, get_computation_cost): Reimplement.
31392 (determine_group_iv_cost_address): Record inv_expr for all uses of
31394 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
31395 (iv_ca_has_deps): Reimplemented to ...
31396 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
31398 (iv_ca_extend): Call iv_ca_more_deps.
31400 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31402 * tree-ssa-address.c (struct mem_address): Move to header file.
31403 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
31404 * tree-ssa-address.h (struct mem_address): Move from C file.
31405 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
31407 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31409 * tree-affine.h (aff_combination_type): New interface.
31410 (aff_combination_zero_p): Remove static.
31411 (aff_combination_const_p): New interface.
31412 (aff_combination_singleton_var_p): New interfaces.
31414 2017-05-11 Richard Biener <rguenther@suse.de>
31416 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31417 Skip unreachable blocks and destinations.
31418 (eliminate): Move stmt removal and fixup ...
31419 (fini_eliminate): ... here. Skip inserted exprs.
31420 (pass_pre::execute): Move fini_pre after fini_eliminate.
31421 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
31422 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
31423 PRE to get rid of dead code that has invalid SSA form and
31424 split critical edges again.
31426 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31428 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31430 2017-05-11 Richard Biener <rguenther@suse.de>
31432 * passes.c (execute_function_todo): Verify loops if they are
31433 said to be up-to-date.
31434 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
31435 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
31437 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
31440 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
31441 handle calling assemble_external ourself.
31444 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
31445 modes with zero size. Enhance comment.
31447 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31449 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
31450 built-ins for vec_xl and vec_xst with short and char pointer
31453 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31455 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
31456 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
31457 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
31458 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
31459 (_mm_maskz_min_round_ss): New intrinsics.
31460 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
31461 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
31462 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
31463 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
31464 (__builtin_ia32_minss_mask_round): New builtins.
31465 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31466 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31467 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
31469 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
31470 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
31472 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
31475 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31477 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
31478 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
31479 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
31480 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
31481 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
31482 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
31483 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
31484 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31485 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31486 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
31487 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
31488 (__builtin_ia32_mulss_mask_round): New builtins.
31489 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31490 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31491 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
31493 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
31494 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31496 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31499 2017-05-10 Julia Koval <julia.koval@intel.com>
31501 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
31502 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
31503 (_mm256_setr_m128i): New intrinsics.
31505 2017-05-10 Julia Koval <julia.koval@intel.com>
31507 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
31508 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
31509 (_mm_maskz_rcp14_ss): New intrinsics.
31510 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
31511 (__builtin_ia32_rcp14ss_mask): New builtins.
31512 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
31514 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
31516 PR tree-optimization/51513
31517 * tree-cfg.c (gimple_seq_unreachable_p): New function.
31518 (assert_unreachable_fallthru_edge_p): Use it.
31519 (group_case_labels_stmt): Likewise.
31520 * tree-cfg.h: Prototype it.
31521 * stmt.c: Include cfghooks.h and tree-cfg.h.
31522 (emit_case_dispatch_table) <gap_label>: New local variable.
31523 Use it to fill dispatch table gaps.
31524 Test for default_label before updating probabilities.
31525 (expand_case) <default_label>: Remove unneeded initialization.
31526 Test for unreachable default case statement and remove its edge.
31527 Set default_label accordingly.
31528 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
31530 2017-05-10 Carl Love <cel@us.ibm.com>
31532 * config/rs6000/rs6000-c: Add support for built-in functions
31533 vector signed char vec_neg (vector signed char)
31534 vector signed short int vec_neg (vector short int)
31535 vector signed int vec_neg (vector signed int)
31536 vector signed long long vec_neg (vector signed long long)
31537 vector float vec_neg (vector float)
31538 vector double vec_neg (vector double)
31539 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
31541 * config/rs6000/altivec.h: Add define for vec_neg
31542 * doc/extend.texi: Update the built-in documentation for the
31543 new built-in functions.
31545 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31547 PR tree-optimization/77644
31548 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
31550 2017-05-10 Nathan Sidwell <nathan@acm.org>
31552 * dumpfile.h (TDI_lang_all): New.
31553 (TDF_KIND): New. Renumber others
31554 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
31556 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
31558 (get_dump_file_name): Adjust suffix generation.
31559 (dump_enable_all): Use TDF_KIND.
31560 * doc/invoke.texi (-fdump-lang-all): Document.
31562 * dumpfile.h: Tabify.
31564 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
31567 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
31568 Move member access before delete.
31570 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
31572 * tree-inline.c (expand_call_inline): Split block at stmt
31575 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
31578 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
31579 are now unused after splitting mov{sf,sd}_hardfloat.
31580 (f32_lr2): Likewise.
31581 (f32_lm): Likewise.
31582 (f32_lm2): Likewise.
31583 (f32_li): Likewise.
31584 (f32_li2): Likewise.
31585 (f32_lv): Likewise.
31586 (f32_sr): Likewise.
31587 (f32_sr2): Likewise.
31588 (f32_sm): Likewise.
31589 (f32_sm2): Likewise.
31590 (f32_si): Likewise.
31591 (f32_si2): Likewise.
31592 (f32_sv): Likewise.
31593 (f32_dm): Likewise.
31594 (f32_vsx): Likewise.
31595 (f32_av): Likewise.
31596 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
31597 For movsf, order stores so the VSX stores occur before the GPR
31598 store which encourages the register allocator to use a traditional
31599 FPR instead of a GPR. For movsd, order the stores so that the GPR
31600 store comes before the VSX stores to allow the power6 to work.
31601 This is due to the power6 not having a 32-bit integer store
31602 instruction from a FPR.
31603 (movsf_hardfloat): Likewise.
31604 (movsd_hardfloat): Likewise.
31606 2017-05-09 Martin Sebor <msebor@redhat.com>
31608 PR translation/80280
31609 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
31612 PR translation/80280
31613 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
31614 data member added in r247778.
31615 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
31617 2017-05-09 Nathan Sidwell <nathan@acm.org>
31619 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
31621 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31624 2017-05-09 Marek Polacek <polacek@redhat.com>
31626 * doc/invoke.texi: Fix typo.
31628 2017-05-09 Richard Biener <rguenther@suse.de>
31630 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31631 (vrp_val_is_min): Likewise.
31632 (set_value_range_to_value): Likewise.
31633 (set_value_range_to_nonnegative): Likewise.
31634 (gimple_assign_nonzero_p): Likewise.
31635 (gimple_stmt_nonzero_p): Likewise.
31636 (vrp_int_const_binop): Likewise. Remove unreachable case.
31637 (adjust_range_with_scev): Adjust comments.
31638 (compare_range_with_value): Likewise.
31639 (extract_range_from_phi_node): Likewise.
31640 (test_for_singularity): Likewise.
31642 2017-05-09 Richard Biener <rguenther@suse.de>
31644 * tree-vrp.c (get_single_symbol): Add assert that we don't
31645 get overflowed constants as invariant part.
31646 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31647 checks. Use wi::cmp instead of recursing for integer constants.
31648 (compare_values): Just ignore whether we assumed undefined
31649 overflow instead of failing the compare.
31650 (extract_range_for_var_from_comparison_expr): Add comment before the
31651 TREE_NO_WARNING sets.
31652 (test_for_singularity): Likewise.
31653 (extract_range_from_comparison): Do not disable optimization
31654 when we assumed undefined overflow.
31655 (extract_range_basic): Remove init of unused var.
31657 2017-05-09 Richard Biener <rguenther@suse.de>
31659 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31660 (extract_range_from_multiplicative_op_1): Adjust.
31661 (extract_range_from_binary_expr_1): Use int_const_binop.
31663 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31666 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31667 rs6000_store_data_bypass_p in seven define_bypass directives and
31668 in several comments.
31669 * config/rs6000/rs6000-protos.h: Add prototype for
31670 rs6000_store_data_bypass_p function.
31671 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31672 function implements slightly different (rs6000-specific) semantics
31673 than store_data_bypass_p, returning false rather than aborting
31674 with assertion error when arguments do not satisfy the
31675 requirements of store data bypass.
31676 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31677 rs6000_store_data_bypass_p.
31679 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31681 * config/xtensa/xtensa-protos.h
31682 (xtensa_initial_elimination_offset): New declaration.
31683 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31684 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31685 macro definition, add case for FRAME_POINTER_REGNUM when
31686 FRAME_GROWS_DOWNWARD.
31687 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31688 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31689 xtensa_initial_elimination_offset.
31691 2017-05-08 Nathan Sidwell <nathan@acm.org>
31693 * doc/invoke.texi: Alphabetize -fdump options.
31695 2017-05-08 Martin Sebor <msebor@redhat.com>
31697 PR translation/80280
31698 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31700 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31702 * target.def (compute_frame_layout): New optional target hook.
31703 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31704 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31705 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31707 * reload1.c (verify_initial_elim_offsets): Likewise.
31708 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31709 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31710 (arm_get_frame_offsets): Split up into this ...
31711 (arm_compute_frame_layout): ... and this function.
31713 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31715 * config/aarch64/constraints.md (Usa): New constraint.
31716 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31718 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31720 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31721 with_multilib_list after it has been checked.
31723 2017-05-08 Richard Biener <rguenther@suse.de>
31725 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31726 (bitmap_set_subtract_values): Likewise.
31728 2017-05-08 Richard Biener <rguenther@suse.de>
31730 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31731 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31733 (gimple_stmt_nonzero_warnv_p): Rename to ...
31734 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31736 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31737 (extract_range_basic): Adjust, do not disable propagation on
31738 strict overflow sensitive simplification.
31739 (vrp_visit_cond_stmt): Likewise.
31741 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31743 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31744 body size unconditionally.
31746 2017-05-07 Jeff Law <law@redhat.com>
31749 2017-05-06 Jeff Law <law@redhat.com>
31750 PR tree-optimization/78496
31751 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31754 PR tree-optimization/78496
31755 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31756 (simplify_stmt_using_ranges): Call it.
31757 (vrp_dom_walker::before_dom_children): Extract equivalences
31758 from an ASSERT_EXPR with an equality comparison against a
31761 2017-05-06 Jeff Law <law@redhat.com>
31763 PR tree-optimization/78496
31764 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31767 PR tree-optimization/78496
31768 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31769 (simplify_stmt_using_ranges): Call it.
31770 (vrp_dom_walker::before_dom_children): Extract equivalences
31771 from an ASSERT_EXPR with an equality comparison against a
31774 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31776 * lra-constraints.c (lra_copy_reg_equiv): New function.
31777 (split_reg): Use it to copy equivalence information from the
31778 original register to the spill register.
31780 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31782 PR rtl-optimization/75964
31783 * simplify-rtx.c (simplify_const_relational_operation): Remove
31784 invalid handling of comparisons of integer ABS.
31786 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31788 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31789 initialize to zero.
31790 (init_regs): Remove declaration.
31791 (function_arg_advance_32): Initialize error_p as boolean variable.
31793 2017-05-05 Nathan Sidwell <nathan@acm.org>
31795 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31796 lines. Use for (;;).
31798 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31800 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31801 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31802 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31803 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31804 VF=2 that require versioning.
31806 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31808 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31811 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31813 * diagnostic.h (diagnostic_override_option_index): Convert from
31814 macro to inline function.
31816 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31818 * diagnostic.c (last_module_changed_p): New function.
31819 (set_last_module): New function.
31820 (diagnostic_report_current_module): Convert macro usage to
31821 the above functions.
31822 * diagnostic.h (diagnostic_context::last_module): Strengthen
31823 from const line_map * to const line_map_ordinary *.
31824 (diagnostic_last_module_changed): Delete macro.
31825 (diagnostic_set_last_module): Delete macro.
31827 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31829 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31830 with diagnostic_report_diagnostic.
31831 (diagnostic_n_impl_richloc): Likewise.
31832 * diagnostic.h (report_diagnostic): Delete macro.
31833 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31834 with diagnostic_report_diagnostic.
31835 * substring-locations.c (format_warning_va): Likewise.
31837 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31839 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31840 save/restor of format_spec. Move option-printing code to...
31841 (print_option_information): ...this new function, and
31842 reimplement by simply printing to the pretty_printer,
31843 rather than appending to the format string.
31845 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31847 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31848 handling logic into...
31849 (update_effective_level_from_pragmas): ...this new function.
31851 2017-05-04 Andrew Waterman <andrew@sifive.com>
31853 * config/riscv/riscv.opt (mstrict-align): New option.
31854 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31855 (SLOW_UNALIGNED_ACCESS): Define.
31856 (riscv_slow_unaligned_access): Declare.
31857 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31859 (riscv_slow_unaligned_access): New variable.
31860 (rocket_tune_info): Set slow_unaligned_access to true.
31861 (optimize_size_tune_info): Set slow_unaligned_access to false.
31862 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31863 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31864 (riscv_option_override): Set riscv_slow_unaligned_access.
31865 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31867 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
31869 * config/riscv/riscv.md: Unify indentation.
31871 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
31876 * config/rs6000/rs6000.md (u code attribute): Add FIX and
31878 (extendsi<mode>2): Add support for doing sign extension via
31879 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
31880 don't have ISA 3.0 instructions.
31881 (extendsi<mode>2 splitter): Likewise.
31882 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
31883 generate the normal insns since SImode can now go in vector
31884 registers. Disallow the special UNSPECs needed for previous
31885 machines to hide SImode being used. Add new insns
31886 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
31887 (fix_trunc<mode>si2_stfiwx): Likewise.
31888 (fix_trunc<mode>si2_internal): Likewise.
31889 (fixuns_trunc<mode>si2): Likewise.
31890 (fixuns_trunc<mode>si2_stfiwx): Likewise.
31891 (fctiw<u>z_<mode>_smallint): Likewise.
31892 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
31893 of floating point to 32-bit integer from doing a direct move to
31894 the GPR registers to do a store.
31895 (fctiwz_<mode>): Break long line.
31897 2017-05-05 Bin Cheng <bin.cheng@arm.com>
31899 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
31900 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
31901 (addr_list, addr_offset_valid_p): New.
31902 (split_address_groups): Check offset validity with above function.
31903 (gt-tree-ssa-loop-ivopts.h): Include header file.
31905 2017-05-05 Nathan Sidwell <nathan@acm.org>
31907 * config.gcc (arm*-*-*): Add missing 'fi'.
31909 2017-05-05 Steve Ellcey <sellcey@cavium.com>
31911 * doc/invoke.texi (-fopt-info): Explicitly say order of options
31912 included in -fopt-info does not matter.
31913 * doc/optinfo.texi (-fopt-info): Fix description of default
31914 behavour. Explicitly say order of options included in -fopt-info
31917 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
31919 * config.gcc: Allow combinations of aprofile and rmprofile values for
31920 --with-multilib-list.
31921 * config/arm/t-multilib: New file.
31922 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
31923 variables. Remove setting of ISA and floating-point ABI in
31924 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
31925 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
31926 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
31927 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
31929 * config/arm/t-rmprofile: Likewise except for the matches changes.
31930 * doc/install.texi (--with-multilib-list): Document the combination of
31931 aprofile and rmprofile values and warn about pitfalls in doing that.
31933 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31935 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
31936 (movdi_aarch64): Likewise.
31938 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31940 PR tree-optimization/80632
31941 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
31943 (build_arrays): Initialize it for virtual phis.
31944 (fix_phi_nodes): Use it for virtual phis.
31946 PR tree-optimization/80558
31947 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
31948 [x, y] op z into [x op, y op z] for op & or | if conditions
31951 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
31952 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31955 * config/arm/arm.md (use_literal_pool): Remove.
31956 (64-bit immediate split): No longer takes cost into consideration
31957 if arm_disable_literal_pool is enabled.
31958 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
31959 used when arm_disable_literal_pool is enabled.
31960 (arm_max_const_double_inline_cost): Remove use of
31961 arm_disable_literal_pool.
31962 (push_minipool_fix): Add assert.
31963 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
31964 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
31965 (no_literal_pool_sf_immediate): New.
31967 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31969 PR tree-optimization/80613
31970 * tree-ssa-dce.c (propagate_necessity): Remove cases for
31971 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
31973 2017-05-05 Richard Biener <rguenther@suse.de>
31975 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
31977 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
31979 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
31980 of this flag from insn conditions due to removal from r247495.
31982 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31984 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
31986 (arm_early_store_addr_dep_ptr): Likewise.
31987 * config/arm/aarch-common-protos.h
31988 (arm_early_load_addr_dep_ptr): Add prototype.
31989 (arm_early_store_addr_dep_ptr): Likewise.
31990 * config/arm/cortex-a53.md: Add new bypasses.
31992 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31994 * tree.c (next_type_uid): Change type to unsigned.
31995 (type_hash_canon): Decrement back next_type_uid if
31996 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
31997 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
32000 2017-05-04 Martin Sebor <msebor@redhat.com>
32002 * builtins.c: Fix a trivial typo in a comment.
32004 PR middle-end/79234
32005 * builtins.c (check_sizes): Adjust to handle reading past the end.
32006 Avoid printing excessive upper bound of ranges. Use %E to print
32007 tree nodes instead of converting them to %wu.
32008 (expand_builtin_memchr): New function.
32009 (compute_dest_size): Rename...
32010 (compute_objsize): ...to this.
32011 (expand_builtin_memcpy): Adjust.
32012 (expand_builtin_mempcpy): Adjust.
32013 (expand_builtin_strcat): Adjust.
32014 (expand_builtin_strcpy): Adjust.
32015 (check_strncat_sizes): Adjust.
32016 (expand_builtin_strncat): Adjust.
32017 (expand_builtin_strncpy): Adjust and simplify.
32018 (expand_builtin_memset): Adjust.
32019 (expand_builtin_bzero): Adjust.
32020 (expand_builtin_memcmp): Adjust.
32021 (expand_builtin): Handle memcmp.
32022 (maybe_emit_chk_warning): Check strncat just once.
32024 2017-05-04 Martin Sebor <msebor@redhat.com>
32026 PR preprocessor/79214
32027 PR middle-end/79222
32028 PR middle-end/79223
32029 * builtins.c (check_sizes): Add inlining context and issue
32030 warnings even when -Wno-system-headers is set.
32031 (check_strncat_sizes): Same.
32032 (expand_builtin_strncat): Same.
32033 (expand_builtin_memmove): New function.
32034 (expand_builtin_stpncpy): Same.
32035 (expand_builtin): Handle memmove and stpncpy.
32037 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32039 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32040 which is not used any more.
32042 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32044 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32046 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32048 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32049 (cortexa53_tunings): Likewise.
32050 (cortexa57_tunings): Likewise.
32051 (cortexa72_tunings): Likewise.
32052 (cortexa73_tunings): Likewise.
32054 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32056 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32057 Set loop alignment to 8.
32059 2017-05-04 Martin Sebor <msebor@redhat.com>
32061 PR translation/80280
32062 * builtins.c (expand_builtin_object_size): Add missing quoting to
32063 %D and like directives.
32064 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32065 (hsa_type_for_tree_type): Same.
32066 (verify_function_arguments): Same.
32067 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32068 * varasm.c (get_section): Same.
32071 2017-05-04 Martin Sebor <msebor@redhat.com>
32073 PR translation/80280
32074 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32076 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32078 * config/aarch64/aarch64.c (generic_addrcost_table):
32079 Change HI/TI mode setting.
32081 2017-05-04 Martin Jambor <mjambor@suse.cz>
32083 PR tree-optimization/80622
32084 * tree-sra.c (comes_initialized_p): New function.
32085 (build_accesses_from_assign): Only set write lazily when
32086 comes_initialized_p is false.
32087 (analyze_access_subtree): Use comes_initialized_p.
32088 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32089 instead of testing for PARM_DECL.
32091 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32093 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32094 constraint on operand 0 to allow more general addressing modes.
32095 Adjust output template.
32096 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32098 * config/aarch64/aarch64-protos.h
32099 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32100 * config/aarch64/constraints.md (Dp): New address constraint.
32101 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32104 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32106 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32107 update use of estimate_ipcp_clone_size_and_time.
32108 (estimate_local_effects): Update use of
32109 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32110 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
32111 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
32112 Return nonspecialized time.
32114 2017-05-04 Richard Biener <rguenther@suse.de>
32116 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
32117 for the last VUSE which def dominates the PHI. Directly call
32119 (get_continuation_for_phi_1): Remove.
32121 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
32123 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
32124 to explain the use of truncating division. Cap the number of
32125 iterations to the maximum given by nb_iterations_upper_bound,
32128 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32130 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
32131 * configure: Regenerate.
32132 * config.in: Regenerate.
32133 * config/i386/driver-mingw32.c: new file.
32134 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
32135 * config.host: Link driver-mingw32.o on MinGW host.
32136 * doc/install.texi: Document new --enable-mingw-wildcard configure
32139 2017-05-04 Marek Polacek <polacek@redhat.com>
32141 PR tree-optimization/80612
32142 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
32144 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32145 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
32147 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
32148 (movt splitter): Likewise.
32149 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
32150 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
32151 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
32152 block for Thumb-1 with MOVT.
32153 (thumb2_legitimate_address_p): Move code block ...
32154 (can_avoid_literal_pool_for_label_p): ... into this new function.
32155 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
32157 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
32158 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
32159 "M-profile targets with the MOVT instruction".
32161 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32163 * config/arm/arm-builtins.c (arm_init_builtins): Rename
32164 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
32165 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
32167 2017-05-04 Martin Liska <mliska@suse.cz>
32169 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
32170 variable cond_code.
32172 2017-05-04 Richard Biener <rguenther@suse.de>
32174 * tree.c (array_at_struct_end_p): Handle arrays at struct
32175 end with flexarrays more conservatively. Refactor and treat
32176 arrays of arrays or aggregates more strict. Fix
32177 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
32178 * tree.c (array_at_struct_end_p): Adjust prototype.
32179 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
32180 * gimple-fold.c (get_range_strlen): Likewise.
32181 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
32183 2017-05-04 Richard Biener <rguenther@suse.de>
32185 PR tree-optimization/31130
32186 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
32188 (supports_overflow_infinity): Likewise.
32189 (is_negative_overflow_infinity): Likewise.
32190 (is_positive_overflow_infinity): Likewise.
32191 (is_overflow_infinity): Likewise.
32192 (stmt_overflow_infinity): Likewise.
32193 (overflow_infinity_range_p): Likewise.
32194 (usable_range_p): Remove as always returning true.
32195 (make_overflow_infinity): Remove.
32196 (negative_overflow_infinity): Likewise.
32197 (positive_overflow_infinity): Likewise.
32198 (avoid_overflow_infinity): Likewise.
32199 (set_value_range): Adjust accordingly.
32200 (set_value_range_to_nonnegative): Likewise, remove now unused
32201 overflow_infinity arg.
32202 (vrp_operand_equal_p): Adjust.
32203 (update_value_range): Likewise.
32204 (range_int_cst_singleton_p): Likewise.
32205 (operand_less_p): Likewise.
32206 (compare_values_warnv): Likewise.
32207 (extract_range_for_var_from_comparison_expr): Likewise.
32208 (vrp_int_const_binop): Likewise.
32209 (zero_nonzero_bits_from_vr): Likewise.
32210 (extract_range_from_multiplicative_op_1): Likewise.
32211 (extract_range_from_binary_expr_1): Likewise.
32212 (extract_range_from_unary_expr): Likewise.
32213 (extract_range_from_comparison): Likewise.
32214 (extract_range_basic): Likewise.
32215 (adjust_range_with_scev): Likewise.
32216 (compare_ranges): Likewise.
32217 (compare_range_with_value): Likewise.
32218 (dump_value_range): Likewise.
32219 (test_for_singularity): Likewise, remove strict_overflow_p parameter
32221 (simplify_cond_using_ranges): Adjust.
32223 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32225 * brig-builtins.def: Added a builtin for class_f64.
32226 * builtin-types.def: Added a builtin type needed by class_f64.
32228 2017-05-03 Jason Merrill <jason@redhat.com>
32230 * timevar.def: Add TV_CONSTEXPR.
32232 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32234 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
32236 2017-05-03 Martin Jambor <mjambor@suse.cz>
32238 * ipa-prop.c (ipa_update_after_lto_read): Removed.
32239 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
32240 * ipa-cp.c (ipcp_propagate_stage): Do not call
32241 ipa_update_after_lto_read.
32242 * ipa-inline.c (ipa_inline): Likewise.
32244 2017-05-03 Martin Jambor <mjambor@suse.cz>
32246 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
32247 tag. Added a default constructor and a destructor.
32248 (ipa_edge_args_sum_t): New class;
32249 (ipa_edge_args_sum): Declare.
32250 (ipa_edge_args_vector): Remove declaration.
32251 (IPA_EDGE_REF): Use ipa_edge_args_sum.
32252 (ipa_free_edge_args_substructures): Remove declaration.
32253 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
32254 (ipa_edge_args_info_available_for_edge_p): Likewise.
32255 * ipa-prop.c (ipa_edge_args_vector): Removed.
32256 (edge_removal_hook_holder): Likewise.
32257 (edge_duplication_hook_holder): Likewise.
32258 (ipa_edge_args_sum): New variable.
32259 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
32260 ipa_edge_args_vector.
32261 (ipa_free_edge_args_substructures): Likewise.
32262 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
32263 ipa_edge_args_vector.
32264 (ipa_edge_removal_hook): Turned into method
32265 ipa_edge_args_sum_t::remove.
32266 (ipa_edge_duplication_hook): Turned into method
32267 ipa_edge_args_sum_t::duplicate.
32268 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
32269 registering edge hooks.
32270 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
32271 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
32272 ipa_edge_args_sum instead of ipa_edge_args_vector.
32273 * ipa-profile.c (ipa_profile): Likewise.
32275 2017-05-03 Martin Jambor <mjambor@suse.cz>
32277 * symbol-summary.h (function_summary): New method exists.
32278 (function_summary::symtab_removal): Deallocate through release.
32279 (call_summary): New class.
32280 (gt_ggc_mx): New overload.
32281 (gt_pch_nx): Likewise.
32282 (gt_pch_nx): Likewise.
32284 2017-05-03 Jeff Law <law@redhat.com>
32286 PR tree-optimization/78496
32287 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
32288 from simplify_cond_using_ranges. Split off code to walk
32289 backwards through casts into ...
32290 (simplify_cond_using_ranges_2): New function.
32291 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
32292 (execute_vrp): After identifying jump threads, call
32293 simplify_cond_using_ranges_2.
32295 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
32298 * ipa-inline.h (inline_summary): Add ctor.
32299 (create_ggc): Do not use ggc_cleared_alloc.
32301 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
32302 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32304 * gcc.c (handle_braces): Support escaping in switch matching
32306 * doc/invoke.texi (Spec Files): Document it.
32307 Remove superfluous @code markup in items.
32309 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32311 * diagnostic-show-locus.c (struct column_range): New struct.
32312 (get_affected_columns): New function.
32313 (get_printed_columns): New function.
32314 (struct correction): New struct.
32315 (correction::ensure_capacity): New function.
32316 (correction::ensure_terminated): New function.
32317 (struct line_corrections): New struct.
32318 (line_corrections::~line_corrections): New dtor.
32319 (line_corrections::add_hint): New function.
32320 (layout::print_trailing_fixits): Reimplement in terms of the new
32322 (selftest::test_overlapped_fixit_printing): New function.
32323 (selftest::diagnostic_show_locus_c_tests): Call it.
32325 2017-05-03 Nathan Sidwell <nathan@acm.org>
32327 Canonicalize canonical type hashing
32328 * tree.h (type_hash_canon_hash): Declare.
32329 * tree.c (type_hash_list, attribute_hash_list): Move into
32330 type_hash_canon_hash.
32331 (build_type_attribute_qual_variant): Break out hash code calc into
32332 type_hash_canon_hash.
32333 (type_hash_canon_hash): New. Generic type hash computation.
32334 (build_range_type_1, build_array_type_1, build_function_type,
32335 build_method_type_directly, build_offset_type, build_complex_type,
32336 make_vector_type): Call it.
32338 2017-05-03 Richard Biener <rguenther@suse.de>
32340 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
32341 When all DRs have unknown misaligned do not always peel
32342 when there is a store but apply the same costing model as if
32343 there were only loads.
32345 2017-05-03 Richard Biener <rguenther@suse.de>
32348 PR tree-optimization/80492
32349 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32350 compare_base_decls returning dont-know properly.
32352 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
32354 * config/arm/iterators.md (CCSI): New mode iterator.
32355 (arch): New mode attribute.
32356 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
32357 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
32358 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
32359 code iterator for success result mode.
32360 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
32361 the corresponding new insn generators.
32363 2017-05-03 Bin Cheng <bin.cheng@arm.com>
32366 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32367 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32369 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
32371 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
32372 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
32373 (DDR_A): Wrap DDR argument in brackets.
32374 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
32375 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
32376 (DDR_REVERSED_P): Likewise.
32378 2017-05-03 Jakub Jelinek <jakub@redhat.com>
32380 PR tree-optimization/79472
32381 * tree-switch-conversion.c (struct switch_conv_info): Add
32382 contiguous_range and default_case_nonstandard fields.
32383 (collect_switch_conv_info): Compute contiguous_range and
32384 default_case_nonstandard fields, don't clear final_bb if
32385 contiguous_range and only the default case doesn't have the required
32387 (check_all_empty_except_final): Set default_case_nonstandard instead
32388 of failing if contiguous_range and the default case doesn't have empty
32390 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
32391 and only the default case doesn't have the required constants. Skip
32393 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
32394 if default_case_nonstandard.
32395 (build_constructors): Build constant 1 just once. Assert that default
32396 values aren't inserted in between cases if contiguous_range. Skip
32398 (build_arrays): Skip virtual phis.
32399 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
32400 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
32401 Handle virtual phis.
32402 (gen_inbound_check): Handle default_case_nonstandard case.
32403 (process_switch): Adjust check_final_bb caller. Call
32404 gather_default_values with the first non-default case instead of
32405 default case if default_case_nonstandard.
32407 2017-05-02 Nathan Sidwell <nathan@acm.org>
32409 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
32410 check. Fix formatting.
32412 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
32414 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
32415 errors when comparing specialized and unspecialized times.
32417 2017-05-02 David Malcolm <dmalcolm@redhat.com>
32419 * diagnostic-show-locus.c
32420 (layout::should_print_annotation_line_p): Make private.
32421 (layout::print_annotation_line): Make private.
32422 (layout::annotation_line_showed_range_p): Make private.
32423 (layout::show_ruler): Make private.
32424 (layout::print_source_line): Make private. Pass in line and
32425 line_width, rather than calling location_get_source_line. Drop
32427 (layout::print_leading_fixits): New method.
32428 (layout::print_any_fixits): Rename to...
32429 (layout::print_trailing_fixits): ...this, and make private.
32430 Don't print newline fixits.
32431 (diagnostic_show_locus): Move logic for printing one row into...
32432 (layout::print_line): ...this new function. Move the
32433 location_get_source_line call and error-handling from
32434 print_source_line to here. Call print_leading_fixits, and rename
32435 print_any_fixits to print_trailing_fixits.
32436 (selftest::test_fixit_insert_containing_newline): Update now that
32437 newlines are partially supported.
32438 (selftest::test_fixit_insert_containing_newline_2): New test.
32439 (selftest::test_fixit_replace_containing_newline): Update comments.
32440 (selftest::diagnostic_show_locus_c_tests): Call the new test.
32441 * edit-context.c (class added_line): New class.
32442 (class edited_line): Describe newline handling in comment.
32443 (edited_line::actually_edited_p): New method.
32444 (edited_line::print_content): Delete redundant decl.
32445 (edited_line::m_predecessors): New field.
32446 (edited_file::print_content): Call edited_line::print_content.
32447 (edited_file::print_diff): Update to support newlines.
32448 (edited_file::print_diff_hunk): Likewise.
32449 (edited_file::print_run_of_changed_lines): New function.
32450 (edited_file::print_diff_line): Convert to...
32451 (print_diff_line): ...this.
32452 (edited_file::get_effective_line_count): New function.
32453 (edited_line::edited_line): Initialize new field m_predecessors.
32454 (edited_line::~edited_line): Clean up m_predecessors.
32455 (edited_line::apply_fixit): Handle newlines.
32456 (edited_line::get_effective_line_count): New function.
32457 (edited_line::print_content): New function.
32458 (edited_line::print_diff_lines): New function.
32459 (selftest::test_applying_fixits_insert_containing_newline): New
32461 (selftest::test_applying_fixits_replace_containing_newline): New
32463 (selftest::insert_line): New function.
32464 (selftest::test_applying_fixits_multiple_lines): Add example of
32466 (selftest::edit_context_c_tests): Call the new tests.
32468 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32470 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
32471 parameter cand. Update dump information.
32472 (get_computation_cost): Update uses.
32474 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32476 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
32477 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
32478 (get_computation_at, rewrite_use_address): Update use of
32479 get_computation_aff.
32481 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32483 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
32484 (get_computation): Delete.
32485 (get_computation_cost): Implement like get_computation_cost_at.
32486 Use get_computation_at.
32487 (get_computation_cost_at): Delete.
32488 (rewrite_use_nonlinear_expr): Use get_computation_at.
32489 (rewrite_use_compare, remove_unused_ivs): Ditto.
32491 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32493 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
32495 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32497 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
32498 (ivopts_global_cost_for_size): Rename parameter and update uses.
32499 (iv_ca_recount_cost): Update uses.
32500 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
32501 candidates seperately in n_invs and n_cands.
32502 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
32504 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32506 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
32507 (find_inv_vars_cb): New.
32508 (find_depends): Renamed to ...
32509 (find_inv_vars): ... this.
32510 (add_candidate_1, force_var_cost): Call find_inv_vars.
32511 (split_address_cost, determine_group_iv_cost_cond): Ditto.
32513 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32515 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
32516 inv_vars. Add inv_exprs.
32517 (struct iv_cand): Rename depends_on to inv_vars.
32518 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
32519 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
32520 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
32521 (dump_cand): Dump inv_vars.
32522 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
32523 (record_invariant, find_depends, add_candidate_1): Ditto.
32524 (set_group_iv_cost, force_var_cost): Ditto.
32525 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
32526 (get_computation_cost_at, get_computation_cost): Ditto.
32527 (determine_group_iv_cost_generic): Ditto.
32528 (determine_group_iv_cost_address): Ditto.
32529 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
32530 (determine_group_iv_costs): Ditto.
32531 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
32532 (iv_ca_set_remove_invariants): Renamed to ...
32533 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
32534 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
32535 (iv_ca_set_add_invariants): Renamed to ...
32536 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
32537 (iv_ca_set_cp): Use iv_ca_set_add_invs.
32538 (iv_ca_has_deps): Support inv_vars and inv_exprs.
32539 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
32540 (create_new_ivs): Remove useless dump.
32542 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32544 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
32546 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
32547 (iv_ca_set_no_cp, create_new_iv): Ditto.
32549 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32551 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32553 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32555 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
32556 function tree_check2.
32558 2017-05-02 Martin Liska <mliska@suse.cz>
32560 * doc/gcov.texi: Add missing preposition.
32561 * gcov.c (function_info::function_info): Properly fill up
32562 all member variables.
32564 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32566 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
32568 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32570 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
32572 2017-05-02 Martin Liska <mliska@suse.cz>
32575 * lto-streamer-in.c (lto_read_tree_1): Remove
32576 LTO_STREAMER_DEBUG.
32577 * lto-streamer.c (struct tree_hash_entry): Likewise.
32578 (struct tree_entry_hasher): Likewise.
32579 (tree_entry_hasher::hash): Likewise.
32580 (tree_entry_hasher::equal): Likewise.
32581 (lto_streamer_init): Likewise.
32582 (lto_orig_address_map): Likewise.
32583 (lto_orig_address_get): Likewise.
32584 (lto_orig_address_remove): Likewise.
32585 * lto-streamer.h: Likewise.
32586 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
32587 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
32589 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
32591 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
32592 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
32593 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
32594 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
32595 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
32596 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
32597 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
32598 (mm_maskz_sub_ss): New intrinsics.
32599 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32600 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32601 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
32602 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
32603 (__builtin_ia32_subss_mask_round): New builtins.
32604 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32605 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32606 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
32608 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
32609 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32611 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32614 2017-05-02 Martin Jambor <mjambor@suse.cz>
32616 PR tree-optimization/78687
32617 * tree-sra.c (access): New field parent.
32618 (process_subtree_disqualification): New function.
32619 (disqualify_candidate): Call it.
32620 (build_accesses_from_assign): Reset write flag if creating an
32622 (build_access_subtree): Fill in parent field and also prpagate
32623 down grp_write flag.
32624 (create_artificial_child_access): New parameter set_grp_write, set
32625 grp_write to its value.
32626 (propagate_subaccesses_across_link): Also propagate grp_write flag
32628 (propagate_all_subaccesses): Push the closest parent back to work
32629 queue if add_access_to_work_queue returned true.
32631 2017-05-02 Richard Biener <rguenther@suse.de>
32633 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32634 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32635 -fstrict-overflow documentation.
32636 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32637 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32638 flag_strict_overflow.
32639 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32640 * lto-opts.c (lto_write_options): Do not stream it.
32641 * lto-wrapper.c (merge_and_complain): Do not handle it.
32642 * opts.c (default_options_table): Do not set -fstrict-overflow.
32643 (finish_options): Likewise do not clear it when sanitizing.
32644 * simplify-rtx.c (simplify_const_relational_operation): Do not
32645 test flag_strict_overflow.
32647 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32649 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32650 using enabled attribute.
32651 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32652 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32653 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32654 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32655 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32656 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32657 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32658 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32659 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32660 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32662 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32664 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32666 2017-05-02 Richard Biener <rguenther@suse.de>
32668 PR tree-optimization/80591
32670 2017-04-10 Richard Biener <rguenther@suse.de>
32672 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32675 2017-05-02 Richard Biener <rguenther@suse.de>
32677 PR tree-optimization/80549
32678 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32679 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32680 headers if they do not have a preheader.
32682 2017-05-02 Martin Liska <mliska@suse.cz>
32685 * common.opt: Fix typo.
32686 * doc/invoke.texi: Likewise.
32688 2017-05-01 Jan Beulich <jbeulich@suse.com>
32690 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32691 swapping, add (x,x,m,x,n) alternative.
32693 2017-05-01 Nathan Sidwell <nathan@acm.org>
32695 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32696 unnecessary unadjusted_alignment check.
32698 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32701 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32703 * gimplify.c (gimplify_cilk_detach): New function.
32704 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32705 * tree-core.h: Document EXPR_CILK_SPAWN.
32706 * tree.h (EXPR_CILK_SPAWN): Define.
32708 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32710 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32711 to use new fixit_hint representation, using the "replace" logic.
32712 (get_line_span_for_fixit_hint): Likewise.
32713 (layout::print_any_fixits): Likewise.
32714 (selftest::test_one_liner_many_fixits): Rename to...
32715 (selftest::test_one_liner_many_fixits_1): ...this, and update
32716 comment and expected output to reflect that the multiple fix-it
32717 hints are now consolidated into one insertion.
32718 (selftest::test_one_liner_many_fixits_2): New test.
32719 (selftest::test_diagnostic_show_locus_one_liner): Update for
32721 (selftest::test_fixit_consolidation): Update for fix-it API
32723 * diagnostic.c (print_parseable_fixits): Likewise.
32724 * edit-context.c (edited_line::m_line_events): Convert from
32725 auto_vec <line_event *> to auto_vec <line_event>.
32726 (class line_event): Convert from abstract base class to a concrete
32727 class, taking over the role of replace_event.
32728 (class insert_event): Delete.
32729 (class replace_event): Rename to class line_event. Convert to
32731 (edit_context::add_fixits): Reimplement.
32732 (edit_context::apply_insert): Delete.
32733 (edit_context::apply_replace): Rename to...
32734 (edit_context::apply_fixit): ...this. Convert to half-open range.
32735 (edited_file::apply_insert): Delete.
32736 (edited_file::apply_replace): Rename to...
32737 (edited_file::apply_fixit): ...this.
32738 (edited_line::~edited_line): Drop deletion of events.
32739 (edited_line::apply_insert): Delete.
32740 (edited_line::apply_replace): Rename to...
32741 (edited_line::apply_fixit): ...this. Convert to half-open range.
32742 Update for change to type of m_line_events.
32743 * edit-context.h (edit_context::apply_insert): Delete.
32744 (edit_context::apply_replace): Rename to...
32745 (edit_context::apply_fixit): ...this.
32747 2017-05-01 Martin Sebor <msebor@redhat.com>
32749 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32752 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32755 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32756 __get_cpuid_max returns 0.
32757 (__get_cpuid_count): Ditto.
32759 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32761 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32762 replacement expression is another instance of one of its arguments.
32764 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32767 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32768 check for stack push/pop autoinc.
32769 * config/i386/i386.c (ix86_agi_dependent): Return false
32770 if the only reason why modified_in_p returned true is that
32771 addr is SP based and set_insn is a push or pop.
32773 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32775 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32778 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32781 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32782 (account_size_time): Use two predicates - exec_pred and
32784 (evaluate_conditions_for_known_args): Compute both clause and
32786 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32787 (inline_summary_t::duplicate): Update.
32788 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32790 (compute_inline_parameters): Likewise.
32791 (estimate_edge_size_and_time): Update caluclation of time.
32792 (estimate_node_size_and_time): Compute both time and nonspecialized
32794 (estimate_ipcp_clone_size_and_time): Update.
32795 (inline_merge_summary): Update.
32796 (do_estimate_edge_time): Update.
32797 (do_estimate_edge_size): Update.
32798 (do_estimate_edge_hints): Update.
32799 (inline_read_section, inline_write_summary): Stream both new predicates.
32800 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32802 (compute_inlined_call_time): Cleanup.
32803 (big_speedup_p): Update.
32804 (edge_badness): Update.
32805 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32806 (size_time_entry): Replace predicate by exec_predicate and
32807 nonconst_predicate.
32808 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32809 (estimate_edge_time): Return also nonspec_time.
32810 (reset_edge_growth_cache): Update.
32812 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32814 PR rtl-optimization/80491
32815 * ifcvt.c (noce_process_if_block): When looking for x setter
32816 with missing else_bb, don't check only the insn right before
32817 cond_earliest, but look for the last insn that x is modified in
32818 within the same bb.
32820 PR rtl-optimization/80491
32821 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32823 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32825 PR tree-optimization/80487
32826 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32828 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32830 PR tree-optimization/79697
32831 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32832 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32833 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32835 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32836 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32838 2017-04-28 Martin Sebor <msebor@redhat.com>
32840 PR tree-optimization/80523
32841 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32842 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32844 (maybe_warn, format_directive, parse_directive): Use new functions.
32845 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32847 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32849 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32851 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32853 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32854 target_header_dir): Set correctly.
32855 * configure: Regenerated.
32856 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32857 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32858 instead of SYSTEM_HEADER_DIR.
32860 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32862 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32863 (estimate_local_effects): Likewise.
32864 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32865 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32866 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32867 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
32868 do_estimate_edge_time, estimate_edge_time): Likewise.
32869 * ipa-inline-analysis.c (estimate_node_size_and_time,
32870 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
32871 (estimate_time_after_inlining): Remove.
32873 2017-04-28 Martin Liska <mliska@suse.cz>
32875 * doc/gcov.texi: Enhance documentation of gcov.
32877 2017-04-28 Martin Liska <mliska@suse.cz>
32879 * doc/gcov.texi: Sort options in alphabetic order.
32880 * doc/gcov-dump.texi: Likewise.
32881 * doc/gcov-tool.texi: Likewise.
32882 * gcov.c (print_usage): Likewise.
32883 * gcov-dump.c (print_usage): Likewise.
32884 * gcov-tool.c (print_merge_usage_message): Likewise.
32885 (print_rewrite_usage_message): Likewise.
32886 (print_overlap_usage_message): Likewise.
32888 2017-04-28 Martin Liska <mliska@suse.cz>
32890 PR gcov-profile/53915
32891 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
32893 2017-04-28 Martin Liska <mliska@suse.cz>
32895 PR gcov-profile/79891
32896 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
32897 is marked by compiler as living on a line.
32898 (get_cycles_count): Remove usage of the union.
32899 (output_intermediate_file): Likewise.
32900 (find_source): Fix GNU coding style.
32901 (accumulate_line_counts): Remove old non-all block mode.
32902 (output_lines): Remove usage of the union.
32903 * profile.c (output_location): Include all BBs, even if
32904 belonging to a same line (and file) as a previous BB.
32906 2017-04-28 Martin Liska <mliska@suse.cz>
32908 * gcov.c (process_args): Handle new argument 'w'.
32909 (read_graph_file): Assign ID to BBs.
32910 (output_branch_count): Display BB # if verbose flag is set.
32911 (output_lines): Likewise for arcs.
32912 (print_usage): Add '--verbose' option help.
32913 * doc/gcov.texi: Document --verbose (-w) option.
32915 2017-04-28 Martin Liska <mliska@suse.cz>
32917 * gcov.c (struct block_location_info): New struct.
32918 (process_file): Fill up the new structure.
32919 (read_graph_file): Replace usage of encoding by the newly added
32921 (add_line_counts): Likewise.
32922 (accumulate_line_counts): Remove usage of the union.
32923 (function_info::function_info): New function.
32924 (function_info::~function_info): Likewise.
32925 (process_file): Call delete instead of release_function.
32926 (release_function): Release the function.
32927 (release_structures): Call delete instead of release_function.
32928 (solve_flow_graph): Replace usage of num_blocks.
32929 (find_exception_blocks): Likewise.
32930 (output_lines): Fix GNU coding style.
32932 2017-04-28 Martin Liska <mliska@suse.cz>
32935 * coverage.c (coverage_remove_note_file): New function.
32936 * coverage.h: Declare the function.
32937 * toplev.c (finalize): Clean if an error has been seen.
32939 2017-04-28 Martin Liska <mliska@suse.cz>
32941 PR gcov-profile/80031
32942 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
32943 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
32944 * gcov.c (read_graph_file): Read just number of blocks.
32945 * profile.c (branch_prob): Do not stream 0 flags per a basic
32948 2017-04-28 Martin Liska <mliska@suse.cz>
32950 * gcov-dump.c (tag_*): Add new argument to declarations.
32951 (dump_gcov_file): Likewise.
32952 (tag_blocks): Add and use new argument depth.
32953 (tag_arcs): Likewise.
32954 (tag_lines): Likewise.
32955 (tag_counters): Likewise.
32956 (tag_summary): Likewise.
32957 (dump_working_sets): Use depth to do a proper indentation.
32959 2017-04-28 Jakub Jelinek <jakub@redhat.com>
32962 * cgraph.h (symtab_node::debug_symtab): No longer inline.
32963 * symtab.c (symtab_node::debug_symtab): Move definition here.
32965 2017-04-28 Richard Biener <rguenther@suse.de>
32967 * lto-streamer.h (LTO_major_version): Bump to 7.
32969 2017-04-28 Richard Biener <rguenther@suse.de>
32971 * tree-vrp.c (assert_info): New struct.
32972 (add_assert_info): New helper.
32973 (register_edge_assert_for_2): Refactor to add asserts to a vector
32975 (register_edge_assert_for_1): Likewise.
32976 (register_edge_assert_for): Likewise.
32977 (finish_register_edge_assert_for): New helper actually registering
32978 asserts where live on edge.
32979 (find_conditional_asserts): Adjust.
32980 (find_switch_asserts): Likewise.
32981 (evrp_dom_walker::try_find_new_range): Generalize.
32982 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
32984 2017-04-27 Marek Polacek <polacek@redhat.com>
32987 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
32988 arg10 and arg11 to itype.
32990 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
32992 * doc/extend.texi (Object Size Checking): Improve grammar.
32994 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
32997 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
32998 that the logic for permitting reciprocal estimates matches that
33001 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33004 * tree.c (type_cache_hasher::equal): Only compare
33005 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33006 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33007 non-aggregate element types.
33008 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33009 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33011 2017-04-27 Richard Biener <rguenther@suse.de>
33013 PR middle-end/80533
33014 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33015 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33016 keeping a reference to a trailing array.
33018 2017-04-27 Richard Biener <rguenther@suse.de>
33020 PR middle-end/80539
33021 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33022 being in loop-closed SSA form conservatively.
33023 (chrec_fold_multiply_poly_poly): Likewise.
33025 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33027 PR middle-end/79665
33028 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33029 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33031 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33034 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33035 (aarch64_function_arg_alignment): Return unsigned int again, but still
33036 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33037 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33038 Don't emit -Wpsabi note.
33039 (aarch64_function_arg_boundary): Likewise.
33040 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33043 2017-04-26 Nathan Sidwell <nathan@acm.org>
33045 * tree.h (crc32_unsigned_n): Declare.
33046 (crc32_unsigned, crc32_unsigned): Make inline.
33047 * tree.c (crc32_unsigned_bits): Replace with ...
33048 (crc32_unsigned_n): ... this.
33049 (crc32_unsigned, crc32_byte): Remove.
33050 (crc32_string): Remove unnecessary braces.
33052 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33054 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33055 * ipa-inline-analysis.c (MAX_TIME): Remove.
33056 (account_size_time): Use sreal for time.
33057 (dump_inline_summary): Update.
33058 (estimate_function_body_sizes): Update.
33059 (estimate_edge_size_and_time): Update.
33060 (estimate_calls_size_and_time): Update.
33061 (estimate_node_size_and_time): Update.
33062 (inline_merge_summary): Update.
33063 (inline_update_overall_summary): Update.
33064 (estimate_time_after_inlining): Update.
33065 (inline_read_section): Update.
33066 (inline_write_summary): Update.
33067 * ipa-inline.c (compute_uninlined_call_time): Update.
33068 (compute_inlined_call_time): Update.
33069 (recursive_inlining): Update.
33070 (inline_small_functions): Update.
33071 (dump_overall_stats): Update.
33072 * ipa-inline.h: Include sreal.h.
33073 (size_time_entry): Turn time to sreal.
33074 (inline_summary): Turn self_time nad time to sreal.
33076 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33078 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33080 (sreal::stream_out, sreal::stream_in): New.
33081 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33083 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33085 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33088 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33091 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33093 (dimode_scalar_chain::compute_convert_gain): Ditto.
33094 (dimode_scalar_chain::make_vector_copies): Ditto.
33095 (dimode_scalar_chain::convert_reg): Ditto.
33096 (dimode_scalar_chain::convert_insn): Ditto.
33097 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33098 (VI248_AVX512BW_1): New mode iterator.
33099 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33100 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33103 2017-04-25 Martin Sebor <msebor@redhat.com>
33105 PR tree-optimization/80497
33106 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33107 constants are representable in HOST_WIDE_INT.
33108 (parse_directive): Ditto.
33110 2017-04-25 Martin Sebor <msebor@redhat.com>
33113 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
33114 (new_zero_array): Adjust signature.
33115 (dom_info::dom_init): Used unsigned rather that size_t.
33116 (dom_info::dom_info): Same.
33118 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33119 Jakub Jelinek <jakub@redhat.com>
33122 * config/arm/arm.c: Include gimple.h.
33123 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33124 returns negative, increment ncrn only if it returned positive.
33125 (arm_needs_doubleword_align): Return int instead of bool,
33126 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
33127 members, but if there is any such non-FIELD_DECL
33128 > PARM_BOUNDARY aligned decl, return -1 instead of false.
33129 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33130 returns negative, increment nregs only if it returned positive.
33131 (arm_setup_incoming_varargs): Likewise.
33132 (arm_function_arg_boundary): Emit -Wpsabi note if
33133 arm_needs_doubleword_align returns negative, return
33134 DOUBLEWORD_ALIGNMENT only if it returned positive.
33136 2017-04-25 Marek Polacek <polacek@redhat.com>
33139 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33140 first argument to type.
33142 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
33145 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
33146 type checks to test for compatibility instead of equality.
33148 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33149 Jakub Jelinek <jakub@redhat.com>
33152 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
33154 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
33155 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
33156 the alignment computation, but return their maximum in warn_alignment.
33157 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33158 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
33160 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
33161 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33164 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33166 * config/arc/simdext.md (dmpyh): Fix typo.
33168 2017-04-25 Richard Biener <rguenther@suse.de>
33170 PR tree-optimization/80492
33171 * alias.c (compare_base_decls): Handle registers with asm
33172 specification conservatively.
33173 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33174 compare_base_decls returning dont-know properly.
33176 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33178 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
33179 (legitimate_offset_address_p): New function.
33180 (arc_legitimate_address_p): Use above function.
33182 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33184 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
33186 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33188 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
33189 ACCH registers whenever they are available.
33191 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33193 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
33194 double regs fix when not used.
33196 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33198 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
33200 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
33201 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
33203 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33205 * config/arc/arc.c (arc_output_addsi): Check for h-register class
33206 when emitting short ADD instructions.
33208 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33210 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
33212 (cmpsi_cc_c_insn): Likewise.
33213 (cbranchsi4_scratch): Compute proper instruction length using
33214 compact_hreg_operand.
33215 * config/arc/predicates.md (compact_hreg_operand): New predicate.
33217 2017-04-25 Richard Biener <rguenther@suse.de>
33219 PR middle-end/80509
33220 * passes.c (pass_manager::pass_manager): Initialize
33221 m_name_to_pass_map.
33223 2017-04-25 Richard Biener <rguenther@suse.de>
33225 PR tree-optimization/79201
33226 * tree-ssa-sink.c (statement_sink_location): Handle calls.
33228 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33231 * config/s390/vector.md: Split MEM->GPR vector moves for
33232 non-s_operand addresses.
33234 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33237 * config/s390/predicates.md (reload_const_wide_int_operand): New
33239 * config/s390/s390.md ("movti"): Remove d/P alternative.
33240 ("movti_bigconst"): New pattern definition.
33242 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33245 * s390-protos.h (s390_expand_cs_hqi): Removed.
33246 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
33247 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
33248 modes as well as CCZ1mode and CCZmode.
33249 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
33250 signature of s390_emit_compare_and_swap.
33251 (s390_expand_cs_hqi): Likewise, make static.
33252 (s390_expand_cs_tdsi): Generate an explicit compare before trying
33253 compare-and-swap, in some cases.
33254 (s390_expand_cs): Wrapper function.
33255 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
33257 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
33258 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
33259 patterns for small and large integers. Forbid symref memory operands.
33260 Move expander to s390.c. Require cc register.
33261 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
33262 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
33263 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
33264 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
33265 symref memory operands. Remove CC mode and call s390_match_ccmode
33267 ("atomic_exchange<mode>"): Allow and implement all integer modes.
33269 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33271 * config/s390/s390.md (define_peephole2): New peephole to help
33272 combining the load-and-test pattern with volatile memory.
33274 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33276 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
33277 with CCZmode for TARGET_Z196.
33279 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33281 PR rtl-optimization/80501
33282 * combine.c (make_compound_operation_int): Set subreg_code to SET
33283 even for AND with mask of the sign bit of mode.
33285 PR rtl-optimization/80500
33286 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
33287 sum's initial value.
33289 2017-04-25 Julian Brown <julian@codesourcery.com>
33290 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33292 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
33294 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
33296 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
33298 2017-04-25 Julian Brown <julian@codesourcery.com>
33299 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33301 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
33302 (thunderx2t99_sha): New Reservation.
33304 2017-04-25 Julian Brown <julian@codesourcery.com>
33305 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33307 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
33308 type for 1-element load.
33310 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
33312 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
33314 2017-04-24 Martin Jambor <mjambor@suse.cz>
33316 PR tree-optimization/80293
33317 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
33318 char arrays not totally scalarizable if it is false.
33319 (analyze_all_variable_accesses): Pass correct value in the new
33320 parameter. Add a statistics counter.
33322 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
33324 PR middle-end/79931
33325 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
33327 2017-04-24 Richard Biener <rguenther@suse.de>
33329 PR tree-optimization/80494
33330 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
33331 out for complex types.
33333 2017-04-24 Richard Biener <rguenther@suse.de>
33335 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
33336 * tree-ssa-sccvn.c (print_scc): Print SCC size.
33337 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
33338 (DFS): Adjust and never fail.
33339 (sccvn_dom_walker::fail): Remove.
33340 (sccvn_dom_walker::before_dom_children): Adjust.
33341 (run_scc_vn): Likewise and never fail.
33342 * tree-ssa-pre.c (pass_pre::execute): Adjust.
33343 (pass_fre::execute): Likewise.
33345 2017-04-24 Richard Biener <rguenther@suse.de>
33347 PR tree-optimization/79725
33348 * tree-ssa-sink.c (statement_sink_location): Return whether
33349 failure reason was zero uses. Move that check later.
33350 (sink_code_in_bb): Deal with zero uses by removing the stmt
33353 2017-04-24 Richard Biener <rguenther@suse.de>
33356 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
33357 pointer-based references.
33359 2017-04-24 Richard Biener <rguenther@suse.de>
33362 * pass_manager.h (pass_manager::operator new): Remove.
33363 (pass_manager::operator delete): Likewise.
33364 * passes.c (pass_manager::operator new): Remove.
33365 (pass_manager::operator delete): Likewise.
33366 (pass_manager::pass_manager): Zero individual pass members.
33368 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
33371 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
33372 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
33373 Check "XEXP (src, 1)" operand here.
33374 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
33375 Check "XEXP (src, 1)" operand here.
33376 (dimode_scalar_chain::make_vector_copies): Detect count register
33377 of a shift instruction. Zero extend count register from QImode
33378 to DImode to satisfy vector shift pattern count operand predicate.
33379 Substitute vector shift count operand with a DImode copy.
33380 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
33383 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
33385 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
33386 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33387 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
33388 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33389 (UNSPEC_NOREX_MEM): Remove definition.
33391 2017-04-21 Richard Biener <rguenther@suse.de>
33393 PR tree-optimization/79547
33394 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
33395 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
33396 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
33397 without any constraints.
33399 2017-04-21 Richard Biener <rguenther@suse.de>
33401 PR tree-optimization/78847
33402 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
33404 2017-04-21 Richard Biener <rguenther@suse.de>
33406 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
33407 (build_distinct_type_copy): Likewise.
33408 (build_variant_type_copy): Likewise.
33409 * tree.c (build_qualified_type): Pass down mem-stat info.
33410 (build_distinct_type_copy): Likewise.
33411 (build_variant_type_copy): Likewise.
33413 2017-04-21 Richard Biener <rguenther@suse.de>
33415 PR tree-optimization/80237
33416 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
33418 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
33419 for a simplified result.
33421 2016-04-21 Richard Biener <rguenther@suse.de>
33423 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
33424 sth as strict as a simple_iv but a chrec without symbols and an
33425 operand defined in the loop we are peeling (and not some subloop).
33426 (propagate_constants_for_unrolling): Propagate all constants.
33428 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33431 * config/i386/i386.c (print_reg): Remove assert for disalowed
33432 regno values, call output_operand_lossage instead.
33434 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33437 * config/i386/constraints.md (Yc): New register constraint.
33438 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
33439 Use Yc constraint for alternative 2 of operand 0. Remove
33440 preferred_for_speed attribute.
33442 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33444 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
33445 lastprivate clauses in SIMT case.
33447 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
33449 * doc/invoke.texi (-Wextra-semi): Document new warning option.
33451 2017-04-20 Richard Biener <rguenther@suse.de>
33453 PR tree-optimization/57796
33454 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
33455 as N scalar stores.
33456 (vect_model_load_cost): Cost gathers as N scalar loads.
33458 2017-04-20 Richard Biener <rguenther@suse.de>
33460 * ggc-page.c (ggc_allocated_p): Rename to ...
33461 (safe_lookup_page_table_entry): ... this and return the lookup
33463 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
33465 2017-04-20 Richard Biener <rguenther@suse.de>
33467 PR tree-optimization/80453
33468 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
33469 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
33470 from the conditions.
33471 (vn_phi_eq): Pass them down.
33472 (vn_phi_lookup): Record them.
33473 (vn_phi_insert): Likewise.
33475 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
33477 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
33478 uninitialized variable warning to avoid buffer overrun.
33480 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33483 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
33484 is suppressed for '{ 0 }' in C.
33486 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33488 * BASE-VER: Set to 8.0.0.
33490 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
33492 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
33493 priority .init_array and .fini_array section with SECTION_NOTYPE
33496 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33498 PR middle-end/80423
33499 * tree.h (build_array_type): Add typeless_storage default argument.
33500 * tree.c (type_cache_hasher::equal): Also compare
33501 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
33502 (build_array_type): Add typeless_storage argument, set
33503 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
33505 (build_nonshared_array_type): Adjust build_array_type_1 caller.
33506 (build_array_type): Likewise. Add typeless_storage argument.
33508 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33509 Jakub Jelinek <jakub@redhat.com>
33511 PR tree-optimization/80426
33512 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
33513 operation on symbolic operands, also compute the overflow for the
33514 invariant part when the operation degenerates into a negation.
33516 2017-04-19 Jakub Jelinek <jakub@redhat.com>
33519 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
33520 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
33523 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
33525 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
33528 * config/avr/avr.c (tree.h): Include it.
33529 (cgraph.h): Include it.
33530 (avr_encode_section_info): Don't warn for uninitialized progmem
33531 variable if it's just an alias.
33533 2017-04-19 Richard Biener <rguenther@suse.de>
33536 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
33537 when needed by AutoPGO.
33539 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
33542 * doc/lto.texi: Remove an extra 'that'.
33544 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
33546 PR rtl-optimization/80429
33547 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
33548 are only used in debug insns.
33550 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33551 Vladimir Makarov <vmakarov@redhat.com>
33553 * config/sparc/predicates.md (input_operand): Add comment. Return
33554 true for any memory operand when LRA is in progress.
33555 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
33557 2017-04-18 Jeff Law <law@redhat.com>
33560 * mips.md ({return,simple_return}_internal): Do not overwrite
33563 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33565 PR tree-optimization/80443
33566 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
33567 instead of adding 1, subtract -1 and similarly instead of subtracting
33570 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
33572 PR rtl-optimization/80357
33573 * haifa-sched.c (tmp_bitmap): New variable.
33574 (model_recompute): Handle duplicate use records.
33575 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
33576 (free_global_sched_pressure_data): Free it.
33578 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
33581 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33582 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33583 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33584 instead of SYSTEM_HEADER_DIR.
33586 2017-04-18 Jeff Law <law@redhat.com>
33588 PR middle-end/80422
33589 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
33590 predecessors after walking up the insn chain.
33592 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33595 * dwarf2out.c (modified_type_die): Try harder not to emit internal
33596 sizetype type into debug info.
33598 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33601 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
33602 unneeded test for TARGET_UPPER_REGS_SF.
33603 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
33605 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33608 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
33609 instead of gsi_after_labels.
33611 2017-04-18 Jeff Law <law@redhat.com>
33613 * regcprop.c (maybe_mode_change): Avoid creating copies of the
33617 2017-04-13 Jeff Law <law@redhat.com>
33618 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33619 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33621 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33624 * config/avr/avr.c (intl.h): Include it.
33625 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33627 2017-04-18 Martin Liska <mliska@suse.cz>
33629 PR gcov-profile/78783
33630 * gcov-tool.c (gcov_output_files): Validate that destination
33631 file is either removed by the tool or by a user.
33633 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33634 Guy Benyei <guybe@mellanox.com>
33636 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33637 block, and do not negate it, the stored id is already negative.
33639 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33641 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33643 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33646 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33647 masks of options that should be turned off if the VSX vector
33648 options are turned off.
33649 (OTHER_P8_VECTOR_MASKS): Likewise.
33650 (OTHER_VSX_VECTOR_MASKS): Likewise.
33651 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33652 rs6000_disable_incompatible_switches to validate no type switches
33654 (rs6000_incompatible_switch): New function to disallow turning on
33655 other vector options if -mno-vsx, -mno-power8-vector, or
33656 -mno-power9-vector are specified.
33658 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33660 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33662 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33664 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33665 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33666 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33667 (ARG_POINTER_CFA_OFFSET): Likewise.
33669 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33671 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33672 conditions to take advantage of various optimizations.
33674 2017-04-13 Jeff Law <law@redhat.com>
33676 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33677 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33678 (zero_extendsidi2_dext): Likewise.
33680 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33683 * fold-const.c (fold_ternary_loc): Revert
33684 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33687 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33689 PR rtl-optimization/80343
33690 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33691 new scratch pseudo.
33693 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33696 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33697 to ubsan_encode_value.
33699 2017-04-13 Jeff Law <law@redhat.com>
33701 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33702 appearing in DEBUG_INSNs.
33704 2017-04-13 Martin Liska <mliska@suse.cz>
33706 PR gcov-profile/80413
33707 * gcov-io.c (gcov_write_string): Copy to buffer just when
33708 allocated size is greater than zero.
33710 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33713 * dwarf2out.c (decls_for_scope): Ignore declarations of
33714 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33716 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33719 * ipa-visibility.c (non_local_p): Fix typos.
33720 (localize_node): When localizing symbol in same comdat group,
33721 dissolve the group only when we know external symbols are going
33723 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33725 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33727 PR tree-optimization/79390
33728 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33729 order does not result in usable sequence, retry with reversed operand
33735 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33736 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33737 op0 instead of fold_convert_loc (loc, type, arg0).
33739 2017-04-12 Jeff Law <law@redhat.com>
33741 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33742 has a delay slot in the generated code.
33744 * config/cris/cris.md (cris_preferred_reload_class): Return
33745 GENNONACR_REGS rather than GENERAL_REGS.
33747 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33750 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33751 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33752 signedness of the result type.
33754 2017-04-12 Richard Biener <rguenther@suse.de>
33755 Jeff Law <law@redhat.com>
33757 PR tree-optimization/80359
33758 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33759 trim stores to TARGET_MEM_REFs.
33761 2017-04-12 Richard Biener <rguenther@suse.de>
33763 PR tree-optimization/79390
33764 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33765 threading case even more.
33767 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33770 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33771 for quad_address_p for TImode, instead of just not indexed_address.
33773 2017-04-12 Richard Biener <rguenther@suse.de>
33774 Bernd Edlinger <bernd.edlinger@hotmail.de>
33776 PR middle-end/79671
33777 * alias.c (component_uses_parent_alias_set_from): Handle
33778 TYPE_TYPELESS_STORAGE.
33779 (get_alias_set): Likewise.
33780 * tree-core.h (tree_type_common): Add typeless_storage flag.
33781 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33782 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33783 for types containing members with TYPE_TYPELESS_STORAGE.
33784 (place_field): Likewise.
33785 (layout_type): Likewise for ARRAY_TYPE.
33786 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33787 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33788 TYPE_TYPELESS_STORAGE.
33789 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33791 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33794 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33795 first argument to type.
33797 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33801 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33802 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33803 (rs6000_expand_binop_builtin): Likewise.
33804 (rs6000_expand_ternop_builtin): Likewise; also add missing
33805 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33806 vshasigma built-ins.
33807 * doc/extend.texi: Document that vec_xxpermdi's third argument
33808 must be a constant.
33810 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33812 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33813 Use shift_const cost parameter when calculating gain of STV shifts.
33815 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33817 PR rtl-optimization/70478
33818 * lra-constraints.c (process_alt_operands): Check memory for
33819 disfavoring memory insn operand.
33821 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33823 PR middle-end/80100
33824 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33825 left shift in unsigned HOST_WIDE_INT type.
33827 PR rtl-optimization/80385
33828 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33829 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33832 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33833 if they have any depend clauses.
33835 2017-04-11 Martin Liska <mliska@suse.cz>
33838 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33839 * ipa-split.c (split_function): Create a local comdat symbol
33840 if caller is in a comdat group.
33842 2017-04-11 Martin Liska <mliska@suse.cz>
33845 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33848 2017-04-11 Martin Sebor <msebor@redhat.com>
33850 PR middle-end/80364
33851 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33852 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33854 (directive::set_width, directive::set_precision, format_character):
33856 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33859 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33862 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33863 conflict, set target->arch_name instead of target->cpu_name.
33865 2017-04-11 Richard Biener <rguenther@suse.de>
33867 PR tree-optimization/80374
33868 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
33869 build_zero_cst, remove fold_convertible_p check again.
33871 2017-04-11 Martin Liska <mliska@suse.cz>
33874 * ubsan.c (instrument_object_size): Do not instrument register
33877 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33880 * config/i386/i386-builtin-types.def
33881 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
33882 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
33883 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
33884 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
33885 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
33886 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
33887 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
33888 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
33889 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
33890 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
33891 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
33892 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
33893 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
33894 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
33895 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
33896 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
33897 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
33898 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
33899 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
33900 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
33901 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
33902 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
33903 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
33904 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
33905 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
33906 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
33907 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
33908 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
33909 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
33910 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
33911 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
33912 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
33913 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
33914 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
33915 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
33916 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
33917 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
33918 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
33919 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
33920 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
33921 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
33922 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
33923 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
33924 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
33925 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
33926 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
33928 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
33929 flag to second_arg_count, handle 4 argument function type _COUNT
33930 aliases, handle second_arg_count on second argument rather than last.
33932 2017-04-10 Jeff Law <law@redhat.com>
33934 PR tree-optimization/80374
33935 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
33936 record anything if we can not convert integer_zero_node to the
33939 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33942 * config/rs6000/rs6000.c (rs6000_option_override_internal):
33943 Enhance special handling given to the TARGET_P9_MINMAX option in
33944 relation to certain other options.
33946 2017-04-10 Bin Cheng <bin.cheng@arm.com>
33948 PR tree-optimization/80153
33949 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
33950 remove POINTER_PLUS_EXPR's base part directly, rather than through
33953 2017-04-10 Richard Biener <rguenther@suse.de>
33954 Bin Cheng <bin.cheng@arm.com>
33956 PR tree-optimization/80153
33957 * tree-affine.c (aff_combination_to_tree): Get base pointer from
33958 the first element of pointer type aff_tree. Build result expr in
33960 (add_elt_to_tree): Convert to type unconditionally. Remove other
33961 fold_convert calls.
33962 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
33963 (rewrite_use_nonlinear_expr): Check invariant using iv information.
33965 2017-04-10 Richard Biener <rguenther@suse.de>
33967 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33970 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
33972 PR rtl-optimization/70478
33973 * lra-constraints.c (curr_small_class_check): New.
33974 (update_and_check_small_class_inputs): New.
33975 (process_alt_operands): Update curr_small_class_check. Disfavor
33976 alternative insn memory operands. Check available regs for small
33979 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33982 * config/mips/mips.opt (-mvirt): Update description.
33983 * doc/invoke.texi (-mvirt): Likewise.
33985 2017-04-10 Richard Biener <rguenther@suse.de>
33987 PR middle-end/80362
33988 * fold-const.c (fold_binary_loc): Look at unstripped ops when
33989 looking for NEGATE_EXPR in -A / -B to A / B folding.
33991 2017-04-10 Martin Liska <mliska@suse.cz>
33993 PR gcov-profile/80224
33994 * gcov.c (print_usage): Fix usage string.
33995 (get_gcov_intermediate_filename): Remove.
33996 (output_gcov_file): Use both for normal and intermediate format.
33997 (generate_results): Do not initialize special file for
33998 intermediate format.
34000 2017-04-10 Richard Biener <rguenther@suse.de>
34002 PR tree-optimization/80304
34003 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34006 2017-04-10 Nathan Sidwell <nathan@acm.org>
34009 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34010 (rs6000_init_builtins): Use it.
34012 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34014 * config/arm/arm.md (<mrc>): Add mode to SET source.
34015 (<mrrc>): Likewise.
34017 2017-04-10 Richard Biener <rguenther@suse.de>
34019 PR middle-end/80344
34020 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34022 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34025 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34026 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34027 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34028 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34029 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34030 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34031 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34032 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34033 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34034 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34035 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34036 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34037 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34038 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34039 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34040 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34041 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34042 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
34043 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
34044 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
34045 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34046 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34047 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34049 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34051 PR rtl-optimization/70478
34052 * lra-constraints.c: Reverse the last patch.
34054 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34056 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34057 Add comment for WCHAR_T.
34059 2017-04-08 Martin Liska <mliska@suse.cz>
34062 2017-04-07 Martin Liska <mliska@suse.cz>
34065 * ipa-split.c (split_function): Add function part to a same comdat
34068 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34071 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34073 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34075 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34076 (vec_pairing): ...to this.
34077 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34078 (rs6000_sched_init): Adjust for name change.
34079 (struct rs6000_sched_context): Likewise.
34080 (rs6000_init_sched_context): Likewise.
34081 (rs6000_set_sched_context): Likewise.
34083 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34089 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34091 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34092 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34093 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34095 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34097 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34099 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34101 PR rtl-optimization/70703
34102 * ira-color.c (update_conflict_hard_regno_costs): Use
34103 int64_t instead of HOST_WIDE_INT.
34105 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34107 PR rtl-optimization/70478
34108 * lra-constraints.c (process_alt_operands): Disfavor alternative
34109 insn memory operands.
34111 2017-04-07 Jeff Law <law@redhat.com>
34113 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
34114 CALL and NOTE_INSN_CALL_ARG_LOCATION.
34116 2017-04-07 Martin Liska <mliska@suse.cz>
34119 * config/aarch64/aarch64.c (aarch64_process_target_attr):
34120 Show error message instead of an ICE.
34122 2017-04-07 Martin Liska <mliska@suse.cz>
34125 * ipa-split.c (split_function): Add function part to a same comdat
34128 2017-04-07 Richard Biener <rguenther@suse.de>
34130 PR middle-end/80341
34131 * tree.c (get_unwidened): Also handle ! for_type case for
34133 * convert.c (do_narrow): Split out from ...
34134 (convert_to_integer_1): ... here. Do not pass final truncation
34135 type to get_unwidened for TRUNC_DIV_EXPR.
34137 2017-04-07 Richard Biener <rguenther@suse.de>
34139 * tree-affine.c (wide_int_ext_for_comb): Take type rather
34141 (aff_combination_const): Adjust.
34142 (aff_combination_scale): Likewise.
34143 (aff_combination_add_elt): Likewise.
34144 (aff_combination_add_cst): Likewise.
34145 (aff_combination_convert): Likewise.
34146 (add_elt_to_tree): Likewise. Remove unused argument.
34147 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
34149 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
34151 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
34153 * config/arm/arm.c (arm_default_short_enums): Use
34154 ARM_DEFAULT_SHORT_ENUMS.
34155 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
34157 2017-04-06 Jakub Jelinek <jakub@redhat.com>
34160 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
34161 members with redundant out-of-class redeclaration.
34163 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34166 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
34167 * config/i386/i386.md (*zero_extendsidi2):
34168 Add (?*x,*x) and (?*v,*v) alternatives.
34170 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34173 * config/i386/i386.c (ix86_expand_builtin)
34174 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
34175 mode from insn data. Convert operands to insn operand mode.
34176 Copy operands that don't satisfy insn predicate to a register.
34178 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
34180 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
34183 2017-04-06 Richard Biener <rguenther@suse.de>
34185 PR tree-optimization/80334
34186 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
34187 preserve alignment of accesses.
34189 2017-04-06 Richard Biener <rguenther@suse.de>
34191 PR tree-optimization/80262
34192 * tree-sra.c (build_ref_for_offset): Preserve address-space
34194 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
34195 Drop useless address-space information on MEM_REF offsets.
34197 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
34199 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
34201 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34203 PR rtl-optimization/70703
34204 * ira-color.c (update_conflict_hard_regno_costs): Use
34205 HOST_WIDE_INT instead of long.
34207 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
34210 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
34211 not defined for x86_64 target. Add -mmmx target option when __SSE2__
34213 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
34214 for x86_64 target. Handle -m3dnowa option.
34216 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34218 PR rtl-optimization/70703
34219 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
34220 (update_conflict_hard_regno_costs): Use long instead of unsigned
34221 arithmetic for cost calculation.
34223 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34224 Bernd Edlinger <bernd.edlinger@hotmail.de>
34227 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
34230 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
34233 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
34234 ptr_mode with Pmode throughout.
34235 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
34236 into probe_stack_range and use DImode.
34238 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
34241 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
34242 call_eh_return is true.
34244 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34246 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
34247 Initialize last_match_fntype_index.
34249 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34252 * tree-nvr.c: Include internal-fn.h.
34253 (pass_return_slot::execute): Ignore internal calls without
34256 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34257 Richard Biener <rguenther@suse.de>
34260 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
34261 captures used multiple times, except for the last use.
34262 * generic-match-head.c: Include gimplify.h.
34264 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34266 PR tree-optimization/79390
34267 * target.h (struct noce_if_info): Declare.
34268 * targhooks.h (default_noce_conversion_profitable_p): Declare.
34269 * target.def (noce_conversion_profitable_p): New target hook.
34270 * ifcvt.h (struct noce_if_info): New type, moved from ...
34271 * ifcvt.c (struct noce_if_info): ... here.
34272 (noce_conversion_profitable_p): Renamed to ...
34273 (default_noce_conversion_profitable_p): ... this. No longer
34275 (noce_try_store_flag_constants, noce_try_addcc,
34276 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
34277 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
34278 instead of noce_conversion_profitable_p.
34279 * config/i386/i386.c: Include ifcvt.h.
34280 (ix86_option_override_internal): Don't override
34281 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
34282 (ix86_noce_conversion_profitable_p): New function.
34283 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
34284 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
34285 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
34286 * doc/tm.texi: Regenerated.
34288 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34290 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
34293 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
34296 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
34297 instructions for small multiply cores.
34299 2017-04-04 Jeff Law <law@redhat.com>
34301 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
34303 (mips_expand_vec_perm_const): Initialize elements in orig_perm
34304 that are not set by the loop over the elements.
34306 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34309 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
34310 int mode, convert_modes it to mode as unsigned, otherwise use
34311 lowpart_subreg to mode rather than SImode.
34312 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
34313 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
34314 Use DImode instead of SImode for the shift count operand.
34315 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
34318 2017-04-04 Richard Biener <rguenther@suse.de>
34320 PR middle-end/80281
34321 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
34322 arithmetic done for the negate or the plus. Simplify.
34323 (A - (-B) -> A + B): Likewise.
34324 * fold-const.c (split_tree): Make sure to not negate pointers.
34326 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
34328 PR rtl-optimization/60818
34329 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
34330 a compare of comparisons with the thing compared if this results
34331 in a different machine mode.
34333 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
34335 * alias.c (base_alias_check): Fix typo in comment.
34336 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
34337 * cgraphunit.c (symbol_table::compile): Likewise.
34338 * collect2.c (maybe_run_lto_and_relink): Likewise.
34339 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
34340 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
34341 * config/avr/avr.c (avr_map_op_t): Likewise.
34342 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
34343 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
34344 * config/epiphany/epiphany.md (movcc): Likewise.
34345 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
34346 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
34348 * config/mips/mips.c (mips_save_restore_reg): Likewise.
34349 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
34350 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
34351 * config/sh/sh.c (sh_rtx_costs): Likewise.
34352 * fold-const.c (fold_truth_andor): Likewise.
34353 * genautomata.c (collapse_flag): Likewise.
34354 * gengtype.h (struct type::u::s): Likewise.
34355 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
34356 * input.c (FORMAT_AMOUNT): Likewise.
34357 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
34358 (known_aggs_to_agg_replacement_list): Likewise.
34359 * ipa-inline-analysis.c: Likewise.
34360 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
34361 * ipa-polymorphic-call.c
34362 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
34363 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
34364 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
34366 * modulo-sched.c (apply_reg_moves): Likewise.
34367 * omp-expand.c (build_omp_regions_1): Likewise.
34368 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
34369 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
34370 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
34371 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
34372 * value-prof.c: Likewise.
34373 * var-tracking.c (val_reset): Likewise.
34375 2017-04-03 Richard Biener <rguenther@suse.de>
34377 PR tree-optimization/80275
34378 * fold-const.c (split_address_to_core_and_offset): Handle
34381 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
34383 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
34384 descriptors is at least equal to that of functions.
34386 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34388 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
34390 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34393 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
34394 (mov<IMOD4:mode>): New expander.
34395 (*mov<IMOD4:mode>_internal): New insn and split pattern.
34397 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
34399 PR rtl-optimization/79405
34400 * fwprop.c (propagations_left): New variable.
34401 (forward_propagate_into): Decrement it.
34402 (fwprop_init): Initialize it.
34403 (fw_prop): If the variable has reached zero, stop propagating.
34404 (fwprop_addr): Ditto.
34406 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34409 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
34410 a FUNCTION_DECL, pass it as decl instead of origin to
34413 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
34415 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
34418 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
34421 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
34422 TARGET_VSX_SMALL_INTEGER.
34424 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34426 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
34427 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
34429 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34431 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
34432 extraction from odd-numbered MSA register.
34434 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34436 PR middle-end/80173
34437 * expmed.c (store_bit_field_1): Don't attempt to create
34438 a word subreg out of hard registers wider than word if they
34439 have HARD_REGNO_NREGS of 1 for their mode.
34441 PR middle-end/80163
34442 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
34443 conversions to integer types wider than word and pointer.
34446 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
34447 (rtx_equal_for_cselib_p): Pass 0 to it.
34448 * cselib.c (cselib_hasher::equal): Likewise.
34449 (rtx_equal_for_cselib_1): Add depth argument. If depth
34450 is 128, don't look up VALUE locs and punt. Increment
34451 depth in recursive calls when walking VALUE locs.
34453 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
34455 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
34456 (make_gcov_file_name): Use the canonical path name for generating
34458 (read_line): Fix handling of files with ascii null bytes.
34460 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
34462 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
34463 to initialise a vector register instead
34464 of using a const_int.
34466 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34468 PR translation/80189
34469 * gimplify.c (omp_default_clause): Use %qs instead of %s in
34470 diagnostic messages.
34472 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
34475 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
34476 (dfp_diex_<mode>): Update mode of operand 1.
34477 * doc/extend.texi (dxex, dxexq): Document change to return type.
34478 (diex, diexq): Document change to argument type.
34480 2017-03-30 Martin Jambor <mjambor@suse.cz>
34483 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
34484 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
34485 it reflects the signature changes performed at the callee side.
34486 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
34487 to cgraph_build_function_type_skip_args.
34488 (build_function_decl_skip_args): Adjust call to the above function.
34490 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34493 * config/i386/sse.md
34494 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
34495 register as dest whenever it is a MEM not rtx_equal_p to the
34496 corresponding dup operand, and when forcing into reg move the
34497 reg into the memory afterwards.
34498 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
34499 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
34500 for the force_reg mode.
34501 (avx512vl_vextractf128<mode>): Use register as dest either
34502 always when a MEM, or when it is a MEM not rtx_equal_p to the
34503 corresponding dup operand, or even not when it is a CONST_VECTOR
34504 depending on the mode and lo vs. hi.
34505 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
34507 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
34508 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
34509 Likewise. Require that operands[2] is even.
34510 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
34511 Remove extraneous parens. Require that operands[2] is a multiple
34513 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
34514 operands[0] is a MEM if <mask_applied>, the predicates/constraints
34515 disallow memory then.
34517 2017-03-30 Richard Biener <rguenther@suse.de>
34519 PR tree-optimization/77498
34520 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
34521 to non-constants over backedges.
34523 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
34525 PR rtl-optimization/80233
34526 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
34527 as last_combined_insn. Do not test for BARRIER_P separately.
34529 2017-03-29 Andreas Schwab <schwab@suse.de>
34532 * calls.c (prepare_call_address): Convert funexp to Pmode before
34533 copying to temp reg.
34535 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34537 PR tree-optimization/80158
34538 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
34539 Handle possible future case of more than one alternate
34541 (replace_rhs_if_not_dup): Likewise.
34542 (replace_one_candidate): Likewise.
34544 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
34546 PR rtl-optimization/80193
34547 * ira.c (ira): Do not check allocation for LRA.
34549 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
34551 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
34552 (nvptx_output_simt_exit): Declare.
34553 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
34554 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
34555 (init_softstack_frame): Move initialization of crtl->is_leaf to...
34556 (nvptx_declare_function_name): ...here. Emit declaration of local
34557 memory space buffer for omp_simt_enter insn.
34558 (nvptx_output_unisimt_switch): New.
34559 (nvptx_output_softstack_switch): New.
34560 (nvptx_output_simt_enter): New.
34561 (nvptx_output_simt_exit): New.
34562 * config/nvptx/nvptx.h (struct machine_function): New fields
34563 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
34564 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
34565 (UNSPECV_SIMT_EXIT): Ditto.
34566 (omp_simt_enter_insn): New insn.
34567 (omp_simt_enter): New expansion.
34568 (omp_simt_exit): New insn.
34569 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
34571 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
34572 (expand_GOMP_SIMT_ENTER_ALLOC): New.
34573 (expand_GOMP_SIMT_EXIT): New.
34574 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
34575 (GOMP_SIMT_ENTER_ALLOC): Ditto.
34576 (GOMP_SIMT_EXIT): Ditto.
34577 * target-insns.def (omp_simt_enter): New insn.
34578 (omp_simt_exit): Ditto.
34579 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
34581 (lower_rec_simd_input_clauses): Implement SIMT privatization.
34582 (lower_rec_input_clauses): Likewise.
34583 (lower_lastprivate_clauses): Handle SIMT privatization.
34585 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
34586 (ompdevlow_adjust_simt_enter): New.
34587 (find_simtpriv_var_op): New.
34588 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
34589 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
34591 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
34592 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
34593 (copy_decl_for_dup_finish): Ditto.
34595 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
34597 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
34600 * config/i386/i386.c (ix86_option_override_internal): Always
34601 allow -mpreferred-stack-boundary=3 for 64-bit targets.
34603 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34605 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
34607 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34609 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
34610 mark new edge's irreducible flag accordign to it.
34611 (vect_do_peeling): Check loop preheader edge's irreducible flag
34612 and pass it to function slpeel_add_loop_guard.
34614 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
34616 PR tree-optimization/80218
34617 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34618 Update block frequencies and counts.
34620 2017-03-28 Richard Biener <rguenther@suse.de>
34622 PR tree-optimization/78644
34623 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34624 of a simplification result we may not use it at all.
34626 2017-03-28 Richard Biener <rguenther@suse.de>
34629 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34630 without arguments, generate default definition of a SSA name.
34632 2017-03-28 Richard Biener <rguenther@suse.de>
34634 PR middle-end/80222
34635 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34636 TYPE_REF_CAN_ALIAS_ALL references.
34637 * fold-const.c (fold_indirect_ref_1): Likewise.
34639 2017-03-28 Martin Liska <mliska@suse.cz>
34642 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34643 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34645 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34646 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34648 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34649 (EXTRA_SPECS): Define.
34650 (SUBTARGET_EXTRA_SPECS): Likewise.
34651 (SUBTARGET_CPP_SPEC): Likewise.
34652 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34653 SUBTARGET_EXTRA_SPECS.
34654 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34656 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34658 * config/arc/simdext.md (vst64_insn): Update pattern.
34659 (vld32wh_insn): Likewise.
34660 (vld32wl_insn): Likewise.
34661 (vld64_insn): Likewise.
34662 (vld32_insn): Likewise.
34664 2017-03-28 Marek Polacek <polacek@redhat.com>
34667 * fold-const.c (fold_comparison): Use protected_set_expr_location
34668 instead of SET_EXPR_LOCATION.
34670 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34672 * tree.c (add_expr): Avoid name lookup warning.
34674 2017-03-27 Jeff Law <law@redhat.com>
34676 PR tree-optimization/80216
34677 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34678 function name. Limit recursion depth.
34679 (record_temporary_equivalences): Corresponding changes.
34681 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34683 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34686 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34689 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34691 * cfgcleanup.c (reg_note_cfa_p): New array.
34692 (insns_have_identical_cfa_notes): New function.
34693 (old_insns_match_p): Don't cross-jump in between /f
34694 and non-/f instructions. If both i1 and i2 are frame related,
34695 verify all CFA notes, their order and content.
34697 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34700 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34701 HImode and SImode with zero extend to DImode to one insn.
34702 (bswap<mode>2_extenddi): Likewise.
34703 (bswapsi2_extenddi): Likewise.
34704 (bswaphi2_extendsi): Likewise.
34705 (bswaphi2): Combine bswap HImode and SImode into one insn.
34706 Separate memory insns from swapping register.
34707 (bswapsi2): Likewise.
34708 (bswap<mode>2): Likewise.
34709 (bswaphi2_internal): Delete, no longer used.
34710 (bswapsi2_internal): Likewise.
34711 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34712 store, and gpr<-gpr swap insns.
34713 (bswap<mode>2_store): Likewise.
34714 (bswaphi2_reg): Register only splitter, combine with the splitter.
34715 (bswaphi2 splitter): Likewise.
34716 (bswapsi2_reg): Likewise.
34717 (bswapsi2 splitter): Likewise.
34718 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34719 the insns into load, store, and register/register insns.
34720 (bswapdi2_ldbrx): Likewise.
34721 (bswapdi2_load): Likewise.
34722 (bswapdi2_store): Likewise.
34723 (bswapdi2_reg): Likewise.
34725 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34727 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34728 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34730 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34733 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34735 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34736 special handling for target option conflicts between dform
34737 options (-mpower9-dform, -mpower9-dform-vector,
34738 -mpower9-dform-scalar) and -mno-direct-move.
34740 2017-03-27 Richard Biener <rguenther@suse.de>
34742 PR tree-optimization/80181
34743 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34745 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34747 * config/arc/predicates.md (move_double_src_operand): Replace the
34748 call to move_double_src_operand with a call to address_operand.
34750 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34752 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34753 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34754 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34756 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34758 * config/arc/predicates.md (long_immediate_loadstore_operand):
34759 Consider scaled addresses cases.
34761 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34763 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34764 restored when in interrupt.
34765 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34766 doesn't have delay slot.
34768 2017-03-27 Richard Biener <rguenther@suse.de>
34771 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34772 inlined thunk clones.
34774 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34777 * asan.c (instrument_derefs): Copy over last operand from
34778 original COMPONENT_REF to the new COMPONENT_REF with
34779 DECL_BIT_FIELD_REPRESENTATIVE.
34780 * ubsan.c (instrument_object_size): Likewise.
34782 2017-03-27 Richard Biener <rguenther@suse.de>
34784 PR tree-optimization/80170
34785 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34786 sure DR/SCEV didnt fold in constants we do not see when looking
34787 at the reference base alignment.
34789 2017-03-27 Richard Biener <rguenther@suse.de>
34791 PR middle-end/80171
34792 * gimple-fold.c (fold_ctor_reference): Properly guard against
34793 NULL return value from canonicalize_constructor_val.
34795 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34798 * config/i386/i386.c (ix86_expand_builtin)
34799 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34800 flags reg setting and flags reg using instructions.
34801 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34802 clobbering instructions to zero extend op2.
34804 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34806 * doc/install.texi (Configuration) <--with-aix-soname>:
34807 Update link to AIX ld.
34809 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34811 PR rtl-optimization/80160
34812 PR rtl-optimization/80159
34813 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34814 reg_alternate_class into account.
34816 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34819 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34820 to consider in curr_insn_transform.
34822 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34824 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34825 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34826 and emit_mode_inner.
34828 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34830 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34831 argument to the overloaded builtin variants. Use the new flag to
34832 deprecate certain builtin variants.
34833 * config/s390/s390-builtin-types.def: Add new builtin types.
34834 * config/s390/s390-builtins.h: Support new flags field for
34835 overloaded builtins.
34836 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34837 (s390_macro_to_expand): Enable vector float data type.
34838 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34839 builtins by incrementing the __VEC__ version number.
34840 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34842 (s390_resolve_overloaded_builtin): Emit error messages depending
34843 on the builtin flags.
34844 * config/s390/s390.c (s390_expand_builtin): Support additional
34845 flags argument. Change error message to match the messages
34846 emitted in s390-c.c.
34847 * config/s390/s390.md: New UNSPEC_* constants.
34848 (op_type): Add new instruction types.
34849 * config/s390/vecintrin.h: Add new builtins and test data class
34851 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34852 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34853 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34854 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34855 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34856 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34858 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34859 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34860 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34861 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34863 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34864 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34865 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34866 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
34868 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
34869 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
34870 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
34871 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
34872 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
34873 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
34874 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
34876 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
34877 ("vec_scatter_element<V_HW_4:mode>_DI")
34878 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
34879 ("vec_fpint<mode>", "vflls")
34880 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
34881 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
34882 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
34883 ("*vec_cmphe<mode>_cc"): ... these.
34885 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
34886 mode constant instead of magic value.
34888 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34890 * config/s390/s390.c (s390_expand_vec_compare): Support other
34891 vector floating point modes than just V2DF.
34892 (s390_expand_vcond): Likewise.
34893 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
34894 (s390_cannot_change_mode_class): Prevent mode changes between TF
34895 and V1TF in vector registers.
34896 * config/s390/s390.md (DF, SF): New mode attributes.
34897 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
34898 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
34899 SFmode support for VRs.
34900 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
34902 (VFT, VF_HW): New mode iterators.
34903 (vw, sdx): New mode attributes.
34904 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
34905 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
34906 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
34907 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
34908 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
34909 also the new vector floating point modes. Renaming to ...
34911 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
34912 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
34913 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
34914 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
34915 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
34916 ("vec_unordered<mode>"): ... these.
34918 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
34919 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
34920 ("*vec_extendv2df"): New insn definitions.
34922 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34924 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
34925 ("mulditi3_2", "*muldi3_sign"): New patterns.
34926 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
34927 rename the pattern definition.
34929 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34931 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
34933 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
34935 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34937 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
34938 instruction if possible.
34939 * config/s390/vector.md (vec_halfnumelts): New mode
34941 ("*vec_vllezlf<mode>"): New pattern.
34943 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34945 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
34946 ("popcountv4si2", "popcountv2di2"): Rename to ...
34947 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
34948 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
34950 ("popcount<mode>2_vxe"): New pattern.
34952 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34954 * common/config/s390/s390-common.c (processor_flags_table): Add
34956 * config.gcc: Add arch12.
34957 * config/s390/driver-native.c (s390_host_detect_local_cpu):
34958 Default to arch12 for unknown CPU model numbers.
34959 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
34960 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
34961 PROCESSOR_max sanity check.
34962 * config/s390/s390-opts.h (enum processor_type): Add
34964 * config/s390/s390.c (processor_table): Add arch12.
34965 (s390_expand_builtin): Add check for B_VXE flag.
34966 (s390_issue_rate): Add PROCESSOR_ARCH12.
34967 (s390_get_sched_attrmask): Likewise.
34968 (s390_get_unit_mask): Likewise.
34969 (s390_sched_score): Enable z13 scheduling for arch12.
34970 (s390_sched_reorder): Likewise.
34971 (s390_sched_variable_issue): Likewise.
34972 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
34974 (s390_tune_attr): Use z13 scheduling also for arch12.
34975 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
34976 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
34977 (TARGET_VXE_P): New macros.
34978 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
34979 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
34980 * config/s390/s390.opt: Add arch12 as processor_type.
34982 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34984 * config/s390/s390.md
34985 ("fixuns_truncdddi2", "fixuns_trunctddi2")
34986 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
34987 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
34989 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
34990 Rename expanders to ...
34992 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
34993 ("fixuns_truncdddi2_emu"): ... these.
34995 ("fixuns_trunc<mode>si2_emu"): New expander.
34997 ("*fixuns_truncdfdi2_z13"): Rename to ...
34998 ("*fixuns_truncdfdi2_vx"): ... this.
35000 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35002 * config/s390/2964.md: Remove the single element vector compare
35003 instructions which are no longer used.
35004 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35006 (s390_canonicalize_comparison): Remove handling of DFmode
35008 (s390_expand_vec_compare_scalar): Remove function.
35009 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35010 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35012 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35014 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35016 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35018 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35019 will anyway by matched by mov<mode>_64dfp.
35021 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35023 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35024 vlef/vstef. Add missing operand to vleif.
35026 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35028 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35029 pair for all vector types with 64 bit elements.
35030 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35031 * config/s390/vector.md (V_HW_64): ... here.
35032 (V_128_NOSINGLE): New mode iterator.
35033 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35034 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35035 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35036 ("*vec_load_pairv2di"): Change to ...
35037 ("*vec_load_pair<mode>"): ... this one.
35039 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35041 * config/s390/constraints.md: Add comments.
35042 (jKK): Reject element sizes > 8 bytes.
35043 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
35045 * config/s390/s390.md: Add the s_operand checks formerly in
35046 s390_split_ok_p to various splitters where they are still
35048 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35049 for 128 bit vectors. Plus two splitters.
35051 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35053 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35056 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35059 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35060 error if the boundary argument is not constant.
35062 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35064 PR rtl-optimization/80112
35065 * loop-doloop.c (doloop_condition_get): Don't check condition
35066 if cmp isn't SET with IF_THEN_ELSE src.
35068 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35070 PR tree-optimization/80158
35071 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35072 replacing a candidate statement, also replace it for the
35073 candidate's alternate interpretation.
35074 (replace_rhs_if_not_dup): Likewise.
35075 (replace_one_candidate): Likewise.
35077 2017-03-24 Richard Biener <rguenther@suse.de>
35079 PR tree-optimization/80167
35080 * graphite-isl-ast-to-gimple.c
35081 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35083 (translate_isl_ast_to_gimple::get_rename): Likewise.
35085 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35087 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35088 handling of certain combinations of target options, including the
35089 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35090 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35092 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35095 * config/arm/arm.md (*load_multiple): Add reload_completed to
35096 matching condition.
35098 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35099 Richard Biener <rguenth@suse.de>
35101 PR tree-optimization/79908
35102 PR tree-optimization/80136
35103 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35104 been cast away, gimplify_and_add suffices.
35106 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35108 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35110 2017-03-23 Richard Biener <rguenther@suse.de>
35112 PR tree-optimization/80032
35113 * gimplify.c (gimple_push_cleanup): Forced unconditional
35114 cleanups still have to go to the conditional_cleanups
35117 2017-03-22 Jakub Jelinek <jakub@redhat.com>
35119 PR tree-optimization/80072
35120 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
35122 (next_operand_entry_id): Change type to unsigned int.
35123 (sort_by_operand_rank): Make sure to return the right return value
35124 even if unsigned fields are bigger than INT_MAX.
35125 (struct oecount): Change cnt and id type to unsigned int.
35126 (oecount_hasher::equal): Formatting fix.
35127 (oecount_cmp): Make sure to return the right return value
35128 even if unsigned fields are bigger than INT_MAX.
35129 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
35132 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
35133 TREE_READONLY on result if writing it more than once.
35136 * doc/invoke.texi (-fsanitize=thread): Document that with
35137 -fnon-call-exceptions atomics are not able to throw
35141 * tsan.c: Include tree-eh.h.
35142 (instrument_builtin_call): Call maybe_clean_eh_stmt or
35143 maybe_clean_or_replace_eh_stmt where needed.
35144 (instrument_memory_accesses): Add cfg_changed argument.
35145 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
35146 if it returned true.
35147 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
35149 PR rtl-optimization/63191
35150 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
35151 wrapper function, moved the whole old content into ...
35152 (ix86_delegitimize_address_1): ... this. New inline function.
35153 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
35154 true as last argument instead of ix86_delegitimize_address.
35156 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35158 * config/aarch64/aarch64.c (generic_branch_cost): Copy
35159 cortexa57_branch_cost.
35161 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35163 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
35165 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35168 * doc/md.texi (Constraints): Document wA constraint.
35169 * config/rs6000/constraints.md (wA): New.
35170 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
35171 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
35172 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
35173 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
35175 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
35178 * gimplify.c (is_oacc_declared): New function.
35179 (oacc_default_clause): Use it to set default flags for acc declared
35180 variables inside parallel regions.
35181 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
35182 declared variables.
35183 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
35184 declare attribute to any decl as necessary.
35186 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
35189 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
35190 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
35191 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
35192 (arm_arch_lpae): This.
35193 * config/arm/arm.c (arm_arch7ve): Rename into ...
35194 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
35195 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
35198 2017-03-22 Martin Liska <mliska@suse.cz>
35201 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
35202 error message instead of an ICE.
35204 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35206 * doc/extend.texi (6.11 Additional Floating Types): Revise.
35208 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
35210 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35212 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35215 2017-03-21 Martin Sebor <msebor@redhat.com>
35217 * doc/extend.texi: Use "cannot" instead of "can't."
35218 * doc/hostconfig.texi: Same.
35219 * doc/install.texi: Same.
35220 * doc/invoke.texi: Same.
35221 * doc/loop.texi: Same.
35222 * doc/md.texi: Same.
35223 * doc/objc.texi: Same.
35224 * doc/rtl.texi: Same.
35225 * doc/tm.texi: Same.
35226 * doc/tm.texi.in: Same.
35227 * doc/trouble.texi: Same.
35229 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
35232 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
35233 (collect_checksum_attributes): Set it.
35234 (die_checksum_ordered): Use it.
35236 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35238 PR tree-optimization/79908
35239 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
35240 change: For a VA_ARG whose LHS has been cast away, use
35241 force_gimple_operand to construct the side effects.
35243 2017-03-21 David Malcolm <dmalcolm@redhat.com>
35245 PR translation/80001
35246 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
35247 more amenable to translation.
35248 (oacc_loop_auto_partitions): Likewise.
35250 2017-03-21 Marek Polacek <polacek@redhat.com>
35251 Martin Sebor <msebor@redhat.com>
35253 PR tree-optimization/80109
35254 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
35255 on INTEGRAL_TYPE_P.
35257 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35258 Segher Boessenkool <segher@kernel.crashing.org>
35261 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
35262 check reg_used_between_p between insn and one of succ or succ2
35263 depending on if succ is artificial insn not inserted into insn
35266 2017-03-21 Martin Liska <mliska@suse.cz>
35268 PR gcov-profile/80081
35269 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
35270 * doc/gcc.texi: Include gcov-dump stuff.
35271 * doc/gcov-dump.texi: New file.
35273 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
35275 PR rtl-optimization/79150
35276 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
35277 conditional jump, if the jump is the last insn of the loop.
35279 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35280 Richard Biener <rguenth@suse.de>
35282 PR tree-optimization/79908
35283 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35284 been cast away, use force_gimple_operand to construct the side
35287 2017-03-21 Martin Liska <mliska@suse.cz>
35289 PR libfortran/79956
35290 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
35293 2017-03-21 Brad Spengler <spender@grsecurity.net>
35296 * plugin.c (htab_hash_plugin): New function.
35297 (add_new_plugin): Use it and adjust.
35298 (parse_plugin_arg_opt): Adjust.
35299 (init_one_plugin): Likewise.
35301 2017-03-21 Richard Biener <rguenther@suse.de>
35303 PR tree-optimization/80032
35304 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
35305 if set force the cleanup to happen unconditionally.
35306 (gimplify_target_expr): Push inserted clobbers with force_uncond
35307 to avoid them being removed by control-dependent DCE.
35309 2017-03-21 Richard Biener <rguenther@suse.de>
35311 PR tree-optimization/80122
35312 * tree-inline.c (copy_bb): Do not expans va-arg packs or
35313 va_arg_pack_len when the inlined call stmt requires pack
35315 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
35317 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35320 * tsan.c (instrument_builtin_call): If the memory model argument
35321 is not a constant, assume it is valid.
35324 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
35327 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35329 PR rtl-optimization/79910
35330 * combine.c (can_combine_p): Do not allow combining an I0 or I1
35331 if its dest is used by an insn before I2 (other than the combined
35332 insns themselves, which are properly handled already).
35334 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35337 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35339 * combine.c (record_used_regs): New static function.
35340 (try_combine): Handle situations where there is an additional
35341 instruction between I2 and I3 which needs to have a LOG_LINK
35345 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35347 * combine.c (try_combine): Delete redundant i1 test. Call
35348 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35350 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35353 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
35354 alternatives 13/14.
35356 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35358 PR tree-optimization/80054
35359 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
35360 the optimization if a PHI or any of its arguments is not dominated
35361 by the candidate's basis. Use gphi* rather than gimple* as
35363 (replace_profitable_candidates): Clean up a gimple* variable that
35364 should be a gphi* variable.
35366 2017-03-20 Martin Sebor <msebor@redhat.com>
35369 * doc/extend.texi (attribute constructor): Document present limitation.
35371 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
35374 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
35375 __POWER9_VECTOR__ #ifdef control, change template definition to
35376 use Power9-specific built-in function.
35377 (vec_any_eq): Likewise.
35378 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
35379 to control outcomes from this test.
35380 (vector_ae_<mode>p): For VEC_F modes, likewise.
35382 2017-03-20 Ian Lance Taylor <iant@google.com>
35384 * config/i386/i386.c (ix86_function_regparm): Save an extra
35385 register for -fsplit-stack with DECL_STATIC_CHAIN.
35387 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35390 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
35391 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
35393 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35395 * config/riscv/riscv.c (riscv_print_operand): Use "fence
35397 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
35400 2017-03-20 Marek Polacek <polacek@redhat.com>
35403 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
35405 2017-03-20 Richard Biener <rguenther@suse.de>
35407 PR tree-optimization/80113
35408 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
35409 allocate extra SSA name for PHI def.
35410 (add_close_phis_to_outer_loops): Likewise.
35411 (add_close_phis_to_merge_points): Likewise.
35412 (copy_loop_close_phi_args): Likewise.
35413 (copy_cond_phi_nodes): Likewise.
35415 2017-03-20 Martin Liska <mliska@suse.cz>
35417 PR middle-end/79753
35418 * tree-chkp.c (chkp_build_returned_bound): Do not build
35419 returned bounds for a LHS that's not a BOUNDED_P type.
35421 2017-03-20 Martin Liska <mliska@suse.cz>
35425 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
35426 COMPLEX_CST and VECTOR_CST.
35428 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35431 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
35432 target operand. A new splitter adds the clobber statement in case
35433 the target operand is dead anyway.
35435 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
35437 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
35438 to age-old versions of binutils and glibc.
35440 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
35442 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
35444 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35446 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
35448 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35450 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
35451 requirement for binutils 2.13.
35453 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35455 * combine.c (try_combine): Delete redundant i1 test. Call
35456 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35458 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
35460 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
35461 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
35463 <riscv64-*-elf>: Re-arrange section
35464 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
35465 <riscv32-*-linux>: Likewise.
35466 <riscv64-*-elf>: Likewise
35467 <riscv64-*-linux>: Likewise.
35469 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
35472 * aarch64.opt(verbose-cost-dump): Fix typo.
35474 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
35477 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
35478 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
35480 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35482 * reload.c (find_reloads): When reloading a nonoffsettable address,
35483 use RELOAD_OTHER for it and its address reloads.
35485 PR rtl-optimization/79910
35486 * combine.c (record_used_regs): New static function.
35487 (try_combine): Handle situations where there is an additional
35488 instruction between I2 and I3 which needs to have a LOG_LINK
35491 2017-03-17 Jeff Law <law@redhat.com>
35493 PR tree-optimization/71437
35494 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
35495 conditional in the hash table first.
35496 (vrp_dom_walker::before_dom_children): Extract condition from
35497 ASSERT_EXPR. Record condition, its inverion and any implied
35498 conditions as well.
35500 2017-03-17 Marek Polacek <polacek@redhat.com>
35501 Markus Trippelsdorf <markus@trippelsdorf.de>
35503 PR tree-optimization/80079
35504 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
35507 2017-03-17 Richard Biener <rguenther@suse.de>
35509 PR middle-end/80075
35510 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
35511 Properly verify the LHS before the RHS possibly claims to be
35513 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
35516 2017-03-17 Martin Jambor <mjambor@suse.cz>
35518 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
35519 (List of -O2 options): Likewise.
35520 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
35523 2017-03-17 Tom de Vries <tom@codesourcery.com>
35525 * gcov-dump.c (print_usage): Print bug_report_url.
35527 2017-03-17 Richard Biener <rguenther@suse.de>
35529 PR middle-end/80050
35530 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
35531 (parser::peek): Likewise.
35533 2017-03-17 Richard Biener <rguenther@suse.de>
35535 PR tree-optimization/80048
35536 * sese.c (free_sese_info): Properly release rename_map and
35537 copied_bb_map elements.
35539 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
35541 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
35542 Add linked-list forward and backlinks. Insert on
35543 construction, remove on destruction.
35544 (class pass_store_merging): Add m_stores_head field.
35545 (pass_store_merging::terminate_and_process_all_chains):
35546 Iterate over m_stores_head list.
35547 (pass_store_merging::terminate_all_aliasing_chains):
35549 (pass_store_merging::execute): Check for debug stmts first.
35550 Push new chains onto the m_stores_head stack.
35552 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
35555 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
35556 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
35557 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
35559 2017-03-16 Jeff Law <law@redhat.com>
35561 PR tree-optimization/71437
35562 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
35563 member function. Implementation moved into after_dom_children
35564 member function and into the threader's thread_outgoing_edges
35566 (dom_opt_dom_walker::after_dom_children): Simplify by moving
35567 some code into new thread_outgoing_edges.
35568 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
35569 definition. Simplify marker handling (do it here). Assume we always
35570 have the available expression and the const/copies tables.
35571 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
35573 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
35574 * tree-vrp.c (equiv_stack): No longer file scoped.
35575 (vrp_dom_walker): New class.
35576 (vrp_dom_walker::before_dom_children): New member function.
35577 (vrp_dom_walker::after_dom_children): Likewise.
35578 (identify_jump_threads): Setup domwalker. Use it rather than
35579 walking edges in a random order by hand. Simplify setup/finalization.
35580 (finalize_jump_threads): Remove.
35581 (vrp_finalize): Do not call identify_jump_threads here.
35582 (execute_vrp): Do it here instead and call thread_through_all_blocks
35585 PR tree-optimization/71437
35586 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
35588 (simplify_stmt_for_jump_threading): Add basic_block argument. All
35590 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
35591 (dom_opt_dom_walker::thread_across_edge): Remove
35592 handle_dominating_asserts argument. All callers changed.
35593 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
35594 changes. Remove calls to lhs_of_dominating_assert. Other
35595 uses of handle_dominating_asserts turn into unconditional code
35596 (simplify_control_stmt_condition_1): Likewise.
35597 (simplify_control_stmt_condition): Likewise.
35598 (thread_through_normal_block, thread_across_edge): Likewise.
35599 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
35600 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
35601 object if it is not an SSA_NAME.
35602 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
35603 before calling into the VRP specific simplifiers.
35604 (identify_jump_threads): Remove handle_dominating_asserts
35607 2017-03-16 Jakub Jelinek <jakub@redhat.com>
35610 * tree-diagnostic.c (default_tree_printer): No longer static.
35611 * tree-diagnostic.h (default_tree_printer): New prototype.
35613 2017-03-16 Tamar Christina <tamar.christina@arm.com>
35615 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
35616 Change ins into fmov.
35618 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35620 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
35621 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
35622 Use h_con constraint for operand 1.
35623 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35624 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35626 2017-03-15 Jeff Law <law@redhat.com>
35628 PR tree-optimization/71437
35629 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35630 (record_temporary_equivalences): Use it.
35632 PR tree-optimization/71437
35633 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35634 tree-ssa-scopedtables.
35635 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35636 (record_conditions, record_cond, vuse_eq): Likewise.
35637 (record_edge_info): Adjust to API tweak of record_conditions.
35638 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35639 (record_temporary_equivalences, optimize_stmt): Likewise.
35640 (eliminate_redundant_computations): Likewise.
35641 (record_equivalences_from_stmt): Likewise.
35642 * tree-ssa-scopedtables.c: Include options.h and params.h.
35643 (vuse_eq): New function, moved from tree-ssa-dom.c
35644 (build_and_record_new_cond): Likewise.
35645 (record_conditions): Likewise. Accept vector of conditions rather
35646 than edge_equivalence structure for first argument.
35647 for the first argument.
35648 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35649 from tree-ssa-dom.c.
35650 (avail_exprs_stack::record_cond): Likewise.
35651 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35652 from tree-ssa-dom.c.
35653 (avail_exprs_stack): Add new member functions lookup_avail_expr
35655 (record_conditions): Declare.
35657 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35660 * lra-constraints.c (process_alt_operands): Increase reject for
35661 reloading an input/output operand.
35663 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35666 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35667 insns to convert from signed/unsigned char/short to IEEE 128-bit
35669 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35671 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35674 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35675 subreg of inner mode for values already in registers.
35677 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35679 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35680 iteration reg is used after the loop.
35682 2017-03-14 Martin Sebor <msebor@redhat.com>
35684 PR tree-optimization/79800
35685 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35686 precision in negative-positive range.
35687 (format_floating): Call non-const overload with adjusted precision.
35689 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35692 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35695 2017-03-14 Martin Sebor <msebor@redhat.com>
35697 PR middle-end/80020
35698 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35699 * builtins.def (aligned_alloc): Use it.
35702 * Makefile.in (GTFILES): Add calls.c.
35703 * calls.c: Include "gt-calls.h".
35705 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35707 PR rtl-optimization/79728
35708 * regs.h (struct target_regs): New field
35709 x_contains_allocatable_regs_of_mode.
35710 (contains_allocatable_regs_of_mode): New macro.
35711 * reginfo.c (init_reg_sets_1): Initialize it, and change
35712 contains_reg_of_mode so it includes global regs as well.
35713 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35714 rather than contains_regs_of_mode.
35716 2017-03-14 Martin Liska <mliska@suse.cz>
35718 * doc/invoke.texi: Document options that can't be combined with
35719 -fcheck-pointer-bounds.
35721 2017-03-14 Martin Liska <mliska@suse.cz>
35723 PR middle-end/79831
35724 * doc/invoke.texi (-Wchkp): Document the option.
35726 2017-03-14 Martin Liska <mliska@suse.cz>
35728 * Makefile.in: Install gcov-dump.
35730 2017-03-14 Martin Liska <mliska@suse.cz>
35732 * multiple_target.c (expand_target_clones): Bail out for
35733 an invalid attribute.
35735 2017-03-14 Richard Biener <rguenther@suse.de>
35737 * alias.c (struct alias_set_entry): Pack properly.
35738 * cfgloop.h (struct loop): Likewise.
35739 * cse.c (struct set): Likewise.
35740 * ipa-utils.c (struct searchc_env): Likewise.
35741 * loop-invariant.c (struct invariant): Likewise.
35742 * lra-remat.c (struct cand): Likewise.
35743 * recog.c (struct change_t): Likewise.
35744 * rtl.h (struct address_info): Likewise.
35745 * symbol-summary.h (function_summary): Likewise.
35746 * tree-loop-distribution.c (struct partition): Likewise.
35747 * tree-object-size.c (struct object_size_info): Likewise.
35748 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35749 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35750 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35751 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35752 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35753 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35754 (struct _stmt_vec_info): Likewise.
35756 2017-03-14 Martin Liska <mliska@suse.cz>
35759 * multiple_target.c (create_dispatcher_calls): Check that
35760 a target can create a function dispatcher.
35762 2017-03-14 Martin Liska <mliska@suse.cz>
35765 * multiple_target.c (expand_target_clones): Drop local.local
35766 flag for default implementation.
35768 2017-03-14 Richard Biener <rguenther@suse.de>
35770 PR tree-optimization/80030
35771 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35773 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35775 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35776 gcc_fallthrough() instead of __attribute__((fallthrough));
35778 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35780 * doc/gcc.texi: Remove "up" link to (DIR).
35781 * doc/gccint.texi: Ditto.
35783 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35785 * doc/install.texi (Specific) <avr>: Remove reference to
35788 2017-03-13 Jeff Law <law@redhat.com>
35790 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35791 attribute rather than comments.
35793 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35794 match_scratch operand is highest.
35796 2017-03-13 Martin Liska <mliska@suse.cz>
35798 PR middle-end/78339
35799 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35800 is a CHKP clone, use original declaration.
35802 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35804 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35805 (arc_conditional_register_usage): Use a different allocation order
35806 when optimizing for size.
35807 * common/config/arc/arc-common.c (arc_option_optimization_table):
35808 Section anchors default on when optimizing for size.
35810 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35812 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35814 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35816 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35817 * config/arc/arc.md (cpu_facility): Add cd variant.
35818 (*movqi_insn): Add code density variant.
35819 (*movhi_insn): Likewise.
35820 (*movqi_insn): Likewise.
35821 (*addsi3_mixed): Likewise.
35822 (subsi3_insn): Likewise.
35824 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35826 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35828 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35830 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35831 expressions with MINUS and UNARY ops.
35833 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35836 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35838 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35839 between vec_select and vector argument.
35840 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35841 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35842 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35843 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35844 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35845 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35847 2017-03-13 Richard Biener <rguenther@suse.de>
35850 * params.def (vect-max-peeling-for-alignment): Fix typo.
35852 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35854 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35855 issue that only occurred with binutils below 2.18.
35857 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35859 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35860 refer to binutils 2.11/2.12 minimum.
35862 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35864 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35865 ftp.kernel.org and simplify binutils requirement.
35867 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
35869 * doc/invoke.texi (Warning Options): Fix spelling of link-time
35871 (Optimize Options): Ditto. Also remove redundancy.
35873 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35875 PR translation/79848
35876 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
35878 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
35879 to G_ to avoid double translation.
35881 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35883 PR translation/79923
35884 * auto-profile.c (get_combined_location): Convert leading
35885 character of diagnostics to lower case and remove trailing period.
35886 (read_profile): Likewise for various diagnostics.
35887 * config/arm/arm.c (arm_option_override): Remove trailing period
35888 from various diagnostics.
35889 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
35890 (msp430_expand_delay_cycles): Likewise.
35892 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35895 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
35896 full command-line argument, rather than just "str".
35897 (aarch64_validate_march): Likewise.
35898 (aarch64_validate_mtune): Likewise.
35900 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
35902 PR rtl-optimization/78911
35903 * lra-assigns.c (must_not_spill_p): New function.
35904 (spill_for): Use it.
35906 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35908 PR tree-optimization/79981
35909 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
35910 ATOMIC_COMPARE_EXCHANGE ifn result.
35911 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
35912 IFN_ATOMIC_COMPARE_EXCHANGE.
35914 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35917 * opts.c (parse_sanitizer_options): Add missing question mark to
35918 "did you mean" message.
35920 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35922 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
35924 (VMULEUH_UNS): Likewise.
35925 (VMULOUB_UNS): Likewise.
35926 (VMULOUH_UNS): Likewise.
35927 * config/rs6000/rs6000.c (builtin_function_type): Remove
35928 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
35930 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35933 * read-rtl-function.c (function_reader::read_rtx_operand): Update
35934 x with result of extra_parsing_for_operand_code_0.
35935 (function_reader::extra_parsing_for_operand_code_0): Convert
35936 return type from void to rtx, returning x. When reading
35937 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
35938 larger size containing struct block_symbol.
35940 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
35942 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
35943 -mfloat128-hardware without -m64.
35945 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
35948 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
35949 entries to the case statement that marks unsigned arguments to
35950 overloaded functions.
35952 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
35954 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
35955 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
35957 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
35960 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
35961 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
35963 2017-03-10 Martin Liska <mliska@suse.cz>
35967 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
35968 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
35971 2017-03-10 Olivier Hainque <hainque@adacore.com>
35973 * tree-switch-conversion (array_value_type): Start by resetting
35974 candidate type to it's main variant.
35976 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35978 PR rtl-optimization/79909
35979 * combine.c (try_combine): Use simplify_replace_rtx on individual
35980 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
35981 of the whole CALL_INSN_FUNCTION_USAGE.
35983 PR tree-optimization/79972
35984 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
35985 get_range_info on SSA_NAMEs. Formatting fixes.
35987 2017-03-10 Richard Biener <rguenther@suse.de>
35988 Jakub Jelinek <jakub@redhat.com>
35990 PR tree-optimization/77975
35991 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
35992 edge to be constant.
35993 (get_val_for): For constant x return it. Formatting fix.
35994 (loop_niter_by_eval): Avoid pointless looping if the next iteration
35995 would use the same bases as the current one.
35997 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35999 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
36000 instead of vec_select for V1TImode.
36001 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
36003 (VSX_LE_128): Add V1TI to this mode iterator.
36004 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36005 (*vsx_le_perm_store_<mode>): Likewise.
36006 (pre-reload splitter for VSX stores): Likewise.
36007 (post-reload splitter for VSX stores): Likewise.
36008 (*vsx_xxpermdi2_le_<mode>): Likewise.
36009 (*vsx_lxvd2x2_le_<mode>): Likewise.
36010 (*vsx_stxvd2x2_le_<mode>): Likewise.
36012 2017-03-09 Michael Eager <eager@eagercon.com>
36014 Correct failures with --enable-checking=yes,rtl.
36016 * config/microblaze/microblaze.c (microblaze_expand_shift):
36017 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36018 test for const0_rtx.
36019 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36020 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36022 2017-03-09 Richard Biener <rguenther@suse.de>
36024 PR tree-optimization/79977
36025 * graphite-scop-detection.c (scop_detection::merge_sese):
36026 Handle the case of extra exits to blocks dominating the entry.
36028 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36030 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36033 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36035 PR rtl-optimization/79949
36036 * lra-constraints.c (process_alt_operands): Check memory when
36037 trying to predict a cycle. Print about the overall increase.
36039 2017-03-09 Richard Biener <rguenther@suse.de>
36041 PR middle-end/79971
36042 * gimple-expr.c (useless_type_conversion_p): Preserve
36043 TYPE_SATURATING for fixed-point types.
36045 2017-03-09 Richard Biener <rguenther@suse.de>
36048 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36049 alignment of BLKmode params.
36051 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36054 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36055 (VALL_NO_V2Q): Likewise.
36057 * config/aarch64/aarch64-simd.md
36058 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36060 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36061 VALL_NO_V2Q mode iterator.
36062 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36064 2017-03-09 Martin Liska <mliska@suse.cz>
36066 PR tree-optimization/79631
36067 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36068 tree_int_cst_sign_bit just for INTEGER constants.
36070 2017-03-09 Martin Liska <mliska@suse.cz>
36074 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36077 2017-03-09 Marek Polacek <polacek@redhat.com>
36080 * tree.c (inchash::add_expr): Handle TREE_VEC.
36082 2017-03-09 Martin Liska <mliska@suse.cz>
36085 (chkp_narrow_size_and_offset): New function.
36086 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36087 (void chkp_parse_bit_field_ref): New function.
36088 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36089 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36091 2017-03-09 Martin Liska <mliska@suse.cz>
36094 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36095 (chkp_find_bounds_1): Remove gcc_unreachable.
36097 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36100 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36101 BUILT_IN_SYNC*, determine the access type from the size suffix and
36102 always build a MEM_REF with that type. Handle forgotten
36103 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36106 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36107 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36108 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36109 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36110 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
36111 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
36112 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
36113 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
36114 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
36115 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
36116 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
36117 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
36118 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
36119 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
36120 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
36121 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
36122 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
36123 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
36124 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
36125 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
36126 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
36127 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
36128 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
36129 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
36130 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
36131 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
36132 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
36133 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
36134 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
36135 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
36136 definitions outside of __OPTIMIZE__ guarded section.
36139 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
36140 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
36141 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
36142 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
36145 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36147 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
36148 ("vfenez<mode>"): Add missing constraints.
36150 2017-03-08 Martin Sebor <msebor@redhat.com>
36153 * config/nds32/nds32.c (nds32_option_override):
36154 Fix misspelled diagnostic.
36156 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36159 * gimplify.c (gimplify_omp_for): Replace index var in outer
36160 taskloop statement with an artificial variable and add
36161 OMP_CLAUSE_PRIVATE clause for it.
36163 2017-03-08 Richard Biener <rguenther@suse.de>
36165 PR tree-optimization/79955
36166 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
36167 for accesses that are completely outside of the variable.
36169 2017-03-08 Andrew Haley <aph@redhat.com>
36171 PR tree-optimization/79943
36172 * tree-ssa-loop-split.c (compute_new_first_bound): When
36173 calculating the new upper bound, (END-BEG) should be added, not
36176 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36178 * config/avr/avr.md (setmemhi): Make sure match_dup
36179 operand number comes before match_scratch.
36181 2017-03-08 Richard Biener <rguenther@suse.de>
36183 PR tree-optimization/79920
36184 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
36185 with ncopies == 1 to ...
36186 (vect_transform_slp_perm_load): ... here. Properly compute
36187 all element loads by iterating VF times over the group. Do
36188 not handle ncopies (computed in a broken way) in
36189 vect_create_mask_and_perm.
36191 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36194 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
36195 is a uniform vector, use uniform_vector_p return value instead of
36196 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
36198 2017-03-07 Marek Polacek <polacek@redhat.com>
36200 PR middle-end/79809
36201 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
36202 (alloca_call_type): Likewise.
36204 2017-03-07 Martin Liska <mliska@suse.cz>
36206 * gcov.c (process_args): Put comment to correct location.
36208 2017-03-07 Martin Liska <mliska@suse.cz>
36210 PR middle-end/68270
36211 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
36212 Use array_at_struct_end_p instead of DECL_CHAIN (field).
36213 (chkp_narrow_bounds_for_field): Likewise.
36214 (chkp_parse_array_and_component_ref): Pass one more argument to
36217 2017-03-07 Richard Biener <rguenther@suse.de>
36219 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
36222 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
36224 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
36225 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
36227 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36230 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
36231 to end of description.
36232 (PARAM_MAX_STORES_TO_MERGE): Likewise.
36234 2017-03-07 Jakub Jelinek <jakub@redhat.com>
36236 PR rtl-optimization/79901
36237 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
36239 (*avx512f_<code><mode>3<mask_name>): ... this.
36240 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
36241 iterator instead of VI8_AVX2_AVX512BW.
36243 PR rtl-optimization/79901
36244 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
36245 min/max expander, expand it using expand_vec_cond_expr.
36248 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
36251 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36254 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
36255 to void * for PCH reasons.
36256 * dwarf2out.c (output_loc_operands, output_die): Cast
36257 v.val_vec.array to unsigned char *.
36259 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
36262 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
36265 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
36267 PR rtl-optimization/79571
36268 * lra-constraints.c (process_alt_operands): Calculate static
36269 reject and subtract it from overall when only addresses will be
36272 2017-03-06 Julia Koval <julia.koval@intel.com>
36275 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
36276 incoming stack boundary to 128 for 64-bit targets.
36278 2017-03-06 Richard Biener <rguenther@suse.de>
36280 PR tree-optimization/79894
36281 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
36282 to NULL after folding it.
36284 2017-03-06 Richard Biener <rguenther@suse.de>
36286 PR tree-optimization/79824
36287 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
36288 check disabling peeling for gaps.
36290 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
36292 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
36293 attributes): Document gettimeofday.
36295 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36297 * config/s390/s390.c (s390_option_override_internal): Set
36298 PARAM_MIN_VECT_LOOP_BOUND
36300 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36302 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
36303 * config/s390/s390.md: Likewise.
36305 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36308 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
36309 (<avx2_avx512>_perm<mode>): Rename to ...
36310 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
36312 (<avx512>_perm<mode>_mask): Rename to ...
36313 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
36315 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
36316 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
36317 instead of VI8F_256_512.
36318 (avx512f_perm<mode>): New define_expand.
36319 (avx512f_perm<mode>_mask): Likewise.
36320 (avx512f_perm<mode>_1<mask_name>): New define_insn.
36321 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
36323 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36325 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
36326 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
36328 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
36330 2017-03-06 Martin Liska <mliska@suse.cz>
36333 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
36334 when having a SSA NAME w/o VAR_DECL assigned to it.
36336 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36338 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
36339 msa_dpsub_<su>_d): Fix MODE for vec_select.
36341 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36343 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
36345 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
36347 2017-03-06 Richard Biener <rguenther@suse.de>
36349 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
36350 * plugin.c (register_plugin_info): Likewise.
36351 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
36353 2017-03-05 Jakub Jelinek <jakub@redhat.com>
36355 * config/i386/sse.md (sse_storehps, sse_storelps,
36356 avx_<castmode><avxsizesuffix>_<castmode>,
36357 avx512f_<castmode><avxsizesuffix>_<castmode>,
36358 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
36359 in condition that at least one operand is not a MEM.
36361 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36363 PR middle-end/79805
36364 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
36365 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
36367 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
36368 gimple_call_nothrow_p flag based on whether original builtin can throw.
36369 If it can, emit following stmts on the fallthrough edge.
36370 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
36371 don't create new bb if inserting just debug stmts on the edge, try to
36372 insert them on the fallthru bb or just reset debug stmts.
36374 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
36377 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
36378 restore recog_data (including the operand rtxes inside it) around
36379 the call to get_insn_template.
36381 2017-03-03 Martin Sebor <msebor@redhat.com>
36383 PR tree-optimization/79699
36384 * context.c (context::~context): Free MPFR caches to avoid
36385 a memory leak on program exit.
36387 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36389 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
36390 Use wide_int::ulow () instead of .elt (0).
36392 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36394 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
36395 (*pushxf): Limit oF constraint to 32bit targets and add oC
36396 constraint for 64bit targets.
36397 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
36398 (*pushdf): Change rmF constraint to rmC.
36400 2017-03-03 Martin Liska <mliska@suse.cz>
36402 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
36403 Remove unused variable.
36405 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36408 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
36409 is a memory operand, increase num_memory.
36410 (ix86_expand_args_builtin): Likewise.
36412 2017-03-03 Jan Hubicka <jh@suse.cz>
36415 * ipa-devirt.c (maybe_record_node): Properly handle
36416 __cxa_pure_virtual visibility.
36418 2017-03-03 Martin Liska <mliska@suse.cz>
36420 PR tree-optimization/79803
36421 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
36423 (pass_loop_prefetch::execute): Disabled optimization if an
36424 assumption about L1 cache size is not met.
36426 2017-03-03 Martin Liska <mliska@suse.cz>
36428 PR rtl-optimization/79574
36429 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
36430 (hash_scan_set): Likewise.
36431 (dump_hash_table): Likewise.
36432 (hoist_code): Likewise.
36434 2017-03-03 Richard Biener <rguenther@suse.de>
36436 * fixed-value.c (fixed_from_string): Restore use of elt (1)
36437 in place of uhigh ().
36438 (fixed_convert_from_real): Likewise.
36440 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36443 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
36445 2017-03-03 Richard Biener <rguenther@suse.de>
36447 PR middle-end/79818
36448 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
36449 TYPE_OVERFLOW_UNDEFINED check.
36451 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36453 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
36455 (vector_ae_<mode>_p): Likewise.
36456 (vector_nez_<mode>_p): Likewise.
36457 (vector_ne_v2di_p): Likewise.
36458 (vector_ae_v2di_p): Likewise.
36459 (vector_ne_<mode>_p): Likewise.
36460 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
36462 (vsx_tsqrt<mode>2_fe): Likewise.
36464 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
36467 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
36469 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36471 PR rtl-optimization/79780
36472 * cprop.c (one_cprop_pass): When second and further conditional trap
36473 in a single basic block is turned into an unconditional trap, turn it
36474 into a deleted note to avoid RTL verification failures.
36476 2017-03-02 Richard Biener <rguenther@suse.de>
36478 * fold-const.c (const_binop): Use ulow () instead of elt (0).
36480 2017-03-02 Richard Biener <rguenther@suse.de>
36482 PR tree-optimization/79345
36484 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
36485 param and abort the walk, returning -1 if it is hit.
36486 (walk_aliased_vdefs): Take a limit param and pass it on.
36487 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
36488 defaulting to 0 and return a signed int.
36489 * tree-ssa-uninit.c (struct check_defs_data): New struct.
36490 (check_defs): New helper.
36491 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
36492 about uninitialized memory.
36493 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
36494 bogus uninitialized warning.
36495 (fixed_convert_from_real): Likewise.
36497 2017-03-02 Bin Cheng <bin.cheng@arm.com>
36499 PR tree-optimization/66768
36500 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
36501 iv_use if base object can't be determined.
36503 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36505 PR tree-optimization/79345
36506 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
36507 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
36508 (get_pattern_stats): Initialize it.
36509 * genemit.c (gen_expand): Verify match_scratch numbers come after
36510 match_operand/match_dup numbers.
36511 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
36512 match_scratch numbers.
36513 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
36515 * config/s390/s390.md (trunctdsd2): Likewise.
36517 2017-03-02 Richard Biener <rguenther@suse.de>
36519 * wide-int.h (wide_int_storage::operator=): Implement in terms
36522 2017-03-02 Richard Biener <rguenther@suse.de>
36524 PR tree-optimization/79777
36525 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
36526 the to insert expression to sth existing.
36528 2017-03-01 Martin Sebor <msebor@redhat.com>
36530 PR middle-end/79692
36531 * gimple-ssa-sprintf.c
36532 (directive::known_width_and_precision): New function.
36533 (format_integer): Use it.
36534 (get_mpfr_format_length): Consider the full range of precision
36535 when computing %g output with the # flag. Set the likely byte
36536 count to 3 rather than 1 when precision is indeterminate.
36537 (format_floating): Correct the lower bound of precision.
36539 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36541 * doc/invoke.texi: Document default code model for 64-bit Linux.
36543 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36546 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
36547 udiv rather than div since input pattern is unsigned.
36549 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
36551 * config/i386/i386.c (print_reg): Warn for values of
36552 unsupported size in integer register.
36554 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
36557 * config/rs6000/predicates.md (current_file_function_operand): Do
36558 not allow self calls to be local if the function is replaceable.
36560 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
36563 * config/rs6000/altivec.h (vec_ctz and others): Change the
36564 preprocessor macro that controls conditional compilation from
36565 _ARCH_PWR9 to __POWER9_VECTOR__.
36566 (vec_all_ne): Change parameterization of __altivec_scalar_pred
36567 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
36568 control (instead of _ARCH_PWR9 control) so that template
36569 definition uses power9-specific function.
36570 (vec_any_eq): Likewise.
36571 (vec_all_ne): Change macro definition to use a power9-specific
36572 expansion under #ifdef __POWER9_VECTOR__ control (instead of
36573 _ARCH_PWR9 control).
36574 (vec_any_eq) Likewise.
36575 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
36576 expansion for CMPNEF to remove support for xvcmpnesp instruction.
36577 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
36578 support for xvcmpnedp instruction.
36579 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
36580 macro expansion so that Power9 implementation of vec_all_ne does
36581 not use the AltiVec predicate framework.
36582 (VCMPNEH_P): Likewise.
36583 (VCMPNEW_P): Likewise.
36584 (VCMPNED_P): Likewise.
36585 (VCMPNEFP_P): Likewise.
36586 (VCMPNEDP_P): Likewise.
36587 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
36588 implementation of vec_any_eq to not use AltiVec predicate
36590 (VCMPAEH_P): Likewise.
36591 (VCMPAEW_P): Likewise.
36592 (VCMPAED_P): Likewise.
36593 (VCMPAEFP_P): Likewise.
36594 (VCMPAEDP_P): Likewise.
36595 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
36596 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
36597 not use the AltiVec predicate framework.
36598 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
36599 of vec_any_eq to not use AltiVec predicate framework.
36600 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36601 support for predefined __POWER9_VECTOR__ macro to indicate that
36602 Power9 instruction selection is enabled.
36603 (altivec_overloaded_builtins): Remove extraneous
36604 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
36605 function argument types RS6000_BTI_bool_V16QI and
36606 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
36607 entry for overloaded function argument types RS6000_BTI_bool_V4SI
36608 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
36609 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
36610 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
36611 Power9 for implementations of vec_cmpne. Change the signature for
36612 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
36613 (representing vec_all_ne) to remove the previously described first
36614 argument of type RS6000_BTI_INTSI, as this was an artifact of
36615 reliance on the AltiVec predicate framework, which is no longer
36616 used in the implementation of these functions. Add
36617 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
36618 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
36619 since, unlike the AltiVec predicate framework implementation, we
36620 do not share function descriptors between vec_alle and vec_anyeq.
36621 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
36622 set of modes that receive special treatment even when
36623 TARGET_P9_VECTOR is true. The special treatment emits code that
36624 does not depend on Power9 instructions.
36625 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36626 define_expand to not rely on AltiVec predicate framework.
36627 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36629 (vector_ne_v2di_p): Change this define_expand to not rely on
36630 AltiVec predicate framework.
36631 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36633 (vector_ne_<mode>_p): Change this define_expand to not rely on
36634 AltiVec predicate framework.
36635 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36637 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36638 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36639 define_insn pattern.
36640 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36641 define_insn pattern because the xvcmpne<VSs>. instruction is not
36643 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36644 instruction is not supported.
36646 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36648 * config/nvptx/nvptx.c: Include intl.h.
36650 2017-03-01 Martin Jambor <mjambor@suse.cz>
36653 * ipa-prop.h (ipa_bits): Removed field known.
36654 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36655 to pointers. Adjusted their comments to warn about their sharing.
36656 (ipcp_transformation_summary): Change bits to a vector of pointers.
36657 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36658 (ipa_get_ipa_bits_for_value): Declare.
36659 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36660 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36661 (ipa_bits_hash_table): Likewise.
36662 (ipa_vr_ggc_hash_traits): Likewise.
36663 (ipa_vr_hash_table): Likewise.
36664 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36665 being pointers and vr_known being removed.
36666 (ipa_set_jf_unknown): Likewise.
36667 (ipa_get_ipa_bits_for_value): New function.
36668 (ipa_set_jfunc_bits): Likewise.
36669 (ipa_get_value_range): New overloaded functions.
36670 (ipa_set_jfunc_vr): Likewise.
36671 (ipa_compute_jump_functions_for_edge): Use the above functions to
36672 construct bits and vr parts of jump functions.
36673 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36674 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36676 (ipcp_grow_transformations_if_necessary): Also allocate
36677 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36679 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36680 them. Fix too long lines.
36681 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36682 vr_known being removed.
36683 (ipa_read_jump_function): Use new setter functions to construct bits
36684 and vr parts of jump functions or set them to NULL.
36685 (write_ipcp_transformation_info): Adjust for bits being pointers.
36686 (read_ipcp_transformation_info): Likewise.
36687 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36689 Include gt-ipa-prop.h.
36690 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36692 (ipcp_store_bits_results): Likewise.
36693 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36694 Do not write to existing jump functions but use a temporary instead.
36696 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36699 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36700 attempt to use its first operand as BIT_FIELD_REF base.
36702 2017-03-01 Richard Biener <rguenther@suse.de>
36704 PR middle-end/79721
36705 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36706 interpolating formula in wrapping arithmetic.
36707 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36709 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36711 PR tree-optimization/79734
36712 * tree-vect-generic.c (expand_vector_condition): Optimize
36713 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36714 Handle VEC_COND_EXPR where comparison has different inner width from
36715 type's inner width.
36717 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36719 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36720 markup, and similar issues. Remove @opindex entries for things
36721 that aren't options. Add missing -mmpy-option entries.
36723 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36725 PR tree-optimization/79737
36726 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36727 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36728 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36729 instead of byte_size. Formatting fix.
36730 (shift_bytes_in_array_right): Formatting fix.
36732 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36735 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36736 condition on optimize for the leaf function test.
36738 2017-02-28 Martin Liska <mliska@suse.cz>
36741 * read-rtl-function.c (function_reader::handle_unknown_directive):
36742 Bail out when one uses -flto.
36744 2017-02-28 Martin Liska <mliska@suse.cz>
36746 * common.opt: Replace space with tabular for options of <number>
36748 * config/i386/i386.opt: Show <number> value for
36749 -mlarge-data-threshold.
36750 * opts.c (print_filtered_help): Do not display number in hexadecimal
36753 2017-02-28 Martin Liska <mliska@suse.cz>
36755 * common.opt: Fix --help=option -Q for options which are of
36758 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36760 * config/i386/i386.c (print_reg): Error out for values
36761 of 8-bit size in invalid integer register.
36763 2017-02-28 Martin Sebor <msebor@redhat.com>
36765 PR tree-optimization/79691
36766 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36768 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36771 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36772 gcc_unreachable with output_operand_lossage.
36774 2017-02-28 Richard Biener <rguenther@suse.de>
36776 PR tree-optimization/79740
36777 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36779 (visit_nary_op): Insert the nary into the hashtable if we
36780 pattern-matched sth.
36781 * tree-ssa-pre.c (eliminate_insert): Robustify.
36783 2017-02-28 Richard Biener <rguenther@suse.de>
36785 PR middle-end/79731
36786 * fold-const.c (decode_field_reference): Reject out-of-bound
36789 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36791 * config/i386/i386.c: Include intl.h.
36792 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36793 instead of just cond ? "..." : "...".
36794 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36795 * coverage.c (read_counts_file): Likewise.
36796 * omp-offload.c: Include intl.h.
36797 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36798 of just cond ? "..." : "...".
36799 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36800 of just cond ? "..." : "...".
36802 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36805 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36807 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36808 'tune for' CPU name.
36809 * config/arm/arm-cpu-data.h: Regenerated.
36811 2017-02-28 Richard Biener <rguenther@suse.de>
36813 PR tree-optimization/79732
36814 * tree-inline.c (expand_call_inline): Do not shadow var.
36816 2017-02-28 Richard Biener <rguenther@suse.de>
36818 PR tree-optimization/79723
36819 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36820 address-space properly.
36822 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36824 * doc/optinfo.texi (Optimization groups): Fix option used for
36826 * doc/invoke.texi (-fopt-info): Document "omp".
36827 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36828 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36829 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36830 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36831 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36833 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36835 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36836 OPTGROUP_OMP with "omp".
36838 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36841 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36842 for arithmetic shift of unsigned V2DI.
36844 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36846 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36847 arc/linux.h headers.
36848 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36849 (LINK_SPEC): Likewise.
36850 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36851 (EXTRA_SPECS): Likewise.
36852 (STARTFILE_SPEC): Likewise.
36853 (ENDFILE_SPEC): Likewise.
36854 (LIB_SPEC): Likewise.
36855 (TARGET_SDATA_DEFAULT): Likewise.
36856 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36857 (MULTILIB_DEFAULTS): Likewise.
36858 (DWARF2_UNWIND_INFO): Likewise.
36859 * config/arc/big.h: New file.
36860 * config/arc/elf.h: Likewise.
36861 * config/arc/linux.h: Likewise.
36862 * config/arc/t-uClibc: Remove.
36864 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36866 PR tree-optimization/77536
36867 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
36868 (tree_transform_and_unroll_loop): Use above function to compute the
36869 estimated niter of unrolled loop and use it when scaling profile.
36870 Also use count info rather than frequency if it's non-zero.
36871 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
36872 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
36873 (vect_transform_loop): Call above function.
36875 2017-02-27 Richard Biener <rguenther@suse.de>
36877 PR tree-optimization/45397
36878 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
36879 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
36880 (visit_nary_op): Add pattern matching for CSEing sign-changed
36881 or truncated operations with wider ones.
36883 2017-02-27 Richard Biener <rguenther@suse.de>
36885 PR tree-optimization/79690
36886 * tree-vect-stmts.c (vectorizable_store): Use vector type
36887 built from the DR with address-space.
36889 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
36891 * doc/invoke.texi (Optimize Options): Refine the description
36892 of asan-use-after-return.
36894 2017-02-25 Alan Modra <amodra@gmail.com>
36896 PR rtl-optimization/79584
36897 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
36898 base, not ad->base_term, the reg within base. Remove assertion
36899 that ad->base == ad->base_term. Replace gen_int_mode using
36900 bogus mode with const0_rtx.
36902 2017-02-25 Jakub Jelinek <jakub@redhat.com>
36904 PR middle-end/79396
36905 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
36906 FMA_EXPR like tcc_binary or tcc_unary.
36908 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
36911 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
36913 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
36914 (output_loc_operands): Handle DW_OP_call_ref and
36915 DW_OP_GNU_variable_value.
36916 (struct variable_value_struct): New type.
36917 (struct variable_value_hasher): Likewise.
36918 (variable_value_hash): New variable.
36919 (string_types): Remove.
36920 (copy_loc_descr): New function.
36921 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
36922 (prepend_loc_descr_to_each): New function.
36923 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
36924 instead of add_loc_descr_to_each if the first argument is single
36925 location list and the second has multiple.
36926 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
36927 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
36928 when looking for variable value which doesn't have other location info.
36929 (loc_list_from_tree): Formatting fix.
36930 (gen_array_type_die): Simplify DW_AT_string_length handling.
36931 (adjust_string_types): Remove.
36932 (gen_subprogram_die): Don't call adjust_string_types nor test/set
36933 string_types. Call resolve_variable_values.
36934 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
36935 (resolve_addr_in_expr): Likewise. Add A argument.
36936 (copy_deref_exprloc): Remove deref argument. Adjust for the
36937 original expression being DW_OP_GNU_variable_value with optionally
36938 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
36939 optionally after it.
36940 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
36941 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
36942 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
36943 (variable_value_hasher::hash, variable_value_hasher::equal): New
36945 (resolve_variable_value_in_expr, resolve_variable_value,
36946 resolve_variable_values, note_variable_value_in_expr,
36947 note_variable_value): New functions.
36948 (dwarf2out_early_finish): Call note_variable_value on all toplevel
36951 2017-02-24 Jakub Jelinek <jakub@redhat.com>
36954 * opts.h (handle_generated_option): Add GENERATED_P argument.
36955 * opts-common.c (handle_option): Adjust function comment.
36956 (handle_generated_option): Add GENERATED_P argument, pass it to
36958 (control_warning_option): Pass false to handle_generated_option
36960 * opts.c (maybe_default_option): Pass true to handle_generated_option
36962 * optc-gen.awk: Likewise.
36964 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36966 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
36967 a REG, look at the REG it is a SUBREG of.
36968 (splitter for cmpeqsi_t): Ditto.
36970 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36972 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
36973 the special USEs with the pattern of the insn, not the insn itself.
36975 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
36978 * doc/invoke.texi: Document -mload-store-pairs.
36980 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36981 Sandra Loosemore <sandra@codesourcery.com>
36983 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
36984 argument isn't a CONST_INT.
36985 (nios2_alternate_compare_const): Assert op is a CONST_INT.
36986 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
36987 (nios2_validate_compare): Bypass alternate compare logic if *op2
36988 is not a CONST_INT.
36989 (ldstwm_operation_p): Return false if first_base is not a REG or
36990 if first_offset is not a CONST_INT.
36992 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36994 * config/cris/cris.md: Use correct operand in a define_peephole2.
36996 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36998 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
37000 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37002 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
37003 this_insn if it is an INSN or JUMP_INSN.
37004 (force_offsettable): Look at base, not at addr.
37005 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37006 on things that aren't necessarily CONST_INTs.
37008 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37010 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37011 -mfpmath=sse is the default also for x86-32 targets with SSE2
37012 instruction set when @option{-ffast-math} is enabled
37014 2017-02-24 Jeff Law <law@redhat.com>
37016 PR rtl-optimizatoin/79286
37017 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37020 2017-02-24 Richard Biener <rguenther@suse.de>
37022 PR tree-optimization/79389
37023 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37026 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37028 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37029 function comment to reflect reality.
37030 (loop_exits_before_overflow): Fix typo in function description.
37032 2017-02-24 Richard Biener <rguenther@suse.de>
37034 PR tree-optimization/79389
37035 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37036 properly that a threading opportunity exists. Detect conditional
37037 copy/constant propagation opportunities.
37039 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37041 * config/visium/visium.md (type): Add trap.
37042 (b): New mode attribute.
37043 (*btst): Rename into...
37044 (*btst<mode>): ...this and adjust.
37045 (*cbranchsi4_btst_insn): Rename into...
37046 (*cbranch<mode>4_btst_insn): ...this and adjust.
37047 (trap): New define_insn.
37049 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37051 PR tree-optimization/79389
37052 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37053 (noce_reversed_cond_code): New function.
37054 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37055 reversed_comparison_code. Formatting fix.
37056 (noce_try_store_flag): Test rev_cond != NULL in addition to
37057 reversed_comparison_code.
37058 (noce_try_store_flag_constants): Likewise.
37059 (noce_try_store_flag_mask): Likewise.
37060 (noce_try_addcc): Use rev_cond if non-NULL instead of
37061 reversed_comparison_code.
37062 (noce_try_cmove_arith): Likewise. Formatting fixes.
37063 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37064 (noce_find_if_block): Initialize rev_cond.
37065 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37066 instead of false as last argument never attempt to reverse it
37069 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37071 PR tree-optimization/79663
37072 * tree-predcom.c (combine_chains): Process refs in reverse order
37073 only for ZERO length chains, and add explaining comment.
37075 2017-02-23 Jeff Law <law@redhat.com>
37077 PR tree-optimization/79578
37078 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37079 in call to operand_equal_p.
37081 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37084 * config/i386/cpuid.h: Fix another undefined behavior.
37086 2017-02-23 Richard Biener <rguenther@suse.de>
37088 PR tree-optimization/79683
37089 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37090 vector types for data-refs.
37092 2017-02-23 Martin Liska <mliska@suse.cz>
37094 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37096 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37098 PR middle-end/79665
37099 * internal-fn.c (get_range_pos_neg): Moved to ...
37100 * tree.c (get_range_pos_neg): ... here. No longer static.
37101 * tree.h (get_range_pos_neg): New prototype.
37102 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37103 are known to be in between 0 and signed maximum inclusive, try to
37104 expand both unsigned and signed divmod and use the cheaper one from
37107 2017-02-22 Jeff Law <law@redhat.com>
37109 PR tree-optimization/79578
37110 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37111 to compare base operands.
37113 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
37116 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
37117 gpc_reg_operand instead of fpr_reg_operand.
37119 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
37121 * config/mips/mips.c (mips_return_in_memory): Force FP
37122 vector types to be returned in memory for o32 ABI.
37124 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37126 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
37127 instead of DW_TAG_member for static data member declarations and don't
37128 set no_linkage_name for static inline data members.
37129 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
37132 2017-02-22 Martin Liska <mliska@suse.cz>
37134 * doc/invoke.texi: Replace inequality signs with square brackets
37137 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
37140 * lra-constraints.c (simplify_operand_subreg): Handle
37141 WORD_REGISTER_OPERATIONS targets.
37143 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37146 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
37147 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
37148 elimination by swapping fld*.
37150 2017-02-22 Richard Biener <rguenther@suse.de>
37152 PR tree-optimization/79673
37153 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
37154 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
37155 irrelevant address-space qualifiers and avoiding a
37156 ADDR_SPACE_CONVERT_EXPR from fold_convert.
37158 2017-02-22 Richard Biener <rguenther@suse.de>
37160 PR tree-optimization/79666
37161 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
37162 to not symbolically negate if that may introduce undefined
37165 2017-02-22 Martin Liska <mliska@suse.cz>
37168 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
37169 * data-streamer-out.c (streamer_write_gcov_count_stream):
37171 * value-prof.c (stream_out_histogram_value): Make assert more
37172 precise based on type of counter.
37174 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
37177 * config/i386/i386.md (standard_x87sse_constant_load splitter):
37178 Use nonimmediate_operand instead of memory_operand for operand 1.
37179 (float-extend standard_x87sse_constant_load splitter): Ditto.
37181 2017-02-21 Jeff Law <law@redhat.com>
37183 PR tree-optimization/79621
37184 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
37185 blocks with edges to themselves.
37187 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37190 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
37191 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
37192 Use gimple_call_builtin_p.
37195 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
37196 on temporarily removed DEBUG_INSNs.
37198 PR tree-optimization/79649
37199 * tree-loop-distribution.c (classify_partition): Give up on
37200 non-generic address space loads/stores.
37202 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
37204 * doc/loop.texi (Loop manipulation): Remove nonexistent
37205 tree_ssa_loop_version from the documentation.
37206 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
37208 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37211 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
37212 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
37213 * config/rs6000/rs6000.c: Include except.h.
37214 (rs6000_expand_split_stack_prologue): Call
37215 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
37217 2017-02-21 Martin Jambor <mjambor@suse.cz>
37220 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
37221 have been analyzed.
37223 2017-02-21 Martin Jambor <mjambor@suse.cz>
37225 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
37226 for backward compatibility only.
37227 * doc/invoke.texi (Option Summary): Remove all references to
37228 -fipa-cp-alignment.
37230 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
37234 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37236 * lra-constraints.c (curr_insn_transform): Handle
37237 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37239 2017-02-21 Martin Liska <mliska@suse.cz>
37241 * config/i386/i386.opt: Replace -masm-dialect with -masm.
37243 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
37245 PR translation/79638
37246 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
37248 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
37251 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
37252 (arm_function_ok_for_sibcall): Return false for an indirect call by
37253 descriptor if all the argument registers are used.
37254 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
37255 alignment of the function.
37257 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37259 PR tree-optimization/61441
37260 * simplify-rtx.c (simplify_const_unary_operation): For
37261 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
37262 the sNaN unmodified.
37264 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
37266 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
37267 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
37268 instead of SYSTEM_HEADER_DIR.
37270 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
37271 Martin Liška <mliska@suse.cz>
37273 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
37274 Fix typos and grammar, use active voice, and clarify.
37276 2017-02-20 Marek Polacek <polacek@redhat.com>
37278 PR middle-end/79537
37279 * gimplify.c (gimplify_expr): Handle unused *&&L;.
37282 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
37284 2017-02-20 Jakub Jelinek <jakub@redhat.com>
37287 * config/i386/i386.c (ix86_expand_builtin): Handle
37288 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
37289 ix86_builtins_isa[fcode].isa as a requirement of those
37290 flags and any other flag in the bitmask.
37291 (ix86_init_mmx_sse_builtins): Use 0 instead of
37292 ~OPTION_MASK_ISA_64BIT as mask.
37293 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
37294 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
37295 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
37296 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
37298 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37301 * lra-constraints.c (split_reg): Check requested split mode
37302 is supported by the register.
37304 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37306 * lra-constraints.c (simplify_operand_subreg): Remove early
37309 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37312 * lra-constraints.c (curr_insn_transform): Tighten condition
37313 for converting SUBREG reloads from OP_OUT to OP_INOUT.
37315 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37318 * lra-constraints.c (curr_insn_transform): Handle
37319 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37321 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
37324 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
37326 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
37328 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
37331 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
37334 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37336 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
37339 2017-02-18 Jakub Jelinek <jakub@redhat.com>
37341 * final.c (last_columnnum, override_columnnum): New variables.
37342 (final_start_function): Set last_columnnum, pass it to begin_prologue
37343 hook and pass 0 to dwarf2out_begin_prologue.
37344 (final_scan_insn): Update override_columnnum. Pass last_columnnum
37345 to source_line debug hook.
37346 (notice_source_line): Compute last_columnnum and for debug_column_info
37347 return true on column changes.
37348 * debug.h (struct gcc_debug_hooks): Add column argument to
37349 source_line and begin_prologue hooks.
37350 (debug_nothing_int_charstar_int_bool): Remove prototype.
37351 (debug_nothing_int_int_charstar,
37352 debug_nothing_int_int_charstar_int_bool): New prototypes.
37353 (dwarf2out_begin_prologue): Add column argument.
37354 * debug.c (do_nothing_debug_hooks): Adjust source_line and
37355 begin_prologue hooks.
37356 (debug_nothing_int_charstar_int_bool): Remove.
37357 (debug_nothing_int_int_charstar,
37358 debug_nothing_int_int_charstar_int_bool): New functions.
37359 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
37360 through to dwarf2out_source_line.
37361 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
37362 (dwarf2out_source_line): Add column argument, emit it if requested.
37363 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
37365 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37366 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37367 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
37368 through to dwarf2out_begin_prologue.
37369 (vmsdbgout_source_line): Add column argument, pass it through to
37370 dwarf2out_source_line.
37371 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
37372 dbxout_source_line caller.
37373 (dbxout_source_line): Add column argument.
37375 * common.opt (gno-column-info, gcolumn-info): New options.
37376 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
37377 (check_die): Also test for multiple DW_AT_decl_column attributes.
37378 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
37379 DW_AT_decl_column if requested.
37380 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
37382 (gen_variable_die): Likewise.
37383 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
37384 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
37387 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
37388 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
37389 (ix86_handle_option): Handle OPT_m3dnowa.
37390 * doc/invoke.texi (-m3dnowa): Document.
37391 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
37392 -m3dnowa instead of -m3dnow -march=athlon.
37395 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
37396 instead of gcc_assert for K, r and R code checks. Formatting fixes.
37398 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37401 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
37402 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
37403 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
37404 generator for vsx_xxpermdi_<mode>_be.
37405 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
37406 force big-endian semantics.
37407 (vsx_xxpermdi_<mode>_be): New define_expand with same
37408 implementation as previous version of vsx_xxpermdi_<mode>.
37410 2017-02-17 Jakub Jelinek <jakub@redhat.com>
37412 PR tree-optimization/79327
37413 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
37414 variable, its initialization and use.
37416 2017-02-17 Julia Koval <julia.koval@intel.com>
37418 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
37419 (OPTION_MASK_ISA_PKU_UNSET): New.
37420 (ix86_handle_option): Handle -mrdpid.
37421 * config/i386/cpuid.h (bit_RDPID): New.
37422 * config/i386/driver-i386.c (host_detect_local_cpu):
37423 Detect RDPID feature.
37424 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
37425 * config/i386/i386-c.c (ix86_target_macros_internal):
37427 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
37428 (ix86_valid_target_attribute_inner_p): Add "rdpid".
37429 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
37430 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
37431 * config/i386/i386.md (define_insn "rdpid"): New.
37432 * config/i386/i386.opt Add -mrdpid.
37433 * config/i386/immintrin.h (_rdpid_u32): New.
37435 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
37437 PR rtl-optimization/79541
37438 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
37439 instead of transforming it into USE.
37441 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
37443 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
37444 If HONOR_SNANS (SFmode) force the input to a register.
37445 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
37446 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
37447 an frsp or similar insn.
37449 2017-02-17 Martin Liska <mliska@suse.cz>
37451 PR rtl-optimization/79577
37452 * params.def (selsched-max-sched-times): Increase minimum to 1.
37454 2017-02-17 Martin Liska <mliska@suse.cz>
37456 PR rtl-optimization/79574
37457 * gcse.c (want_to_gcse_p): Prevent integer overflow.
37459 2017-02-17 Martin Liska <mliska@suse.cz>
37461 PR tree-optimization/79529
37462 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
37463 ssa_defined_default_def_p to handle cases which are implicitly
37465 * tree-ssa.c (ssa_defined_default_def_p): New function.
37466 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
37467 which are implicitly defined.
37468 * tree-ssa.h (ssa_defined_default_def_p): Declare.
37470 2017-02-17 Richard Biener <rguenther@suse.de>
37472 PR middle-end/79576
37473 * params.def (max-ssa-name-query-depth): Limit to 10.
37475 2017-02-17 Richard Biener <rguenther@suse.de>
37477 PR tree-optimization/79552
37478 * tree-ssa-structalias.c (visit_loadstore): Properly verify
37481 2017-02-17 Richard Biener <rguenther@suse.de>
37484 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
37486 2017-02-17 Marek Polacek <polacek@redhat.com>
37488 PR middle-end/79536
37489 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
37490 (fold_negate_expr): New wrapper.
37492 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
37494 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
37495 Correct terminology and de-emphasize pre-standard behavior.
37497 2017-02-16 Alan Modra <amodra@gmail.com>
37499 PR rtl-optimization/79286
37500 * ira.c (def_dominates_uses): New function.
37501 (update_equiv_regs): Don't create an equivalence for insns that
37502 may trap where the register def does not dominate the use.
37504 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
37506 PR rtl-optimization/78127
37507 * lra.c (lra): Call lra_eliminate before finish the loop after
37510 2017-02-16 Richard Biener <rguenther@suse.de>
37512 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
37514 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
37515 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
37516 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
37517 (isl_val_int_from_wi): New function.
37518 (extract_affine_gmp): Rename to ...
37519 (extract_affine_wi): ... this, take a widest_int.
37520 (extract_affine_int): Just wrap extract_affine_wi.
37521 (add_param_constraints): Use isl_val_int_from_wi.
37522 (add_loop_constraints): Likewise, and extract_affine_wi.
37524 2017-02-15 Jeff Law <law@redhat.com>
37526 PR middle-end/79521
37527 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
37528 ira_init_register_move_cost_if_necessary.
37530 2017-02-15 Martin Sebor <msebor@redhat.com>
37532 PR middle-end/32003
37533 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
37534 removed in a prior commit.
37536 2017-02-15 Bin Cheng <bin.cheng@arm.com>
37538 PR tree-optimization/79347
37539 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
37540 counters during peeling.
37542 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
37544 * Makefile.in (site.exp): Remove "set ISLVER".
37546 2017-02-15 Jakub Jelinek <jakub@redhat.com>
37549 * real.c (real_from_integer): Call real_convert even for decimal.
37551 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
37554 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
37556 2017-02-14 Andrew Pinski <apinski@cavium.com>
37558 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
37559 cores and change the partno/implementer to be correct.
37560 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
37561 the 'B" as the implementer.
37562 * config/aarch64/aarch64-tune.md: Regenerate.
37564 2017-02-14 Carl Love <cel@us.ibm.com>
37566 * config/rs6000/rs6000.c: Add case statement entry to make the
37567 xvcvuxdsp built-in argument unsigned.
37568 * config/rs6000/vsx.md: Fix the source and return operand types so they
37569 match the instruction definitions from the ISA document. Fix typo
37570 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
37573 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
37576 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
37577 member early_clobber_alts.
37578 * lra-lives.c (reg_early_clobber_p): New.
37579 (process_bb_lives): Use it.
37580 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
37581 (debug_operand_data): Initialize early_clobber_alts.
37582 (setup_operand_alternative): Set up early_clobber_alts.
37583 (collect_non_operand_hard_regs): Ditto. Pass early clobber
37584 alternatives to new_insn_reg.
37585 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
37587 (lra_update_insn_regno_info): Pass the new arg.
37589 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37591 PR middle-end/79505
37592 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
37593 (new_oacc_loop_raw): Don't clear already cleared fields.
37596 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
37597 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
37598 _mm512_prefetch_i64gather_ps): New inline functions and macros.
37600 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
37603 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
37605 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
37608 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
37609 the extra instruction to the right place to store 128-bit constant
37612 2017-02-14 Martin Sebor <msebor@redhat.com>
37614 PR middle-end/79448
37615 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
37616 warning for strings of unknown length.
37618 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
37620 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
37622 2017-02-14 Jeff Law <law@redhat.com>
37625 * ira-costs.c (scan_one_insn): Initialize register move costs
37626 for pseudos seen in USE/CLOBBER insns.
37628 PR tree-optimization/79095
37629 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37630 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37631 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37632 if the operands are known to be not equal, then the resulting range
37634 (intersect_ranges): If the new range is ~[0,0] and the old range is
37635 wide, then prefer ~[0,0].
37636 * tree-vrp.c (overflow_comparison_p_1): New function.
37637 (overflow_comparison_p): New function.
37638 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37639 if NAME is used in an overflow test.
37640 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37641 overflow check that can be expressed as an equality test, then adjust
37642 ops to be that equality test.
37644 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37646 * config/s390/s390-builtin-types.def: Remove flags argument.
37647 * config/s390/s390.c (s390_init_builtins): Likewise.
37649 2017-02-14 Martin Liska <mliska@suse.cz>
37651 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37652 vector. Fix trailing white spaces.
37654 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37656 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37659 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37661 PR rtl-optimization/68664
37662 * config/arm/arm.c (arm_sched_can_speculate_insn):
37663 New function. Declare prototype.
37664 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37666 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37668 PR rtl-optimization/68664
37669 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37671 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37673 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37675 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37676 max skip bytes for function, loop and jump.
37678 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37680 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37681 ABS_EXPR for gimple dump.
37683 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37686 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37688 PR tree-optimization/79408
37689 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37690 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37691 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37692 also if rhs1 is INTEGER_CST.
37694 2017-02-14 Richard Biener <rguenther@suse.de>
37696 PR middle-end/79432
37697 * tree-into-ssa.c (insert_phi_nodes): When the function can
37698 have abnormal edges rewrite SSA names with broken use-def
37699 dominance out of SSA and register them for PHI insertion.
37701 2017-02-13 Martin Sebor <msebor@redhat.com>
37703 PR middle-end/79496
37704 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37705 clearing info.nowrite flag when snprintf size argument is a range.
37707 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37709 * cprop.c (cprop_jump): Add missing space in string literal.
37710 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37711 (get_constraint_for_component_ref): Likewise.
37712 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37713 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37714 * lra-constraints.c (process_alt_operands): Likewise.
37715 * ipa-inline.c (inline_small_functions): Likewise.
37716 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37717 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37718 * trans-mem.c (diagnose_tm_1_op): Likewise.
37719 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37720 (grid_parallel_clauses_gridifiable): Likewise.
37722 * config/nvptx/mkoffload.c (process): Add space in between
37725 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37726 "MOD4_SSE_REGS" and "ALL_REGS".
37728 * spellcheck.c (test_data): Add , in between "foo" and "food".
37730 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37733 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37734 boundary crossing check and subsequent code generation agree.
37736 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37738 * config/aarch64/aarch64.c (has_memory_op): Delete.
37739 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37742 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37744 PR rtl-optimization/79388
37745 PR rtl-optimization/79450
37746 * combine.c (distribute_notes): When removing TEM_INSN for which
37747 corresponding dest has last value recorded, invalidate that last
37750 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37752 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37753 of explicit '@'. Add missing assembly comment marker on branch costs
37756 2017-02-13 Nathan Sidwell <nathan@acm.org>
37758 * gengtype-lex.l (<in_struct>): Add '/'.
37760 2017-02-13 Martin Liska <mliska@suse.cz>
37763 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37765 2017-02-13 Richard Biener <rguenther@suse.de>
37767 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37769 * configure: Re-generate.
37770 * config.in: Likewise.
37771 * graphite-dependences.c: Simplify as if
37772 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37773 * graphite-isl-ast-to-gimple.c: Likewise.
37774 * graphite-optimize-isl.c: Likewise.
37775 * graphite-poly.c: Likewise.
37776 * graphite-sese-to-poly.c: Likewise.
37777 * graphite.h: Likewise.
37778 * toplev.c: Include isl/version.h and use isl_version () for
37779 printing the ISL version.
37780 * doc/install.texi: Update ISL requirement.
37782 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37784 * doc/standards.texi (Standards): Update reference to
37787 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37789 * doc/extend.texi (Named Address Spaces): sourceware.org now
37791 * doc/install.texi (Binaries): Ditto.
37794 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37796 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37797 standard terminology "stringize"/"stringizing" throughout.
37798 * doc/cppinternals.texi: Likewise.
37800 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37802 * doc/extend.texi: Fix some spelling mistakes and typos.
37803 * doc/invoke.texi: Likewise.
37805 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37808 * params.def (inline-min-speedup) Change from 10 to 8.
37810 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37812 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37815 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37818 * ipa-inline-analysis.c (get_minimal_bb): New function.
37819 (record_modified): Use it.
37820 (remap_edge_change_prob): Handle also ancestor functions.
37822 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37824 * doc/contrib.texi (Contributors): Remove broken link into
37825 the Mauve CVS repository.
37827 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37829 PR middle-end/79454
37830 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37831 result computation whenever lhs doesn't have vector mode, not
37832 just when it has BLKmode.
37834 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37836 * doc/makefile.texi (profiledbootstrap): Refer to the
37837 installation instructions only in textual form.
37839 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37842 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37844 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37846 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37847 (Specific): Update mingw-w64 reference.
37849 (Specific): Remove broken link to Renesas RX processor.
37851 2017-02-10 Richard Biener <rguenther@suse.de>
37853 * toplev.c (process_options): Do not mention obsolete graphite
37854 options when printing sorry message about missing graphite support.
37855 Mention -floop-nest-optimize.
37857 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37859 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37860 (vtst_p16): Likewise.
37861 (vtstq_p8): Likewise.
37862 (vtstq_p16): Likewise.
37864 (vtstq_p64): Likewise.
37865 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37866 (vset_lane_p64): New.
37867 (vsetq_lane_p64): New.
37869 2017-02-10 Jakub Jelinek <jakub@redhat.com>
37871 PR tree-optimization/79411
37872 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
37873 stmt operands are SSA_NAMEs used in abnormal phis.
37874 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
37877 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37880 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
37883 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37885 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
37887 2017-02-09 Jakub Jelinek <jakub@redhat.com>
37889 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
37893 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
37894 not arbitrary TREE_CONSTANT.
37897 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
37898 "omp declare target link" attribute unless is_global_var.
37899 * omp-offload.c (find_link_var_op): Likewise.
37901 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
37902 Chung-Lin Tang <cltang@codesourcery.com>
37904 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
37906 (gimplify_adjust_omp_clauses): Don't delete TILE.
37907 (gimplify_omp_for): Deal with TILE.
37908 * internal-fn.c (expand_GOACC_TILE): New function.
37909 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
37911 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
37912 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
37914 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
37915 avoid DIV for outermost collapse var.
37916 (expand_oacc_for): Insert tile element loop as needed. Adjust.
37917 Remove out of date comments, fix whitespace.
37918 * omp-general.c (omp_extract_for_data): Deal with tiling.
37919 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
37920 adjust OLF_DIM_BASE value.
37921 (struct omp_for_data): Add tiling field.
37922 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
37923 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
37924 for auto loops. Remove default auto determining, moved to
37925 oacc_loop_fixed_partitions.
37926 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
37927 stmts, add e_mask field.
37928 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
37929 (oacc_thread_numbers): Use oacc_dim_call.
37930 (oacc_xform_tile): New.
37931 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
37932 (finish_oacc_loop): Adjust for ifns vector.
37933 (oacc_loop_discover_walk): Append loop abstraction sites to list,
37934 add case for GOACC_TILE fns.
37935 (oacc_loop_xform_loop): Delete.
37936 (oacc_loop_process): Iterate over call list directly, and add
37937 handling for GOACC_TILE fns.
37938 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
37940 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
37941 vector partitioning to outer loops. Assign 2 partitions to loops
37942 when available. Add TILE handling.
37943 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
37944 (execite_oacc_device_lower): Process GOACC_TILE fns,
37945 ignore unknown specs.
37946 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
37947 * tree.c (omp_clause_num_ops): Adjust TILE ops.
37948 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
37950 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
37952 * configure.ac (ACX_BUGURL): Update.
37953 * configure: Regenerate.
37955 2017-02-09 Richard Biener <rguenther@suse.de>
37957 PR tree-optimization/69823
37958 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
37959 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
37961 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37963 * config/arc/arc-c.def: Add __NPS400__ definition.
37964 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
37965 (TARGET_NPS400): Define.
37967 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37969 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
37971 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
37972 pointer, arch_info.
37973 (arc_cpu_types): Fill the arch_info field with a pointer into the
37974 arc_arch_types table.
37975 (arc_selected_cpu): Declare.
37976 * config/arc/arc.c (arc_selected_cpu): Make global.
37977 (arc_selected_arch): Delete.
37978 (arc_base_cpu): Delete.
37979 (arc_override_options): Remove references to deleted variables,
37980 update access to arch information.
37981 (ARC_OPT): Update access to arch information.
37982 (ARC_OPTX): Likewise.
37983 * config/arc/arc.h (arc_base_cpu): Remove declaration.
37984 (TARGET_ARC600): Update access to arch information.
37985 (TARGET_ARC601): Likewise.
37986 (TARGET_ARC700): Likewise.
37987 (TARGET_EM): Likewise.
37988 (TARGET_HS): Likewise.
37989 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
37992 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
37995 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
37996 condition/operands for integer GE/LE/GEU/LEU operations.
37998 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
38000 PR translation/79397
38001 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
38004 2017-02-08 Martin Jambor <mjambor@suse.cz>
38007 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38008 whether allocation happened.
38009 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38010 nothing was allocated.
38012 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38014 PR tree-optimization/79408
38015 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38016 constant, but SSA_NAME with a known integer range, use the minimum
38017 of that range instead of op1 to determine if modulo can be replaced
38018 with its first operand.
38020 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38022 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38024 2017-02-08 Richard Biener <rguenther@suse.de>
38026 PR tree-optimization/71824
38027 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38028 Check all loops contained in the merged region.
38030 2017-02-07 Andrew Pinski <apinski@cavium.com>
38032 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38034 2017-02-07 Andrew Pinski <apinski@cavium.com>
38036 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38037 (thunderxt88): Likewise.
38038 (thunderxt81): Disable LSE and change v8.1 to v8.
38039 (thunderxt83): Likewise.
38041 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38042 Richard Biener <rguenther@suse.de>
38044 PR middle-end/79399
38045 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38046 type from int to size_t.
38047 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38049 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38051 PR rtl-optimization/79386
38052 * cprop.c (bypass_conditional_jumps): Initialize
38053 bypass_last_basic_block already before splitting bbs after
38054 unconditional traps...
38055 (bypass_conditional_jumps): ... rather than here.
38058 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38059 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38060 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38061 fix -masm=intel patterns.
38063 2017-02-07 Richard Biener <rguenther@suse.de>
38065 PR tree-optimization/79256
38066 PR middle-end/79278
38067 * builtins.c (get_object_alignment_2): Use min_align_of_type
38068 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38069 and ADJUST_FIELD_ALIGN.
38071 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38073 * doc/tm.texi: Regenerate.
38074 * stor-layout.c (layout_decl): Adjust.
38075 (update_alignment_for_field): Likewise.
38076 (place_field): Likewise.
38077 (min_align_of_type): Likewise.
38078 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38079 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38080 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38081 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38082 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38083 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38084 * config/i386/i386.c (x86_field_alignment): Likewise.
38085 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38086 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38087 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38088 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38089 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38090 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38094 2017-01-30 Richard Biener <rguenther@suse.de>
38096 PR tree-optimization/79256
38097 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38098 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38101 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38103 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38104 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38105 builtins to SImode and emit a zero-extend, if necessary.
38107 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38109 * docs/invoke.texi (RISC-V Options): Alphabetize.
38111 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38113 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
38116 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38118 * config/riscv/riscv.c: New file.
38119 * common/config/riscv/riscv-common.c: Likewise.
38120 * config.gcc: Likewise.
38121 * config/riscv/constraints.md: Likewise.
38122 * config/riscv/elf.h: Likewise.
38123 * config/riscv/generic.md: Likewise.
38124 * config/riscv/linux.h: Likewise.
38125 * config/riscv/multilib-generator: Likewise.
38126 * config/riscv/peephole.md: Likewise.
38127 * config/riscv/pic.md: Likewise.
38128 * config/riscv/predicates.md: Likewise.
38129 * config/riscv/riscv-builtins.c: Likewise.
38130 * config/riscv/riscv-c.c: Likewise.
38131 * config/riscv/riscv-ftypes.def: Likewise.
38132 * config/riscv/riscv-modes.def: Likewise.
38133 * config/riscv/riscv-opts.h: Likewise.
38134 * config/riscv/riscv-protos.h: Likewise.
38135 * config/riscv/riscv.h: Likewise.
38136 * config/riscv/riscv.md: Likewise.
38137 * config/riscv/riscv.opt: Likewise.
38138 * config/riscv/sync.md: Likewise.
38139 * config/riscv/t-elf-multilib: Likewise.
38140 * config/riscv/t-linux: Likewise.
38141 * config/riscv/t-linux-multilib: Likewise.
38142 * config/riscv/t-riscv: Likewise.
38143 * configure.ac: Likewise.
38144 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
38145 Waterman as RISC-V maintainers.
38146 * doc/install.texi: Add RISC-V entries.
38147 * doc/invoke.texi: Add RISC-V options section.
38148 * doc/md.texi: Add RISC-V constraints section.
38149 * configure: Regenerated.
38151 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
38154 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
38155 false values to be constant vectors with all 0 or all 1 bits set.
38156 (vcondu<mode><mode>): Likewise.
38157 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
38159 (fpmask_comparison_operator): Update comment.
38160 (vecint_comparison_operator): New predicate.
38161 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
38162 vector conditionals when the true and false values are constant
38163 vectors with all 0 bits or all 1 bits set.
38165 2017-02-06 Martin Sebor <msebor@redhat.com>
38167 PR tree-optimization/79376
38168 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
38170 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
38172 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
38173 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
38174 to simplify split condition.
38176 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38178 * omp-expand.c (oxpand_omp_atomic_fetch_op,
38179 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
38182 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
38184 PR rtl-optimization/68664
38185 * target.def (can_speculate_insn): New hook.
38186 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
38187 * doc/tm.texi: Regenerate.
38188 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
38189 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
38190 (rs6000_sched_can_speculate_insn): New function.
38192 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38194 PR tree-optimization/79284
38195 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
38196 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
38197 vectorizable_mask_load_store, vectorizable_operation,
38198 vect_is_simple_cond, get_same_sized_vectype): Use it instead
38199 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
38200 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
38201 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
38202 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
38203 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
38204 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
38205 is_gimple_assign (stmt). Replace another such test with
38206 is_gimple_assign (stmt).
38208 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38211 * config/avr/avr.c (rtl-iter.h): Include it.
38212 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
38213 (avr_legitimate_combined_insn): ...and implementation.
38215 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
38217 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
38218 * config/s390/s390.c (s390_const_operand_ok)
38219 (s390_canonicalize_comparison, s390_extract_part)
38220 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
38221 (s390_contiguous_bitmask_p, s390_rtx_costs)
38222 (legitimize_pic_address): Likewise.
38223 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
38224 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
38225 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
38226 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
38227 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
38229 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38231 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
38232 REGNO($0) == REGNO($1).
38234 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38236 * config/s390/linux.h(SIZE_TYPE): Add comment.
38238 2017-02-06 Julian Brown <julian@codesourcery.com>
38239 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38240 Virendra Pathak <virendra.pathak@broadcom.com>
38242 * config/aarch64/aarch64-cores.def: Change the scheduler
38244 * config/aarch64/aarch64.md: Include thunderx2t99.md.
38245 * config/aarch64/thunderx2t99.md: New file.
38247 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38249 * doc/standards.texi (Go Language): Update link to language
38252 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38254 * tree-eh.c (lower_resx): Sanitize profile.
38255 (cleanup_empty_eh_move_lp): Likewise.
38257 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38260 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
38262 * cfgloopmanip.h (loop_version): Update prototype.
38263 * modulo-sched.c (sms_schedule): Update call of loop_version.
38264 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
38265 * tree-parloops.c (gen_parallel_loop): Likewise.
38266 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
38267 * tree-ssa-loop-split.c (split_loop): Likewise.
38268 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
38269 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
38271 2017-02-05 Martin Liska <mliska@suse.cz>
38274 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
38276 (print_operand_address): Initialize a struct to zero.
38278 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38280 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
38281 garbage collector only in textual form.
38283 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38285 * doc/extend.texi (x86 specific memory model extensions for
38286 transactional memory): Simplify a phrase.
38288 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
38291 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
38292 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
38293 (atomic_storedi_1): Likewise.
38295 2017-02-04 Jakub Jelinek <jakub@redhat.com>
38297 PR tree-optimization/79338
38298 * tree-parloops.c (gather_scalar_reductions): Don't call
38299 vect_analyze_loop_form for loop->inner before destroying loop's
38302 2017-02-03 Martin Sebor <msebor@redhat.com>
38304 PR tree-optimization/79327
38305 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
38306 when precision has resulted in leading zeros.
38307 (format_integer): Adjust the likely counter to assume an unknown
38308 argument that may be zero is non-zero.
38310 2017-02-03 Jason Merrill <jason@redhat.com>
38313 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
38314 avoid copying non-taken branch.
38316 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38318 PR tree-optimization/79340
38319 * tree-vect-loop.c (vectorizable_reduction): Release
38320 vec_defs elements after safe_splicing them into other vectors.
38323 PR tree-optimization/79327
38324 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
38325 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
38327 (format_integer): Use wide_int_to_tree instead of build_int_cst
38328 + to_?hwi. If argmin is NULL, just set argmin and argmax to
38329 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
38330 of shortest and longest sequence.
38332 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
38334 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
38335 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
38337 2017-02-03 Walter Lee <walt@tilera.com>
38340 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
38341 after initial stackframe link reg save.
38342 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
38344 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38347 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
38348 wu for stxssp alternative.
38350 2017-02-03 Martin Sebor <msebor@redhat.com>
38352 PR tree-optimization/79352
38353 * gimple-fold.c (get_range_strlen): Add argument.
38354 (get_range_strlen): Change return type to bool.
38355 (get_maxval_strlen): Pass in a dummy argument.
38356 * gimple-fold.h (get_range_strlen): Change return type to bool.
38357 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
38358 * tree.h (array_at_struct_end_p): Add argument.
38359 * tree.c (array_at_struct_end_p): Handle it.
38361 2017-02-03 Martin Liska <mliska@suse.cz>
38364 * multiple_target.c (create_dispatcher_calls): Redirect edge
38365 from a caller of a dispatcher.
38366 (expand_target_clones): Make the clones local.
38367 (ipa_target_clone): Do both target clones and resolvers.
38368 (ipa_dispatcher_calls): Remove the pass.
38369 (pass_dispatcher_calls::gate): Likewise.
38370 (make_pass_dispatcher_calls): Likewise.
38371 * passes.def (pass_target_clone): Put as very first IPA early
38374 2017-02-03 Martin Liska <mliska@suse.cz>
38376 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
38377 in case of a function with ifunc attribute.
38379 2017-02-03 Martin Liska <mliska@suse.cz>
38381 * cgraph.c (cgraph_node::dump): Dump function version info.
38382 * symtab.c (symtab_node::dump_base): Add missing new line.
38384 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38386 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
38387 (ifcombine_ifandif): Use it.
38389 2017-02-03 Martin Liska <mliska@suse.cz>
38391 * doc/invoke.texi: Document default value for
38392 use-after-scope-direct-emission-threshold.
38394 2017-02-03 Martin Liska <mliska@suse.cz>
38396 PR tree-optimization/79339
38397 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
38398 (format_floating): Likewise.
38400 2017-02-03 Martin Liska <mliska@suse.cz>
38403 * ipa-prop.c (ipa_node_params_t::insert): Remove current
38405 (ipa_node_params_t::remove): Likewise.
38406 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
38407 initialization from removed ipa_node_params_t::insert.
38408 (ipa_node_params::~ipa_node_params): Move from removed
38409 ipa_node_params_t::release.
38410 * symbol-summary.h (symbol_summary::m_released): New member.
38411 Do not release a summary twice. Do not allow to call finalizer
38412 for types of a summary that live in GGC memory.
38414 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38416 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
38419 2017-02-02 Martin Sebor <msebor@redhat.com>
38421 PR middle-end/79275
38422 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
38423 (format_string): Tighten up the range of output for non-constant
38424 strings and correct the expected range for wide non-constant strings.
38426 2017-02-02 Martin Sebor <msebor@redhat.com>
38428 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
38430 PR middle-end/32003
38431 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
38433 (-fdump-tree-@var): Add to index and document how to come up
38434 with pass-specific option and dump file names.
38435 (-fdump-passes): Clarify where to look for output.
38437 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38439 PR middle-end/77445
38440 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
38441 statistics of the analyzed path; allow threading for speed when
38442 any of BBs along the path are optimized for speed.
38444 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
38446 PR middle-end/78468
38447 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
38448 settings of the virtual registers.
38451 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
38453 * explow.c (get_dynamic_stack_size): Take known alignment of stack
38454 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
38457 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38459 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
38460 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
38462 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38464 * config/s390/s390.md: Add missing comments with the expanded
38466 * config/s390/vector.md: Likewise.
38467 * config/s390/vx-builtins.md: Likewise.
38469 2017-02-02 Jakub Jelinek <jakub@redhat.com>
38472 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
38473 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
38474 conditions on a single line.
38476 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38478 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38479 __S390_VX__ to __VX__.
38481 2017-02-01 Andrew Pinski <apinski@cavium.com>
38483 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
38484 stmt_info to record_stmt_cost.
38485 (vect_get_known_peeling_cost): Pass stmt_info if known to
38487 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
38488 cpu_vector_cost field into
38489 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
38490 field into vec_int_stmt_cost and vec_fp_stmt_cost.
38491 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
38492 splitting of scalar_stmt_cost and vec_stmt_cost.
38493 (thunderx_vector_cost): Likewise.
38494 (cortexa57_vector_cost): LIkewise.
38495 (exynosm1_vector_cost): Likewise.
38496 (xgene1_vector_cost): Likewise.
38497 (thunderx2t99_vector_cost): Improve after the splitting of the two
38499 (aarch64_builtin_vectorization_cost): Update for the splitting of
38500 scalar_stmt_cost and vec_stmt_cost.
38502 2017-02-01 Torvald Riegel <triegel@redhat.com>
38503 Richard Henderson <rth@redhat.com>
38505 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
38506 conditional on existance of a fast atomic load.
38507 * optabs-query.c (can_atomic_load_p): New function.
38508 * optabs-query.h (can_atomic_load_p): Declare it.
38509 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
38510 no fast atomic load is available for the particular size of access.
38511 (expand_atomic_compare_and_swap): Likewise.
38512 (expand_atomic_load): Likewise.
38513 (expand_atomic_store): Likewise.
38514 (expand_atomic_fetch_op): Likewise.
38515 * testsuite/lib/target-supports.exp
38516 (check_effective_target_sync_int_128): Remove x86 because it provides
38517 no fast atomic load.
38518 (check_effective_target_sync_int_128_runtime): Likewise.
38520 2017-02-01 Richard Biener <rguenther@suse.de>
38522 * graphite.c: Include tree-vectorizer.h for find_loop_location.
38523 (graphite_transform_loops): Provide opt-info for optimized nests.
38524 * tree-parloop.c (parallelize_loops): Provide opt-info for
38525 parallelized loops.
38527 2017-02-01 Richard Biener <rguenther@suse.de>
38529 PR middle-end/79315
38530 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
38531 was not set before.
38533 2017-02-01 Richard Biener <rguenther@suse.de>
38535 PR tree-optimization/71824
38536 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38537 Verify the loops are valid in the merged SESE region.
38538 (scop_detection::can_represent_loop_1): Check analyzing the
38539 evolution of the number of iterations in the region succeeds.
38541 2017-01-31 Ian Lance Taylor <iant@golang.org>
38543 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
38544 REG_ARGS_SIZE note to 32-bit push insns and call insn.
38546 2017-01-31 David Malcolm <dmalcolm@redhat.com>
38548 PR preprocessor/79210
38549 * input.c (get_substring_ranges_for_loc): Replace line_width
38550 assertion with error-handling.
38552 2017-01-31 Richard Biener <rguenther@suse.de>
38554 PR tree-optimization/77318
38555 * graphite-sese-to-poly.c (extract_affine): Fix assert.
38556 (create_pw_aff_from_tree): Take loop parameter.
38557 (add_condition_to_pbb): Pass loop of the condition to
38558 create_pw_aff_from_tree.
38560 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38562 * config/s390/s390.c (s390_asan_shadow_offset): New function.
38563 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
38565 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
38569 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
38571 (convert_int_to_float128): Likewise.
38572 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
38573 (convert_int_to_float128): Likewise.
38574 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
38575 (UNSPEC_IEEE128_CONVERT): Likewise.
38576 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
38577 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
38578 Use local variables for IBM extended format.
38579 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
38580 (fix_trunc<mode>si2_fprs): Likewise.
38581 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
38582 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
38583 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
38584 to know that we can now have integers of all sizes in vector
38586 (fix<uns>_<mode>di2_hw): Likewise.
38587 (float<uns>_<mode>si2_hw): Likewise.
38588 (fix_<mode>si2_hw): Likewise.
38589 (fixuns_<mode>si2_hw): Likewise.
38590 (float<uns>_<mode>di2_hw): Likewise.
38591 (float_<mode>di2_hw): Likewise.
38592 (float_<mode>si2_hw): Likewise.
38593 (floatuns_<mode>di2_hw): Likewise.
38594 (floatuns_<mode>si2_hw): Likewise.
38595 (xscvqp<su>wz_<mode>): Delete, no longer used.
38596 (xscvqp<su>dz_<mode>): Likewise.
38597 (xscv<su>dqp_<mode>): Likewise.
38598 (ieee128_mfvsrd_64bit): Likewise.
38599 (ieee128_mfvsrd_32bit): Likewise.
38600 (ieee128_mfvsrwz): Likewise.
38601 (ieee128_mtvsrw): Likewise.
38602 (ieee128_mtvsrd_64bit): Likewise.
38603 (ieee128_mtvsrd_32bit): Likewise.
38605 2017-01-31 Martin Liska <mliska@suse.cz>
38608 * ipa-prop.c (ipa_free_all_node_params): Call release method
38609 instead of ~sumbol_summary to not to trigger double times
38612 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
38614 PR tree-optimization/71691
38615 * bitmap.h (class auto_bitmap): New.
38616 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
38617 is_maybe_undefined instead of ssa_undefined_value_p.
38619 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38621 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38622 __S390_ARCH_LEVEL__ to __ARCH__.
38624 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38626 PR tree-optimization/79267
38627 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38628 if should_remove_lhs_p is true.
38630 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38633 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38634 (add_alignment_attribute): New.
38635 (base_type_die): Add alignment attribute.
38636 (subrange_type_die): Likewise.
38637 (modified_type_die): Likewise.
38638 (gen_array_type_die): Likewise.
38639 (gen_descr_array_type_die: Likewise.
38640 (gen_enumeration_type_die): Likewise.
38641 (gen_subprogram_die): Likewise.
38642 (gen_variable_die): Likewise.
38643 (gen_field_die): Likewise.
38644 (gen_ptr_to_mbr_type_die): Likewise.
38645 (gen_struct_or_union_type_die): Likewise.
38646 (gen_subroutine_type_die): Likewise.
38647 (gen_typedef_die): Likewise.
38648 (base_type_cmp): Compare alignment attribute.
38650 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38653 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38654 (setb_unsigned) New pattern for setb with CCUNS.
38655 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38656 subfc./subfe sequence to avoid overflow problems. Generate a
38657 shorter sequence with cmpld/setb for power9.
38658 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38659 for generating subfc. instruction.
38660 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38661 now uses this instruction.
38663 2017-01-30 Ian Lance Taylor <iant@google.com>
38666 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38667 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38669 2017-01-30 Martin Sebor <msebor@redhat.com>
38671 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38672 Move constant to the right of a relational operator.
38673 (get_mpfr_format_length, format_character, format_string): Ditto.
38674 (should_warn_p, maybe_warn): Same.
38676 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38678 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38681 * asan.c (get_translation_unit_decl): Remove function.
38682 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38684 2017-01-30 Martin Liska <mliska@suse.cz>
38686 PR gcov-profile/79259
38687 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38688 -fprofile-generate.
38690 2017-01-30 Martin Liska <mliska@suse.cz>
38693 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38694 Initialize variables with NULL value.
38696 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38699 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38701 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38703 2017-01-30 Richard Biener <rguenther@suse.de>
38705 PR tree-optimization/79276
38706 * tree-vrp.c (process_assert_insertions): Properly adjust common
38707 when removing a duplicate.
38709 2017-01-30 Richard Biener <rguenther@suse.de>
38711 PR tree-optimization/79256
38712 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38713 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38715 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38717 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38720 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38721 ("*r<noxa>sbg_<mode>_sll_bitmask")
38722 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38723 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38724 Use contiguous_bitmask_nowrap_operand.
38726 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38729 * config/rs6000/altivec.h (vec_xl): Revise #define.
38730 (vec_xst): Likewise.
38732 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38734 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38736 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38738 PR rtl-optimization/79194
38739 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38740 traps before call to bypass_conditional_jumps.
38742 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38744 PR tree-optimization/71374
38745 * lra-constraints.c (check_conflict_input_operands): New.
38746 (match_reload): Use it.
38748 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38751 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38752 account to calculate conflict_set.
38754 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38756 PR rtl-optimization/78559
38757 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38758 other_insn in combine.
38760 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38762 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38763 uint16_type_node for BT_UINT16.
38765 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38767 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38768 "RTL Tests" to menu.
38769 (GIMPLE Tests): New node.
38770 (RTL Tests): New node.
38772 2017-01-27 Richard Biener <rguenther@suse.de>
38774 PR tree-optimization/79245
38775 * tree-loop-distribution.c (distribute_loop): Apply cost
38776 modeling also to detected patterns.
38778 2017-01-27 Richard Biener <rguenther@suse.de>
38780 PR tree-optimization/71433
38781 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38782 (compare_assert_loc): New function.
38783 (process_assert_insertions): Sort and optimize assert locations
38784 to remove duplicates and push down identical assertions on
38785 edges to their destination block.
38787 2017-01-27 Richard Biener <rguenther@suse.de>
38789 PR tree-optimization/79244
38790 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38791 out SSA names even if abnormal.
38793 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38795 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38796 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38797 instead of MPFR_RNDN.
38799 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38802 * arm.c (arm_option_override): Don't call build_target_option_node
38803 until after doing all option overrides.
38804 (arm_valid_target_attribute_tree): Likewise.
38806 2017-01-27 Martin Liska <mliska@suse.cz>
38808 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38809 for {cd}tors and C++ {cd}tors.
38811 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38813 * config/s390/s390.md ("*setmem_long_and")
38814 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38816 2017-01-26 Martin Sebor <msebor@redhat.com>
38818 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38821 2017-01-26 Martin Sebor <msebor@redhat.com>
38823 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38824 HAVE_DFmode before using XFmode or DFmode.
38825 (parse_directive): Avoid using the z length modifier to avoid
38826 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38828 PR middle-end/78703
38829 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38830 to accept adjustment as an array.
38831 (get_int_range): New function.
38832 (struct directive): Make width and prec arrays.
38833 (directive::set_width, directive::set_precision): Call get_int_range.
38834 (format_integer, format_floating): Handle width and precision ranges.
38835 (format_string, parse_directive): Same.
38837 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38840 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38841 comdat_type_p set, just clone them, but keep the children in the
38845 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38846 which have direct callers with -fvar-tracking-assignments enabled
38848 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38849 inside of type units.
38851 2017-01-26 Martin Sebor <msebor@redhat.com>
38853 PR middle-end/78703
38854 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38856 (struct format_result): Replace number_chars, number_chars_min,
38857 and number_chars_max with a single member of struct result_range.
38859 (format_result::operator+=): Adjust.
38860 (struct fmtresult): Remove bounded. Handle likely and unlikely
38862 (fmtresult::adjust_for_width_or_precision): New function.
38863 (fmtresult:type_max_digits): New function.
38864 (bytes_remaining): Handle likely and unlikely counters.
38865 (min_bytes_remaining): Remove.
38866 (format_percent): Simplify.
38867 (format_integer, format_floating): Set likely and unlikely counters.
38868 (get_string_length, format_character, format_string): Same.
38869 (format_plain, should_warn_p): New function.
38870 (maybe_warn): Call should_warn_p. Update diagnostic messages
38871 and handle those for all directives, including plain strings.
38872 (format_directive): Handle likely and unlikely counters.
38873 Remove unnecessary quoting from diagnostics. Add an informational
38875 (add_bytes): Remove.
38876 (pass_sprintf_length::compute_format_length): Simplify.
38877 (try_substitute_return_value): Handle likely and unlikely counters.
38879 2017-01-26 Carl Love <cel@us.ibm.com>
38881 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
38882 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
38884 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
38887 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
38888 endianess for subregs into account.
38889 * lra-constraints.c (lra_constraints): Do risky transformations
38890 always on the first iteration.
38891 * lra-lives.c (check_pseudos_live_through_calls): Add arg
38892 last_call_used_reg_set.
38893 (process_bb_lives): Define and use last_call_used_reg_set.
38894 * lra.c (lra): Always continue after lra_constraints on the first
38897 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
38899 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
38901 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
38903 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38905 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
38906 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
38907 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
38908 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
38909 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
38910 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
38911 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
38912 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
38913 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
38915 2017-01-26 Marek Polacek <polacek@redhat.com>
38918 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
38919 for the third operand.
38921 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38923 PR middle-end/79236
38924 * omp-low.c (struct omp_context): Add simt_stmt field.
38925 (scan_omp_for): Return omp_context *.
38926 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
38927 context to the _simt_ SIMD stmt.
38928 (lower_omp_for): For combined SIMD with sibling _simt_
38929 SIMD, make sure to use the same decls in _looptemp_
38930 clauses as in the sibling.
38932 2017-01-26 David Sherwood <david.sherwood@arm.com>
38934 PR middle-end/79212
38935 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
38938 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38941 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
38942 emit fld b; fld a; if possible.
38944 * brig-builtins.def: Update copyright years.
38945 * config/arm/arm_acle_builtins.def: Update copyright years.
38947 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
38950 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
38951 constraint instead of o for the stxsd instruction.
38953 2017-01-25 Carl Love <cel@us.ibm.com>
38955 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
38956 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
38958 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
38960 * doc/invoke.texi (C++ Dialect Options): Fix typo.
38962 2017-01-25 Richard Biener <rguenther@suse.de>
38964 PR tree-optimization/69264
38965 * target.def (vector_alignment_reachable): Improve documentation.
38966 * doc/tm.texi: Regenerate.
38967 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
38969 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
38970 earlier changes with respect to TYPE_USER_ALIGN.
38971 (vector_alignment_reachable_p): Likewise. Improve dumping.
38973 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38976 * config/arm/arm.md (xordi3): Force constant operand into a register
38979 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38981 * doc/invoke.texi (-fstore-merging): Correct default optimization
38982 levels at which it is enabled.
38983 (-O): Move -fstore-merging from list to...
38986 2017-01-25 Richard Biener <rguenther@suse.de>
38989 * omp-expand.c: Include debug.h.
38990 (expand_omp_taskreg): Make sure to generate early debug before
38991 outlining anything from a function.
38992 (expand_omp_target): Likewise.
38993 (grid_expand_target_grid_body): Likewise.
38995 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
38998 * asan.c (get_translation_unit_decl): New function.
38999 (asan_add_global): Extract modules file name from globals
39000 TRANSLATION_UNIT_DECL name.
39002 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39005 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39006 for long calls with APCS frame and VFP.
39008 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39010 * cfg.c (original_copy_tables_initialized_p): New function.
39011 * cfg.h (original_copy_tables_initialized_p): New decl.
39012 * cfgrtl.c (relink_block_chain): Guard the call to
39013 free_original_copy_tables with a call to
39014 original_copy_tables_initialized_p.
39015 * cgraph.h (symtab_node::native_rtl_p): New decl.
39016 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39017 (symtab_node::needed_p): Don't assert for early assembly output
39018 for __RTL functions.
39019 (cgraph_node::finalize_function): Set "force_output" for __RTL
39021 (cgraph_node::analyze): Bail out early for __RTL functions.
39022 (analyze_functions): Update assertion to support __RTL functions.
39023 (cgraph_node::expand): Bail out early for __RTL functions.
39024 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39026 * function.h (struct function): Update comment for field
39028 * gimple-expr.c: Include "tree-pass.h".
39029 (gimple_has_body_p): Return false for __RTL functions.
39030 * Makefile.in (OBJS): Add run-rtl-passes.o.
39031 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39033 (gcc::pass_manager::get_clean_slate): New accessor.
39034 * passes.c: Include "insn-addr.h".
39035 (should_skip_pass_p): Add logging. Update logic for running
39036 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39037 property-provider override so it is only done for gimple passes.
39039 (skip_pass): New function.
39040 (execute_one_pass): Call skip_pass when skipping passes.
39041 * read-md.c (md_reader::read_char): Support filtering
39042 the input to a subset of line numbers.
39043 (md_reader::md_reader): Initialize fields
39044 m_first_line and m_last_line.
39045 (md_reader::read_file_fragment): New function.
39046 * read-md.h (md_reader::read_file_fragment): New decl.
39047 (md_reader::m_first_line): New field.
39048 (md_reader::m_last_line): New field.
39049 * read-rtl-function.c (function_reader::create_function): Only
39050 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39051 curr_properties. Set DECL_INITIAL to a dummy block.
39052 (read_rtl_function_body_from_file_range): New function.
39053 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39055 * run-rtl-passes.c: New file.
39056 * run-rtl-passes.h: New file.
39058 2017-01-24 Jeff Law <law@redhat.com>
39060 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39063 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39065 PR tree-optimization/79159
39066 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39067 (record_nonwrapping_iv): Improve boundary using above function if no
39068 value range information.
39070 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39071 Martin Jambor <mjambor@suse.cz>
39073 * brig-builtins.def: New file.
39074 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39075 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39076 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39077 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39078 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39079 * builtin-types.def (BT_INT8): New.
39080 (BT_INT16): Likewise.
39081 (BT_UINT8): Likewise.
39082 (BT_UINT16): Likewise.
39083 (BT_FN_ULONG): Likewise.
39084 (BT_FN_UINT_INT): Likewise.
39085 (BT_FN_UINT_ULONG): Likewise.
39086 (BT_FN_UINT_LONG): Likewise.
39087 (BT_FN_UINT_PTR): Likewise.
39088 (BT_FN_ULONG_PTR): Likewise.
39089 (BT_FN_INT8_FLOAT): Likewise.
39090 (BT_FN_INT16_FLOAT): Likewise.
39091 (BT_FN_UINT32_FLOAT): Likewise.
39092 (BT_FN_UINT16_FLOAT): Likewise.
39093 (BT_FN_UINT8_FLOAT): Likewise.
39094 (BT_FN_UINT64_FLOAT): Likewise.
39095 (BT_FN_UINT16_UINT32): Likewise.
39096 (BT_FN_UINT32_UINT16): Likewise.
39097 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39098 (BT_FN_INT_PTR_INT): Likewise.
39099 (BT_FN_UINT_PTR_UINT): Likewise.
39100 (BT_FN_LONG_PTR_LONG): Likewise.
39101 (BT_FN_ULONG_PTR_ULONG): Likewise.
39102 (BT_FN_VOID_UINT64_UINT64): Likewise.
39103 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39104 (BT_FN_INT8_INT8_INT8): Likewise.
39105 (BT_FN_INT16_INT16_INT16): Likewise.
39106 (BT_FN_INT_INT_INT): Likewise.
39107 (BT_FN_UINT_FLOAT_UINT): Likewise.
39108 (BT_FN_FLOAT_UINT_UINT): Likewise.
39109 (BT_FN_ULONG_UINT_UINT): Likewise.
39110 (BT_FN_ULONG_UINT_PTR): Likewise.
39111 (BT_FN_ULONG_ULONG_ULONG): Likewise.
39112 (BT_FN_UINT_UINT_UINT): Likewise.
39113 (BT_FN_VOID_UINT_PTR): Likewise.
39114 (BT_FN_UINT_UINT_PTR: Likewise.
39115 (BT_FN_UINT32_UINT64_PTR): Likewise.
39116 (BT_FN_INT_INT_UINT_UINT): Likewise.
39117 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
39118 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
39119 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
39120 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
39121 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
39122 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
39123 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
39124 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
39125 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
39126 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
39127 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
39128 * doc/frontends.texi: List BRIG FE.
39129 * doc/install.texi (Testing): Add BRIG tesring requirements.
39130 * doc/invoke.texi (Overall Options): Mention BRIG.
39131 * doc/standards.texi (Standards): Doucment BRIG HSA version.
39133 2017-01-24 Richard Biener <rguenther@suse.de>
39135 PR translation/79208
39136 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
39138 2017-01-24 Martin Jambor <mjambor@suse.cz>
39141 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
39142 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
39143 and known_contexts.
39145 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
39147 PR middle-end/79123
39148 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
39149 casts from signed to unsigned really don't have a range.
39151 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
39153 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
39154 GMP_RNDx for compatiblity.
39156 2017-01-24 Martin Liska <mliska@suse.cz>
39159 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
39160 that would prevent us to call alloca with -1 as argument.
39162 2017-01-24 Jakub Jelinek <jakub@redhat.com>
39164 * dwarf2out.c (output_compilation_unit_header, output_file_names):
39165 Avoid -Wformat-security warning.
39167 2017-01-23 Andrew Pinski <apinski@cavium.com>
39169 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
39172 2017-01-23 Martin Sebor <msebor@redhat.com>
39174 PR middle-end/78703
39175 * gimple-ssa-sprintf.c (warn_level): New global.
39176 (format_integer): Use it here and throughout the rest of the file.
39177 Use the same switch to compute sign as base.
39178 (maybe_warn): New function.
39179 (format_directive): Factor out warnings into maybe_warn.
39180 Add debugging output. Use warn_level.
39181 (add_bytes): Use warn_level.
39182 (pass_sprintf_length::compute_format_length): Add debugging output.
39183 (try_substitute_return_value): Same.
39184 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
39186 PR middle-end/78703
39187 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
39188 (struct fmtresult, format_integer, format_floating): Adjust.
39189 (fmtresult::fmtresult): Set max correctly in two argument ctor.
39190 (get_string_length, format_string,format_directive): Same.
39191 (pass_sprintf_length::compute_format_length): Same.
39192 (try_substitute_return_value): Simplify slightly.
39194 PR middle-end/78703
39195 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
39196 (fmtresult::operator+=): Outlined.
39197 (struct fmtresult): Add ctors.
39198 (struct conversion_spec): Rename...
39199 (struct directive): ...to this. Add and remove data members.
39200 (directive::set_width, directive::set_precision): New functions.
39201 (format_percent): Use fmtresult ctor.
39202 (get_width_and_precision): Remove.
39203 (format_integer): Make naming changes. Avoid computing width and
39205 (format_floating): Same. Adjust indentation.
39206 (format_character, format_none): New functions.
39207 (format_string): Moved character handling to format_character.
39208 (format_directive): Remove arguments, change return type.
39209 (parse_directive): New function.
39210 (pass_sprintf_length::compute_format_length): Move directive
39211 parsing to parse_directive.
39213 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39215 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
39216 (assign_assembler_name_if_needed): ... this.
39217 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
39218 (assign_assembler_name_if_needed): ... this.
39219 (free_lang_data_in_cgraph): Adjust callers.
39220 * cgraphunit.c (cgraph_node::analyze): Likewise.
39221 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
39224 2017-01-23 Richard Biener <rguenther@suse.de>
39226 PR tree-optimization/79088
39227 PR tree-optimization/79188
39228 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
39229 resetting loop bounds after last path deletion. Reset loop
39230 bounds of the target loop, make code match the comments.
39231 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
39232 Make sure loops need no fixups.
39234 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
39236 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
39237 exponent support with double type for first argument.
39238 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
39239 type returned by __builtin_vec_extract_sig,
39240 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
39241 functions from "vector int" to "vector unsigned int" or from
39242 "vector long long int" to "vector unsigned long long int".
39243 Changed type returned by __builtin_vec_extract_exp,
39244 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
39245 functions from "vector int" to "vector unsigned int" or from
39246 "vector long long int" to "vector unsigned long long int".
39247 Changed return type of __builtin_vec_test_data_class,
39248 __builtin_vec_test_data_class_sp, and
39249 __builtin_vec_test_data_class_dp from "vector int" to
39250 "vector bool int" or from "vector long long int" to "vector bool
39251 long long int" and changed second argument type from "unsigned
39252 int" to "int". Added new overloaded function forms "vector float
39253 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
39254 "vector float __builtin_vec_insert_exp_sp (vector float, vector
39255 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
39256 double, vector unsigned long long int)" and "vector double
39257 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
39258 long int)". Changed return type of
39259 __builtin_scalar_test_data_class and
39260 __builtin_scalar_test_data_class_sp and
39261 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
39262 int" and changed second argument from "unsigned int" to "int".
39263 Changed type returned by __builtin_scalar_test_neg,
39264 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
39265 from "int" to "bool int". Added new overloaded function form
39266 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
39267 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
39268 exponent double-precision with floating point first argument.
39269 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
39270 documentation of scalar_test_data_class, scalar_test_neg,
39271 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
39272 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
39273 vec_test_data_class built-in functions to reflect refinements in
39274 their type signatures.
39276 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
39278 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
39280 (aarch64_elf_asm_destructor): Likewise.
39282 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
39284 PR rtl-optimization/78634
39285 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
39286 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
39287 * ifcvt.c (noce_try_cmove): Add missing cost check.
39289 PR rtl-optimization/71724
39290 * combine.c (if_then_else_cond): Look for situations where it is
39291 beneficial to undo the work of one of the recursive calls.
39293 2017-01-23 Bin Cheng <bin.cheng@arm.com>
39295 PR tree-optimization/70754
39296 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
39297 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
39298 combined stmt before it if not NULL.
39299 (combine_chains): Process refs reversely and compute dominance point
39302 2017-01-23 Martin Liska <mliska@suse.cz>
39304 PR tree-optimization/79196
39305 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
39306 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
39308 (strlen_optimize_stmt): Call the renamed function.
39310 2017-01-23 Michael Matz <matz@suse.de>
39312 PR tree-optimization/78384
39313 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
39315 2017-01-23 Richard Biener <rguenther@suse.de>
39317 PR tree-optimization/79186
39318 * tree-vrp.c (register_new_assert_for): Make sure we've seen
39319 both incoming edges before moving an assert.
39321 2017-01-23 Martin Jambor <mjambor@suse.cz>
39323 * ipa-prop.c (load_from_param_1): Removed.
39324 (load_from_unmodified_param): Bits from load_from_param_1 put back
39326 (load_from_param): Removed.
39327 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
39328 with stmt. Reverted back to use of load_from_unmodified_param.
39330 2017-01-23 Martin Jambor <mjambor@suse.cz>
39333 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
39334 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
39335 field a pointer to garbage collected vector, mark lattices and
39336 ipcp_orig_node with GTY((skip)).
39337 (ipa_get_param_count): Adjust to descriptors being a pointer.
39338 (ipa_get_param): Likewise.
39339 (ipa_get_type): Likewise.
39340 (ipa_get_param_move_cost): Likewise.
39341 (ipa_set_param_used): Likewise.
39342 (ipa_get_controlled_uses): Likewise.
39343 (ipa_set_controlled_uses): Likewise.
39344 (ipa_is_param_used): Likewise.
39345 (ipa_node_params_t): Move into garbage collector. New methods insert
39347 (ipa_node_params_sum): Annotate wth GTY(()).
39348 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
39350 (ipa_load_from_parm_agg): Adjust declaration.
39351 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
39352 * ipa-profile.c (ipa_profile): Likewise.
39353 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
39354 (ipa_populate_param_decls): Make descriptors parameter garbage
39356 (ipa_dump_param): Adjust to descriptors being a pointer.
39357 (ipa_alloc_node_params): Likewise.
39358 (ipa_initialize_node_params): Likewise.
39359 (load_from_param_1): Make descriptors parameter garbage collected.
39360 (load_from_unmodified_param): Likewise.
39361 (load_from_param): Likewise.
39362 (ipa_load_from_parm_agg): Likewise.
39363 (ipa_node_params::~ipa_node_params): Removed.
39364 (ipa_free_all_node_params): Remove call to delete operator.
39365 (ipa_node_params_t::insert): New.
39366 (ipa_node_params_t::remove): Likewise.
39367 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
39368 copy known_csts and known_contexts vectors.
39369 (ipa_read_node_info): Adjust to descriptors being a pointer.
39370 (ipcp_modif_dom_walker): Make m_descriptors field garbage
39372 (ipcp_transform_function): Make descriptors variable garbage
39375 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
39377 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
39378 * config/i386/avx512dqintrin.h: Ditto.
39379 * config/i386/avx512fintrin.h: Ditto.
39380 * config/i386/i386.c: Handle new builtins.
39381 * config/i386/i386-builtin.def: Add new builtins.
39382 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
39383 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
39385 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39386 Martin Liska <mliska@suse.cz>
39388 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
39389 * asan.c (asan_expand_poison_ifn): Support stores and use
39390 appropriate ASAN report function.
39391 * internal-fn.c (expand_ASAN_POISON_USE): New function.
39392 * internal-fn.def (ASAN_POISON_USE): Declare.
39393 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
39394 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
39395 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
39396 ASAN_POISON calls w/o LHS.
39397 * tree-ssa.c (execute_update_addresses_taken): Create clobber
39398 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
39399 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
39400 * gimplify.c (asan_poison_variables): Add attribute
39401 use_after_scope_memory to variables that really needs to live
39403 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
39404 having the attribute.
39406 2017-01-23 Martin Liska <mliska@suse.cz>
39408 * asan.c (create_asan_shadow_var): New function.
39409 (asan_expand_poison_ifn): Likewise.
39410 * asan.h (asan_expand_poison_ifn): New declaration.
39411 * internal-fn.c (expand_ASAN_POISON): Likewise.
39412 * internal-fn.def (ASAN_POISON): New builtin.
39413 * sanopt.c (pass_sanopt::execute): Expand
39414 asan_expand_poison_ifn.
39415 * tree-inline.c (copy_decl_for_dup_finish): Make function
39417 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
39418 * tree-ssa.c (is_asan_mark_p): New function.
39419 (execute_update_addresses_taken): Rewrite local variables
39420 (identified just by use-after-scope as addressable) into SSA.
39422 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39424 * doc/install.texi (Specific): opensource.apple.com uses https
39425 now. Remove trailing slash.
39427 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39429 * README.Portability: Remove note on an Irix compatibility issue.
39431 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
39433 * gcov.c (INCLUDE_ALGORITHM): Define.
39434 (INCLUDE_VECTOR): Define.
39435 No longer include <vector> and <algorithm> directly.
39437 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
39439 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
39441 * doc/invoke.texi (Code Gen Options): Ditto.
39443 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
39446 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
39448 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
39450 rtl-optimization/79125
39451 * cprop.c (local_cprop_pass): Handle cases where we make an
39452 unconditional trap.
39454 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
39458 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
39459 read from, for big endian.
39461 2017-01-20 Jiong Wang <jiong.wang@arm.com>
39463 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
39464 register pauth builtins for LP64 only.
39466 2017-01-20 Marek Polacek <polacek@redhat.com>
39469 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
39472 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39474 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
39476 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
39477 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
39478 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
39480 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39483 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
39484 in big-endian mode when they are not a single duplicated value.
39486 2017-01-20 Richard Biener <rguenther@suse.de>
39488 * BASE-VER: Bump to 7.0.1.
39490 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39492 * omp-low.c (omplow_simd_context): New struct. Use it...
39493 (lower_rec_simd_input_clauses): ...here and...
39494 (lower_rec_input_clauses): ...here to hold common data. Adjust all
39495 references to idx, lane, max_vf, is_simt.
39497 2017-01-20 Graham Markall <graham.markall@embecosm.com>
39499 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
39502 2017-01-20 Martin Jambor <mjambor@suse.cz>
39504 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
39505 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
39507 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
39508 (GTFILES): Rename hsa.c to hsa-common.c.
39509 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
39510 * hsa-dump.c: Likewise.
39511 * hsa-gen.c: Likewise.
39512 * hsa-regalloc.c: Likewise.
39513 * ipa-hsa.c: Likewise.
39514 * omp-expand.c: Likewise.
39515 * omp-low.c: Likewise.
39516 * toplev.c: Likewise.
39518 2017-01-20 Marek Polacek <polacek@redhat.com>
39521 * doc/invoke.texi: Document -Wduplicated-branches.
39522 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
39523 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
39524 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
39525 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
39526 return 0 only when not OEP_LEXICOGRAPHIC.
39527 (fold_build_cleanup_point_expr): Use the expression
39528 location when building CLEANUP_POINT_EXPR.
39529 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
39530 * tree.c (add_expr): Handle error_mark_node.
39532 2017-01-20 Martin Liska <mliska@suse.cz>
39535 * tree-profile.c (init_ic_make_global_vars): Do not call
39537 (gimple_init_gcov_profiler): Likewise.
39539 2017-01-20 Martin Liska <mliska@suse.cz>
39542 * cgraph.h (maybe_create_reference): Remove argument and
39544 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
39546 * ipa-cp.c (create_specialized_node): Likewise.
39547 * symtab.c (symtab_node::maybe_create_reference): Handle
39548 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
39550 2017-01-20 Martin Liska <mliska@suse.cz>
39552 * read-rtl-function.c (function_reader::create_function): Use
39553 build_decl instread of build_decl_stat.
39555 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
39557 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
39558 * config/i386/avx512dqintrin.h: Ditto.
39559 * config/i386/avx512fintrin.h: Ditto.
39560 * config/i386/i386-builtin-types.def: Add new types.
39561 * config/i386/i386.c: Handle new types.
39562 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
39563 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
39564 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
39565 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
39566 (__builtin_ia32_kshiftridi): New.
39567 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
39569 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
39573 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
39574 define to rs6000_init_stack_protect_guard.
39575 (rs6000_init_stack_protect_guard): New function.
39577 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39578 Yunqiang Su <yunqiang.su@imgtec.com>
39580 * config.gcc (supported_defaults): Add madd4.
39581 (with_madd4): Add validation.
39582 (all_defaults): Add madd4.
39583 * config/mips/mips.opt (mmadd4): New option.
39584 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39586 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39588 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
39589 (ISA_HAS_FUSED_MADD4): Likewise.
39590 * doc/invoke.texi (-mmadd4): Document the new option.
39591 * doc/install.texi (--with-madd4): Document the new option.
39593 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39595 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
39596 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
39597 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
39598 (aarch64_init_pauth_hint_builtins): New.
39599 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
39600 (aarch64_expand_builtin): Expand new builtins.
39602 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39604 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
39605 * combine-stack-adj.c (no_unhandled_cfa): Handle
39606 REG_CFA_TOGGLE_RA_MANGLE.
39607 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
39608 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
39609 info for return address signing.
39610 (aarch64_expand_epilogue): Likewise.
39612 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39614 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
39615 * config/aarch64/aarch64-protos.h
39616 (aarch64_return_address_signing_enabled): New declaration.
39617 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
39619 (aarch64_expand_prologue): Sign return address before it's pushed onto
39621 (aarch64_expand_epilogue): Authenticate return address fetched from
39623 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39624 (aarch64_attributes): New function attributes for "sign-return-address".
39625 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39626 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39627 ("*do_return"): Generate combined instructions according to key index.
39628 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39629 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39631 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39632 * config/aarch64/aarch64.opt (msign-return-address=): New.
39633 * doc/extend.texi (AArch64 Function Attributes): Documents
39634 "sign-return-address=".
39635 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39637 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39639 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39640 overall option summary.
39642 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39644 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39645 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39646 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39647 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39649 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39651 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39652 -mpower9-minmax by default for -mcpu=power9.
39653 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39654 128-bit floating point.
39656 2017-01-20 Alan Modra <amodra@gmail.com>
39658 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39659 optimizing for size.
39661 2017-01-20 Alan Modra <amodra@gmail.com>
39664 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39665 for strcmp and strncmp from corresponding builtin decl.
39667 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39669 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39670 instead of i386/rtems-64.h.
39671 * config/i386/rtems-64.h: Remove.
39673 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39677 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39679 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39681 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39683 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39684 Change int to HOST_WIDE_INT.
39685 * config/aarch64/aarch64-protos.h
39686 (aarch64_simd_gen_const_vector_dup): Likewise.
39687 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39689 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39691 * langhooks-def.h (lhd_type_for_size): New decl.
39692 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39693 * langhooks.c (lhd_type_for_size): New function, taken from
39696 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39698 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39699 define_bypass for CR latency.
39700 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39701 (power9-alu2): Add define_bypass for CR latency.
39703 (power9-mul): Update insn latency.
39704 (power9-mul-compare): Update insn latency, bypass latency and remove
39707 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39709 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39711 * config/aarch64/aarch64.md
39712 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39713 aarch64_nopcrelative_literal_loads.
39714 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39716 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39718 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39719 TARGET_LOONGSON_3A.
39720 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39722 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39725 * config.gcc (supported_defaults): Add lxc1-sxc1.
39726 (with_lxc1_sxc1): Add validation.
39727 (all_defaults): Add lxc1-sxc1.
39728 * config/mips/mips.opt (mlxc1-sxc1): New option.
39729 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39731 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39732 __mips_no_lxc1_sxc1.
39733 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39734 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39735 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39737 2017-01-19 Richard Biener <rguenther@suse.de>
39739 PR tree-optimization/72488
39740 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39741 sure to restore SSA info.
39742 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39744 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39746 PR rtl-optimization/79121
39747 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39748 of the inner type when shifting an extended value.
39750 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39753 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39754 interposable aliases.
39756 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39759 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39760 Use the evmergelohi instruction.
39761 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39762 (mov_sitf_e500_subreg8_2_be): Likewise.
39763 (mov_sitf_e500_subreg12_2_le): Likewise.
39764 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39765 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39766 (mov_sitf_e500_subreg8_2_le): Likewise.
39767 (mov_sitf_e500_subreg12_2_be): Likewise.
39769 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39771 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39772 attribute from vecsimple to vecperm.
39773 (altivec_vbpermq2): Likewise.
39775 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39778 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39780 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39781 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39782 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39783 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39784 case where N arg is SIZE_MAX.
39785 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39786 (cmpstrsi): Add pattern.
39788 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39790 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39791 __builtin_vec_revb builtins.
39792 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39793 built-in functions to support generation of the ISA 3.0 XXBR<x>
39794 vector byte reverse instructions.
39795 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39796 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39797 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39798 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39799 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39800 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39801 (P9V_BUILTIN_VEC_REVB): Likewise.
39802 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39803 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39804 (p9_xxbrq_v16qi): Likewise.
39805 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39806 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39807 (p9_xxbrh_v8hi): Likewise.
39808 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39809 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39810 vec_revb built-in functions.
39812 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39814 PR rtl-optimization/78952
39815 * config/i386/i386.md (any_extract): New code iterator.
39816 (*insvqi_2): Use any_extract for source operand.
39817 (*insvqi_3): Use any_shiftrt for source operand.
39819 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39821 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39823 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39825 2017-01-18 Matthias Klose <doko@ubuntu.com>
39827 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39829 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39831 * config/rs6000/altivec.h (vec_bperm): Change #define.
39832 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39833 (altivec_vbpermq2): New define_insn.
39834 (altivec_vbpermd): Likewise.
39835 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39836 function interface.
39837 (VBPERMD): Likewise.
39838 (VBPERM): New polymorphic function interface.
39839 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39840 Add entries for P9V_BUILTIN_VEC_VBPERM.
39841 * doc/extend.texi: Add interfaces for vec_bperm.
39843 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39845 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39846 first letter of error messages.
39847 (s390_resolve_overloaded_builtin): Likewise.
39848 * config/s390/s390.c (s390_expand_builtin): Likewise.
39849 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39850 (s390_valid_target_attribute_inner_p): Likewise.
39851 * config/s390/s390.md ("tabort"): Likewise.
39853 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39855 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39856 (ISA_AVOID_DIV_HILO): New macro.
39857 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39858 (ISA_HAS_DDIV): Likewise.
39860 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39862 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39864 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39866 * doc/invoke.texi (fabi-version): Spelling fix.
39868 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39871 * doc/invoke.texi (fabi-version): Mention mangling fix for
39874 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39877 * doc/invoke.texi (fabi-version): Document discriminator mangling.
39879 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
39882 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
39883 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
39885 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
39887 (stack_protect_test): Ditto.
39888 * config/rs6000/rs6000.opt (mstack-protector-guard=,
39889 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
39891 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
39892 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
39893 -mstack-protector-guard-offset=.
39894 (RS/6000 and PowerPC Options): Ditto.
39896 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39898 * config/i386/i386.h (MASK_CLASS_P): New define.
39899 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
39900 there are no registers from different register sets also when
39901 mask registers are used. Update function comment.
39902 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
39903 to (*k/*r) and (*k/*km) alternatives.
39905 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
39907 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
39908 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
39909 (EH_RETURN_HANDLER_RTX): New define.
39910 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
39911 Force frame pointer in EH return functions.
39912 (aarch64_expand_epilogue): Add barrier for eh_return.
39913 (aarch64_final_eh_return_addr): Remove.
39914 (aarch64_eh_return_handler_rtx): New function.
39915 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
39917 (aarch64_eh_return_handler_rtx): New prototype.
39919 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39921 * config/rs6000/altivec.h (vec_rlmi): New #define.
39922 (vec_vrlnm): Likewise.
39923 (vec_rlnm): Likewise.
39924 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
39925 (UNSPEC_VRLNM): Likewise.
39926 (VIlong): New mode iterator.
39927 (altivec_vrl<VI_char>mi): New define_insn.
39928 (altivec_vrl<VI_char>nm): Likewise.
39929 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
39931 (VRLDNM): Likewise.
39932 (RLNM): New polymorphic function entry.
39933 (VRLWMI): New monomorphic function entry.
39934 (VRLDMI): Likewise.
39935 (RLMI): New polymorphic function entry.
39936 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
39937 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
39938 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
39941 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39944 * dwarf2out.c (field_byte_offset): Restore the
39945 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
39946 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
39947 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
39950 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39953 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
39955 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39958 * dwarf2out.c (add_data_member_location_attribute): For constant
39959 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
39960 instead of DW_AT_data_member_location, DW_AT_bit_offset and
39961 DW_AT_byte_size attributes.
39963 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39965 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
39966 after forcing to constant memory when the code model is medium.
39968 2017-01-17 Julia Koval <julia.koval@intel.com>
39971 * config/i386/avx512fintrin.h
39972 (_mm512_i32gather_ps): Change __addr type to void const*.
39973 (_mm512_mask_i32gather_ps): Ditto.
39974 (_mm512_i32gather_pd): Ditto.
39975 (_mm512_mask_i32gather_pd): Ditto.
39976 (_mm512_i64gather_ps): Ditto.
39977 (_mm512_mask_i64gather_ps): Ditto.
39978 (_mm512_i64gather_pd): Ditto.
39979 (_mm512_mask_i64gather_pd): Ditto.
39980 (_mm512_i32gather_epi32): Ditto.
39981 (_mm512_mask_i32gather_epi32): Ditto.
39982 (_mm512_i32gather_epi64): Ditto.
39983 (_mm512_mask_i32gather_epi64): Ditto.
39984 (_mm512_i64gather_epi32): Ditto.
39985 (_mm512_mask_i64gather_epi32): Ditto.
39986 (_mm512_i64gather_epi64): Ditto.
39987 (_mm512_mask_i64gather_epi64): Ditto.
39988 (_mm512_i32scatter_ps): Change __addr type to void*.
39989 (_mm512_mask_i32scatter_ps): Ditto.
39990 (_mm512_i32scatter_pd): Ditto.
39991 (_mm512_mask_i32scatter_pd): Ditto.
39992 (_mm512_i64scatter_ps): Ditto.
39993 (_mm512_mask_i64scatter_ps): Ditto.
39994 (_mm512_i64scatter_pd): Ditto.
39995 (_mm512_mask_i64scatter_pd): Ditto.
39996 (_mm512_i32scatter_epi32): Ditto.
39997 (_mm512_mask_i32scatter_epi32): Ditto.
39998 (_mm512_i32scatter_epi64): Ditto.
39999 (_mm512_mask_i32scatter_epi64): Ditto.
40000 (_mm512_i64scatter_epi32): Ditto.
40001 (_mm512_mask_i64scatter_epi32): Ditto.
40002 (_mm512_i64scatter_epi64): Ditto.
40003 (_mm512_mask_i64scatter_epi64): Ditto.
40004 * config/i386/avx512pfintrin.h
40005 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40006 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40007 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40008 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40009 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40010 (_mm512_prefetch_i32scatter_ps): Ditto.
40011 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40012 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40013 (_mm512_prefetch_i64scatter_pd): Ditto.
40014 (_mm512_prefetch_i64scatter_ps): Ditto.
40015 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40016 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40017 * config/i386/avx512vlintrin.h
40018 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40019 (_mm_mmask_i32gather_ps): Ditto.
40020 (_mm256_mmask_i32gather_pd): Ditto.
40021 (_mm_mmask_i32gather_pd): Ditto.
40022 (_mm256_mmask_i64gather_ps): Ditto.
40023 (_mm_mmask_i64gather_ps): Ditto.
40024 (_mm256_mmask_i64gather_pd): Ditto.
40025 (_mm_mmask_i64gather_pd): Ditto.
40026 (_mm256_mmask_i32gather_epi32): Ditto.
40027 (_mm_mmask_i32gather_epi32): Ditto.
40028 (_mm256_mmask_i32gather_epi64): Ditto.
40029 (_mm_mmask_i32gather_epi64): Ditto.
40030 (_mm256_mmask_i64gather_epi32): Ditto.
40031 (_mm_mmask_i64gather_epi32): Ditto.
40032 (_mm256_mmask_i64gather_epi64): Ditto.
40033 (_mm_mmask_i64gather_epi64): Ditto.
40034 (_mm256_i32scatter_ps): Change __addr type to void*.
40035 (_mm256_mask_i32scatter_ps): Ditto.
40036 (_mm_i32scatter_ps): Ditto.
40037 (_mm_mask_i32scatter_ps): Ditto.
40038 (_mm256_i32scatter_pd): Ditto.
40039 (_mm256_mask_i32scatter_pd): Ditto.
40040 (_mm_i32scatter_pd): Ditto.
40041 (_mm_mask_i32scatter_pd): Ditto.
40042 (_mm256_i64scatter_ps): Ditto.
40043 (_mm256_mask_i64scatter_ps): Ditto.
40044 (_mm_i64scatter_ps): Ditto.
40045 (_mm_mask_i64scatter_ps): Ditto.
40046 (_mm256_i64scatter_pd): Ditto.
40047 (_mm256_mask_i64scatter_pd): Ditto.
40048 (_mm_i64scatter_pd): Ditto.
40049 (_mm_mask_i64scatter_pd): Ditto.
40050 (_mm256_i32scatter_epi32): Ditto.
40051 (_mm256_mask_i32scatter_epi32): Ditto.
40052 (_mm_i32scatter_epi32): Ditto.
40053 (_mm_mask_i32scatter_epi32): Ditto.
40054 (_mm256_i32scatter_epi64): Ditto.
40055 (_mm256_mask_i32scatter_epi64): Ditto.
40056 (_mm_i32scatter_epi64): Ditto.
40057 (_mm_mask_i32scatter_epi64): Ditto.
40058 (_mm256_i64scatter_epi32): Ditto.
40059 (_mm256_mask_i64scatter_epi32): Ditto.
40060 (_mm_i64scatter_epi32): Ditto.
40061 (_mm_mask_i64scatter_epi32): Ditto.
40062 (_mm256_i64scatter_epi64): Ditto.
40063 (_mm256_mask_i64scatter_epi64): Ditto.
40064 (_mm_i64scatter_epi64): Ditto.
40065 (_mm_mask_i64scatter_epi64): Ditto.
40066 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40067 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40068 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40069 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40070 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40071 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40072 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40073 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40074 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40075 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40076 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40077 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40078 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40079 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40080 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40081 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40082 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40083 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40084 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40085 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40086 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40087 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40088 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40089 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40090 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40091 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40092 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40093 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40094 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40095 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40096 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40097 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40098 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40099 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40100 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40101 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40102 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40103 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40104 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40105 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40106 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40107 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40108 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40109 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40110 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
40111 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
40112 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
40113 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
40114 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
40115 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
40116 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
40117 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
40118 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
40119 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
40120 definitions accordingly.
40122 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
40123 Kuan-Lin Chen <kuanlinchentw@gmail.com>
40126 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
40129 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
40132 * ira-conflicts.c (ira_build_conflicts): Update total conflict
40133 hard regs for inner regno.
40135 2017-01-17 Martin Liska <mliska@suse.cz>
40138 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
40139 assumption and add comment.
40141 2017-01-17 Nathan Sidwell <nathan@acm.org>
40143 * ipa-visibility.c (localize_node): New function, broken out of ...
40144 (function_and_variable_visibility): ... here. Call it.
40146 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40148 PR middle-end/77445
40149 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
40150 correctly set frequency of oudgoing edge.
40151 (duplicate_thread_path): Fix profile updating.
40153 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40156 * configure.ac: Add GCC_BASE_VER.
40157 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
40158 version from BASE-VER file.
40159 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
40160 (gcc.o): Depend on $(BASEVER).
40161 * common.opt (dumpfullversion): New option.
40162 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
40163 * doc/invoke.texi: Document -dumpfullversion.
40164 * doc/install.texi: Document --with-gcc-major-version-only.
40165 * configure: Regenerated.
40167 2017-01-17 Richard Biener <rguenther@suse.de>
40169 PR tree-optimization/71433
40170 * tree-vrp.c (register_new_assert_for): Merge same asserts
40171 on all incoming edges.
40172 (process_assert_insertions_for): Handle insertions at the
40175 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
40177 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
40178 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
40180 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
40183 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
40186 2017-01-17 Alan Modra <amodra@gmail.com>
40189 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
40190 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
40191 symbolic stack limit when pic.
40193 2017-01-16 Martin Sebor <msebor@redhat.com>
40195 PR tree-optimization/78608
40196 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
40198 2017-01-16 Jeff Law <law@redhat.com>
40201 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
40202 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
40203 for several include directories that may be relative to sysroot.
40204 * config/i386/x-mingw32 (gplus_includedir): Define.
40205 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
40206 (native_system_includedir): Likewise.
40207 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
40208 override if TARGET_SYSTEM_ROOT is defined.
40209 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
40211 PR tree-optimization/79090
40212 PR tree-optimization/33562
40213 PR tree-optimization/61912
40214 PR tree-optimization/77485
40215 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
40216 and computed trims into the dump file.
40218 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40220 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
40222 2017-01-16 Jakub Jelinek <jakub@redhat.com>
40225 * gimplify.c (gimplify_init_constructor): If want_value and
40226 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
40230 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
40231 sequence. Formatting fixes.
40232 (doloop_optimize): Formatting fixes.
40235 * gcc.c (debug_level_greater_than_spec_func): New function.
40236 (static_spec_functions): Add debug-level-gt spec function.
40237 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
40239 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40240 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40241 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
40242 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
40243 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
40244 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
40246 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
40248 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
40249 QImode fixups to general and mask registers only.
40251 2017-01-16 Carl Love <cel@us.ibm.com>
40253 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
40254 for built-in functions
40255 vector signed char vec_nabs (vector signed char)
40256 vector signed short vec_nabs (vector signed short)
40257 vector signed int vec_nabs (vector signed int)
40258 vector signed long long vec_nabs (vector signed long long)
40259 vector float vec_nabs (vector float)
40260 vector double vec_nabs (vector double)
40261 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
40263 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
40264 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
40265 * doc/extend.texi: Update the documentation file for the new built-in
40268 2017-01-16 Martin Sebor <msebor@redhat.com>
40270 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
40273 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40275 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
40276 UNSPEC_VSX__XXSPLTD to require special splat handling.
40278 2017-01-16 David Malcolm <dmalcolm@redhat.com>
40281 * system.h: Poison strndup.
40283 2017-01-16 Alan Modra <amodra@gmail.com>
40286 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
40289 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
40291 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
40293 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
40295 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
40296 call recog here. Assert that INSN_CODE (insn) is non-negative.
40298 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40301 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
40303 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
40304 in the currently scheduled RTL fragment.
40306 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40308 PR rtl-optimization/78751
40309 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
40312 2017-01-14 Jeff Law <law@redhat.com>
40314 PR tree-optimization/79090
40315 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
40316 variable length stores.
40317 (compute_trims): Delete dead assignment to *trim_tail.
40318 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
40321 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
40323 PR rtl-optimization/78626
40324 PR rtl-optimization/78727
40325 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
40326 of a block, and split such blocks after everything else is finished.
40328 2017-01-14 Alan Modra <amodra@gmail.com>
40331 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
40332 target legitimate_combined_insn.
40333 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
40334 (rs6000_legitimate_combined_insn): New function.
40335 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
40337 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
40338 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
40339 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
40341 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
40343 * doc/frontends.texi (G++ and GCC): Remove references to Java.
40345 2017-01-13 Jeff Law <law@redhat.com>
40347 PR tree-optimization/33562
40348 PR tree-optimization/61912
40349 PR tree-optimization/77485
40350 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
40352 (delete_dead_assignment): Likewise.
40353 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
40354 statement to delete_dead_call and delete_dead_assignment.
40356 2017-01-13 David Malcolm <dmalcolm@redhat.com>
40359 * substring-locations.c (format_warning_va): Strengthen case 1 so
40360 that both endpoints of the substring must be within the format
40361 range for just the substring to be printed.
40363 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
40365 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
40366 * config/i386/i386.c (ix86_target_string): Add missing options
40367 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
40368 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
40369 flags_other and ix86_target_other to flags2_other. Display unknown
40371 (ix86_valid_target_attribute_inner_p): Add missing options and
40372 reorder options by implied ISAs, as in ix86_target_string.
40374 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40376 * hash-table.h (hash_table::too_empty_p): New function.
40377 (hash_table::expand): Use it.
40378 (hash_table::traverse): Likewise.
40379 (hash_table::empty_slot): Use sizeof (value_type) instead of
40380 sizeof (PTR) to convert bytes to elements. Shrink the table
40381 if the current size is excessive for the current number of
40384 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40386 * ira-costs.c (record_reg_classes): Break from the inner loop
40387 early once alt_fail is known to be true. Update outer loop
40388 handling accordingly.
40390 2017-01-13 Jeff Law <law@redhat.com>
40392 * tree-ssa-dse.c (decrement_count): New function.
40393 (increment_start_addr, maybe_trim_memstar_call): Likewise.
40394 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
40395 when we know the partially dead statement is a mem* function.
40397 PR tree-optimization/61912
40398 PR tree-optimization/77485
40399 * tree-ssa-dse.c: Include expr.h.
40400 (maybe_trim_constructor_store): New function.
40401 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
40403 PR tree-optimization/33562
40404 PR tree-optimization/61912
40405 PR tree-optimization/77485
40406 * doc/invoke.texi: Document new dse-max-object-size param.
40407 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
40408 * tree-ssa-dse.c: Include params.h.
40409 (dse_store_status): New enum.
40410 (initialize_ao_ref_for_dse): New, partially extracted from
40412 (valid_ao_ref_for_dse, normalize_ref): New.
40413 (setup_live_bytes_from_ref, compute_trims): Likewise.
40414 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
40415 (maybe_trim_partially_dead_store): Likewise.
40416 (maybe_trim_complex_store): Likewise.
40417 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
40418 Track what bytes live from the original store. Return tri-state
40419 for dead, partially dead or live.
40420 (dse_dom_walker): Add constructor, destructor and new private members.
40421 (delete_dead_call, delete_dead_assignment): New extracted from
40423 (dse_optimize_stmt): Make a member of dse_dom_walker.
40424 Use initialize_ao_ref_for_dse.
40426 PR tree-optimization/33562
40427 PR tree-optimization/61912
40428 PR tree-optimization/77485
40429 * sbitmap.h (bitmap_count_bits): Prototype.
40430 (bitmap_clear_range, bitmap_set_range): Likewise.
40431 * sbitmap.c (bitmap_clear_range): New function.
40432 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
40434 2017-01-13 Martin Liska <mliska@suse.cz>
40437 * function.c (set_cfun): Add new argument force.
40438 * function.h (set_cfun): Likewise.
40439 * ipa-inline-transform.c (inline_call): Use the function when
40440 strict alising from is dropped for function we inline to.
40442 2017-01-13 Richard Biener <rguenther@suse.de>
40444 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
40445 for dumping GIMPLE INTEGER_CSTs.
40447 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
40449 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
40450 to 201112L since C++17.
40452 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
40455 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
40456 if -fsanitize=kernel-address is present.
40458 2017-01-13 Richard Biener <rguenther@suse.de>
40460 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
40461 as _Literal ( type ) number in case usual suffixes do not
40462 preserve all information.
40464 2017-01-13 Richard Biener <rguenther@suse.de>
40466 PR tree-optimization/77283
40467 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
40468 and ssa-iterators.h.
40469 (is_feasible_trace): Implement a cost model based on joiner
40472 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
40475 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
40476 char or short to __float128/_Float128 directly.
40478 2017-01-12 Martin Sebor <msebor@redhat.com>
40480 to -Wformat-overflow.
40481 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
40482 (min_bytes_remaining): Same.
40483 (get_string_length): Same.
40484 (format_string): Same.
40485 (format_directive): Same.
40487 (pass_sprintf_length::handle_gimple_call): Same.
40489 2017-01-12 Jakub Jelinek <jakub@redhat.com>
40491 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
40492 info.nowrite calls with no lhs that can't throw. Return bool
40493 whether gsi_remove has been called or not.
40494 (pass_sprintf_length::handle_gimple_call): Return bool whether
40495 try_substitute_return_value called gsi_remove. Formatting fix.
40496 (pass_sprintf_length::execute): Don't use gsi_remove if
40497 handle_gimple_call returned true.
40500 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
40501 be removed due to side-effects, don't remove following barrier nor
40502 turn the successor edge into fallthru edge.
40504 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40507 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
40508 element-reversing loads and stores as not swappable.
40510 2017-01-12 Nathan Sidwell <nathan@acm.org>
40511 Nicolai Stange <nicstange@gmail.com>
40513 * combine.c (try_combine): Don't ignore result of overlap checking
40514 loop. Combine overlap & asm check into single loop.
40516 2017-01-12 Richard Biener <rguenther@suse.de>
40518 * tree-pretty-print.c (dump_generic_node): Provide -gimple
40519 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
40521 2017-01-12 Richard Biener <rguenther@suse.de>
40523 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
40524 and TS_TARGET_OPTION directly derive from TS_BASE.
40525 * tree-core.h (tree_optimization_option): Derive from tree_base.
40526 (tree_target_option): Likewise.
40528 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
40530 * config/i386/i386.c (memory_address_length): Increase len
40531 only when rip_relative_addr_p returns false.
40533 2017-01-11 Julia Koval <julia.koval@intel.com>
40535 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
40536 (OPTION_MASK_ISA_SGX_SET): New.
40537 (ix86_handle_option): Handle OPT_msgx.
40538 * config.gcc: Added sgxintrin.h.
40539 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
40540 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
40541 * config/i386/i386.c (ix86_target_string): Add -msgx.
40543 (ix86_option_override_internal): Handle new options.
40544 (ix86_valid_target_attribute_inner_p): Add sgx.
40545 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
40546 * config/i386/i386.opt: Add msgx.
40547 * config/i386/sgxintrin.h: New file.
40548 * config/i386/x86intrin.h: Add sgxintrin.h.
40550 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40553 * fold-const.c (maybe_nonzero_address): Return 1 for function
40555 (tree_single_nonzero_warnv_p): Don't handle function local objects
40559 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
40562 2017-01-11 David Malcolm <dmalcolm@redhat.com>
40565 * opts.c: Include "spellcheck.h"
40566 (struct string_fragment): New struct.
40567 (struct edit_distance_traits<const string_fragment &>): New
40569 (get_closest_sanitizer_option): New function.
40570 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
40572 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40574 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
40576 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
40577 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
40578 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
40579 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
40580 for initial die_offset if dwarf_split_debug_info.
40581 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
40582 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
40584 (output_skeleton_debug_sections): Formatting fix. Use
40585 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
40586 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
40588 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
40590 * config/arm/cortex-a53.md: Add bypasses for
40591 cortex_a53_r2f_cvt.
40592 (cortex_a53_r2f): Only use for transfers.
40593 (cortex_a53_f2r): Likewise.
40594 (cortex_a53_r2f_cvt): Add reservation for conversions.
40595 (cortex_a53_f2r_cvt): Likewise.
40597 2017-01-11 Tamar Christina <tamar.christina@arm.com>
40599 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
40600 to all inlined functions, change static to extern.
40602 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
40605 * config/arm/arm.c (legitimize_pic_address): Handle reference to
40607 (arm_assemble_integer): Likewise.
40609 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
40611 * config.gcc: Use new awk script to check CPU, FPU and architecture
40612 parameters for --with-... options.
40613 * config/arm/parsecpu.awk: New file
40614 * config/arm/arm-cpus.in: New file.
40615 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
40617 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
40619 * config/arm/t-arm: Update dependency rules.
40620 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
40621 of processing .def files.
40622 * config/arm/genopt.sh: Deleted.
40623 * config/arm/gentune.sh: Deleted.
40624 * config/arm/arm-cores.def: Deleted.
40625 * config/arm/arm-arches.def: Deleted.
40626 * config/arm/arm-fpus.def: Deleted.
40627 * config/arm/arm-tune.md: Regenerated.
40628 * config/arm/arm-tables.opt: Regenerated.
40629 * config/arm/arm-cpu.h: New generated file.
40630 * config/arm/arm-cpu-data.h: New generated file.
40631 * config/arm/arm-cpu-cdata.h: New generated file.
40633 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40636 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40638 (input_varpool_node): Unpack dynamically_initialized bit.
40640 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40642 PR rtl-optimization/79032
40643 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40644 the alignment of the adjusted memory reference against that of MODE,
40645 instead of the alignment of the original memory reference.
40647 2017-01-11 Martin Jambor <mjambor@suse.cz>
40649 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40651 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40652 decorated functions.
40654 2017-01-11 Richard Biener <rguenther@suse.de>
40656 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40657 set range/nonnull info for PHI results. Do not set it on
40658 stmts marked for removal.
40660 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40662 * expr.c (store_field): In the bitfield case, fetch the return value
40663 from the registers before applying a single big-endian adjustment.
40664 Always do a final load for a BLKmode value not larger than a word.
40666 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40669 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40670 that we correctly handle column numbers greater than
40671 LINE_MAP_MAX_COLUMN_NUMBER.
40673 2017-01-10 Martin Sebor <msebor@redhat.com>
40675 PR middle-end/78245
40676 * gimple-ssa-sprintf.c (get_destination_size): Call
40677 {init,fini}object_sizes.
40678 * tree-object-size.c (addr_object_size): Adjust.
40679 (pass_through_call): Adjust.
40680 (pass_object_sizes::execute): Adjust.
40681 * tree-object-size.h (fini_object_sizes): Declare.
40683 2017-01-10 Martin Sebor <msebor@redhat.com>
40685 PR tree-optimization/78775
40686 * builtins.c (get_size_range): Move...
40687 * calls.c: ...to here.
40688 (alloc_max_size): Accept zero argument.
40689 (operand_signed_p): Remove.
40690 (maybe_warn_alloc_args_overflow): Call get_size_range.
40691 * calls.h (get_size_range): Declare.
40693 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40695 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40696 from TI's devices.csv file as of September 2016.
40697 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40699 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40701 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40702 * doc/invoke.texi: Likewise.
40703 * doc/md.texi: Likewise.
40704 * doc/objc.texi: Likewise.
40706 2017-01-10 Joshua Conner <joshconner@google.com>
40708 * config/arm/fuchsia-elf.h: New file.
40709 * config/fuchsia.h: New file.
40710 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40711 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40713 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40715 2016-01-10 Richard Biener <rguenther@suse.de>
40717 PR tree-optimization/79034
40718 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40719 Propagate out degenerate PHIs in the joiner.
40721 2017-01-10 Martin Liska <mliska@suse.cz>
40723 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40724 (sort_congruence_classes_by_decl_uid): Likewise.
40725 (sort_congruence_class_groups_by_decl_uid): Likewise.
40726 (sem_item_optimizer::merge_classes): Sort class, groups in these
40727 classes and members in the groups by DECL_UID of declarations.
40728 This would make merge operations stable.
40730 2017-01-10 Martin Liska <mliska@suse.cz>
40732 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40733 usage of m_classes_vec.
40734 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40735 (sem_item_optimizer::get_group_by_hash): Likewise.
40736 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40737 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40738 (sem_item_optimizer::verify_classes): Likewise.
40739 (sem_item_optimizer::process_cong_reduction): Likewise.
40740 (sem_item_optimizer::dump_cong_classes): Likewise.
40741 (sem_item_optimizer::merge_classes): Likewise.
40742 * ipa-icf.h (congruence_class_hash): Rename from
40743 congruence_class_group_hash. Remove declaration of m_classes_vec.
40745 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40747 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40748 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40749 * config.gcc: Add avx512vpopcntdqintrin.h.
40750 * config/i386/avx512vpopcntdqintrin.h: New.
40751 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40752 * config/i386/i386-builtin-types.def: Add new types.
40753 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40754 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40755 __builtin_ia32_vpopcountq_v8di_mask): New.
40756 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40757 __AVX512VPOPCNTDQ__.
40758 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40759 (PTA_AVX512VPOPCNTDQ): Define.
40760 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40761 TARGET_AVX512VPOPCNTDQ_P): Define.
40762 * config/i386/i386.opt: Add mavx512vpopcntdq.
40763 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40764 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40766 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40768 PR middle-end/77484
40769 * predict.def (PRED_CALL): Set to 67.
40771 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40773 * expr.c (store_field): In the bitfield case, if the value comes from
40774 a function call and is of an aggregate type returned in registers, do
40775 not modify the field mode; extract the value in all cases if the mode
40776 is BLKmode and the size is not larger than a word.
40778 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40781 * config/i386/cpuid.h: Fix undefined behavior.
40783 2017-01-04 Jeff Law <law@redhat.com>
40785 PR tree-optimization/79007
40786 PR tree-optimization/67955
40787 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40788 conservative for pt.null when flag_non_call_exceptions is on.
40790 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40792 PR translation/79019
40793 PR translation/79020
40794 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40795 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40796 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40798 * config/avr/avr.opt (maccumulate-args): Likewise.
40799 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40800 * common.opt (freport-bug): Likewise.
40801 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40802 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40803 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40804 translatable string.
40805 * config/i386/i386.c (function_value_32): Likewise.
40806 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40807 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40809 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40810 * common/config/msp430/msp430-common.c (msp430_handle_option):
40812 * symtab.c (symtab_node::verify_base): Likewise.
40813 * opts.c (set_debug_level): Likewise.
40814 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40815 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40816 missing whitespace to translatable strings.
40817 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40818 * config/sh/superh.h: Likewise.
40819 * config/i386/xopintrin.h: Likewise.
40820 * config/i386/znver1.md: Likewise.
40821 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40822 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40823 * double-int.h (struct double_int): Likewise.
40824 * double-int.c (div_and_round_double): Likewise.
40825 * wide-int.cc: Likewise.
40826 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40827 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40828 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40829 (crossjumps_occurred): ... this.
40830 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40833 PR tree-optimization/78899
40834 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40835 returning bool return struct loop *, NULL for failure and the new
40837 (versionable_outer_loop_p): Don't version outer loop if it has
40838 dont_vectorized bit set.
40839 (tree_if_conversion): When versioning outer loop, ensure
40840 tree_if_conversion is performed also on the inner loop of the
40841 non-vectorizable outer loop copy.
40842 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40843 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40844 prevent vectorization of it.
40845 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40846 the outer loop vectorization of the non-scalar version is attempted
40847 before vectorization of the inner loop in scalar version. If
40848 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40849 vectorization of its inner loop.
40850 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40851 has 2 inner loops, rename also on edges from bb whose single pred
40852 is outer_loop->header. Fix typo in function comment.
40854 2017-01-09 Martin Sebor <msebor@redhat.com>
40857 * asan.c (asan_emit_stack_protection): Increase local buffer size
40858 to avoid snprintf truncation warning.
40860 2017-01-09 Andrew Pinski <apinski@cavium.com>
40862 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40863 to reference thunderx2t99 for the tuning structure
40864 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40866 (thunderx2t99_extra_costs): This.
40867 * config/aarch64/aarch64-tune.md: Regenerate.
40868 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
40869 (vulcan_addrcost_table): This.
40870 (vulcan_regmove_cost): Rename to ...
40871 (thunderx2t99_regmove_cost): This.
40872 (vulcan_vector_cost): Rename to ...
40873 (thunderx2t99_vector_cost): this.
40874 (vulcan_branch_cost): Rename to ...
40875 (thunderx2t99_branch_cost): This.
40876 (vulcan_tunings): Rename to ...
40877 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
40878 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
40880 2017-01-09 Martin Jambor <mjambor@suse.cz>
40884 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
40885 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
40886 (propagate_vr_accross_jump_function): Use the above function for all
40887 value range computations for pass-through jump functions and type
40888 converasion from explicit value range values.
40889 (ipcp_propagate_stage): Do not attempt to deduce types of formal
40890 parameters from TYPE_ARG_TYPES.
40891 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
40892 (ipa_write_node_info): Stream type of the actual argument.
40893 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
40895 2017-01-09 Martin Liska <mliska@suse.cz>
40898 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
40899 (lookup_compiler): Do not show error message with have_E.
40901 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40903 PR tree-optimization/78938
40904 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
40905 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
40906 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
40907 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
40910 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40912 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
40915 2017-01-09 Richard Biener <rguenther@suse.de>
40917 PR tree-optimization/78997
40918 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
40919 name condition properly.
40921 2017-01-09 Richard Biener <rguenther@suse.de>
40924 * dwarf2out.c (is_cxx): New overload with context.
40925 (is_naming_typedef_decl): Use it.
40927 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40929 * invoke.texi (Option Summary): Correct spacing in option lists
40930 and add line breaks to fix over-long lines.
40932 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40934 PR middle-end/17660
40936 * extend.texi (Common Variable Attributes): Add xref to GCC
40937 Internals manual to explain mode attribute keywords.
40939 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40942 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
40943 and Preprocessor Options.
40944 (Options for Linking): Document -pthread here....
40945 (RS/6000 and PowerPC Options): ...not here.
40946 (Solaris 2 Options): ...or here.
40947 * doc/cppopts.texi: Document -pthread.
40949 2017-01-08 Martin Sebor <msebor@redhat.com>
40951 PR middle-end/77708
40952 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
40953 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
40954 New member functions.
40955 (format_directive): Used them.
40957 (pass_sprintf_length::handle_gimple_call): Same.
40958 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
40959 to avoid truncation for any argument.
40960 (extract_affine_mul): Same.
40961 * tree.c (get_file_function_name): Same.
40963 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40965 PR middle-end/77484
40966 * predict.def (PRED_INDIR_CALL): Set to 86.
40968 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40970 PR preprocessor/54124
40971 * doc/cppopts.texi: Reformat -d subtable to list the full name
40972 of the options. Add cross-reference to the docs for the general
40973 compiler -d options.
40974 * doc/invoke.texi (Developer Options): Add cross-reference to the
40975 preprocessor-specific -d option documentation.
40977 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40979 PR preprocessor/13498
40980 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
40981 redudant material, and reflect new command-line options.
40982 (System Headers): Likewise.
40984 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40986 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
40987 -isystem, and -idirafter. Copy-edit.
40988 * doc/cppopts.texi: Copy-edit. Remove contradiction about
40989 default for -ftrack-macro-expansion. Delete obsolete and
40990 badly-formatted implementation details about -fdebug-cpp output.
40991 * doc/cppwarnopts.texi: Copy-edit.
40993 2017-01-07 David Malcolm <dmalcolm@redhat.com>
40996 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40997 that the transition from a max line width >= 1<<10 to narrower
40998 lines works correctly.
41000 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
41002 * doc/options.texi (PerFunction): New.
41003 * opt-functions.awk (switch_flags): Map both Optimization and
41004 PerFunction to CL_OPTIMIZATION.
41005 * opth-gen.awk: Test for PerFunction flag along with
41007 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41008 it only when the latter is present. Skip those that don't in
41009 the hash function generator.
41010 * common.opt (fvar-tracking): Mark as PerFunction instead of
41012 (fvar-tracking-assignments): Likewise.
41013 (fvar-tracking-assignments-toggle): Likewise.
41014 (fvar-tracking-uninit): Likewise.
41016 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41018 PR translation/79018
41019 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41022 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41025 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41026 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41027 TARGET_LONG_JUMP_TABLE_OFFSETS.
41028 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41029 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41030 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41031 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41032 * config/m68k/m68k.md (tablejump expander): Likewise.
41033 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41034 TARGET_LONG_JUMP_TABLE_OFFSETS.
41035 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41036 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41038 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41039 David Holsgrove <david.holsgrove@xilinx.com>
41041 * common/config/microblaze/microblaze-common.c
41042 (TARGET_EXCEPT_UNWIND_INFO): Remove.
41043 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
41045 * config/microblaze/microblaze.c (microblaze_must_save_register)
41046 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41048 (microblaze_eh_return): New function.
41049 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41050 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41051 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41052 * config/microblaze/microblaze.md (eh_return): New pattern.
41054 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41056 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41057 GCC_DIAGNOSTIC_STRINGIFY): Define.
41059 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41061 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41063 * config/arm/arm.md (<mcrr>): New.
41065 * config/arm/arm.c (arm_arch5te): New.
41066 (arm_option_override): Set arm_arch5te.
41067 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41069 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41070 (arm_mcrr_qualifiers): ... this. New.
41071 (MRRC_QUALIFIERS): Define to...
41072 (arm_mrrc_qualifiers): ... this. New.
41073 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41075 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41076 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41077 (MRRCI, mrrc, MRRC): New.
41078 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41079 VUNSPEC_MRRC2): New.
41081 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41083 * config/arm/arm.md (<mcr>): New.
41085 * config/arm/arm.c (arm_coproc_builtin_available): Add
41086 support for mcr, mrc, mcr2 and mrc2.
41087 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41088 (arm_mcr_qualifiers): ... this. New.
41089 (MRC_QUALIFIERS): Define to ...
41090 (arm_mrc_qualifiers): ... this. New.
41091 (MCR_QUALIFIERS): Define to ...
41092 (arm_mcr_qualifiers): ... this. New.
41093 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41095 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41096 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41097 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41098 VUNSPEC_MRC2): New.
41100 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41102 * config/arm/arm.md (*ldc): New.
41106 * config/arm/arm.c (arm_coproc_builtin_available): Add
41107 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41108 (arm_coproc_ldc_stc_legitimate_address): New.
41109 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41110 'qualifier_const_pointer'.
41111 (LDC_QUALIFIERS): Define to...
41112 (arm_ldc_qualifiers): ... this. New.
41113 (STC_QUALIFIERS): Define to...
41114 (arm_stc_qualifiers): ... this. New.
41115 * config/arm/arm-protos.h
41116 (arm_coproc_ldc_stc_legitimate_address): New.
41117 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
41118 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
41119 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
41120 stc2, stcl, stc2l): New.
41121 * config/arm/constraints.md (Uz): New.
41122 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
41123 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
41124 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
41125 VUNSPEC_STC2L): New.
41127 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41129 * config/arm/arm.md (<cdp>): New.
41130 * config/arm/arm.c (neon_const_bounds): Rename this ...
41131 (arm_const_bounds): ... this.
41132 (arm_coproc_builtin_available): New.
41133 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
41134 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
41135 (CDP_QUALIFIERS): Define to...
41136 (arm_cdp_qualifiers): ... this. New.
41138 (arm_expand_builtin_args): Add case for 6 arguments.
41139 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
41140 (arm_const_bounds): ... this.
41141 (arm_coproc_builtin_available): New.
41142 * config/arm/arm_acle.h (__arm_cdp): New.
41144 * config/arm/arm_acle_builtins.def (cdp): New.
41146 * config/arm/iterators.md (CDPI,CDP,cdp): New.
41147 * config/arm/neon.md: Rename all 'neon_const_bounds' to
41148 'arm_const_bounds'.
41149 * config/arm/types.md (coproc): New.
41150 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
41151 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
41152 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
41153 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
41155 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41157 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
41158 (UBINOP_QUALIFIERS): New.
41160 (acle_builtin_data): New. Change comment.
41161 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
41162 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
41163 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
41164 arm_acle_builtins.def.
41165 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
41166 (arm_init_acle_builtins): New.
41167 (CRC32_BUILTIN): Remove.
41168 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
41169 crc32cb, crc32ch and crc32cw.
41170 (arm_init_crc32_builtins): Remove.
41171 (arm_init_builtins): Use arm_init_acle_builtins rather
41172 than arm_init_crc32_builtins.
41173 (arm_expand_acle_builtin): New.
41174 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
41175 * config/arm/arm_acle_builtins.def: New.
41177 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41179 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
41180 (arm_builtin_datum): ... this.
41181 (arm_init_neon_builtin): Rename to ...
41182 (arm_init_builtin): ... this. Add a new parameters PREFIX
41183 and USE_SIG_IN_NAME.
41184 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
41185 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
41186 'arm_builtin_datum'.
41187 (arm_init_vfp_builtins): Likewise.
41188 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
41189 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
41190 (arm_expand_neon_args): Rename to ...
41191 (arm_expand_builtin_args): ... this. Rename builtin_arg
41192 enum values and differentiate between ARG_BUILTIN_MEMORY
41193 and ARG_BUILTIN_NEON_MEMORY.
41194 (arm_expand_neon_builtin_1): Rename to ...
41195 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
41196 values, arm_expand_builtin_args and add bool parameter NEON.
41197 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
41198 (arm_expand_vfp_builtin): Likewise.
41199 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
41201 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41203 PR middle-end/77484
41204 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
41205 * predict.c (tree_estimate_probability_bb): Reverse direction of
41206 polymorphic call predictor.
41208 2017-01-06 David Malcolm <dmalcolm@redhat.com>
41210 * passes.c (execute_one_pass): Split out pass-skipping logic into...
41211 (determine_pass_name_match): ...this new function and...
41212 (should_skip_pass_p): ...this new function.
41214 2017-01-06 Nathan Sidwell <nathan@acm.org>
41216 * ipa-visibility.c (function_and_variable_visibility): Reformat
41217 comments and long lines. Remove extrneous if.
41218 * symtab.c (symtab_node::make_decl_local): Fix code format.
41219 (symtab_node::set_section_for_node): Fix comment typo.
41221 2017-01-06 Martin Liska <mliska@suse.cz>
41224 * lra-constraints.c: Rename invariant to lra_invariant.
41225 * predict.c (set_even_probabilities): Initialize e to NULL.
41227 2017-01-05 Martin Sebor <msebor@redhat.com>
41229 PR tree-optimization/78910
41230 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
41231 (format_integer): Correct off-by-one error in the handling
41232 of precision with negative numbers in signed conversions..
41234 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
41236 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
41238 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41240 PR tree-optimization/71016
41241 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
41242 factor_out_conditional_conversion. Formatting fix.
41243 (factor_out_conditional_conversion): Add cond_stmt argument.
41244 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
41245 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
41248 2017-01-05 David Malcolm <dmalcolm@redhat.com>
41250 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
41251 read-rtl-function.o, and selftest-rtl.o.
41252 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
41253 (selftest::aarch64_test_loading_full_dump): New function.
41254 (selftest::aarch64_run_selftests): New function.
41255 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
41256 selftest::aarch64_run_selftests.
41257 * config/i386/i386.c
41258 (selftest::ix86_test_loading_dump_fragment_1): New function.
41259 (selftest::ix86_test_loading_call_insn): New function.
41260 (selftest::ix86_test_loading_full_dump): New function.
41261 (selftest::ix86_test_loading_unspec): New function.
41262 (selftest::ix86_run_selftests): Call the new functions.
41263 * emit-rtl.c (maybe_set_max_label_num): New function.
41264 * emit-rtl.h (maybe_set_max_label_num): New decl.
41265 * function.c (instantiate_decls): Guard call to
41266 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
41267 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
41269 * gensupport.c (gen_reader::gen_reader): Pass "false"
41270 for new "compact" param of rtx_reader.
41271 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
41272 rather than an empty string for NULL strings.
41273 * read-md.c: Potentially include config.h rather than bconfig.h.
41274 Wrap include of errors.h with #ifdef GENERATOR_FILE.
41275 (have_error): New global, copied from errors.c.
41276 (md_reader::read_name): Rename to...
41277 (md_reader::read_name_1): ...this, adding "out_loc" param,
41278 and converting "missing name or number" to returning false, rather
41280 (md_reader::read_name): Reimplement in terms of read_name_1.
41281 (md_reader::read_name_or_nil): New function.
41282 (md_reader::read_string): Handle "(nil)" by returning NULL.
41283 (md_reader::md_reader): Add new param "compact".
41284 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
41285 (md_reader::read_file): New method.
41286 * read-md.h (md_reader::md_reader): Add new param "compact".
41287 (md_reader::read_file): New method.
41288 (md_reader::is_compact): New accessor.
41289 (md_reader::read_name): Convert return type from void to file_location.
41290 (md_reader::read_name_or_nil): New decl.
41291 (md_reader::read_name_1): New decl.
41292 (md_reader::m_compact): New field.
41293 (noop_reader::noop_reader): Pass "false" for new "compact" param
41295 (rtx_reader::rtx_reader): Add new "compact" param.
41296 (rtx_reader::read_rtx_operand): Make virtual and convert return
41297 type from void to rtx.
41298 (rtx_reader::read_until): New decl.
41299 (rtx_reader::handle_any_trailing_information): New virtual function.
41300 (rtx_reader::postprocess): New virtual function.
41301 (rtx_reader::finalize_string): New virtual function.
41302 (rtx_reader::m_in_call_function_usage): New field.
41303 (rtx_reader::m_reuse_rtx_by_id): New field.
41304 * read-rtl-function.c: New file.
41305 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
41306 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
41307 (selftest::verify_three_block_rtl_cfg): New decl.
41308 * read-rtl-function.h: New file.
41309 * read-rtl.c: Potentially include config.h rather than bconfig.h.
41310 For host, include function.h, memmodel.h, and emit-rtl.h.
41311 (one_time_initialization): New function.
41312 (struct compact_insn_name): New struct.
41313 (compact_insn_names): New array.
41314 (find_code): Handle insn codes in compact dumps.
41315 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
41316 (bind_subst_iter_and_attr): Likewise.
41317 (add_condition_to_string): Likewise.
41318 (add_condition_to_rtx): Likewise.
41319 (apply_attribute_uses): Likewise.
41320 (add_current_iterators): Likewise.
41321 (apply_iterators): Likewise.
41322 (initialize_iterators): Guard usage of apply_subst_iterator with
41323 #ifdef GENERATOR_FILE.
41324 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
41325 (md_reader::read_mapping): Likewise.
41326 (add_define_attr_for_define_subst): Likewise.
41327 (add_define_subst_attr): Likewise.
41328 (read_subst_mapping): Likewise.
41329 (check_code_iterator): Likewise.
41330 (rtx_reader::read_rtx): Likewise. Move one-time initialization
41332 (one_time_initialization): New function.
41333 (rtx_reader::read_until): New method.
41334 (read_flags): New function.
41335 (parse_reg_note_name): New function.
41336 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
41337 Handle reuse_rtx ids.
41338 Wrap iterator lookup within #ifdef GENERATOR_FILE.
41339 Add parsing support for RTL dumps, mirroring the special-cases in
41340 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
41341 values, and calling handle_any_trailing_information.
41342 (rtx_reader::read_rtx_operand): Convert return type from void
41343 to rtx, returning return_rtx. Handle case 'e'. Call
41344 finalize_string on XSTR and XTMPL fields.
41345 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
41346 "(nil)" values were omitted. Call the postprocess vfunc on the
41348 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
41349 class ctor. Initialize m_in_call_function_usage. Call
41350 one_time_initialization.
41351 * rtl-tests.c (selftest::test_uncond_jump): Call
41352 set_new_first_and_last_insn.
41353 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
41354 * selftest-rtl.c: New file.
41355 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
41356 (selftest::get_insn_by_uid): New decl.
41357 * selftest-run-tests.c (selftest::run_tests): Call
41358 read_rtl_function_c_tests.
41359 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
41360 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
41363 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
41365 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
41366 operands in a special way. Assert that pos+len <= mode precision.
41368 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41370 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
41371 3 argument Alias with unlimited for the negative form.
41372 (fno-vect-cost-model): Removed.
41374 2017-01-05 Martin Liska <mliska@suse.cz>
41376 * hsa-gen.c (gen_hsa_divmod): New function.
41377 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
41379 2017-01-05 Martin Liska <mliska@suse.cz>
41382 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
41385 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41387 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
41388 small constant length operands.
41390 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41392 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
41393 between loop iterations.
41395 2017-01-05 Martin Liska <mliska@suse.cz>
41398 * gimplify.c (gimplify_decl_expr): Compare to
41399 asan_poisoned_variables instread of checking flags.
41400 (gimplify_target_expr): Likewise.
41401 (gimplify_expr): Likewise.
41402 (gimplify_function_tree): Conditionally initialize
41403 asan_poisoned_variables.
41405 2017-01-04 Jeff Law <law@redhat.com>
41407 PR tree-optimizatin/78812
41408 * rtl.h (contains_mem_rtx_p): Prototype.
41409 * ifcvt.c (containts_mem_rtx_p): Move from here to...
41410 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
41411 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
41412 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
41413 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
41415 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41417 * input.c (assert_char_at_range): Default-initialize actual_range.
41419 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41421 * df-scan.c (df_ref_create_structure): Make regno unsigned,
41422 to match the caller.
41424 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41426 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
41427 insns after final jump in test to emit dummy move.
41429 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41431 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
41432 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
41434 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41436 * multiple_target.c (create_dispatcher_calls): Init e_next.
41437 * tree-ssa-loop-split.c (split_loop): Init border.
41438 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
41441 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
41446 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
41447 (altivec_register_operand): Do not return true if the operand
41448 contains a SUBREG mixing SImode and SFmode.
41449 (vsx_register_operand): Likewise.
41450 (vsx_reg_sfsubreg_ok): New predicate.
41451 (vfloat_operand): Do not return true if the operand contains a
41452 SUBREG mixing SImode and SFmode.
41453 (vint_operand): Likewise.
41454 (vlogical_operand): Likewise.
41455 (gpc_reg_operand): Likewise.
41456 (int_reg_operand): Likewise.
41457 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
41458 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
41459 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
41461 (rs6000_emit_move_si_sf_subreg): New helper function.
41462 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
41463 fixup SUBREGs involving SImode and SFmode.
41464 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
41465 numbers for the new peephole2 optimization.
41466 (peephole2 for SFmode unions): New peephole2 to optimize cases in
41467 the GLIBC math library that do AND/IOR/XOR operations on single
41468 precision floating point.
41469 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
41470 target macros to say whether we need to avoid SUBREGs mixing
41472 (TARGET_ALLOW_SF_SUBREG): Likewise.
41473 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
41474 (UNSPEC_SI_FROM_SF): Likewise.
41475 (iorxor): Change spacing.
41476 (and_ior_xor): New iterator for AND, IOR, and XOR.
41477 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
41478 (movdi_from_sf_zero_ext): Likewise.
41479 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
41480 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
41481 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
41482 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
41483 (fms<mode>4): Likewise.
41484 (fnma<mode>4): Likewise.
41485 (fnms<mode>4): Likewise.
41486 (nfma<mode>4): Likewise.
41487 (nfms<mode>4): Likewise.
41489 2017-01-04 Marek Polacek <polacek@redhat.com>
41492 * doc/invoke.texi: Document -Wpointer-compare.
41494 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41496 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
41499 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
41500 descriptions for -gdwarf-5 and emit them as uleb128 instead of
41503 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
41506 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
41507 documentation of the powerpc_popcntb_ok attribute.
41508 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
41509 code to issue warning messages if a requested CPU configuration is
41510 not supported by the binary (assembler and loader) toolchain.
41511 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
41512 made to define a built-in function that has been disabled.
41513 (paired_init_builtins): Add assertion to prevent ICE if attempt is
41514 made to define a built-in function that has been disabled.
41515 (altivec_init_builtins): Add comment explaining why definition
41516 of the DST built-in functions is not preceded by an assertion
41517 check. Add assertions to prevent ICE if attempts are made to
41518 define an altivec predicate or an abs* built-in function that has
41520 (htm_init_builtins): Add comment explaining why definition of the
41521 htm built-in functions is not preceded by an assertion check.
41523 2017-01-04 Jeff Law <law@redhat.com>
41525 PR tree-optimizatin/67955
41526 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
41527 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
41528 the points-to solution does not include pt_null. Use DECL_PT_UID
41531 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
41533 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
41534 Use gen_int_mode instead of gen_lopwart for const_int operands.
41536 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41538 PR tree-optimization/71563
41539 * match.pd: Simplify X << Y into X if Y is known to be 0 or
41540 out of range value - has low bits known to be zero.
41542 2017-01-04 Alan Modra <amodra@gmail.com>
41544 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
41545 * configure: Regenerate.
41546 * config.in: Regenerate.
41548 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41551 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
41552 a substring of the message, but strcmp with the whole message. Ifdef
41553 ENABLE_NLS, translate the message first using dgettext.
41555 2017-01-03 Jeff Law <law@redhat.com>
41557 PR tree-optimizatin/78856
41558 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
41559 (mark_threaded_blocks): Remove code to truncate thread paths that
41560 cross multiple loop headers. Instead invalidate the cached loop
41561 iteration information and handle case of a thread path walking
41562 into an irreducible region.
41564 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
41567 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
41568 assertions. Add support for doing the signbit if the IEEE 128-bit
41569 floating point value is in a GPR.
41570 * config/rs6000/rs6000.md (Fsignbit): Delete.
41571 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
41572 Update the length attribute if the value is in a GPR.
41573 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
41574 the sign or zero extension instruction, since the value is always 0/1.
41575 (signbit<mode>2_dm2): Delete using <Fsignbit>.
41578 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
41579 extracting SImode to a GPR register so that we can generate a
41580 store, limit the vector to be in a traditional Altivec register
41581 for the vextuwrx instruction.
41583 2017-01-03 Ian Lance Taylor <iant@google.com>
41585 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
41587 2017-01-03 Martin Sebor <msebor@redhat.com>
41589 PR tree-optimization/78696
41590 * gimple-ssa-sprintf.c (format_floating): Correct handling of
41591 precision. Use MPFR for %f for greater fidelity. Correct handling
41593 (pass_sprintf_length::compute_format_length): Set width and precision
41594 specified by asrerisk to void_node for vararg functions.
41595 (try_substitute_return_value): Adjust dump output.
41597 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
41599 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
41601 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
41603 * doc/invoke.texi (SPARC options): Document -mlra as the default.
41604 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
41605 -mlra/-mno-lra was passed to the compiler.
41607 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
41609 PR rtl-optimization/65618
41610 * emit-rtl.c (try_split): Move initialization of "before" and
41611 "after" to just before the call to emit_insn_after_setloc.
41613 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
41615 * doc/md.texi (Standard Names): Remove reference to Java frontend.
41617 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
41619 * dwarf2out.c (gen_enumeration_type_die): When
41620 -gno-strict-dwarf, add a DW_AT_encoding attribute.
41622 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41624 PR tree-optimization/78965
41625 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41626 Change first argument from const call_info & to call_info &. For %n
41627 set info.nowrite to false.
41629 PR middle-end/78901
41630 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41631 possibly throwing calls.
41633 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41634 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41635 and fns handling, rather than in a separate case SSA_NAME.
41637 2017-01-02 Jeff Law <law@redhat.com>
41639 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41640 fixes for first_period and second_period variables.
41642 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41645 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41646 (*insvqi_1): New insn pattern.
41647 (*insvqi_1_mem_rex64): Ditto.
41648 (*insvqi_2): Ditto.
41649 (*insvqi_3): Rename from *insvqi.
41651 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41653 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41655 * doc/cfg.texi (Edges): Remove reference to Java.
41656 (Maintaining the CFG): Ditto.
41658 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41660 PR middle-end/77674
41661 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41662 transparent aliases.
41664 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41666 PR middle-end/77484
41667 * predict.def (PRED_CALL): Update hitrate.
41668 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41669 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41670 into direct/indirect/polymorphic variants.
41672 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41674 Update copyright years.
41676 * gcc.c (process_command): Update copyright notice dates.
41677 * gcov-dump.c (print_version): Ditto.
41678 * gcov.c (print_version): Ditto.
41679 * gcov-tool.c (print_version): Ditto.
41680 * gengtype.c (create_file): Ditto.
41681 * doc/cpp.texi: Bump @copying's copyright year.
41682 * doc/cppinternals.texi: Ditto.
41683 * doc/gcc.texi: Ditto.
41684 * doc/gccint.texi: Ditto.
41685 * doc/gcov.texi: Ditto.
41686 * doc/install.texi: Ditto.
41687 * doc/invoke.texi: Ditto.
41689 Copyright (C) 2017 Free Software Foundation, Inc.
41691 Copying and distribution of this file, with or without modification,
41692 are permitted in any medium without royalty provided the copyright
41693 notice and this notice are preserved.