1 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
3 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
4 (aarch64_ld1x2<VDC:mode>): Likewise.
5 (aarch64_simd_ld1<mode>_x2): Likewise.
6 (aarch64_simd_ld1<mode>_x2): Likewise.
7 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
8 (vld1_s8_x2): Likewise.
9 (vld1_u16_x2): Likewise.
10 (vld1_s16_x2): Likewise.
11 (vld1_u32_x2): Likewise.
12 (vld1_s32_x2): Likewise.
13 (vld1_u64_x2): Likewise.
14 (vld1_s64_x2): Likewise.
15 (vld1_f16_x2): Likewise.
16 (vld1_f32_x2): Likewise.
17 (vld1_f64_x2): Likewise.
18 (vld1_p8_x2): Likewise.
19 (vld1_p16_x2): Likewise.
20 (vld1_p64_x2): Likewise.
21 (vld1q_u8_x2): Likewise.
22 (vld1q_s8_x2): Likewise.
23 (vld1q_u16_x2): Likewise.
24 (vld1q_s16_x2): Likewise.
25 (vld1q_u32_x2): Likewise.
26 (vld1q_s32_x2): Likewise.
27 (vld1q_u64_x2): Likewise.
28 (vld1q_s64_x2): Likewise.
29 (vld1q_f16_x2): Likewise.
30 (vld1q_f32_x2): Likewise.
31 (vld1q_f64_x2): Likewise.
32 (vld1q_p8_x2): Likewise.
33 (vld1q_p16_x2): Likewise.
34 (vld1q_p64_x2): Likewise.
36 2017-12-27 Martin Liska <mliska@suse.cz>
38 PR tree-optimization/83552
39 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
40 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
42 2017-12-27 Tom de Vries <tom@codesourcery.com>
44 * config/nvptx/nvptx.c (nvptx_option_override): Disable
45 -gstatement-frontiers.
47 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
49 PR rtl-optimization/83513
50 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
51 before priority comparison.
53 2017-12-25 Jakub Jelinek <jakub@redhat.com>
56 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
57 ix86_isa_flags2 to ix86_isa_flags.
58 * config/i386/i386-c.c (ix86_target_macros_internal): Test
59 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
60 isa_flags rather than isa_flags2.
61 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
62 and -mavx512bitalg from isa2_opts to isa_opts.
63 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
64 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
65 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
67 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
68 OPTION_MASK_ISA_AVX512F ored with another option similarly to
69 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
70 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
71 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
72 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
73 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
74 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
75 builtins from bdesc_args2 to bdesc_args section.
76 (__builtin_ia32_compressstoreuqi512_mask,
77 __builtin_ia32_compressstoreuhi512_mask,
78 __builtin_ia32_compressstoreuqi256_mask,
79 __builtin_ia32_expandloadqi512_mask,
80 __builtin_ia32_expandloadqi512_maskz,
81 __builtin_ia32_expandloadhi512_mask,
82 __builtin_ia32_expandloadhi512_maskz,
83 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
84 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
85 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
86 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
87 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
88 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
89 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
90 __builtin_ia32_vpshldv_v32hi_maskz,
91 __builtin_ia32_vpopcountb_v64qi_mask,
92 __builtin_ia32_vpopcountw_v32hi_mask,
93 __builtin_ia32_vpshufbitqmb512_mask,
94 __builtin_ia32_vpshufbitqmb256_mask): Add
95 " | OPTION_MASK_ISA_AVX512BW".
96 (__builtin_ia32_expandloadqi256_mask,
97 __builtin_ia32_expandloadqi256_maskz,
98 __builtin_ia32_vpopcountb_v32qi_mask): Add
99 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
100 (__builtin_ia32_expandloadhi256_mask,
101 __builtin_ia32_expandloadhi256_maskz,
102 __builtin_ia32_expandloadqi128_mask,
103 __builtin_ia32_expandloadqi128_maskz,
104 __builtin_ia32_expandloadhi128_mask,
105 __builtin_ia32_expandloadhi128_maskz,
106 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
107 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
108 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
109 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
110 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
111 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
112 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
113 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
114 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
115 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
116 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
117 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
118 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
119 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
120 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
121 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
122 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
123 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
124 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
125 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
126 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
127 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
128 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
129 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
130 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
131 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
132 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
133 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
134 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
135 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
136 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
137 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
138 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
139 __builtin_ia32_vpopcountw_v8hi_mask): Add
140 " | OPTION_MASK_ISA_AVX512VL".
141 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
142 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
143 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
144 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
145 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
146 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
147 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
148 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
149 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
150 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
151 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
152 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
153 _mm_mask_bitshuffle_epi64_mask): Likewise.
154 * common/config/i386/i386-common.c
155 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
156 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
157 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
158 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
159 OPTION_MASK_ISA_AVX512BITALG_UNSET.
160 (OPTION_MASK_ISA2_AVX512F_UNSET,
161 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
162 (ix86_handle_option): For -mno-general-regs-only, clear from
163 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
164 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
165 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
166 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
167 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
168 for moving from ix86_isa_flags2 to ix86_isa_flags.
170 2017-12-23 Jakub Jelinek <jakub@redhat.com>
173 * fold-const.c (struct contains_label_data): New type.
174 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
175 inside of a SWITCH_BODY seen during the walk.
176 (contains_label_p): Use walk_tree instead of
177 walk_tree_without_duplicates, prepare data for contains_label_1 and
180 2017-12-22 Jakub Jelinek <jakub@redhat.com>
183 * tree-iterator.c (alloc_stmt_list): Start with cleared
184 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
188 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
189 enable avx512f explicitly in #pragma GCC target.
190 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
191 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
192 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
193 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
194 __builtin_ia32_vpdpbusds_v8si_mask,
195 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
196 __builtin_ia32_vpdpbusds_v4si_mask,
197 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
198 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
199 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
200 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
201 __builtin_ia32_vpdpwssds_v8si_mask,
202 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
203 __builtin_ia32_vpdpwssds_v4si_mask,
204 __builtin_ia32_vpdpwssds_v4si_maskz): Use
205 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
206 just OPTION_MASK_ISA_AVX512VNNI.
208 2017-12-22 Martin Jambor <mjambor@suse.cz>
211 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
214 2017-12-22 Julia Koval <julia.koval@intel.com>
215 Sebastian Peryt <sebastian.peryt@intel.com>
217 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
218 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
219 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
220 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
221 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
222 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
223 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
224 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
225 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
226 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
227 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
228 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
229 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
230 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
231 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
232 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
233 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
234 _mm256_maskz_popcnt_epi64): New intrinsics.
235 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
236 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
237 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
238 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
239 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
240 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
241 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
242 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
243 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
244 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
245 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
246 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
247 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
248 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
249 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
250 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
251 __builtin_ia32_vpshufbitqmb256_mask,
252 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
253 * config/i386/i386-c.c (__AVX512BITALG__): New.
254 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
255 (ix86_valid_target_attribute_inner_p): Ditto.
256 (ix86_expand_args_builtin): Handle new types.
257 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
258 * config/i386/i386.opt: Add -mavx512bitalg.
259 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
260 avx512bitalgintrin.h.
261 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
262 (vpopcount<mode><mask_name>): Add more types.
263 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
264 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
266 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
268 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
269 Or in OPTION_MASK_ISA_AVX512F_SET.
270 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
271 OPTION_MASK_ISA_AVX512VNNI_UNSET.
272 (ix86_handle_option): Adjust for
273 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
274 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
276 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
277 isa_flag instead of isa_flag2.
278 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
279 isa_opts2 to isa_opts.
280 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
283 2017-12-22 Mike Stump <mikestump@comcast.net>
284 Eric Botcazou <ebotcazou@adacore.com>
286 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
288 2017-12-21 Jakub Jelinek <jakub@redhat.com>
291 * config/i386/i386.c (ix86_function_arg_boundary): Return
292 PARM_BOUNDARY for TYPE_EMPTY_P types.
295 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
296 if navail is >= dir.len.
298 2017-12-21 Steve Ellcey <sellcey@cavium.com>
300 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
303 2017-12-21 Jakub Jelinek <jakub@redhat.com>
305 PR rtl-optimization/80747
306 PR rtl-optimization/83512
307 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
308 succ edge from ENTRY, copy partition from e->dest to the newly
310 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
311 ENTRY, use BB_PARTITION of its successor block as current_partition.
312 Don't copy partition when splitting succ edge from ENTRY.
314 PR tree-optimization/83523
315 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
316 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
317 (convert_mult_to_fma): Likewise.
319 PR tree-optimization/83521
320 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
321 gimple_build_assign without code on result of
322 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
325 2017-12-21 Andrew Pinski <apinski@cavium.com>
326 Steve Ellcey <sellcey@cavium.com>
328 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
329 multi-arch for ilp32.
331 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
334 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
335 constraints to operand 2.
336 (*ashl<mode>3_mask_1): Ditto.
337 (*<shift_insn><mode>3_mask): Ditto.
338 (*<shift_insn><mode>3_mask_1): Ditto.
339 (*<rotate_insn><mode>3_mask): Ditto.
340 (*<rotate_insn><mode>3_mask_1): Ditto.
342 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
344 * reorg.c (make_return_insns): Reemit each insn with its own location.
346 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
348 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
349 generation for cases where splatting a value is not useful.
350 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
351 across a vec_duplicate and a paradoxical subreg forming a vector
352 mode to a vec_concat.
354 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
356 * combine.c (simplify_set): Do not transform subregs to zero_extends
357 if the destination is not a scalar int mode.
359 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
362 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
363 shared zero if the input has overflowed.
365 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
367 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
368 support for Cortex-A55 and Cortex-A75.
370 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
372 * common/config/arm/arm-common.c (compare_opt_names): Add function
373 comment. Use strcmp instead of manual loop.
375 2017-12-21 Martin Liska <mliska@suse.cz>
377 PR gcov-profile/83509
378 * gcov-dump.c (dump_gcov_file): Do not read info about
379 support_unexecuted_blocks for gcda files.
381 2017-12-21 Jakub Jelinek <jakub@redhat.com>
383 PR rtl-optimization/82973
384 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
385 (valid_for_const_vector_p): ... this.
386 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
387 (valid_for_const_vector_p): ... this. Adjust function comment.
388 (gen_vec_duplicate): Adjust caller.
389 * optabs.c (expand_vector_broadcast): Likewise.
390 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
391 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
393 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
394 instead of GET_CODE == CONST_FIXED.
395 (simplify_subreg): Use CONST_FIXED_P macro instead of
396 GET_CODE == CONST_FIXED.
399 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
400 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
401 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
402 (ix86_option_override_internal): Adjust for
403 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
404 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
405 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
406 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
407 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
408 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
409 def_builtin2 for CET builtins.
410 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
411 up formatting in IX86_BUILTIN_RDPID code.
412 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
413 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
414 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
415 ix86_isa_flags2 to ix86_isa_flags.
416 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
418 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
419 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
420 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
422 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
423 Or in OPTION_MASK_ISA_AVX512F_SET.
424 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
425 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
426 (ix86_handle_option): Adjust for
427 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
428 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
430 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
431 Alan Hayward <alan.hayward@arm.com>
432 David Sherwood <david.sherwood@arm.com>
434 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
435 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
436 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
437 Track polynomial offsets.
439 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
440 Alan Hayward <alan.hayward@arm.com>
441 David Sherwood <david.sherwood@arm.com>
443 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
444 Treat drb->init as a poly_int. Fail if its misalignment wrt
445 vector_alignment isn't known.
447 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
448 Alan Hayward <alan.hayward@arm.com>
449 David Sherwood <david.sherwood@arm.com>
451 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
453 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
454 (vect_loop_versioning): Take the loop versioning threshold as a
456 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
457 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
458 versioning_threshold.
459 (vect_analyze_loop_2): Compute the loop versioning threshold
460 whenever loop versioning is needed, and store it in the new
461 field rather than combining it with the cost model threshold.
462 (vect_transform_loop): Update call to vect_loop_versioning.
463 Try to combine the loop versioning and cost thresholds here.
465 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
466 Alan Hayward <alan.hayward@arm.com>
467 David Sherwood <david.sherwood@arm.com>
469 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
470 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
471 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
472 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
473 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
474 (group_compare_offset): Update accordingly.
475 (split_small_address_groups_p): Likewise.
476 (record_use): Take addr_offset as a poly_uint64 rather than
477 an unsigned HOST_WIDE_INT.
478 (strip_offset): Return the offset as a poly_uint64 rather than
479 an unsigned HOST_WIDE_INT.
480 (record_group_use, split_address_groups): Track polynomial offsets.
481 (add_iv_candidate_for_use): Likewise.
482 (addr_offset_valid_p): Take the offset as a poly_int64 rather
483 than a HOST_WIDE_INT.
484 (strip_offset_1): Return the offset as a poly_int64 rather than
487 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
488 Alan Hayward <alan.hayward@arm.com>
489 David Sherwood <david.sherwood@arm.com>
491 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
492 rather than a HOST_WIDE_INT.
493 * tree.c (get_binfo_at_offset): Likewise.
495 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
496 Alan Hayward <alan.hayward@arm.com>
497 David Sherwood <david.sherwood@arm.com>
499 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
500 rather than a HOST_WIDE_INT.
501 * tree-sra.c (build_ref_for_offset): Likewise.
503 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
504 Alan Hayward <alan.hayward@arm.com>
505 David Sherwood <david.sherwood@arm.com>
507 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
509 * tree.c (mem_ref_offset): Likewise.
510 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
511 * builtins.c (get_object_alignment_2): Likewise.
512 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
513 * gimple-fold.c (get_base_constructor): Likewise.
514 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
515 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
517 * ipa-polymorphic-call.c
518 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
519 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
520 (get_ancestor_addr_info): Likewise.
521 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
522 * match.pd: Likewise.
523 * tree-data-ref.c (dr_analyze_innermost): Likewise.
524 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
525 * tree-eh.c (tree_could_trap_p): Likewise.
526 * tree-object-size.c (addr_object_size): Likewise.
527 * tree-ssa-address.c (copy_ref_info): Likewise.
528 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
529 (indirect_refs_may_alias_p): Likewise.
530 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
531 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
532 (non_rewritable_mem_ref_base): Likewise.
533 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
534 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
535 * varasm.c (decode_addr_const): Likewise.
537 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
538 Alan Hayward <alan.hayward@arm.com>
539 David Sherwood <david.sherwood@arm.com>
541 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
542 offsets for MEM_REFs.
544 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
545 Alan Hayward <alan.hayward@arm.com>
546 David Sherwood <david.sherwood@arm.com>
548 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
549 as a poly_uint64 rather than an unsigned int.
550 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
552 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
553 Alan Hayward <alan.hayward@arm.com>
554 David Sherwood <david.sherwood@arm.com>
556 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
558 (decode_addr_const): Update accordingly.
560 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
561 Alan Hayward <alan.hayward@arm.com>
562 David Sherwood <david.sherwood@arm.com>
564 * tree.h (bit_field_size, bit_field_offset): New functions.
565 * hsa-gen.c (gen_hsa_addr): Use them.
566 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
567 (simplify_vector_constructor): Likewise.
568 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
569 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
570 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
571 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
574 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
575 Alan Hayward <alan.hayward@arm.com>
576 David Sherwood <david.sherwood@arm.com>
578 * expr.h (emit_group_load, emit_group_load_into_temps)
579 (emit_group_store): Take the size as a poly_int64 rather than an int.
580 * expr.c (emit_group_load_1, emit_group_load): Likewise.
581 (emit_group_load_into_temp, emit_group_store): Likewise.
583 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
584 Alan Hayward <alan.hayward@arm.com>
585 David Sherwood <david.sherwood@arm.com>
587 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
588 an unsigned int to a poly_uint64.
589 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
590 sizes as poly_uint64s rather than unsigned ints.
591 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
594 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
595 Alan Hayward <alan.hayward@arm.com>
596 David Sherwood <david.sherwood@arm.com>
598 * lra-constraints.c (emit_inc): Change inc_amount from an int
601 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
602 Alan Hayward <alan.hayward@arm.com>
603 David Sherwood <david.sherwood@arm.com>
605 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
607 (add_stack_var, stack_var_cmp, partition_stack_vars)
608 (dump_stack_var_partition): Update accordingly.
609 (alloc_stack_frame_space): Take the size as a poly_int64 rather
610 than a HOST_WIDE_INT.
611 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
612 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
613 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
614 rather than a HOST_WIDE_INT.
616 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
617 Alan Hayward <alan.hayward@arm.com>
618 David Sherwood <david.sherwood@arm.com>
620 * target.def (return_pops_args): Treat both the input and output
621 sizes as poly_int64s rather than HOST_WIDE_INTS.
622 * targhooks.h (default_return_pops_args): Update accordingly.
623 * targhooks.c (default_return_pops_args): Likewise.
624 * doc/tm.texi: Regenerate.
625 * emit-rtl.h (incoming_args): Change pops_args, size and
626 outgoing_args_size from int to poly_int64_pod.
627 * function.h (expr_status): Change x_pending_stack_adjust and
628 x_stack_pointer_delta from int to poly_int64.
629 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
630 (ARGS_SIZE_RTX): Update accordingly.
631 * calls.c (highest_outgoing_arg_in_use): Change from int to
633 (stack_usage_watermark, stored_args_watermark): New variables.
634 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
635 (emit_call_1): Change the stack_size and rounded_stack_size
636 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
638 (save_fixed_argument_area): Check stack_usage_watermark.
639 (initialize_argument_information): Change old_pending_adj from
640 a HOST_WIDE_INT * to a poly_int64_pod *.
641 (compute_argument_block_size): Return the size as a poly_int64
643 (finalize_must_preallocate): Track polynomial argument sizes.
644 (compute_argument_addresses): Likewise.
645 (internal_arg_pointer_based_exp): Track polynomial offsets.
646 (mem_overlaps_already_clobbered_arg_p): Rename to...
647 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
648 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
649 Check stored_args_used_watermark.
650 (load_register_parameters): Update accordingly.
651 (check_sibcall_argument_overlap_1): Likewise.
652 (combine_pending_stack_adjustment_and_call): Take the unadjusted
653 args size as a poly_int64 rather than an int. Return a bool
654 indicating whether the optimization was possible and return
655 the new adjustment by reference.
656 (check_sibcall_argument_overlap): Track polynomail argument sizes.
657 Update stored_args_watermark.
658 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
659 (expand_call): Likewise. Maintain stack_usage_watermark and
660 stored_args_watermark. Update calls to
661 combine_pending_stack_adjustment_and_call.
662 (emit_library_call_value_1): Handle polynomial argument sizes.
663 Call stack_region_maybe_used_p and mark_stack_region_used.
664 Maintain stack_usage_watermark.
665 (store_one_arg): Likewise. Update call to
666 mem_overlaps_already_clobbered_arg_p.
667 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
669 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
670 * config/microblaze/microblaze.c (microblaze_function_prologue):
672 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
673 TARGET_RETURN_POPS_ARGS interface.
674 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
676 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
677 * config/i386/i386.c (ix86_return_pops_args): Update for new
678 TARGET_RETURN_POPS_ARGS interface.
679 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
680 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
681 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
682 TARGET_RETURN_POPS_ARGS interface.
683 * config/vax/vax.c (vax_return_pops_args): Likewise.
684 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
685 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
686 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
687 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
688 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
689 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
690 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
691 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
692 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
693 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
694 and x_stack_pointer_delta from int to poly_int64.
695 * dojump.c (do_pending_stack_adjust): Update accordingly.
696 * explow.c (allocate_dynamic_stack_space): Handle polynomial
697 stack_pointer_deltas.
698 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
699 (pad_to_arg_alignment): Track polynomial offsets.
700 (assign_parm_find_stack_rtl): Likewise.
701 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
702 * toplev.c (output_stack_usage): Update reference to
703 current_function_pushed_stack_size.
705 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
706 Alan Hayward <alan.hayward@arm.com>
707 David Sherwood <david.sherwood@arm.com>
709 * function.c (in_arg_offset, var_offset, dynamic_offset)
710 (out_arg_offset, cfa_offset): Change from int to poly_int64.
711 (instantiate_new_reg): Return the new offset as a poly_int64_pod
712 rather than a HOST_WIDE_INT.
713 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
714 (instantiate_virtual_regs_in_insn): Likewise.
716 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
717 Alan Hayward <alan.hayward@arm.com>
718 David Sherwood <david.sherwood@arm.com>
720 * rtl.h (get_args_size, add_args_size_note): New functions.
721 (find_args_size_adjust): Return a poly_int64 rather than a
723 (fixup_args_size_notes): Likewise. Make the same change to the
724 end_args_size parameter.
725 * rtlanal.c (get_args_size, add_args_size_note): New functions.
726 * builtins.c (expand_builtin_trap): Use add_args_size_note.
727 * calls.c (emit_call_1): Likewise.
728 * explow.c (adjust_stack_1): Likewise.
729 * cfgcleanup.c (old_insns_match_p): Update use of
730 find_args_size_adjust.
731 * combine.c (distribute_notes): Track polynomial arg sizes.
732 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
733 end_true_args_size, beg_delay_args_size and end_delay_args_size
734 from HOST_WIDE_INT to poly_int64.
735 (add_cfi_args_size): Take the args_size as a poly_int64 rather
736 than a HOST_WIDE_INT.
737 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
738 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
739 polynomial arg sizes.
740 * emit-rtl.c (try_split): Use get_args_size.
741 * recog.c (peep2_attempt): Likewise.
742 * reload1.c (reload_as_needed): Likewise.
743 * expr.c (find_args_size_adjust): Return the adjustment as a
744 poly_int64 rather than a HOST_WIDE_INT.
745 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
746 to a poly_int64 and change the return type in the same way.
747 (emit_single_push_insn): Track polynomial arg sizes.
749 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
750 Alan Hayward <alan.hayward@arm.com>
751 David Sherwood <david.sherwood@arm.com>
753 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
754 from HOST_WIDE_INT to poly_int64.
755 * expr.c (push_block, emit_push_insn): Likewise.
757 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
758 Alan Hayward <alan.hayward@arm.com>
759 David Sherwood <david.sherwood@arm.com>
761 * function.h (frame_space): Change start and length from HOST_WIDE_INT
763 (get_frame_size): Return the size as a poly_int64 rather than a
765 (frame_offset_overflow): Take the offset as a poly_int64 rather
766 than a HOST_WIDE_INT.
767 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
768 (assign_stack_temp): Likewise for the size.
769 * function.c (get_frame_size): Return a poly_int64 rather than
771 (frame_offset_overflow): Take the offset as a poly_int64 rather
772 than a HOST_WIDE_INT.
773 (try_fit_stack_local): Take the start, length and size as poly_int64s
774 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
775 rather than a HOST_WIDE_INT.
776 (add_frame_space): Take the start and end as poly_int64s rather than
778 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
779 (assign_stack_temp): Likewise for the size.
780 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
782 (find_temp_slot_from_address): Handle polynomial offsets.
783 (combine_temp_slots): Likewise.
784 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
786 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
787 poly_int64 rather than a HOST_WIDE_INT.
788 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
789 than a HOST_WIDE_INT.
790 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
791 polynomial frame offsets.
792 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
793 as a poly_int64 rather than an int.
794 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
795 * config/v850/v850-protos.h (compute_frame_size): Likewise.
796 * config/v850/v850.c (compute_frame_size): Likewise.
797 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
798 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
799 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
800 * config/pa/pa.c (pa_compute_frame_size): Likewise.
801 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
802 rather than a HOST_WIDE_INT.
803 * explow.c (get_dynamic_stack_base): Likewise.
804 * final.c (final_start_function): Use the constant lower bound
805 of the frame size for -Wframe-larger-than.
806 * ira.c (do_reload): Adjust for new get_frame_size return type.
807 * lra.c (lra): Likewise.
808 * reload1.c (reload): Likewise.
809 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
810 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
811 * rtlanal.c (get_initial_register_offset): Return the offset as
812 a poly_int64 rather than a HOST_WIDE_INT.
814 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
815 Alan Hayward <alan.hayward@arm.com>
816 David Sherwood <david.sherwood@arm.com>
818 * reload1.c (elim_table): Change initial_offset, offset and
819 previous_offset from HOST_WIDE_INT to poly_int64_pod.
820 (offsets_at): Change the target array's element type from
821 HOST_WIDE_INT to poly_int64_pod.
822 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
823 (elimination_costs_in_insn, update_eliminable_offsets)
824 (verify_initial_elim_offsets, set_offsets_for_label)
825 (init_eliminable_invariants): Update after above changes.
827 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
828 Alan Hayward <alan.hayward@arm.com>
829 David Sherwood <david.sherwood@arm.com>
831 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
832 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
833 (decomposition): Change start and end from HOST_WIDE_INT
835 (decompose, immune_p): Update accordingly.
836 (find_inc_amount): Return a poly_int64 rather than an int.
837 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
840 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
841 Alan Hayward <alan.hayward@arm.com>
842 David Sherwood <david.sherwood@arm.com>
844 * tree.h (get_inner_reference): Return the bitsize and bitpos
845 as poly_int64_pods rather than HOST_WIDE_INT.
846 * fold-const.h (ptr_difference_const): Return the pointer difference
847 as a poly_int64_pod rather than a HOST_WIDE_INT.
848 * expr.c (get_inner_reference): Return the bitsize and bitpos
849 as poly_int64_pods rather than HOST_WIDE_INT.
850 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
852 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
853 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
854 (optimize_bit_field_compare): Update call to get_inner_reference.
855 (decode_field_reference): Likewise.
856 (fold_unary_loc): Track polynomial offsets and sizes.
857 (split_address_to_core_and_offset): Return the bitpos as a
858 poly_int64_pod rather than a HOST_WIDE_INT.
859 (ptr_difference_const): Likewise for the pointer difference.
860 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
861 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
862 * dbxout.c (dbxout_expand_expr): Likewise.
863 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
864 (loc_list_from_tree_1, fortran_common): Likewise.
865 * gimple-laddress.c (pass_laddress::execute): Likewise.
866 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
867 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
868 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
869 * tree-affine.c (tree_to_aff_combination): Likewise.
870 (get_inner_reference_aff): Likewise.
871 * tree-data-ref.c (split_constant_offset_1): Likewise.
872 (dr_analyze_innermost): Likewise.
873 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
874 * tree-sra.c (ipa_sra_check_caller): Likewise.
875 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
876 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
877 (instrument_bool_enum_load, instrument_object_size): Likewise.
878 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
879 to get_inner_reference.
880 * hsa-gen.c (gen_hsa_addr): Likewise.
881 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
882 * tsan.c (instrument_expr): Likewise.
883 * match.pd: Update call to ptr_difference_const.
885 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
886 Alan Hayward <alan.hayward@arm.com>
887 David Sherwood <david.sherwood@arm.com>
889 * fold-const.c (fold_comparison): Track sizes and offsets as
890 poly_int64s rather than HOST_WIDE_INTs when folding address
893 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
894 Alan Hayward <alan.hayward@arm.com>
895 David Sherwood <david.sherwood@arm.com>
897 * expr.h (get_bit_range): Return the bitstart and bitend as
898 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
899 as a poly_int64 rather than a HOST_WIDE_INT.
900 * expr.c (get_bit_range): Likewise.
901 (expand_assignment): Update call accordingly.
902 * fold-const.c (optimize_bit_field_compare): Likewise.
904 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
905 Alan Hayward <alan.hayward@arm.com>
906 David Sherwood <david.sherwood@arm.com>
908 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
909 (aarch64_legitimate_address_p): Use it instead of an rtx code,
910 as an optional final parameter.
911 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
912 (aarch64_legitimate_address_p): Likewise.
913 (aarch64_print_address_internal): Take an aarch64_addr_query_type
914 instead of an rtx code.
915 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
916 (aarch64_legitimate_address_hook_p): Likewise.
917 (aarch64_print_ldpstp_address): Likewise.
918 (aarch64_print_operand_address): Likewise.
919 (aarch64_address_cost): Likewise.
920 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
921 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
922 (aarch64_mem_pair_lanes_operand): Likewise.
924 2017-12-20 Richard Biener <rguenther@suse.de>
926 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
927 update_stmt_if_modified.
929 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
931 PR tree-optimization/83491
932 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
933 before walking uses. Improve coding style and comments.
935 2017-12-20 Tom de Vries <tom@codesourcery.com>
937 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
939 2017-12-20 Jakub Jelinek <jakub@redhat.com>
942 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
944 (pass_ipa_free_fn_summary): Add small_p private data member,
945 initialize to false in the ctor.
946 (pass_ipa_free_fn_summary::clone,
947 pass_ipa_free_fn_summary::set_pass_param,
948 pass_ipa_free_fn_summary::gate): New methods.
949 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
950 | TODO_dump_symtab if small_p.
951 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
952 entry and add another instance of the pass with false parm after
954 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
955 ipa_free_fn_summary here.
957 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
959 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
961 2017-12-20 Martin Sebor <msebor@redhat.com>
964 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
965 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
966 (expand_builtin_strncmp): Same.
968 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
971 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
972 allowed debug stmts before labels.
973 (expand_gimple_basic_block): Likewise.
974 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
975 * gimple-iterator.h (gsi_after_labels): Likewise.
976 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
977 rename reused variable, and simplify using gsi_move_before.
978 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
979 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
980 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
981 (gimple_verify_flow_info, gimple_block_label): Likewise.
982 (make_blocks): Move debug markers after adjacent labels.
983 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
984 allowed debug insns outside blocks.
985 * df-scan.c (df_insn_delete): Likewise.
986 * lra-constraints.c (update_ebb_live_info): Likewise.
987 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
988 (vt_initialize, delete_vta_debug_insns): Likewise.
989 (reemit_marker_as_note): Drop BB parm. Adjust callers.
991 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
992 Alan Hayward <alan.hayward@arm.com>
993 David Sherwood <david.sherwood@arm.com>
995 * poly-int-types.h (round_down_to_byte_boundary): New macro.
996 (round_up_to_byte_boundary): Likewise.
997 * expr.h (get_bit_range): Add temporary shim.
998 * gimple-ssa-store-merging.c (store_operand_info): Change the
999 bitsize, bitpos, bitregion_start and bitregion_end fields from
1000 unsigned HOST_WIDE_INT to poly_uint64.
1001 (merged_store_group): Likewise load_align_base.
1002 (compatible_load_p, compatible_load_p): Update accordingly.
1003 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1004 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1005 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1006 bitregion_start and bitregion_end as poly_uint64s rather than
1007 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1008 (handled_load): Take the bitsize, bitpos,
1009 bitregion_start and bitregion_end as poly_uint64s rather than
1010 unsigned HOST_WIDE_INTs.
1011 (pass_store_merging::process_store): Update call to
1012 mem_valid_for_store_merging.
1014 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1015 Alan Hayward <alan.hayward@arm.com>
1016 David Sherwood <david.sherwood@arm.com>
1018 * builtins.c (get_object_alignment_2): Track polynomial offsets
1019 and sizes. Update the alignment handling.
1021 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1022 Alan Hayward <alan.hayward@arm.com>
1023 David Sherwood <david.sherwood@arm.com>
1025 * tree.h (get_inner_reference): Add a version that returns the
1026 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1027 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1028 the case in which bitpos is not associated with the first byte.
1030 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1031 Alan Hayward <alan.hayward@arm.com>
1032 David Sherwood <david.sherwood@arm.com>
1034 * tree-affine.h (get_inner_reference_aff): Return the size as a
1036 * tree-affine.c (get_inner_reference_aff): Likewise.
1037 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1038 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
1040 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1041 Alan Hayward <alan.hayward@arm.com>
1042 David Sherwood <david.sherwood@arm.com>
1044 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1045 HOST_WIDE_INT rather than a poly_int64.
1047 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1048 Alan Hayward <alan.hayward@arm.com>
1049 David Sherwood <david.sherwood@arm.com>
1051 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1052 HOST_WIDE_INT to poly_int64_pod.
1053 (perform_symbolic_merge): Update accordingly.
1054 (bswap_replace): Likewise.
1056 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1057 Alan Hayward <alan.hayward@arm.com>
1058 David Sherwood <david.sherwood@arm.com>
1060 * tree-affine.h (aff_tree::offset): Change from widest_int
1062 (wide_int_ext_for_comb): Delete.
1063 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1064 constants as poly_widest_int rather than widest_int.
1065 (aff_combination_constant_multiple_p): Return the multiplier
1066 as a poly_widest_int.
1067 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1069 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1070 version static and add an overload for poly_widest_int.
1071 (aff_combination_const, aff_combination_add_cst)
1072 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1073 the constants as poly_widest_int rather than widest_int.
1074 (tree_to_aff_combination): Generalize INTEGER_CST case to
1076 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1077 (aff_combination_add_product, aff_combination_mult): Handle
1079 (aff_combination_constant_multiple_p): Return the multiplier
1080 as a poly_widest_int.
1081 * tree-predcom.c (determine_offset): Return the offset as a
1083 (split_data_refs_to_components, suitable_component_p): Update
1085 (valid_initializer_p): Update call to
1086 aff_combination_constant_multiple_p.
1087 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1088 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1089 as a poly_int64 rather than a HOST_WIDE_INT.
1090 (get_address_cost): Handle polynomial offsets.
1091 (iv_elimination_compare_lt): Likewise.
1092 (rewrite_use_nonlinear_expr): Likewise.
1094 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1095 Alan Hayward <alan.hayward@arm.com>
1096 David Sherwood <david.sherwood@arm.com>
1098 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1099 as a poly_int64_pod rather than a HOST_WIDE_INT.
1100 (get_addr_base_and_unit_offset): Likewise.
1101 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1102 (get_addr_base_and_unit_offset): Likewise.
1103 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1104 to poly_int64 in example.
1105 * fold-const.c (fold_binary_loc): Update call to
1106 get_addr_base_and_unit_offset.
1107 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1108 (maybe_canonicalize_mem_ref_addr): Likewise.
1109 (gimple_fold_stmt_to_constant_1): Likewise.
1110 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1112 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
1113 * match.pd: Likewise.
1114 * omp-low.c (lower_omp_target): Likewise.
1115 * tree-sra.c (build_ref_for_offset): Likewise.
1116 (build_debug_ref_for_model): Likewise.
1117 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
1118 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
1119 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
1120 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1121 (constant_pointer_difference): Likewise.
1122 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
1123 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
1124 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
1125 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1126 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
1127 (set_ssa_val_to): Likewise.
1128 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
1129 (maybe_diag_stxncpy_trunc): Likewise.
1130 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1131 * tree.c (build_simple_mem_ref_loc): Likewise.
1132 (array_at_struct_end_p): Likewise.
1134 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1135 Alan Hayward <alan.hayward@arm.com>
1136 David Sherwood <david.sherwood@arm.com>
1138 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
1139 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1140 (get_ref_base_and_extent_hwi): Declare.
1141 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
1142 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1143 (get_ref_base_and_extent_hwi): New function.
1144 * cfgexpand.c (expand_debug_expr): Update call to
1145 get_ref_base_and_extent.
1146 * dwarf2out.c (add_var_loc_to_decl): Likewise.
1147 * gimple-fold.c (get_base_constructor): Return the offset as a
1148 poly_int64_pod rather than a HOST_WIDE_INT.
1149 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
1150 * ipa-polymorphic-call.c
1151 (ipa_polymorphic_call_context::set_by_invariant)
1152 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
1153 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
1154 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
1155 rather than get_ref_base_and_extent.
1156 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
1157 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
1158 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
1160 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
1161 call to get_ref_base_and_extent.
1162 * tree-sra.c (create_access, get_access_for_expr): Likewise.
1163 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
1164 (stmt_kills_ref_p): Likewise.
1165 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1166 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
1168 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
1169 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
1170 when calling native_encode_expr.
1171 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
1172 call to get_ref_base_and_extent.
1173 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
1174 get_ref_base_and_extent.
1175 * var-tracking.c (track_expr_p): Likewise.
1177 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1178 Alan Hayward <alan.hayward@arm.com>
1179 David Sherwood <david.sherwood@arm.com>
1181 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
1182 HOST_WIDE_INT to poly_int64_pod.
1183 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
1184 polynomail parameter offsets.
1186 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1187 Alan Hayward <alan.hayward@arm.com>
1188 David Sherwood <david.sherwood@arm.com>
1190 * gengtype.c (main): Handle poly_int64_pod.
1191 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
1192 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
1193 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
1194 from HOST_WIDE_INT to poly_int64_pod.
1195 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
1196 (copy_cfa): New function.
1197 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
1198 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
1199 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
1200 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
1201 (initial_return_save): Treat offsets as poly_ints.
1202 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
1203 offset is nonconstant.
1204 (reg_save): Take the offset as a poly_int64. Fall back to
1205 DW_CFA_expression for nonconstant offsets.
1206 (queue_reg_save): Take the offset as a poly_int64.
1207 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
1209 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1210 Alan Hayward <alan.hayward@arm.com>
1211 David Sherwood <david.sherwood@arm.com>
1213 * rtl.h (operand_subword, operand_subword_force): Take the offset
1214 as a poly_uint64 an unsigned int.
1215 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
1217 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1218 Alan Hayward <alan.hayward@arm.com>
1219 David Sherwood <david.sherwood@arm.com>
1221 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
1222 'p' format code. Use INT_LIST rather than SUBREG as the example of
1223 a code with an XINT and an XEXP. Remove the implication that
1224 accessing an rtx field using XINT is expected to work.
1225 * rtl.def (SUBREG): Change format from "ei" to "ep".
1226 * rtl.h (rtunion::rt_subreg): New field.
1227 (XCSUBREG): New macro.
1228 (SUBREG_BYTE): Use it.
1229 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
1230 Update constructor accordingly.
1231 (subreg_shape::operator ==): Update accordingly.
1232 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
1233 than an unsigned int.
1234 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
1235 a poly_uint64 rather than an unsigned int.
1236 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
1237 than an unsigned int.
1238 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
1239 (subreg_size_highpart_offset): Return a poly_uint64 rather than
1240 an unsigned int. Take the sizes as poly_uint64s.
1241 (subreg_offset_from_lsb): Return a poly_uint64 rather than
1242 an unsigned int. Take the shift as a poly_uint64 rather than
1244 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
1245 as a poly_uint64 rather than an unsigned int.
1246 (simplify_subreg_regno): Likewise.
1247 (byte_lowpart_offset): Return the memory offset as a poly_int64
1249 (subreg_memory_offset): Likewise. Take the subreg offset as a
1250 poly_uint64 rather than an unsigned int.
1251 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
1252 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
1253 poly_uint64 rather than an unsigned int.
1254 * rtl.c (rtx_format): Describe 'p' in comment.
1255 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
1256 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
1257 offset as a poly_uint64 rather than an unsigned int.
1258 (byte_lowpart_offset): Return the memory offset as a poly_int64
1260 (subreg_memory_offset): Likewise. Take the subreg offset as a
1261 poly_uint64 rather than an unsigned int.
1262 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
1263 mode sizes as poly_uint64s rather than unsigned ints. Return a
1264 poly_uint64 rather than an unsigned int.
1265 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
1266 (copy_insn_1): Handle 'p'.
1267 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
1268 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
1269 an unsigned int. Return the shift in the same way.
1270 (subreg_lsb): Return the shift as a poly_uint64 rather than an
1272 (subreg_size_offset_from_lsb): Take the sizes and shift as
1273 poly_uint64s rather than unsigned ints. Return the offset as
1275 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
1276 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
1278 * rtlhash.c (add_rtx): Handle 'p'.
1279 * genemit.c (gen_exp): Likewise.
1280 * gengenrtl.c (type_from_format, gendef): Likewise.
1281 * gensupport.c (subst_pattern_match, get_alternatives_number)
1282 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
1283 (subst_dup): Likewise.
1284 * gengtype.c (adjust_field_rtx_def): Likewise.
1285 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
1286 (match_pattern_2): Likewise.
1287 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
1288 (rtx_test::subreg_field): New function.
1289 (operator ==, safe_to_hoist_p, transition_parameter_type)
1290 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
1291 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
1292 * genpeep.c (match_rtx): Likewise.
1293 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
1294 (rtx_writer::print_rtx_operand): Handle 'p'.
1295 (print_value): Handle SUBREG.
1296 * read-rtl.c (apply_int_iterator): Likewise.
1297 (rtx_reader::read_rtx_operand): Handle 'p'.
1298 * alias.c (rtx_equal_for_memref_p): Likewise.
1299 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
1300 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
1302 * calls.c (expand_call): Likewise.
1303 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
1304 (make_extraction, gen_lowpart_for_combine): Likewise.
1305 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
1307 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
1308 rather than an unsigned int. Treat subreg offsets as poly_ints.
1309 (exp_equiv_p): Handle 'p'.
1310 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
1311 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
1312 * dse.c (find_shift_sequence): Likewise.
1313 * dwarf2out.c (rtl_for_decl_location): Likewise.
1314 * expmed.c (extract_low_bits): Likewise.
1315 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
1316 (expand_expr_real_2): Likewise.
1317 * final.c (alter_subreg): Likewise.
1318 (leaf_renumber_regs_insn): Handle 'p'.
1319 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
1320 Treat subreg offsets as poly_ints.
1321 * fwprop.c (forward_propagate_and_simplify): Likewise.
1322 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
1323 * ira.c (get_subreg_tracking_sizes): Likewise.
1324 * ira-conflicts.c (go_through_subreg): Likewise.
1325 * ira-lives.c (process_single_reg_class_operands): Likewise.
1326 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
1327 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
1328 as a poly_uint64 rather than an unsigned int.
1329 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
1330 subreg offsets as poly_ints.
1331 * lra-constraints.c (operands_match_p): Handle 'p'.
1332 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
1333 * lra-spills.c (assign_mem_slot): Likewise.
1334 * postreload.c (move2add_valid_value_p): Likewise.
1335 * recog.c (general_operand, indirect_operand): Likewise.
1336 * regcprop.c (copy_value, maybe_mode_change): Likewise.
1337 (copyprop_hardreg_forward_1): Likewise.
1338 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
1339 (record_subregs_of_mode): Likewise.
1340 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
1341 * reload.c (operands_match_p): Handle 'p'.
1342 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
1343 * reload1.c (alter_reg, choose_reload_regs): Likewise.
1344 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
1345 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
1346 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
1347 (simplify_const_poly_int_tests<N>::run): Likewise.
1348 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
1349 a poly_uint64 rather than an unsigned int.
1350 * valtrack.c (debug_lowpart_subreg): Likewise.
1351 * var-tracking.c (var_lowpart): Likewise.
1352 (loc_cmp): Handle 'p'.
1354 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1355 Alan Hayward <alan.hayward@arm.com>
1356 David Sherwood <david.sherwood@arm.com>
1358 * ira.c (get_subreg_tracking_sizes): New function.
1359 (init_live_subregs): Take an integer size rather than a register.
1360 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
1361 to init_live_subregs.
1363 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1364 Alan Hayward <alan.hayward@arm.com>
1365 David Sherwood <david.sherwood@arm.com>
1367 * expr.c (store_constructor_field): Change bitsize from a
1368 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
1369 HOST_WIDE_INT to a poly_int64.
1370 (store_constructor): Change size from a HOST_WIDE_INT to
1372 (store_field): Likewise bitsize and bitpos.
1374 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1375 Alan Hayward <alan.hayward@arm.com>
1376 David Sherwood <david.sherwood@arm.com>
1378 * expmed.h (store_bit_field): Change bitregion_start and
1379 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
1380 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
1381 (store_bit_field_1, store_integral_bit_field, store_bit_field)
1382 (store_fixed_bit_field, store_split_bit_field): Likewise.
1383 * expr.c (store_constructor_field, store_field): Likewise.
1384 (optimize_bitfield_assignment_op): Likewise. Make the same change
1385 to bitsize and bitpos.
1386 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
1387 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
1388 same change in the constructor arguments.
1389 (get_best_mode): Change bitregion_start and bitregion_end from
1390 unsigned HOST_WIDE_INT to poly_uint64.
1391 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1392 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
1394 (bit_field_mode_iterator::next_mode): Update for new types
1395 of m_bitregion_start and m_bitregion_end.
1396 (get_best_mode): Change bitregion_start and bitregion_end from
1397 unsigned HOST_WIDE_INT to poly_uint64.
1399 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1400 Alan Hayward <alan.hayward@arm.com>
1401 David Sherwood <david.sherwood@arm.com>
1403 * rtl.h (simplify_gen_subreg): Add a temporary overload that
1404 accepts poly_uint64 offsets.
1405 * expmed.h (extract_bit_field): Take bitsize and bitnum as
1406 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1407 * expmed.c (lowpart_bit_field_p): Likewise.
1408 (extract_bit_field_as_subreg): New function, split out from...
1409 (extract_bit_field_1): ...here. Take bitsize and bitnum as
1410 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
1411 extractions, check that BITSIZE matches the size of the extracted
1412 value and that BITNUM is an exact multiple of that size.
1413 If all else fails, try forcing the value into memory if
1414 BITNUM is variable, and adjusting the address so that the
1415 offset is constant. Split the part that can only handle constant
1416 bitsize and bitnum out into...
1417 (extract_integral_bit_field): ...this new function.
1418 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
1419 rather than unsigned HOST_WIDE_INTs.
1421 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1422 Alan Hayward <alan.hayward@arm.com>
1423 David Sherwood <david.sherwood@arm.com>
1425 * expmed.h (store_bit_field): Take bitsize and bitnum as
1426 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1427 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
1428 that returns the byte size.
1429 (store_bit_field_1): Take bitsize and bitnum as
1430 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
1431 to simple_mem_bitfield_p. Split the part that can only handle
1432 constant bitsize and bitnum out into...
1433 (store_integral_bit_field): ...this new function.
1434 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
1435 than unsigned HOST_WIDE_INTs.
1436 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
1438 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1439 Alan Hayward <alan.hayward@arm.com>
1440 David Sherwood <david.sherwood@arm.com>
1442 * lra-int.h (lra_reg): Change offset from int to poly_int64.
1443 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
1445 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1446 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
1447 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
1448 offset as a poly_int64 rather than an int.
1449 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
1450 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1451 * lra-constraints.c (equiv_address_substitution): Track offsets
1453 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
1454 (curr_insn_transform): Handle the new form of sp_offset.
1455 * lra-eliminations.c (lra_elim_table): Change previous_offset
1456 and offset from HOST_WIDE_INT to poly_int64.
1457 (print_elim_table, update_reg_eliminate): Update accordingly.
1458 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
1459 (get_elimination): Update accordingly.
1460 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
1461 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1462 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
1463 poly_int64 offsets generally.
1464 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
1465 (mark_not_eliminable, init_elimination): Update accordingly.
1466 (remove_reg_equal_offset_note): Return a bool and pass the new
1467 offset back by pointer as a poly_int64.
1468 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
1469 rather than a HOST_WIDE_INT.
1470 (do_remat): Track offsets poly_int64s.
1471 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
1473 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1474 Alan Hayward <alan.hayward@arm.com>
1475 David Sherwood <david.sherwood@arm.com>
1477 * rtl.h (mem_attrs): Add a default constructor. Change size and
1478 offset from HOST_WIDE_INT to poly_int64.
1479 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
1480 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
1481 (widen_memory_access): Take the sizes and offsets as poly_int64s
1482 rather than HOST_WIDE_INTs.
1483 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
1484 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
1486 (adjust_offset_for_component_ref): Change the offset from a
1487 HOST_WIDE_INT to a poly_int64.
1488 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
1489 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
1490 * dce.c (find_call_stack_args): Likewise.
1491 * dse.c (record_store): Likewise.
1492 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
1493 * print-rtl.c (rtx_writer::print_rtx): Likewise.
1494 * read-rtl-function.c (test_loading_mem): Likewise.
1495 * rtlanal.c (may_trap_p_1): Likewise.
1496 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1497 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
1498 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
1499 (mem_attrs::mem_attrs): New function.
1500 (set_mem_attributes_minus_bitpos): Change bitpos from a
1501 HOST_WIDE_INT to poly_int64.
1502 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
1503 (clear_mem_offset, clear_mem_size, change_address)
1504 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
1505 initialize mem_attrs.
1506 (set_mem_offset, set_mem_size, adjust_address_1)
1507 (adjust_automodify_address_1, offset_address, widen_memory_access):
1508 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
1510 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1511 Alan Hayward <alan.hayward@arm.com>
1512 David Sherwood <david.sherwood@arm.com>
1514 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
1515 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
1516 rather than 0 to represent an unknown size. Assert that the size
1517 is known when the mode isn't BLKmode.
1518 (may_trap_p_1): Use -1 for unknown sizes.
1519 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
1521 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1522 Alan Hayward <alan.hayward@arm.com>
1523 David Sherwood <david.sherwood@arm.com>
1525 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
1526 to poly_int64. Update commentary for positions_needed.large.
1527 (read_info_type): Change offset and width from HOST_WIDE_INT
1529 (set_usage_bits): Likewise.
1530 (canon_address): Return the offset as a poly_int64 rather than
1531 a HOST_WIDE_INT. Use strip_offset_and_add.
1532 (set_all_positions_unneeded, any_positions_needed_p): Use
1533 positions_needed.large to track stores with non-constant widths.
1534 (all_positions_needed_p): Likewise. Take the offset and width
1535 as poly_int64s rather than ints. Assert that rhs is nonnull.
1536 (record_store): Cope with non-constant offsets and widths.
1537 Nullify the rhs of an earlier store if we can't tell which bytes
1539 (find_shift_sequence): Take the access_size and shift as poly_int64s
1541 (get_stored_val): Take the read_offset and read_width as poly_int64s
1542 rather than HOST_WIDE_INTs.
1543 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
1544 non-constant offsets and widths.
1546 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1547 Alan Hayward <alan.hayward@arm.com>
1548 David Sherwood <david.sherwood@arm.com>
1550 * inchash.h (inchash::hash::add_poly_int): New function.
1551 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1552 Use poly_int64 rather than HOST_WIDE_INT.
1553 (ao_ref::max_size_known_p): New function.
1554 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1555 rather than HOST_WIDE_INT.
1556 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
1557 to temporaries until its interface is adjusted to match.
1558 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
1559 (aliasing_component_refs_p, decl_refs_may_alias_p)
1560 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
1561 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
1562 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
1564 * alias.c (ao_ref_from_mem): Likewise.
1565 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1566 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
1567 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
1568 (maybe_trim_complex_store, maybe_trim_constructor_store)
1569 (live_bytes_read, dse_classify_store): Likewise.
1570 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
1571 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
1572 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
1573 (vn_reference_lookup_3): Likewise.
1574 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
1576 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1577 Alan Hayward <alan.hayward@arm.com>
1578 David Sherwood <david.sherwood@arm.com>
1580 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
1581 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
1582 instead of ranges_overlap_p.
1584 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1585 Alan Hayward <alan.hayward@arm.com>
1586 David Sherwood <david.sherwood@arm.com>
1588 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
1589 sizes as poly_int64s rather than HOST_WIDE_INTs.
1591 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1592 Alan Hayward <alan.hayward@arm.com>
1593 David Sherwood <david.sherwood@arm.com>
1595 * gimple-fold.h (fold_ctor_reference): Take the offset and size
1596 as poly_uint64 rather than unsigned HOST_WIDE_INT.
1597 * gimple-fold.c (fold_ctor_reference): Likewise.
1599 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1600 Alan Hayward <alan.hayward@arm.com>
1601 David Sherwood <david.sherwood@arm.com>
1603 * target.def (dwarf_poly_indeterminate_value): New hook.
1604 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
1605 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
1606 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
1607 * doc/tm.texi: Regenerate.
1608 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
1609 offset as a poly_int64.
1610 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
1611 offset as a poly_int64.
1612 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
1613 Take the offset as a poly_int64.
1614 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1615 (frame_pointer_fb_offset): Change to a poly_int64.
1616 (int_loc_descriptor): Take the offset as a poly_int64. Use
1617 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1618 (based_loc_descr): Take the offset as a poly_int64.
1619 Use strip_offset_and_add to handle (plus X (const)).
1620 Use new_reg_loc_descr instead of an open-coded version of the
1621 previous implementation.
1622 (mem_loc_descriptor): Handle CONST_POLY_INT.
1623 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1624 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1626 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1627 Alan Hayward <alan.hayward@arm.com>
1628 David Sherwood <david.sherwood@arm.com>
1630 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1631 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1632 * inchash.h (inchash::hash::add_poly_hwi): New function.
1633 * gengtype.c (main): Register poly_int64.
1634 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1635 offset as a poly_int.
1636 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1637 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1638 offset as a poly_int64.
1639 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1640 * print-rtl.c (print_poly_int): New function.
1641 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1643 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1645 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1646 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1647 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1648 Enforce track_offset_p for parts of a PARALLEL.
1649 (vt_add_function_parameter): Use const_offset for the final
1650 offset to track. Use get_tracked_reg_offset for the parts
1653 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1654 Alan Hayward <alan.hayward@arm.com>
1655 David Sherwood <david.sherwood@arm.com>
1657 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1658 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1659 * doc/tm.texi: Regenerate.
1660 * hooks.h (hook_bool_uint_uint_true): Delete.
1661 (hook_bool_puint64_puint64_true): Declare.
1662 * hooks.c (hook_bool_uint_uint_true): Delete.
1663 (hook_bool_puint64_puint64_true): New function.
1664 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1665 instead of unsigned ints.
1666 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1667 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1669 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1670 Alan Hayward <alan.hayward@arm.com>
1671 David Sherwood <david.sherwood@arm.com>
1673 * optabs.h (expand_operand): Add an int_value field.
1674 (create_expand_operand): Add an int_value parameter and use it
1675 to initialize the new expand_operand field.
1676 (create_integer_operand): Replace with a declaration of a function
1677 that accepts poly_int64s. Move the implementation to...
1678 * optabs.c (create_integer_operand): ...here.
1679 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1680 the mode preserves the value of int_value, instead of calling
1681 const_int_operand on the rtx. Use gen_int_mode to generate
1684 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1685 Alan Hayward <alan.hayward@arm.com>
1686 David Sherwood <david.sherwood@arm.com>
1688 * dumpfile.h (dump_dec): Declare.
1689 * dumpfile.c (dump_dec): New function.
1690 * pretty-print.h (pp_wide_integer): Turn into a function and
1691 declare a poly_int version.
1692 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1694 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1695 Alan Hayward <alan.hayward@arm.com>
1696 David Sherwood <david.sherwood@arm.com>
1698 * doc/generic.texi (POLY_INT_CST): Document.
1699 * tree.def (POLY_INT_CST): New tree code.
1700 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1701 * tree-core.h (tree_poly_int_cst): New struct.
1702 (tree_node): Add a poly_int_cst field.
1703 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1704 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1705 instead of a wide_int_ref.
1706 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1708 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1709 instead of an unsigned HOST_WIDE_INT.
1710 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1711 (ptrdiff_tree_p): Declare.
1712 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1713 extern inline implementations if the target doesn't use POLY_INT_CST.
1714 (poly_int_tree_p): New function.
1715 (wi::unextended_tree): New class.
1716 (wi::int_traits <unextended_tree>): New override.
1717 (wi::extended_tree): Add a default constructor.
1718 (wi::extended_tree::get_tree): New function.
1719 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1720 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1721 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1722 (wi::tree_to_poly_wide_ref): New typedefs.
1723 (wi::ints_for): Provide overloads for extended_tree and
1725 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1726 (wi::to_wide): New functions.
1727 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1728 * tree.c (poly_int_cst_hasher): New struct.
1729 (poly_int_cst_hash_table): New variable.
1730 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1731 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1733 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1734 (init_ttree): Initialize poly_int_cst_hash_table.
1735 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1736 a poly_int64 instead of a HOST_WIDE_INT.
1737 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1738 instead of an unsigned HOST_WIDE_INT.
1739 (wide_int_to_tree): Rename to...
1740 (wide_int_to_tree_1): ...this.
1741 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1742 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1743 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1744 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1745 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1746 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1748 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1749 (streamer_read_tree_body): Likewise.
1750 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1751 (streamer_write_tree_body): Likewise.
1752 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1753 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1754 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1755 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1756 * gimple-expr.h (is_gimple_constant): Likewise.
1757 * gimplify.c (maybe_with_size_expr): Likewise.
1758 * print-tree.c (print_node): Likewise.
1759 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1760 * tree-pretty-print.c (dump_generic_node): Likewise.
1761 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1762 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1763 * tree-vrp.c (compare_values_warnv): Likewise.
1764 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1765 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1766 (force_expr_to_var_cost): Likewise.
1767 * tree-ssa-loop.c (for_each_index): Likewise.
1768 * fold-const.h (build_invariant_address, size_int_kind): Take a
1769 poly_int64 instead of a HOST_WIDE_INT.
1770 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1771 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1773 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1774 (int_const_binop_2): New function, split out from...
1775 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1776 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1777 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1778 * gimple-ssa-strength-reduction.c (slsr_process_add)
1779 (slsr_process_mul): Check for INTEGER_CSTs before using them
1781 * stor-layout.c (bits_from_bytes): New function.
1782 (bit_from_pos): Use it.
1783 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1784 by BITS_PER_UNIT to get the TYPE_SIZE.
1785 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1786 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1788 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1789 Alan Hayward <alan.hayward@arm.com>
1790 David Sherwood <david.sherwood@arm.com>
1792 * doc/rtl.texi (const_poly_int): Document. Also document the
1793 rtl sharing behavior.
1794 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1795 * rtl.h (const_poly_int_def): New struct.
1796 (rtx_def::u): Add a cpi field.
1797 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1798 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1799 (wi::rtx_to_poly_wide_ref): New typedef
1800 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1801 (poly_int_rtx_p): New functions.
1802 (trunc_int_for_mode): Declare a poly_int64 version.
1803 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1804 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1806 (strip_offset): Declare.
1807 (strip_offset_and_add): New function.
1808 * rtl.def (CONST_POLY_INT): New rtx code.
1809 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1810 (shared_const_p): Use poly_int_rtx_p.
1811 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1813 (gen_int_shift_amount): Likewise.
1814 * emit-rtl.c (const_poly_int_hasher): New class.
1815 (const_poly_int_htab): New variable.
1816 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1817 (const_poly_int_hasher::hash): New function.
1818 (const_poly_int_hasher::equal): Likewise.
1819 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1820 (immed_wide_int_const): Rename to...
1821 (immed_wide_int_const_1): ...this and make static.
1822 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1823 instead of a wide_int_ref.
1824 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1825 (gen_lowpart_common): Handle CONST_POLY_INT.
1826 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1827 * cselib.c (cselib_hash_rtx): Likewise.
1828 * dwarf2out.c (const_ok_for_output_1): Likewise.
1829 * expr.c (convert_modes): Likewise.
1830 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1831 * rtlhash.c (add_rtx): Likewise.
1832 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1833 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1834 Handle existing CONST_POLY_INT rtxes.
1835 * expmed.h (expand_shift): Take a poly_int64 instead of a
1837 * expmed.c (expand_shift): Likewise.
1838 * rtlanal.c (strip_offset): New function.
1839 (commutative_operand_precedence): Give CONST_POLY_INT the same
1840 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1842 * rtl-tests.c (const_poly_int_tests): New struct.
1843 (rtl_tests_c_tests): Use it.
1844 * simplify-rtx.c (simplify_const_unary_operation): Handle
1846 (simplify_const_binary_operation): Likewise.
1847 (simplify_binary_operation_1): Fold additions of symbolic constants
1848 and CONST_POLY_INTs.
1849 (simplify_subreg): Handle extensions and truncations of
1851 (simplify_const_poly_int_tests): New struct.
1852 (simplify_rtx_c_tests): Use it.
1853 * wide-int.h (storage_ref): Add default constructor.
1854 (wide_int_ref_storage): Likewise.
1855 (trailing_wide_ints): Use GTY((user)).
1856 (trailing_wide_ints::operator[]): Add a const version.
1857 (trailing_wide_ints::get_precision): New function.
1858 (trailing_wide_ints::extra_size): Likewise.
1860 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1861 Alan Hayward <alan.hayward@arm.com>
1862 David Sherwood <david.sherwood@arm.com>
1864 * emit-rtl.h (gen_int_shift_amount): Declare.
1865 * emit-rtl.c (gen_int_shift_amount): New function.
1866 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
1868 * calls.c (shift_return_value): Likewise.
1869 * cse.c (fold_rtx): Likewise.
1870 * dse.c (find_shift_sequence): Likewise.
1871 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
1872 (expand_shift, expand_smod_pow2): Likewise.
1873 * lower-subreg.c (shift_cost): Likewise.
1874 * optabs.c (expand_superword_shift, expand_doubleword_mult)
1875 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
1876 (expand_vec_perm_var): Likewise.
1877 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1878 (simplify_binary_operation_1): Likewise.
1879 * combine.c (try_combine, find_split_point, force_int_to_mode)
1880 (simplify_shift_const_1, simplify_shift_const): Likewise.
1881 (change_zero_ext): Likewise. Use simplify_gen_binary.
1883 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1885 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
1887 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1889 * doc/invoke.texi (ARM Options): Document accepted extension options
1890 for -march=armv8.3-a.
1892 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
1895 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
1896 or --with-float=softfp, set the default CPU to arm10e.
1898 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
1900 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
1901 * config/visium/predicates.md (const_shift_operand): Likewise.
1902 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
1903 (visium_legitimize_reload_address): Likewise.
1905 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1908 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
1910 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1913 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
1915 2017-12-20 Julia Koval <julia.koval@intel.com>
1917 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
1918 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
1919 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
1920 * config.gcc: Include vpclmulqdqintrin.h.
1921 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
1922 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
1924 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
1925 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
1926 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
1927 * config/i386/i386.c (isa2_opts): Add -mcx16.
1928 (isa_opts): Add -mpclmulqdq, remove -mcx16.
1929 (ix86_option_override_internal): Move mcx16 to flags2.
1930 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
1931 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
1932 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
1933 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
1934 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
1935 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
1936 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
1937 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
1938 * doc/invoke.texi: Add -mvpclmulqdq.
1940 2017-12-20 Tom de Vries <tom@codesourcery.com>
1943 * config/i386/i386.c (ix86_static_chain): Move
1944 DECL_STATIC_CHAIN test ...
1945 * calls.c (rtx_for_static_chain): ... here. New function.
1946 * calls.h (rtx_for_static_chain): Declare.
1947 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
1948 instead of targetm.calls.static_chain.
1949 * df-scan.c (df_get_entry_block_def_set): Same.
1951 2017-12-19 Tom de Vries <tom@codesourcery.com>
1953 PR tree-optimization/83493
1954 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
1957 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1959 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
1960 inform with hardcoded english plural handling.
1962 2017-12-18 Jeff Law <law@redhat.com>
1964 PR tree-optimization/83477
1965 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
1966 a non-virtual PHI, always push a new range.
1968 2017-12-19 Martin Sebor <msebor@redhat.com>
1971 * builtins.c (compute_objsize): Handle non-constant offsets.
1973 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1975 PR tree-optimization/83444
1976 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
1977 character load case, if get_stridx on MEM_REF's operand doesn't
1978 look usable, retry with get_addr_stridx.
1980 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
1983 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
1984 (variable_tracking_main_1): Keep markers even when VTA fails.
1987 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
1988 even if there are markers before it.
1989 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
1991 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1993 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
1994 typical order conditions.
1995 * sel-sched.c (extract_new_fences_from): Likewise.
1996 * config/visium/constraints.md (J, K, L): Likewise.
1997 * config/visium/predicates.md (const_shift_operand): Likewise.
1998 * config/visium/visium.c (visium_legitimize_address,
1999 visium_legitimize_reload_address): Likewise.
2000 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2001 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2002 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2003 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2004 SET_DUMP_DETAIL): Likewise.
2005 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2006 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2007 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2008 avr_set_core_architecture, avr_set_current_function,
2009 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2010 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2011 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2012 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2013 avr_map_decompose, avr_fold_builtin): Likewise.
2014 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2015 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2016 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2017 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2018 * config/m32c/m32c.c (m32c_conditional_register_usage,
2019 m32c_address_cost): Likewise.
2020 * config/m32c/predicates.md (shiftcount_operand,
2021 longshiftcount_operand): Likewise.
2022 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2023 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2024 can_use_cdx_ldstw): Likewise.
2025 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2026 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2028 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2029 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2030 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2031 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2032 * config/vax/vax.c (adjacent_operands_p): Likewise.
2033 * config/ft32/constraints.md (L, b, KA): Likewise.
2034 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2036 * cfgexpand.c (expand_stack_alignment): Likewise.
2037 * gcse.c (insert_expr_in_table): Likewise.
2038 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2039 * cgraphunit.c (cgraph_node::expand): Likewise.
2040 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2041 * emit-rtl.c (add_insn): Likewise.
2042 * input.c (dump_location_info): Likewise.
2043 * passes.c (NEXT_PASS): Likewise.
2044 * read-rtl-function.c (parse_note_insn_name,
2045 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2047 * sched-rgn.c (sched_rgn_init): Likewise.
2048 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2049 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2050 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2052 * explow.c (eliminate_constant_term): Likewise.
2053 * final.c (leaf_renumber_regs_insn): Likewise.
2054 * cfgrtl.c (print_rtl_with_bb): Likewise.
2055 * genhooks.c (emit_init_macros): Likewise.
2056 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2057 * tree-data-ref.c (conflict_fn): Likewise.
2058 * selftest.c (assert_streq): Likewise.
2059 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2060 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2061 fold_binary_loc, multiple_of_p): Likewise.
2062 * reload.c (push_reload, find_equiv_reg): Likewise.
2063 * et-forest.c (et_nca, et_below): Likewise.
2064 * dbxout.c (dbxout_symbol_location): Likewise.
2065 * reorg.c (relax_delay_slots): Likewise.
2066 * dojump.c (do_compare_rtx_and_jump): Likewise.
2067 * gengtype-parse.c (type): Likewise.
2068 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2069 simplify_const_relational_operation): Likewise.
2070 * reload1.c (do_output_reload): Likewise.
2071 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2072 * gengtype.c (type_for_name): Likewise.
2073 * gimple-ssa-sprintf.c (format_directive): Likewise.
2075 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2078 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2079 accessing it. Adjust comment.
2081 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2084 * predict.c (zero_one_minusone): New function.
2085 (apply_return_prediction): Avoid return prediction for functions
2086 returning only -1, 0 and 1 values, unless they only return -1 and 0
2089 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2091 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2092 fall-through warning.
2093 (arc_compute_frame_size): Remove unused variables.
2094 (arc_print_operand): Fix fprintif format.
2095 (arc_can_follow_jump): Clean fall-through warning.
2097 2017-12-19 Marek Polacek <polacek@redhat.com>
2100 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2103 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2105 * config/arc/arc.c (overriderregs): New variable.
2106 (arc_override_options): Track fixed/call saved/call options.
2107 (arc_conditional_register_usage): Check against overrideregs
2108 variable whenever we change register properties.
2110 2017-12-19 Nathan Sidwell <nathan@acm.org>
2112 * opts.c (finish_options): Don't prefix dump_base_name if it
2113 already contains directories.
2115 2017-12-19 Martin Liska <mliska@suse.cz>
2117 PR rtl-optimization/82675
2118 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
2119 more element in sbitmap.
2121 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
2123 * gimplify.c (gimplify_expr): Use error_operand_p.
2125 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
2128 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
2131 2017-12-19 Daniel Cederman <cederman@gaisler.com>
2133 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
2134 the jump is to a label.
2136 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2138 PR tree-optimization/83444
2139 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
2144 * ipa-inline.c (flatten_remove_node_hook): New function.
2145 (ipa_inline): Keep only nodes with flatten attribute at the end of
2146 the array in the order from ipa_reverse_postorder, only walk that
2147 portion of array for flattening, if there is more than one such
2148 node, temporarily register a removal hook and ignore removed nodes.
2150 PR tree-optimization/80631
2151 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
2152 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
2153 IFN_REDUC_MAX or IFN_REDUC_MIN.
2155 2017-12-18 Jeff Law <law@redhat.com>
2157 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
2158 record symbolic equivalences from backedges in the CFG.
2161 2017-11-19 Jeff Law <law@redhat.com>
2163 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2164 of degenerates resulting from ignoring an edge.
2166 2017-12-18 Martin Sebor <msebor@redhat.com>
2169 PR tree-optimization/78450
2170 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
2171 (handle_builtin_strlen): Call it.
2173 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
2175 PR rtl-optimization/83424
2176 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
2178 2017-12-18 Marek Polacek <polacek@redhat.com>
2181 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
2182 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
2185 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2187 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
2188 (umaddsidi4, umaddsidi_split): Likewise.
2190 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2192 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
2195 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2198 * config/s390/s390.c (s390_option_override): Avoid strncpy.
2200 2017-12-18 Richard Biener <rguenther@suse.de>
2202 PR tree-optimization/81877
2203 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
2204 (outermost_indep_loop): Adjust.
2205 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
2206 (can_sm_ref_p): Adjust.
2208 2017-12-18 Richard Biener <rguenther@suse.de>
2211 * tree.c (array_at_struct_end_p): Return true if the underlying
2212 object has space for at least one element in excess of what
2213 the array domain specifies.
2215 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2217 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
2218 example to avoid overfull hbox.
2219 * doc/invoke.texi (Option Summary): Add missing @gol.
2220 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
2223 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2224 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2226 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
2227 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
2228 (AArch64 Options): Document -mverbose-cost-dump.
2229 (ARM Options): Likewise, plus -mflip-thumb.
2231 2017-12-17 Martin Sebor <msebor@redhat.com>
2234 * gimple-ssa-warn-restrict.c
2235 (builtin_memref::offset_out_of_bounds): Correct the handling of
2238 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2240 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
2242 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
2244 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
2246 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
2248 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
2251 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
2253 * doc/invoke.texi: Fix some typos.
2255 2017-12-16 Martin Sebor <msebor@redhat.com>
2257 PR tree-optimization/78918
2258 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
2259 * builtins.c (check_sizes): Rename...
2260 (check_access): ...to this. Rename function arguments for clarity.
2261 (check_memop_sizes): Adjust names.
2262 (expand_builtin_memchr, expand_builtin_memcpy): Same.
2263 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
2264 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
2265 (check_strncat_sizes, expand_builtin_strncat): Same.
2266 (expand_builtin_strncpy, expand_builtin_memset): Same.
2267 (expand_builtin_bzero, expand_builtin_memcmp): Same.
2268 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
2269 (maybe_emit_sprintf_chk_warning): Same.
2270 (expand_builtin_strcpy): Adjust.
2271 (expand_builtin_stpcpy): Same.
2272 (expand_builtin_with_bounds): Detect out-of-bounds accesses
2273 in pointer-checking forms of memcpy, memmove, and mempcpy.
2274 (gcall_to_tree_minimal, max_object_size): Define new functions.
2275 * builtins.h (max_object_size): Declare.
2276 * calls.c (alloc_max_size): Call max_object_size instead of
2277 hardcoding ssizetype limit.
2278 (get_size_range): Handle new argument.
2279 * calls.h (get_size_range): Add a new argument.
2280 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
2281 * doc/invoke.texi (-Wrestrict): Adjust, add example.
2282 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
2284 (gimple_fold_builtin_memory_chk): Same.
2285 (gimple_fold_builtin_stxcpy_chk): New function.
2286 * gimple-ssa-warn-restrict.c: New source.
2287 * gimple-ssa-warn-restrict.h: New header.
2288 * gimple.c (gimple_build_call_from_tree): Propagate location.
2289 * passes.def (pass_warn_restrict): Add new pass.
2290 * tree-pass.h (make_pass_warn_restrict): Declare.
2291 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
2293 (handle_builtin_strcat): Same.
2294 (strlen_optimize_stmt): Rename...
2295 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
2296 stpncpy, strncpy, and their checking forms.
2298 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
2300 PR rtl-optimization/82849
2301 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
2302 and get_max_loop_iterations_int.
2304 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2305 Alan Hayward <alan.hayward@arm.com>
2306 David Sherwood <david.sherwood@arm.com>
2308 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
2309 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
2310 size as a poly_uint64.
2311 (mode_for_vector, mode_for_int_vector): Take the number of vector
2312 elements as a poly_uint64.
2313 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
2314 size as a poly_uint64.
2315 (mode_for_vector, mode_for_int_vector): Take the number of vector
2316 elements as a poly_uint64.
2318 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2319 Alan Hayward <alan.hayward@arm.com>
2320 David Sherwood <david.sherwood@arm.com>
2322 * machmode.h (MACRO_MODE): New macro.
2323 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
2324 * targhooks.c (default_libcall_value, default_secondary_reload)
2325 (default_memory_move_cost, default_register_move_cost)
2326 (default_class_max_nregs): Likewise.
2328 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2329 Alan Hayward <alan.hayward@arm.com>
2330 David Sherwood <david.sherwood@arm.com>
2332 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
2333 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
2335 * genautomata.c (main): Likewise.
2336 * genconditions.c (write_header): Likewise.
2337 * genemit.c (main): Likewise.
2338 * genextract.c (print_header): Likewise.
2339 * genopinit.c (main): Likewise.
2340 * genoutput.c (output_prologue): Likewise.
2341 * genpeep.c (main): Likewise.
2342 * genpreds.c (write_insn_preds_c): Likewise.
2343 * genrecog.c (writer_header): Likewise.
2344 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
2345 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
2346 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
2347 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
2348 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
2349 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
2350 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
2351 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
2352 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
2353 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
2354 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
2355 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
2356 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
2357 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
2358 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
2359 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
2360 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
2361 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
2362 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
2363 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
2364 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
2365 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
2366 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
2367 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
2368 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
2369 * config/darwin.c (IN_TARGET_CODE): Likewise.
2370 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
2371 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
2372 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
2373 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
2374 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
2375 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
2376 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
2377 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
2378 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
2379 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
2380 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
2381 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
2382 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
2383 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
2384 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
2385 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
2386 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
2387 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
2388 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
2389 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
2390 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
2391 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
2392 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
2393 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
2394 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
2395 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
2396 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
2397 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
2398 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
2399 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
2400 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
2401 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
2402 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
2403 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
2404 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
2405 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
2406 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
2407 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
2408 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
2409 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
2410 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
2411 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
2412 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
2413 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
2414 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
2415 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
2416 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
2417 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
2418 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
2419 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
2420 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
2421 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
2422 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
2423 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
2424 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
2425 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
2426 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
2427 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
2428 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
2429 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
2430 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2431 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
2432 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
2433 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
2434 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
2435 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
2436 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
2437 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
2438 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
2439 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
2440 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
2441 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2442 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
2443 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
2444 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
2445 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
2446 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
2447 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
2448 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
2449 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
2450 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
2451 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
2452 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
2453 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
2454 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
2455 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
2456 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
2457 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
2458 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
2459 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
2460 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
2461 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
2462 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
2463 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
2464 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
2465 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
2466 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
2467 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
2468 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
2469 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
2470 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
2471 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
2472 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
2473 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
2474 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
2475 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
2477 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2478 Alan Hayward <alan.hayward@arm.com>
2479 David Sherwood <david.sherwood@arm.com>
2481 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
2484 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2485 Alan Hayward <alan.hayward@arm.com>
2486 David Sherwood <david.sherwood@arm.com>
2488 * doc/generic.texi (VEC_SERIES_EXPR): Document.
2489 * doc/md.texi (vec_series@var{m}): Document.
2490 * tree.def (VEC_SERIES_EXPR): New tree code.
2491 * tree.h (build_vec_series): Declare.
2492 * tree.c (build_vec_series): New function.
2493 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
2494 * tree-pretty-print.c (dump_generic_node): Likewise.
2495 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
2496 * tree-inline.c (estimate_operator_cost): Likewise.
2497 * expr.c (expand_expr_real_2): Likewise.
2498 * optabs-tree.c (optab_for_tree_code): Likewise.
2499 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
2500 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
2501 * expmed.c (make_tree): Handle VEC_SERIES.
2502 * optabs.def (vec_series_optab): New optab.
2503 * optabs.h (expand_vec_series_expr): Declare.
2504 * optabs.c (expand_vec_series_expr): New function.
2505 * tree-vect-generic.c (expand_vector_operations_1): Check that
2506 the operands also have vector type.
2508 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2509 Alan Hayward <alan.hawyard@arm.com>
2510 David Sherwood <david.sherwood@arm.com>
2512 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
2513 (VEC_COND_EXPR): Add missing @tindex.
2514 * doc/md.texi (vec_duplicate@var{m}): Document.
2515 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
2516 * tree.c (build_vector_from_val): Add stubbed-out handling of
2517 variable-length vectors, using VEC_DUPLICATE_EXPR.
2518 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
2519 * cfgexpand.c (expand_debug_expr): Likewise.
2520 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2521 * tree-inline.c (estimate_operator_cost): Likewise.
2522 * tree-pretty-print.c (dump_generic_node): Likewise.
2523 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
2524 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
2525 (test_vec_duplicate_folding): New function.
2526 (fold_const_c_tests): Call it.
2527 * optabs.def (vec_duplicate_optab): New optab.
2528 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
2529 * optabs.h (expand_vector_broadcast): Declare.
2530 * optabs.c (expand_vector_broadcast): Make non-static. Try using
2531 vec_duplicate_optab.
2532 * expr.c (store_constructor): Try using vec_duplicate_optab for
2534 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
2536 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
2539 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
2540 div/mod latencies a bit.
2542 2017-12-15 Jeff Law <law@redhat.com>
2544 PR tree-optimization/36550
2545 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
2546 (mark_threaded_blocks): Rewrite code to avoid block copying when
2547 optimizing for size. Don't pessimize blocks which will be
2548 copied, but all the statements will be dead.
2550 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
2552 PR tree-optimization/81165
2553 * tree-ssa-threadupdate.c (uses_in_bb): New.
2554 (estimate_threading_killed_stmts): New.
2555 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
2556 * tree-ssa-threadedge.c
2557 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
2560 2017-12-15 Jeff Law <law@redhat.com>
2562 PR tree-optimization/83410
2563 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
2564 threads when parallelizing loops.
2566 2017-12-15 Jakub Jelinek <jakub@redhat.com>
2568 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
2570 * config/alpha/alpha.c (vms_attribute_table): Swap
2571 affects_type_identity and handler fields, adjust comments.
2572 * config/mips/mips.c (mips_attribute_table): Likewise.
2573 * config/visium/visium.c (visium_attribute_table): Likewise.
2574 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
2575 * config/microblaze/microblaze.c (microblaze_attribute_table):
2577 * config/spu/spu.c (spu_attribute_table): Likewise.
2578 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
2579 * config/arc/arc.c (arc_attribute_table): Likewise.
2580 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
2581 * config/v850/v850.c (v850_handle_interrupt_attribute,
2582 v850_handle_data_area_attribute): Formatting fixes.
2583 (v850_attribute_table): Swap affects_type_identity and handler
2584 fields, adjust comments.
2585 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
2586 * config/arm/arm.c (arm_attribute_table): Likewise.
2587 * config/avr/avr.c (avr_attribute_table): Likewise.
2588 * config/s390/s390.c (s390_attribute_table): Likewise.
2589 * config/sh/sh.c (sh_attribute_table): Likewise.
2590 * config/i386/i386.c (ix86_handle_cconv_attribute,
2591 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
2592 (ix86_attribute_table): Swap affects_type_identity and handler
2593 fields, adjust comments.
2594 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2595 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
2596 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
2597 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
2598 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
2599 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2600 * config/rx/rx.c (rx_attribute_table): Likewise.
2601 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
2602 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
2603 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
2604 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
2605 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2606 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
2607 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
2608 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
2609 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2610 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2611 * doc/plugins.texi (user_attr): Likewise. Add NULL for
2613 * attribs.c (empty_attribute_table): Swap affects_type_identity and
2615 (register_scoped_attributes, decl_attributes): Formatting fixes.
2617 PR tree-optimization/83269
2618 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2619 subtraction in arg0's type if type is signed and arg0 is unsigned.
2623 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2624 convert? on @0 instead of convert. Check type of @1, not @0.
2625 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2626 convert? on @0 instead of convert. Check type of @1, not @0.
2627 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2628 only optimize if either both @1 and @2 types are narrower
2629 precision, or both are wider or equal precision, and in the former
2630 case only if both have undefined overflow.
2632 2017-12-15 Richard Biener <rguenther@suse.de>
2635 * internal-fn.def (IFN_NOP): Add.
2636 * internal-fn.c (expand_NOP): Do nothing.
2637 * lto-streamer-in.c (input_function): Instead of removing
2638 sanitizer calls replace them with IFN_NOP calls.
2640 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2641 Alan Hayward <alan.hayward@arm.com>
2642 David Sherwood <david.sherwood@arm.com>
2644 * dse.c (store_info, read_info_type): Replace begin and end with
2646 (print_range): New function.
2647 (set_all_positions_unneeded, any_positions_needed_p)
2648 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2650 (record_store): Likewise. Optimize the case in which all positions
2652 (get_stored_val): Replace read_begin and read_end with read_offset
2654 (replace_read): Update call accordingly.
2656 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2658 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2659 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2660 (loop_cand::loop_cand): Initialize above members.
2661 (loop_cand::supported_operations): Delete.
2662 (loop_cand::can_interchange_p): Inline above function.
2663 (loop_cand::classify_simple_reduction): Record number of constant
2664 initialized simple reductions.
2665 (should_interchange_loops): New parameters. Check stmt cost of loops
2667 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2668 Update call to should_interchange_loops.
2669 (should_interchange_loop_nest): Update call to
2670 should_interchange_loops.
2672 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2675 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2676 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2677 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2679 2017-12-15 Julia Koval <julia.koval@intel.com>
2681 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2682 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
2684 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2685 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2686 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2688 2017-12-15 Julia Koval <julia.koval@intel.com>
2690 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2691 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2692 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2693 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2694 _mm_aesenc_epi128): New intrinsics.
2696 2017-12-15 Julia Koval <julia.koval@intel.com>
2698 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2699 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
2701 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2702 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2703 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2705 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2707 * gimple-ssa-strength-reduction.c (analyze_increments):
2708 Distinguish replacement costs for constant strides from those for
2711 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2713 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2715 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2717 * doc/invoke.texi: Document -Wcast-function-type.
2718 * recog.h (stored_funcptr): Change signature.
2719 * tree-dump.c (dump_node): Avoid warning.
2720 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2722 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2725 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2728 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2730 PR tree-optimization/83312
2731 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2732 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2733 "val" param, and to cope with arbitrary basic blocks.
2734 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2735 handle NULL_TREE for "val", dropping "bb" param.
2736 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2737 drop "bb" param. Handle NULL_TREE for "val".
2738 (find_case_label_for_value): Make "switch_stmt" param const.
2739 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2741 (vrp_prop::check_all_array_refs): Reimplement as...
2742 (check_array_bounds_dom_walker::before_dom_children): ...this new
2743 vfunc. Replace linear search through BB block list, excluding
2744 those with non-executable in-edges via dominator walk.
2746 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2748 * config/arm/arm.opt (mverbose-cost-dump): New option.
2749 * config/arm/arm.c (arm_rtx_costs): Use it.
2751 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2754 * reload1.c (emit_input_reload_insns): Skip debug markers.
2756 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2758 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2762 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2763 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2764 * config/sh/sh.c: Skip debug insns besides notes.
2765 * config/sh/sh.md: Likewise.
2766 * config/sh/sh_treg_combine.cc: Likewise.
2767 * config/sh/sync.md: Likewise.
2769 2017-12-14 Tom de Vries <tom@codesourcery.com>
2771 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2772 Add item for weak_undefined.
2774 2017-12-14 Richard Biener <rguenther@suse.de>
2776 PR tree-optimization/67842
2777 * sese.h (bb_in_region): Remove #if 0'ed code.
2779 2017-12-14 Richard Biener <rguenther@suse.de>
2781 PR tree-optimization/83326
2782 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2783 may_be_zero parameter and handle it by not marking the first
2784 peeled copy as not exiting the loop.
2785 (try_peel_loop): Likewise.
2786 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2787 to handle the case of constant or zero iterations and perform
2788 loop header copying on-the-fly.
2790 2017-12-14 Richard Biener <rguenther@suse.de>
2792 PR tree-optimization/83418
2793 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2794 Instead of asserting we don't get unfolded comparisons deal with them.
2796 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2799 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2800 basic blocks. Assert debug bind insns don't appear outside of bbs,
2801 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2804 PR tree-optimization/83198
2805 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2806 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2807 value if arg is a REAL_CST with incompatible type.
2809 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2810 Bin Cheng <bin.cheng@arm.com>
2813 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2815 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2818 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2820 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2821 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2823 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2824 Alan Hayward <alan.hayward@arm.com>
2825 David Sherwood <david.sherwood@arm.com>
2827 * poly-int.h: New file.
2828 * poly-int-types.h: Likewise.
2829 * coretypes.h: Include them.
2830 (POLY_INT_CONVERSION): Define.
2831 * target.def (estimated_poly_value): New hook.
2832 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2833 * doc/tm.texi: Regenerate.
2834 * doc/poly-int.texi: New file.
2835 * doc/gccint.texi: Include it.
2836 * doc/rtl.texi: Describe restrictions on subreg modes.
2837 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2838 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2839 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2840 * targhooks.h (default_estimated_poly_value): Declare.
2841 * targhooks.c (default_estimated_poly_value): New function.
2842 * target.h (estimated_poly_value): Likewise.
2843 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2844 (wi::unary_traits): Delete.
2845 (wi::binary_traits::signed_shift_result_type): Define for
2846 offset_int << HOST_WIDE_INT, etc.
2847 (generic_wide_int::operator <<=): Define for all types and use
2848 wi::lshift instead of <<.
2849 (wi::hwi_with_prec): Add a default constructor.
2850 (wi::ints_for): New class.
2851 (operator <<): Define for all wide-int types.
2852 (operator /): New function.
2853 (operator %): Likewise.
2854 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2855 (ASSERT_MAYBE_NE_AT): New macros.
2857 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2858 Dominik Vogt <vogt@linux.vnet.ibm.com>
2861 * emit-rtl.c (init_emit): Remove ??? comment.
2862 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2863 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
2864 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
2865 alignment of 3 virtual registers to BITS_PER_WORD.
2867 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
2869 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
2871 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
2872 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
2874 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2877 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
2879 2017-12-13 Jakub Jelinek <jakub@redhat.com>
2881 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
2882 with the exception of debug begin stmt markers appear before
2886 * final.c (rest_of_handle_final): Call variable_tracking_main only
2887 if !flag_var_tracking.
2889 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2890 Jakub Jelinek <jakub@redhat.com>
2894 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
2895 labels debug stmts that can only appear after labels.
2897 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
2899 PR rtl-optimization/82398
2900 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
2901 EXPR_USEFULNESS in priority comparison.
2903 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
2905 PR rtl-optimization/83393
2906 * combine.c (move_deaths): If reg_stat points to a too new insn in
2907 last_death, do not use it: find the proper insn instead.
2909 2017-12-12 Jeff Law <law@redhat.com>
2911 PR tree-optimization/83298
2912 PR tree-optimization/83362
2913 PR tree-optimization/83383
2914 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
2915 push_value_range a public interface. Add new argument to
2916 record_ranges_from_stmt.
2917 * gimple-ssa-evrp-analyze.c
2918 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
2919 Update comments. Handle recording temporary equivalences.
2920 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
2921 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
2922 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
2923 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
2924 gimple-ssa-evrp-analyze.h.
2925 (record_temporary_equivalences_from_phis): Add new argument. When
2926 the PHI arg is an SSA_NAME, set the result's range to the range
2928 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
2929 from statements too.
2930 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
2931 Pass it down to children as needed.
2932 (thread_outgoing_edges): Likewise.
2933 (thread_across_edge): Likewise. Push/pop range state as needed.
2934 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
2936 2017-12-12 Julia Koval <julia.koval@intel.com>
2938 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
2939 (PTA_CANNONLAKE): Remove PTA_CLWB.
2941 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2944 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
2945 (vcondv2div2df): Likewise.
2946 (vconduv2dfv2di): Likewise.
2948 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2950 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
2951 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
2952 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
2953 ATTR_MATHFN_FPROUNDING_ERRNO.
2955 2017-12-12 Richard Biener <rguenther@suse.de>
2957 PR tree-optimization/83385
2958 * graphite-scop-detection.c (get_order, order): Remove.
2959 (bb_to_rpo): New global.
2961 (build_scops): Sort pbbs in RPO order.
2963 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
2965 * combine.c (simplify_set): Do not transform subregs to zero_extends
2966 if the destination mode is a vector mode.
2968 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2970 PR tree-optimization/83359
2971 * tree-cfg.h (fold_loop_internal_call): Declare.
2972 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
2973 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
2974 (find_loop_dist_alias): New function.
2975 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
2976 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
2979 PR tree-optimization/80631
2980 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
2981 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
2982 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
2983 hardcoding zero as the value if COND_EXPR is never true. For
2984 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
2985 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
2986 hardcoding MAX_EXPR as the reduction operation.
2987 (is_nonwrapping_integer_induction): Allow negative step.
2988 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
2989 vect_create_epilog_for_reduction, if no value is suitable, don't
2990 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
2992 2017-12-12 Richard Biener <rguenther@suse.de>
2994 PR tree-optimization/81889
2995 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
2996 range info from the non-wrapping IV instead of just the range
2999 2017-12-12 Julia Koval <julia.koval@intel.com>
3001 * config.gcc: Add vaesintrin.h.
3002 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3004 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3005 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3007 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3008 * config/i386/immintrin.h: Include vaesintrin.h.
3009 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3010 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3011 _mm_aesdec_epi128): New intrinsics.
3013 2017-12-12 Julia Koval <julia.koval@intel.com>
3015 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3016 OPTION_MASK_ISA_VAES_UNSET): New.
3017 (ix86_handle_option): Handle -mvaes.
3018 * config/i386/cpuid.h: Define bit_VAES.
3019 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3020 * config/i386/i386-c.c (__VAES__): New.
3021 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3022 (ix86_valid_target_attribute_inner_p): Ditto.
3023 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3024 * config/i386/i386.opt: Add -mvaes.
3025 * doc/invoke.texi: Ditto.
3027 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3029 * debug.h (gcc_debug_hooks): Add inline_entry.
3030 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3031 * debug.c (do_nothing_debug_hooks): Likewise.
3032 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3033 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3034 (dwarf2_lineno_debug_hooks): Likewise.
3036 * common.opt (gstatement-frontiers): New, setting
3037 debug_nonbind_markers_p.
3038 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
3039 * toplev.c (process_options): Autodetect value for debug statement
3041 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3042 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3044 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3045 markers. Integrate source bind into debug stmt expand loop.
3046 (pass_expand::execute): Check debug marker limit. Avoid deep
3047 TER and expand debug locations for debug bind insns only.
3048 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3050 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3051 * final.c (reemit_insn_block_notes): Take current block from
3052 nonbind markers. Declare note where it's first set.
3053 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3054 begin stmt markers if enabled.
3055 (notice_source_line): Handle nonbind markers. Fail if their
3056 location is unknown or that of builtins.
3057 (rest_of_handle_final): Convert begin stmt markers to notes if
3058 var-tracking didn't run.
3059 (rest_of_clean_state): Skip begin stmt markers.
3060 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3062 * function.c (allocate_struct_function): Set begin_stmt_markers.
3063 * function.h (struct function): Add debug_marker_count counter
3064 and debug_nonbind_markers flag.
3065 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3066 * gimple-low.c (lower_function_body): Adjust
3067 debug_nonbind_markers.
3068 (lower_stmt): Drop or skip gimple debug stmts.
3069 (lower_try_catch): Skip debug stmts.
3070 * gimple.c (gimple_build_debug_begin_stmt): New.
3071 (gimple_copy): Increment debug_marker_count if copying one.
3072 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3073 * gimplify.c (rexpr_location): New.
3074 (rexpr_has_location): New.
3075 (warn_switch_unreachable_r): Handle gimple debug stmts.
3076 (shortcut_cond_r): Call expr_location.
3078 (find_goto_label): New.
3079 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3081 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3083 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3084 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3085 (LANG_HOOKS_INITIALIZER): ... this.
3086 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3087 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3089 (update_ebb_live_info): Skip debug insn markers.
3090 * lra.c (debug_insn_static_data): Rename to...
3091 (debug_bind_static_data): ... this.
3092 (debug_marker_static_data): New.
3093 (lra_set_insn_recog_data): Select one of the above depending
3095 (lra_update_isn_regno_info): Don't assume debug insns have
3097 (push_insns): Skip debug insns.
3098 * lto-streamer-in.c (input_function): Drop debug stmts
3099 depending on active options. Adjust debug_nonbind_markers.
3100 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3101 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3102 begin stmt marker notes.
3103 (print_insn): Likewise.
3104 * recog.c (extract_insn): Recognize rtl for debug markers.
3105 * rtl.def (DEBUG_MARKER): New.
3106 * tree-inline.c: Include params.h.
3107 (remap_gimple_stmt): Handle nonbind markers.
3108 (maybe_move_debug_stmts_to_successors): Likewise.
3109 (copy_debug_stmt): Likewise.
3110 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
3111 markers regardless of no side effects.
3112 (tsi_link_before): Don't update container's side effects when adding
3113 a begin stmt marker.
3114 (tsi_link_after): Likewise.
3115 (expr_first): Skip begin stmt markers.
3116 (expr_last): Likewise.
3117 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
3118 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3119 Disregard nonbind markers.
3120 * tree.c (make_node_stat): Don't set side effects for begin stmt
3122 (build1_stat): Likewise.
3123 * tree.def (DEBUG_BEGIN_STMT): New.
3124 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
3125 * var-tracking.c (delete_debug_insns): Renamed to...
3126 (delete_vta_debug_insns): ... this.
3127 (reemit_marker_as_note): New.
3128 (vt_initialize): Reemit markers.
3129 (delete_vta_debug_insns): Likewise.
3130 (vt_debug_insns_local): Reemit or delete markers.
3131 (variable_tracking_main_1): Likewise.
3132 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
3133 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
3134 (gimple_debug_nonbind_marker_p): New.
3135 (gimple_build_debug_bind): Adjust.
3136 (gimple_build_debug_begin_stmt): New.
3137 * doc/invoke.texi (max-debug-marker-count): New param.
3138 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
3139 (debug_parameter_ref, debug_marker): New.
3140 (NOTE_INSN_BEGIN_STMT): New.
3141 (DEBUG_INSN): Describe begin stmt markers.
3143 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
3144 without debug insns, we wouldn't, but clean up debug insns
3145 after a control flow insn nevertheless.
3147 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
3148 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
3149 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
3150 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
3151 (rtl_tidy_fallthru_edge): Likewise.
3152 (rtl_verify_fallthru): Likewise.
3153 (rtl_verify_bb_layout): Likewise.
3154 (skip_insns_after_block): Likewise.
3155 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
3156 * dwarf2out.c: Include print-rtl.h.
3157 (dwarf2out_next_real_insn): New.
3158 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
3159 Dump debug binds in asm comments.
3160 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
3161 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
3162 callers to use gsi_start_nondebug_bb instead.
3163 (gsi_after_labels): Skip gimple debug stmts.
3164 (gsi_start_nondebug): New.
3165 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
3166 (proper_loop_form_for_interchange): Adjust.
3167 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
3168 * gimple.h (gimple_seq_last_nondebug_stmt): New.
3169 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
3170 (collect_fallthrough_labels): Likewise.
3171 (should_warn_for_implicit_fallthrough): Likewise.
3172 (warn_implicit_fallthrough_r): Likewise.
3173 (expand_FALLTHROUGH_r): Likewise.
3174 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
3175 (graphite_copy_stmts_from_block): Skip nonbind markers.
3176 * haifa-sched.c (sched_extend_bb): Skip debug insns.
3177 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
3178 * jump.c (clean_barriers): Skip debug insns.
3179 * omp-expand.c (expand_parallel_call): Skip debug insns.
3180 (expand_task_call): Likewise.
3181 (remove_exit_barrier): Likewise.
3182 (expand_omp_taskreg): Likewise.
3183 (expand_omp_for_init_counts): Likewise.
3184 (expand_omp_for_generic): Likewise.
3185 (expand_omp_for_static_nochunk): Likewise.
3186 (expand_omp_for_static_chunk): Likewise.
3187 (expand_omp_simd): Likewise.
3188 (expand_omp_taskloop_for_outer): Likewise.
3189 (expand_omp_taskloop_for_inner): Likewise.
3190 (expand_oacc_for): Likewise.
3191 (expand_omp_sections): Likewise.
3192 (expand_omp_single): Likewise.
3193 (expand_omp_synch): Likewise.
3194 (expand_omp_atomic_load): Likewise.
3195 (expand_omp_atomic_store): Likewise.
3196 (expand_omp_atomic_fetch_op): Likewise.
3197 (expand_omp_atomic_pipeline): Likewise.
3198 (expand_omp_atomic_mutex): Likewise.
3199 (expand_omp_target): Likewise.
3200 (grid_expand_omp_for_loop): Likewise.
3201 (grid_expand_target_grid_body): Likewise.
3202 (build_omp_regions_1): Likewise.
3203 * omp-low.c (check_combined_parallel): Skip debug stmts.
3204 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
3205 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
3206 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
3208 * tree-cfg.c (make_blobs_1): Skip debug stmts.
3209 (make_edges): Likewise.
3210 (cleanup_dead_labels): Likewise.
3211 (gimple_can_merge_blocks_p): Likewise.
3212 (stmt_starts_bb_p): Likewise.
3213 (gimple_block_label): Likewise.
3214 (gimple_redirect_edge_and_branch): Likewise.
3215 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
3217 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
3219 * tree-pretty-print (print_declaration): Omit initializer in slim
3221 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
3223 (eliminate_unnecessary_stmts): Stabilize block removal order.
3224 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
3225 * var-tracking.c (get_first_insn): New.
3226 (vt_emit_notes): Call it.
3227 (vt_initialize): Walk any insns before the first BB.
3228 (delete_debug_insns): Likewise.
3230 * gimple.h (enum gimple_debug_subcode): Add
3231 GIMPLE_DEBUG_BEGIN_STMT.
3232 (gimple_debug_begin_stmt_p): New.
3233 (gimple_debug_nonbind_marker_p): New.
3234 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
3235 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
3236 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
3237 * insn-notes.def (BEGIN_STMT): New.
3238 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
3239 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
3240 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
3241 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
3242 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
3243 (INSN_DEBUG_MARKER_KIND): New.
3244 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
3245 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
3246 (INSN_VAR_LOCATION_PTR): New.
3247 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
3249 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
3250 deps only if debug bind insns are enabled.
3251 (pass_expand::execute): Avoid deep TER and expand
3252 debug locations for debug bind insns only.
3253 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
3254 debug stmts special handling down to debug bind stmts.
3255 * combine.c (try_combine): Narrow debug insns special handling
3256 down to debug bind insns.
3257 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
3258 Narrow debug insns preexisting special handling down to debug
3260 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
3261 handling down to debug bind insns.
3262 * function.c (instantiate_virtual_regs): Skip debug markers,
3263 adjust handling of debug binds.
3264 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
3265 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
3266 * haifa-sched.c (schedule_insn): Narrow special handling of debug
3267 insns to debug bind insns.
3268 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
3269 special handling of debug stmts to debug bind stmts.
3270 * ipa-split.c (split_function): Likewise.
3271 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
3272 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
3274 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
3275 * regrename.c (build_def_use): Likewise.
3276 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
3277 (pass_cprop_hardreg): Narrow special casing of debug insns to
3279 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
3280 * reload1.c (reload): Likewise.
3281 * sese.c (sese_insert_phis_for_liveouts): Narrow special
3282 casing of debug stmts to debug bind stmts.
3283 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
3284 * ssa-iterators.h (num_imm_uses): Likewise.
3285 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
3286 debug stmts to debug bind stmts.
3287 * tree-inline.c (tree_function_versioning): Narrow special casing
3288 of debug stmts to debug bind stmts.
3289 * tree-loop-distribution.c (generate_loops_for_partition):
3290 Narrow special casing of debug stmts to debug bind stmts.
3291 * tree-sra.c (analyze_access_subtree): Narrow special casing
3292 of debug stmts to debug bind stmts.
3293 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
3294 stmts to debug bind stmts.
3295 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
3296 casing of debug stmts to debug bind stmts.
3297 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
3298 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
3299 casing of debug stmts to debug bind stmts.
3300 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3302 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
3303 debug stmts to debug bind stmts.
3304 (gimple_replace_ssa_lhs): Likewise.
3305 (insert_debug_temp_for_var_def): Likewise.
3306 (insert_debug_temps_for_defs): Likewise.
3307 (reset_debug_uses): Likewise.
3308 * tree-ssanames.c (release_ssa_name_fn): Likewise.
3309 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
3310 (adjust_debug_stmts): Likewise.
3311 (adjust_phi_and_debug_stmts): Likewise.
3312 (vect_do_peeling): Likewise.
3313 * tree-vect-loop.c (vect_transform_loop): Likewise.
3314 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
3315 * var-tracking.c (adjust_mems): Narrow special casing of debug
3316 insns to debug bind insns.
3317 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
3318 (compute_bb_dataflow, vt_find_locations): Likewise.
3319 (vt_expand_loc, emit_notes_for_changes): Likewise.
3320 (vt_init_cfa_base): Likewise.
3321 (vt_emit_notes): Likewise.
3322 (vt_initialize): Likewise.
3323 (vt_finalize): Likewise.
3325 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
3326 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
3327 (next_nonnote_nondebug_insn_bb): New.
3328 (prev_nonnote_nondebug_insn_bb): New.
3329 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
3330 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
3331 (prev_nonnote_nondebug_insn_bb): Declare.
3332 (next_nonnote_nondebug_insn_bb): Declare.
3333 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
3334 * cfgrtl.c (get_last_bb_insn): Likewise.
3335 * lra.c (push_insns): Likewise.
3337 2017-12-11 David Malcolm <dmalcolm@redhat.com>
3340 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
3342 * selftest.c (selftest::test_runner::test_runner): New ctor.
3343 (selftest::test_runner::~test_runner): New dtor.
3344 * selftest.h (class selftest::test_runner): New class.
3346 2017-12-11 Carl Love <cel@us.ibm.com>
3348 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
3349 vec_extract_fp32_from_shortl]): Add #defines.
3350 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
3351 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
3352 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
3353 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
3354 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
3355 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
3356 * doc/extend.texi: Add documentation for the added builtins.
3358 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
3360 PR rtl-optimization/80693
3361 PR rtl-optimization/81019
3362 PR rtl-optimization/81020
3363 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
3364 are not mentioned in i3. Place the REG_UNUSED note on i2,
3365 possibly modified to REG_DEAD, if it did not originate in i3.
3367 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3369 * recog.c (store_data_bypass_p_1): New function.
3370 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
3371 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
3373 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
3375 PR rtl-optimization/83361
3376 * ifcvt.c (if_convert): Call fixup_partitions.
3378 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
3380 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3381 early folding of splat_u{8,16,32}.
3383 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3385 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
3386 output_operand_lossage first argument with capital letter.
3387 (aarch64_override_options): Don't start error and sorry first argument
3388 with capital letter.
3390 2017-12-11 Andi Kleen <ak@linux.intel.com>
3392 PR gcov-profile/83355
3393 * auto-profile.c (string_table::get_index_by_decl): Don't
3394 recurse when abstract origin points to itself.
3396 2017-12-11 Bin Cheng <bin.cheng@arm.com>
3398 PR tree-optimization/83320
3399 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
3400 (prune_datarefs_not_in_loop): Ditto.
3402 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
3404 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
3406 2017-12-10 Jakub Jelinek <jakub@redhat.com>
3408 PR tree-optimization/83337
3409 * gimple-loop-interchange.cc (compute_access_stride): Handle
3410 bitfield DRs properly.
3412 2017-12-09 Jakub Jelinek <jakub@redhat.com>
3414 PR tree-optimization/83338
3415 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
3416 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
3417 vector of unsigned integers to vector of signed integers.
3419 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
3421 PR rtl-optimization/83317
3422 * lra-constraints.c (process_address_1): Add insn code check.
3424 2017-12-08 Michael Matz <matz@suse.de>
3426 Fix PR tree-optimization/83323
3427 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
3428 head-controlled loops and loop BBs.
3429 * common.opt (funroll-and-jam): Remove, instead ...
3430 (floop-unroll-and-jam): ... reuse this option.
3431 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
3432 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
3433 (-floop-unroll-and-jam): ... this option.
3435 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3437 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
3438 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
3439 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
3441 2017-12-08 Richard Biener <rguenther@suse.de>
3444 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
3445 handle accesses outside of zero-sized vars.
3447 2017-12-08 Martin Jambor <mjambor@suse.cz>
3449 PR tree-optimization/83141
3450 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
3451 test for MEM_REFs implicitely changing types with padding. Remove
3453 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
3455 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3457 * config/arc/arc.c (arc_attribute_table): Add exclusions to
3459 * config/avr/avr.c (avr_attribute_table): Likewise.
3460 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3461 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3462 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3463 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
3464 of struct attribute_spec.
3465 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3467 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
3470 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
3472 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
3475 * profile-count.c (profile_count::from_gcov_type): Move from
3476 profile-count.h; handle overflow.
3477 * profile-count.h (profile_count::from_gcov_type): Move offline.
3479 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
3481 PR rtl-optimization/83304
3482 * combine.c (move_deaths): If we do not know where a register died,
3485 2017-12-08 Richard Biener <rguenther@suse.de>
3487 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
3488 Provide -fopt-info-loop feedback when we interchange in a nest.
3490 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3492 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
3493 for armv6 ARM CPU IDs.
3495 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3497 * common/config/arm/arm-common.c: Include <algorithm>.
3498 (INCLUDE_VECTOR): Define.
3499 (compare_opt_names): New function.
3500 (arm_rewrite_selected_arch): Only strip out extensions that can be
3501 expressed through -mfpu. Sort the remaining extensions
3504 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3506 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
3507 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
3508 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
3509 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
3511 2017-06-08 Tristan Gingold <gindold@adacore.com>
3514 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
3515 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
3517 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3519 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
3520 of struct attribute_spec.
3522 2017-12-08 Julia Koval <julia.koval@intel.com>
3524 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
3525 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
3526 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
3527 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
3528 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
3529 _mm_maskz_dpwssds_epi32): Ditto.
3531 2017-12-08 Richard Biener <rguenther@suse.de>
3533 PR tree-optimization/81303
3534 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
3535 conditions try to create a comparison vector type matching
3536 the data vector type.
3537 (vectorizable_condition): Adjust.
3538 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3539 Leave invariant conditions alone in case we can vectorize those.
3541 2017-12-08 Julia Koval <julia.koval@intel.com>
3543 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
3544 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
3545 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
3546 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
3547 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
3549 2017-12-08 Julia Koval <julia.koval@intel.com>
3551 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
3552 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
3553 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
3554 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
3555 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
3556 _mm_maskz_dpbusds_epi32): New intrinsics.
3558 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3560 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
3562 (ld<bh>io_signed): Likewise.
3563 (st<bhw>io): Likewise.
3564 * config/nios2/predicates.md (ldstio_memory_operand): Allow
3565 SMALL_INT12 constant integer operand.
3567 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3569 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
3571 (nios2_large_constant_allowed): ...this. Adjust uses.
3572 (nios2_plus_symbolic_constant_p): Rename to...
3573 (nios2_plus_large_constant_p): ...this. Adjust uses.
3574 (nios2_legitimate_address_p): Correct CONST_INT handling.
3575 (nios2_symbolic_memory_operand_p): Rename to...
3576 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
3577 (nios2_large_constant_p): Check for large constant integers too.
3578 (nios2_split_large_constant): Handle constant integers.
3579 (nios2_split_symbolic_memory_operand): Rename to...
3580 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
3581 (nios2_legitimize_constant_address): Handle constant integers.
3582 (r0rel_constant_p): Handle small constant integers.
3583 (nios2_print_operand_address): Handle r0-relative integer addresses.
3584 * config/nios2/nios2-protos.h: Adjust for renamed functions.
3585 * config/nios2/nios2.md: Adjust for renamed functions.
3587 2017-12-07 Andrew Waterman <andrew@sifive.com>
3589 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
3590 (TARGET_HAVE_SRODATA_SECTION): New define.
3591 (riscv_select_section): New function.
3593 2017-12-08 Joseph Myers <joseph@codesourcery.com>
3594 Alexander Monakov <amonakov@ispras.ru>
3595 Jakub Jelinek <jakub@redhat.com>
3598 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
3600 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3602 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
3603 a bool success value. Don't call output_operand_lossage here.
3604 (aarch64_print_ldpstp_address): Return a bool success value.
3605 (aarch64_print_operand_address): Call output_addr_const if
3606 aarch64_print_address_internal fails.
3607 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
3608 'y'; call output_operand_lossage instead. Call output_operand_lossage
3609 if aarch64_print_ldpstp_address fails.
3611 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3613 * tree-vector-builder.h
3614 (tree_vector_builder::binary_encoded_nelts): Declare.
3615 * tree-vector-builder.c
3616 (tree_vector_builder::binary_encoded_nelts): New function.
3617 * fold-const.c (negate_expr_p): Likewise.
3618 (operand_equal_p, fold_checksum_tree): Likewise.
3619 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3620 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3621 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3622 (uniform_vector_p): Likewise.
3623 * varasm.c (const_hash_1, compare_constant): Likewise.
3624 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3625 (valid_lattice_transition): Operate directly on the VECTOR_CST
3627 * ipa-icf.c: Include tree-vector-builder.h.
3628 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3629 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3631 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3633 * tree.c (build_vector): Delete.
3634 * tree.h (build_vector): Make static and move into the self-testing
3637 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3639 * vector-builder.h (vector_builder::derived): New const overload.
3640 (vector_builder::elt): New function.
3641 * tree-vector-builder.h (tree_vector_builder::type): New function.
3642 (tree_vector_builder::apply_step): Declare.
3643 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3645 * gimple-fold.h (tree_vector_builder): Declare.
3646 (gimple_build_vector): Take a tree_vector_builder instead of a
3647 type and vector of elements.
3648 * gimple-fold.c (gimple_build_vector): Likewise.
3649 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3651 (get_initial_defs_for_reduction): Likewise.
3652 (vectorizable_induction): Likewise.
3654 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3656 * tree-vector-builder.h
3657 (tree_vector_builder::new_binary_operation): Declare.
3658 * tree-vector-builder.c
3659 (tree_vector_builder::new_binary_operation): New function.
3660 * fold-const.c (fold_relational_const): Use it.
3661 (const_binop): Likewise. Check that both input vectors have
3662 the same number of elements, thus excluding things like WIDEN_SUM.
3663 Check whether it is possible to operate directly on the encodings
3666 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3668 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3669 new_unary_operation, operating only on the encoded elements.
3670 (const_unop): Likewise.
3671 (exact_inverse): Likewise.
3672 (distributes_over_addition_p): New function.
3673 (const_binop): Use tree_vector_builder and new_unary_operation
3674 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3675 on the encoded elements unless the encoding is strided and the
3676 operation does not distribute over addition.
3677 (fold_convert_const): Use tree_vector_builder and
3678 new_unary_operation. Operate only on the encoded elements
3679 for truncating integer conversions, or for non-stepped encodings.
3681 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3683 * config/sparc/sparc.c: Include tree-vector-builder.h.
3684 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3685 * expmed.c: Include tree-vector-builder.h.
3686 (make_tree): Use tree_vector_builder instead of build_vector.
3687 * fold-const.c: Include tree-vector-builder.h.
3688 (const_binop): Use tree_vector_builder instead of build_vector.
3689 (const_unop): Likewise.
3690 (native_interpret_vector): Likewise.
3691 (fold_vec_perm): Likewise.
3692 (fold_ternary_loc): Likewise.
3693 * gimple-fold.c: Include tree-vector-builder.h.
3694 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3696 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3697 (simplify_vector_constructor): Use tree_vector_builder instead
3699 * tree-vect-generic.c: Include tree-vector-builder.h.
3700 (add_rshift): Use tree_vector_builder instead of build_vector.
3701 (expand_vector_divmod): Likewise.
3702 (optimize_vector_constructor): Likewise.
3703 * tree-vect-loop.c: Include tree-vector-builder.h.
3704 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3705 of build_vector. Explicitly use a stepped encoding for
3707 * tree-vect-slp.c: Include tree-vector-builder.h.
3708 (vect_get_constant_vectors): Use tree_vector_builder instead
3710 (vect_transform_slp_perm_load): Likewise.
3711 (vect_schedule_slp_instance): Likewise.
3712 * tree-vect-stmts.c: Include tree-vector-builder.h.
3713 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3714 (vect_gen_perm_mask_any): Likewise.
3715 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3716 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3718 (build_vector_from_val): Likewise. Explicitly use a duplicate
3721 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3723 * doc/generic.texi (VECTOR_CST): Describe new representation of
3725 * vector-builder.h: New file.
3726 * tree-vector-builder.h: Likewise.
3727 * tree-vector-builder.c: Likewise.
3728 * Makefile.in (OBJS): Add tree-vector-builder.o.
3729 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3730 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3731 (tree_vector): Change the number of elements to
3732 vector_cst_encoded_nelts.
3733 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3734 (VECTOR_CST_ELTS): Delete.
3735 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3736 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3737 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3738 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3739 (VECTOR_CST_ENCODED_ELT): Likewise.
3740 (vector_cst_encoded_nelts): New function.
3741 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3742 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3743 (vector_cst_int_elt, vector_cst_elt): Declare.
3744 * tree.c: Include tree-vector-builder.h.
3745 (tree_code_size): Abort if passed VECTOR_CST.
3746 (tree_size): Update for new VECTOR_CST layout.
3747 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3748 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3749 (build_vector): Use tree_vector_builder.
3750 (vector_cst_int_elt, vector_cst_elt): New functions.
3751 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3752 encoded elements and then create the vector in the canonical form.
3753 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3754 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3755 (tree_c_tests): Call test_vector_cst_patterns.
3756 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3758 (hash_tree): Likewise.
3759 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3760 (streamer_write_tree_header): Likewise.
3761 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3762 (streamer_alloc_tree): Likewise. Update call to make_vector.
3763 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3765 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3767 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3768 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3769 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3771 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3772 Richard Biener <rguenther@suse.de>
3774 PR tree-optimization/81303
3775 * Makefile.in (gimple-loop-interchange.o): New object file.
3776 * common.opt (floop-interchange): Reuse the option from graphite.
3777 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3778 -floop-interchange and mention it for -O3.
3779 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3780 * gimple-loop-interchange.cc: New file.
3781 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3782 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3783 * passes.def (pass_linterchange): New pass.
3784 * timevar.def (TV_LINTERCHANGE): New time var.
3785 * tree-pass.h (make_pass_linterchange): New declaration.
3786 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3787 interchange. Record IV before/after increment in new parameters.
3788 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3789 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3791 (check_reduction_path): ...New function here.
3792 * tree-vectorizer.h (check_reduction_path): New declaration.
3794 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3797 PR rtl-optimization/80818
3798 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3799 always early clobbered.
3800 * lra-lives.c (process_bb_lives): Check input hard regs for early
3801 clobbered non-operand hard reg.
3803 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3806 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3807 types_compatible_p, just that TYPE_MODE is the same.
3809 2017-12-07 Martin Sebor <msebor@redhat.com>
3812 * attribs.c (empty_attribute_table): Initialize new member of
3813 struct attribute_spec.
3814 (decl_attributes): Add argument. Handle mutually exclusive
3815 combinations of attributes.
3816 (selftests::test_attribute_exclusions): New function.
3817 (selftests::attribute_c_tests): Ditto.
3818 * attribs.h (decl_attributes): Add default argument.
3819 * selftest.h (attribute_c_tests): Declare.
3820 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3821 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3823 * doc/extend.texi (Common Function Attributes): Update const and pure.
3824 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3825 of struct attribute_spec.
3826 * config/arc/arc.c (arc_attribute_table): Same.
3827 * config/arm/arm.c (arm_attribute_table): Same.
3828 * config/avr/avr.c ( avr_attribute_table): Same.
3829 * config/bfin/bfin.c (bfin_attribute_table): Same.
3830 * config/cr16/cr16.c (cr16_attribute_table): Same.
3831 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3832 * config/h8300/h8300.c (h8300_attribute_table): Same.
3833 * config/i386/i386.c (ix86_attribute_table): Same.
3834 * config/ia64/ia64.c (ia64_attribute_table): Same.
3835 * config/m32c/m32c.c (m32c_attribute_table): Same.
3836 * config/m32r/m32r.c (m32r_attribute_table): Same.
3837 * config/m68k/m68k.c (m68k_attribute_table): Same.
3838 * config/mcore/mcore.c (mcore_attribute_table): Same.
3839 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3840 * config/mips/mips.c (mips_attribute_table): Same.
3841 * config/msp430/msp430.c (msp430_attribute_table): Same.
3842 * config/nds32/nds32.c (nds32_attribute_table): Same.
3843 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3844 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3845 * config/rl78/rl78.c (rl78__attribute_table): Same.
3846 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3847 * onfig/rx/rx.c (rx_attribute_table): Same.
3848 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3849 * config/sh/sh.c (sh_attribute_table): Same.
3850 * config/sparc/sparc.c (sparc_attribute_table): Same.
3851 * config/spu/spu.c (spu_attribute_table): Same.
3852 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3853 * config/v850/v850.c (v850_attribute_table): Same.
3854 * config/visium/visium.c (visium_attribute_table): Same.
3856 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3859 * config/arm/arm.c (INCLUDE_STRING): Define.
3860 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3861 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
3864 2017-12-07 Michael Matz <matz@suse.de>
3866 Add unroll and jam pass
3868 * gimple-loop-jam.c: New file.
3869 * Makefile.in (OBJS): Add gimple-loop-jam.o.
3870 * common.opt (funroll-and-jam): New option.
3871 * opts.c (default_options_table): Add unroll-and-jam at -O3.
3872 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
3873 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
3874 * passes.def: Add pass_loop_jam.
3875 * timevar.def (TV_LOOP_JAM): Add.
3876 * tree-pass.h (make_pass_loop_jam): Declare.
3877 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
3878 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
3879 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
3880 to flow_loop_tree_node_add.
3881 (duplicate_subloops, copy_loops_to): Append to sibling list.
3882 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
3883 * doc/invoke.texi (-funroll-and-jam): Document new option.
3884 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
3886 2017-12-07 Richard Biener <rguenther@suse.de>
3888 PR tree-optimization/83296
3889 PR tree-optimization/67769
3890 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
3891 flow sensitive info in an unrelated BB.
3892 (value_replacement): Use reset_flow_sensitive_info.
3893 (minmax_replacement): Reset flow sensitive info on the def
3894 we move. Do not reset flow sensitive info in the whole BB
3895 we move the stmt to.
3896 (abs_replacement): Likewise.
3898 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3901 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3902 rs6000_cpu to the given -mcpu=, or to the default processor.
3904 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3906 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
3907 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
3908 instead of rs6000_cpu_attr.
3909 (rs6000_adjust_cost): Ditto.
3910 (is_microcoded_insn): Ditto.
3911 (rs6000_adjust_priority): Ditto.
3912 (rs6000_issue_rate): Ditto.
3913 (rs6000_use_sched_lookahead): Ditto.
3914 (rs6000_use_sched_lookahead_guard): Ditto.
3915 (rs6000_sched_reorder): Ditto.
3916 (force_new_group): Ditto.
3917 * config/rs6000/rs6000.md (cpu attribute): Ditto.
3918 (group_ending_nop): Ditto.
3920 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3922 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
3923 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
3924 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
3925 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
3927 (rs6000_reassociation_width): Ditto.
3928 (rs6000_emit_epilogue): Ditto.
3929 (rs6000_adjust_cost): Ditto.
3930 (is_microcoded_insn): Ditto.
3931 (is_cracked_insn): Ditto.
3932 (rs6000_adjust_priority): Ditto.
3933 (rs6000_sched_reorder): Ditto.
3934 (rs6000_sched_reorder2): Ditto.
3935 (insn_must_be_first_in_group): Ditto.
3936 (insn_must_be_last_in_group): Ditto.
3937 (rs6000_register_move_cost): Ditto.
3938 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
3941 2017-12-07 Julia Koval <julia.koval@intel.com>
3943 * config.gcc: Add vaesintrin.h.
3944 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3946 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3947 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3949 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3950 * config/i386/immintrin.h: Include vaesintrin.h.
3951 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3952 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3953 _mm_aesdec_epi128): New intrinsics.
3955 2017-12-06 David Malcolm <dmalcolm@redhat.com>
3957 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
3958 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
3959 c-family/c-spellcheck.cc.
3960 (best_macro_match::best_macro_match): Likewise.
3962 (struct edit_distance_traits<cpp_hashnode *>): Move to
3963 c-family/c-spellcheck.h.
3964 (class best_macro_match): Likewise.
3966 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3968 PR tree-optimization/83293
3969 * gimple-ssa-strength-reduction.c (insert_initializers): Use
3970 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
3971 might insert into empty bb.
3974 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
3975 simplify for plus with :c added, and pointer_plus without that.
3976 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
3977 with undefined overflow and the conversion is not widening,
3978 perform negation in utype and only convert to type afterwards.
3979 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
3980 simplify for plus with :c added, and pointer_plus without that.
3981 If type is integral with undefined overflow and the conversion is
3982 not widening, perform minus in utype and only convert to type
3983 afterwards. Move the last pointer_diff_expr simplify into the
3986 2017-12-06 Martin Sebor <msebor@redhat.com>
3988 PR tree-optimization/82646
3989 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
3990 strncpy, not maxlen.
3992 2017-12-06 Martin Sebor <msebor@redhat.com>
3994 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
3997 PR tree-optimization/83075
3998 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
3999 strncat/strncpy don't change length of source string.
4001 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4004 2017-11-29 Martin Aberg <maberg@gaisler.com>
4006 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4007 to prevent b2bst errata sequence.
4008 (sqrtdf2_fix): Likewise.
4010 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4012 PR tree-optimization/81945
4013 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4014 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4015 to dest_cfun has orig_loop_num set, either remap it to the new
4016 loop number if the loop got moved too, or clear it.
4018 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4020 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4022 (thunderx2t99_nothing): New insn reservation.
4023 (thunderx2t99_mrs): New insn reservation.
4024 (thunderx2t99_multiple): New insn reservation.
4025 (thunderx2t99_alu_basi): Add bfx to reservation.
4026 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4028 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4031 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4033 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4035 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4036 tree-ssa-pre.c::remove_dead_inserted_code.
4037 * tree-ssa-dce.h: New file.
4038 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
4039 (remove_dead_inserted_code): Move and rename to function
4040 tree-ssa-dce.c::simple_dce_from_worklist.
4041 (pass_pre::execute): Update use.
4043 2017-12-05 Richard Biener <rguenther@suse.de>
4045 PR tree-optimization/83277
4046 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4047 to code-gen liveout vars.
4049 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4051 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4052 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4053 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4056 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4058 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4060 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4061 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4064 2017-12-05 Richard Biener <rguenther@suse.de>
4066 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4067 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4068 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4069 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4071 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4073 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4075 2017-12-05 Martin Liska <mliska@suse.cz>
4076 Jakub Jelinek <jakub@redhat.com>
4078 * doc/invoke.texi: Document the options.
4079 * flag-types.h (enum sanitize_code): Add
4080 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4081 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4082 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4083 * opts.c: Define new sanitizer options.
4084 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4085 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4087 2017-12-05 Julia Koval <julia.koval@intel.com>
4089 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4090 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4091 (ix86_handle_option): Handle -mavx512vnni.
4092 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4093 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4094 * config/i386/i386-c (__AVX512VNNI__): New.
4095 * config/i386/i386.c (ix86_target_string): Handle new option.
4096 (ix86_valid_target_attribute_inner_p): Handle new option.
4097 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4098 * config/i386/i386.opt (mavx512vnni): New option.
4100 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4103 * athlon.md: Disable for generic.
4104 * haswell.md: Enable for generic.
4105 * i386.c (ix86_sched_init_global): Add core hooks for generic.
4106 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
4108 (ix86_adjust_cost): Move generic to haswell path.
4110 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
4112 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
4113 instead of MEM_P in a couple more places. Fix formatting issues.
4115 2017-12-04 Jim Wilson <jimw@sifive.com>
4117 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
4118 instead of GP_REG_LAST-1.
4119 (riscv_adjust_libcall_cfi_prologue): Likewise.
4120 (riscv_adjust_libcall_cri_epilogue): Likewise.
4121 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
4124 2017-12-04 Luis Machado <luis.machado@linaro.org>
4126 * ipa-pure-const.c (check_decl): Add missing newline.
4127 (state_from_flags): Likewise.
4129 2017-12-04 Jeff Law <law@redhat.com>
4131 PR tree-optimizatin/78496
4132 * gimple-ssa-evrp-analyze.h
4133 (evrp_range_analyzer::get_vr_values): Simplify.
4134 * gimple-ssa-evrp-analyze.c: Corresponding changes.
4135 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
4136 and gimple-ssa-evrp-analyze.h.
4137 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
4138 (simplify_stmt_for_jump_threading): Copy a blob of code from
4139 tree-vrp.c to use ranges to simplify statements.
4140 (dom_opt_dom_walker::before_dom_children): Call
4141 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
4142 (dom_opt_dom_walker::after_dom_children): Similarly for
4143 evrp_range_analyzer::leave.
4144 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
4147 * gimple-ssa-evrp-analyze.c
4148 (evrp_range_analyzer::extract_range_from_stmt): Always use
4149 vr_values::update_value_range so preexisting range info is
4150 medged with new range info, even if the new range is VR_VARYING.
4152 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4154 * combine.c: Adjust comment.
4155 (use_crosses_set_p): Delete.
4156 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
4157 (try_combine): Ditto.
4159 2017-12-04 Richard Biener <rguenther@suse.de>
4161 PR tree-optimization/83255
4162 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
4163 Re-add zero-iteration check.
4165 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4167 PR rtl-optimization/83245
4168 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
4169 hard registers as earlyclobber, also if not in an asm.
4171 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4175 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4178 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4179 rs6000_cpu based on cpu_index, not tune_index.
4181 2017-12-04 Richard Biener <rguenther@suse.de>
4183 PR tree-optimization/83238
4184 * graphite-scop-detection.c (scop_detection::merge_sese): Make
4185 code match comment, rejecting invalid SESE regions.
4187 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
4189 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
4190 require base operand is a REG_POINTER prior to reload on targets
4191 with non-equivalent space registers.
4193 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4195 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
4196 (update_specialized_profile): Fix updating of counts.
4197 (perhaps_add_new_callers): Likewise.
4199 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4202 * x86-tune.def: Remove obsolette FIXMEs.
4203 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
4204 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
4205 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
4207 (X86_TUNE_PAD_RETURNS): Disable for generic.
4209 2017-12-02 Jakub Jelinek <jakub@redhat.com>
4211 PR tree-optimization/83170
4212 PR tree-optimization/83241
4213 * gimple-ssa-store-merging.c
4214 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
4215 gimple_vuse (ins_stmt) in case it has changed.
4216 (imm_store_chain_info::output_merged_store): Likewise.
4218 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
4222 * tree-cfg.c (pass_warn_function_return::execute): Handle
4223 __builtin_ubsan_handle_missing_return like __builtin_unreachable
4224 with BUILTINS_LOCATION.
4228 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
4229 is BLKmode for vector field with vector raw mode, use TYPE_MODE
4230 instead of DECL_MODE.
4232 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
4233 last argument to rtx pointer.
4234 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
4235 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
4236 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
4237 depending on the chosen ISAs.
4238 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
4239 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
4240 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
4242 * config/i386/sse.md (mov<mode>_internal): Likewise.
4243 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
4245 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4247 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
4249 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4251 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
4252 parameters from prototype.
4253 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
4254 parameters. Don't print a comment.
4255 (emit_fusion_gpr_load): Adjust.
4256 (emit_fusion_load_store): Adjust.
4257 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
4258 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
4259 comment on the second line.
4261 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4264 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4265 rs6000_cpu based on cpu_index, not tune_index.
4267 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4269 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
4270 which_alternative instead of which_alternative + 1.
4271 (output_asm_insn): Print an extra tab if the template is short.
4273 2017-12-01 Jim Wilson <jimw@sifive.com>
4275 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
4277 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
4278 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
4280 * doc/tm.texi: Regenerate.
4282 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
4285 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
4286 whether we can allocate pseudos before trying to fix an address.
4287 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
4288 memory address is indexed or indirect.
4289 (floatuns_<mode>si2_hw2): Likewise.
4291 2017-12-01 Jason Merrill <jason@redhat.com>
4293 * Makefile.in (TAGS): Add c-family/*.cc.
4295 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
4297 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
4298 (call_value_insn): Likewise.
4299 (sibcall_insn): Likewise.
4300 (sibcall_value_insn): Likewise.
4301 (movsi_aarch64): Likewise.
4302 (movdi_aarch64): Likewise.
4303 (add_losym_): Likewise.
4304 (ldr_got_small_): Likewise.
4305 (ldr_got_small_sidi): Likewise.
4306 (ldr_got_small_28k_): Likewise.
4307 (ldr_got_small_28k_sidi): Likewise.
4308 * config/aarch64/aarch64.c (aarch64_print_address_internal):
4309 Move output_addr_const to symbolic case. Add error check.
4311 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4313 * config/s390/predicates.md (plus16_Q_operand): New predicate.
4314 * config/s390/s390.md: Disable MVC merging peephole if it would
4315 disable operand forwarding.
4316 (new peephole2): Split MVCs if it would turn them into up to 2
4319 2017-12-01 Richard Biener <rguenther@suse.de>
4321 PR tree-optimization/83232
4322 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
4323 detection of same access. Instead of breaking the group here
4324 do not consider the duplicate. Add comment explaining real fix.
4326 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
4328 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
4330 2017-12-01 Sudakshina Das <sudi.das@arm.com>
4332 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
4334 2017-12-01 Jakub Jelinek <jakub@redhat.com>
4336 * function.h (struct function): Remove cilk_frame_decl,
4337 is_cilk_function and calls_cilk_spawn fields.
4338 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
4340 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
4341 cilk_elemental field.
4342 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
4343 * target.def: Adjust comment.
4344 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
4345 Don't test cilk_elemental.
4347 PR tree-optimization/83233
4348 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
4349 bswap_stat name for the struct.
4352 * tree.h (SWITCH_BREAK_LABEL_P): Define.
4353 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
4354 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
4355 SWITCH_BREAK_LABEL_P set on the label.
4356 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
4357 added for default case if it was missing and not all cases covered.
4358 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
4359 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
4361 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
4362 to avoid -Wimplicit-fallthrough warning.
4363 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
4364 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
4366 PR tree-optimization/83221
4367 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
4369 (init_reassoc): Formatting fix.
4372 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
4373 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
4377 * tree-cfg.c: Include asan.h.
4378 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
4380 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4382 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
4384 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4386 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
4387 vpcmpeqd instruction.
4389 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4391 * config/i386/i386.c (standard_sse_constant_opcode): Fix
4392 registers type for 128bit mode.
4394 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4396 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
4398 * gengtype-state.c (read_state_pair): Likewise.
4399 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
4400 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
4401 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
4402 * ifcvt.c (noce_try_store_flag_constants): Likewise.
4403 * tree-ssa-ccp.c (ccp_finalize): Likewise.
4404 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
4405 * builtins.c (fold_builtin_3): Likewise.
4406 * graphite-scop-detection.c
4407 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
4408 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
4411 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
4414 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
4415 parameter. Return true if ignore_decl_rtl_set_p is true and other
4416 conditions are satisfied.
4417 * asan.h (asan_protect_global): Add new parameter.
4418 * varasm.c (categorize_decl_for_section): Pass true as second parameter
4419 to asan_protect_global calls.
4421 2017-11-30 Jim Wilson <jimw@sifive.com>
4423 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
4424 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
4425 -mstrict-align, add info on default value. Delete redundant lines for
4426 -mabi. Add missing -mexplicit-relocs docs.
4428 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4430 * config/arc/arc.md (trap): New pattern.
4432 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4434 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
4435 instruction to end into a delay slot.
4436 * config/arc/arc.md (cond_delay_insn): Check if the instruction
4437 can be placed into a delay slot against reg_note.
4439 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4441 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
4442 labels number of usages.
4444 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4446 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
4448 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
4450 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4453 * internal-fn.c (expand_mul_overflow): Optimize unsigned
4454 multiplication by power of 2 constant into two shifts + comparison.
4456 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
4459 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
4461 2017-11-30 Richard Biener <rguenther@suse.de>
4463 PR tree-optimization/83202
4464 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
4465 allow_peel argument and guard peeling.
4466 (canonicalize_loop_induction_variables): Likewise.
4467 (canonicalize_induction_variables): Pass false.
4468 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
4469 peeling from cunrolli.
4471 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4473 * combine.c (try_combine): Print a message to dump file whenever
4474 I0, I1, or I2 cannot be combined into I3.
4476 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4478 PR rtl-optimization/83156
4479 PR rtl-optimization/82621
4480 * combine.c (try_combine): Don't split an I2 if one of the dests is
4481 set again before I3. Allow unused dests.
4483 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4485 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
4487 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
4489 PR rtl-optimization/80818
4490 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
4491 recursively. Use insn code for clobber.
4492 (lra_set_insn_recog_data): Pass the new arg to
4493 collect_non_operand_hard_regs.
4494 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
4496 (lra_update_insn_regno_info): Pass insn to
4497 add_regs_to_insn_regno_info.
4499 2017-11-29 Jim Wilson <jimw@sifive.com>
4500 Andrew Waterman <andrew@sifive.com>
4502 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
4503 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
4504 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
4505 use SINGLE_SHIFT_COST.
4506 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
4507 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
4509 2017-11-29 Julia Koval <julia.koval@intel.com>
4511 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
4512 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
4513 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
4514 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
4515 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
4516 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
4517 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
4518 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
4519 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
4520 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
4521 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
4522 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
4523 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
4524 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
4525 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
4526 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
4527 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
4528 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
4529 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
4530 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
4531 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
4532 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
4533 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
4534 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
4535 __builtin_ia32_vpshldv_v2di_mask,
4536 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
4537 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
4538 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
4540 2017-11-29 Julia Koval <julia.koval@intel.com>
4542 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
4543 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
4544 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
4545 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
4546 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
4547 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
4548 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
4549 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
4550 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
4551 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
4552 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
4553 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
4554 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
4555 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
4556 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
4557 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
4558 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
4559 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
4560 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4561 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
4562 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
4564 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4566 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
4567 movsi_pic_gotdata_op instruction as a load for the UT699 errata
4570 2017-11-29 Martin Aberg <maberg@gaisler.com>
4572 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4573 to prevent b2bst errata sequence.
4574 (sqrtdf2_fix): Likewise.
4576 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4578 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
4579 (div_sqrt_insn_p): New function.
4580 (sparc_do_work_around_errata): Insert NOP instructions to
4581 prevent sequences that could trigger the TN-0013 errata for
4582 certain LEON3 processors.
4583 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
4584 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
4585 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
4586 (in_branch_delay): Prevent div and sqrt in delay slot if
4588 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
4590 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4592 * config/sparc/sparc.c (atomic_insn_p): New function.
4593 (sparc_do_work_around_errata): Insert NOP instructions to
4594 prevent sequences that could trigger the TN-0010 errata for
4596 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
4597 instruction referable in atomic_insns_p.
4599 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4601 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
4602 (atomic_compare_and_swap_leon3_1): Likewise.
4605 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4607 * config/sparc/sparc.c (fpop_insn_p): New function.
4608 (sparc_do_work_around_errata): Insert NOP instructions to
4609 prevent sequences that could trigger the TN-0012 errata for
4611 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
4612 * config/sparc/sparc.md (fix_gr712rc): New attribute.
4613 (in_branch_annul_delay): Prevent floating-point instructions
4614 in delay slot of annulled integer branch.
4616 2017-11-29 Richard Biener <rguenther@suse.de>
4618 PR tree-optimization/83202
4619 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4621 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4622 nodes vectorized to the same stmts multiple times.
4623 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4624 (vect_analyze_slp_instance): Adjust.
4625 (scalar_stmts_to_slp_tree_map_t): New typedef.
4626 (vect_schedule_slp_instance): Add a map recording the SLP node
4627 representing the vectorized stmts for a set of scalar stmts.
4628 Avoid code-generating redundancies.
4629 (vect_schedule_slp): Allocate map and pass it down.
4631 2017-11-29 Nathan Sidwell <nathan@acm.org>
4634 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4637 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4639 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4640 cases for printing LDP/STP memory addresses.
4641 (aarch64_print_address_internal): Renamed from
4642 aarch64_print_operand_address, added parameter, add Pmode check.
4643 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4644 (aarch64_print_operand_address): Indirect to
4645 aarch64_print_address_internal.
4646 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4647 'y' operand output specifier.
4649 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4652 * tree.c (build_simple_mem_ref_loc): Handle
4653 get_addr_base_and_unit_offset returning a MEM_REF.
4656 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4659 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4662 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4664 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4666 (layout::move_to_column): Likewise.
4668 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4671 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4672 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4674 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4675 Martin Jambor <mjambor@suse.cz>
4678 * tree.h (expr_type_first_operand_type_p): Declare
4679 * tree.c (expr_type_first_operand_type_p): New function.
4680 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4681 (ipa_value_from_jfunc): Adjust declaration.
4682 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4683 Use it as result type for arithmetics, unless it is NULL in which case
4684 be more conservative.
4685 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4686 ipa_get_jf_pass_through_result.
4687 (propagate_vals_across_pass_through): Likewise.
4688 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4689 is to propagate_vals_across_pass_through.
4690 (propagate_constants_across_call): Pass PARM_TYPE to
4691 propagate_scalar_across_jump_function.
4692 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4693 ipa_value_from_jfunc.
4694 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4695 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4696 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4697 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4698 target_type, pass it to ipa_value_from_jfunc.
4699 (update_indirect_edges_after_inlining): Pass parameter type to
4700 try_make_edge_direct_simple_call.
4702 2017-11-28 Jeff Law <law@redhat.com>
4704 * gimple-ssa-evrp-analyze.c
4705 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4706 refine ranges if scev_initialized_p returns true.
4707 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4709 2017-11-28 Julia Koval <julia.koval@intel.com>
4711 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4712 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4713 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4714 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4715 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4716 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4717 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4718 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4719 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4720 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4721 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4722 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4723 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4724 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4725 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4726 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4727 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4728 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4729 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4730 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4731 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4732 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4734 2017-11-28 Julia Koval <julia.koval@intel.com>
4736 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4737 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4738 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4739 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4740 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4741 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4742 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4743 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4744 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4745 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4746 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4747 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4748 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4749 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4750 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4751 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4752 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4753 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4754 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4755 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4756 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4757 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4758 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4759 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4760 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4761 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4762 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4763 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4764 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4765 __builtin_ia32_vpshld_v2di_mask): New builtins.
4766 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4767 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4769 2017-11-28 Richard Biener <rguenther@suse.de>
4771 PR tree-optimization/80776
4772 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4774 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4776 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4777 If the incoming edge is an effective fallthru because the other
4778 edge only reaches a __builtin_unreachable () then record ranges
4779 derived from the controlling condition in SSA info.
4780 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4781 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4783 2017-11-28 Olivier Hainque <hainque@adacore.com>
4785 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4786 on mingw build hosts.
4788 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4791 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4793 2017-11-28 Richard Biener <rguenther@suse.de>
4796 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4797 copies generated from memcpy use a character array as reference
4800 2017-11-28 Julia Koval <julia.koval@intel.com>
4801 Sebastian Peryt <sebastian.peryt@intel.com>
4803 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4804 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4805 cilk-common.o, cilk.h, cilk-common.c): Remove.
4807 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4808 * builtins.c (is_builtin_name): Remove cilkplus condition.
4809 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4810 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4811 cilk-builtins.def, cilkplus.def): Remove.
4812 * cif-code.def (CILK_SPAWN): Remove.
4813 * cilk-builtins.def: Delete.
4814 * cilk-common.c: Ditto.
4816 * cilkplus.def: Ditto.
4817 * config/darwin.h (fcilkplus): Delete.
4818 * cppbuiltin.c: Ditto.
4819 * doc/extend.texi: Remove cilkplus doc.
4820 * doc/generic.texi: Ditto.
4821 * doc/invoke.texi: Ditto.
4822 * doc/passes.texi: Ditto.
4823 * gcc.c (fcilkplus): Remove.
4824 * gengtype.c (cilk.h): Remove.
4825 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4827 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4829 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4831 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4832 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4833 cilkplus conditions.
4834 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4835 inline_read_section): Ditto.
4836 * ipa-inline-analysis.c (cilk.h): Remove.
4837 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4838 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4839 append_linker_options): Remove condition for fcilkplus.
4840 * lto/lto-lang.c (cilk.h): Remove.
4841 (lto_init): Remove condition for fcilkplus.
4842 * omp-expand.c (expand_cilk_for_call): Delete.
4843 (expand_omp_taskreg, expand_omp_for_static_chunk,
4844 expand_omp_for): Remove cilkplus
4846 (expand_cilk_for): Delete.
4847 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4848 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4849 execute_lower_omp, diagnose_sb_0): Ditto.
4850 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4851 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4852 * tree-nested.c: Ditto.
4853 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4854 (dump_generic_node): Ditto.
4855 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4856 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4858 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4860 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4862 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
4863 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
4864 for vector mode and !TARGET_SIMD.
4866 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4868 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
4870 * tree.h (SWITCH_LABELS): Remove.
4871 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
4872 assert SWITCH_BODY is non-NULL.
4873 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
4875 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
4877 PR tree-optimization/80788
4878 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
4879 has TREE_OVERFLOW set, call drop_tree_overflow.
4881 2017-11-28 Richard Biener <rguenther@suse.de>
4883 PR tree-optimization/83158
4884 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
4886 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
4889 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
4890 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
4892 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
4894 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
4895 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
4897 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4900 * varasm.c (compare_constant): Take the mode of the constants into
4901 account when comparing floating point constants.
4903 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
4905 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
4906 from explicit instantiation of debug_helper.
4907 * vec.h (DEFINE_DEBUG_VEC): Ditto.
4909 2017-11-27 Richard Biener <rguenther@suse.de>
4911 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
4914 2017-11-27 Richard Biener <rguenther@suse.de>
4916 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
4917 (type_hash_canon): Also clear int_cst_hash_table entry for
4919 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
4921 2017-11-27 Tamar Christina <tamar.christina@arm.com>
4923 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4924 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4925 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4927 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4929 * hash-map.h (gt_cleare_cache): Avoid UB.
4931 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
4933 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
4934 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
4935 (decide_unroll_runtime_iterations): Likewise.
4936 (decide_unroll_stupid): Likewise.
4938 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4941 * config/i386/i386.md: Add a loop with incssp.
4943 2017-11-27 Martin Jambor <mjambor@suse.cz>
4945 PR tree-optimization/81248
4946 * tree-sra.c (splice_param_accesses): Remove size check.
4947 (decide_one_param_reduction): Fix size check.
4948 * gimple-pretty-print.c (dump_profile): Silence warning.
4949 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
4951 2017-11-27 Jakub Jelinek <jakub@redhat.com>
4954 * dbxout.c (lastlineno): New variable.
4955 (dbx_debug_hooks): Use dbxout_switch_text_section as
4956 switch_text_section debug hook.
4957 (dbxout_function_end): Switch to current_function_section
4958 rather than function_section. If crtl->has_bb_partition,
4959 output just one N_FUN, depending on in_cold_section_p.
4960 (dbxout_source_line): Remember last lineno in lastlineno.
4961 (dbxout_switch_text_section): New function.
4962 (dbxout_function_decl): Adjust dbxout_block caller.
4963 (dbx_block_with_cold_children): New function.
4964 (dbxout_block): Return true if any LBRAC/RBRAC have been
4965 emitted. Use dbx_block_with_cold_children at depth == 0
4966 in second partition. Add PARENT_BLOCKNUM argument, pass
4967 it optionally adjusted to children. Output LBRAC/RBRAC
4968 around recursive call only if the block is in the current
4969 partition, if not and anything was output, emit empty
4971 * final.c (final_scan_insn): Compute cold_function_name
4972 before calling switch_text_section debug hook. Call
4973 that hook even if dwarf2out_do_frame if not emitting
4977 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
4978 TREE_READONLY decls.
4980 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4982 PR rtl-optimization/82488
4983 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
4985 2017-11-26 Julia Koval <julia.koval@intel.com>
4987 * config/i386/i386.c (processor_target_table): Add skylake_cost for
4989 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
4992 2017-11-26 Julia Koval <julia.koval@intel.com>
4994 * config/i386/driver-i386.c (host_detect_local_cpu):
4995 Detect skylake-avx512.
4997 2017-11-26 Julia Koval <julia.koval@intel.com>
4999 * config.gcc: Add -march=cannonlake.
5000 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5001 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5002 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5003 (PTA_CANNONLAKE): New.
5004 (processor_target_table): Add cannonlake.
5005 (ix86_option_override_internal): Ditto.
5006 (fold_builtin_cpu): Ditto.
5007 (get_builtin_code_for_version): Handle cannonlake.
5008 (M_INTEL_COREI7_CANNONLAKE): New.
5009 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5010 * doc/invoke.texi: Add -march=cannonlake.
5012 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5014 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5015 (try_init_one_plugin): Alternative implementation for MinGW.
5016 * Makefile.in (plugin_implib): New.
5017 (gengtype-lex.c): Fix broken AIX workaround.
5018 * configure: Regenerate.
5019 * doc/plugins.texi: Document support for MinGW.
5021 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5023 PR rtl-optimization/81553
5024 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5025 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5026 is a shift where C1 has different mode than the whole shift, use C1's
5027 mode for MULT rather than the shift's mode.
5030 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5031 builtins not enabled in the currently selected ISA.
5033 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5035 PR tree-optimization/71026
5036 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5037 (insert_reciprocals): Change to insert reciprocals before a division
5038 by a square and to insert the square of a reciprocal.
5039 (execute_cse_reciprocals_1): Change to consider division by a square.
5040 (register_division_in): Add importance parameter.
5042 2017-11-24 Richard Biener <rguenther@suse.de>
5044 PR tree-optimization/82402
5045 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5046 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5048 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5050 * match.pd (0-ptr): New transformation.
5052 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5055 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5056 when skipping checking.
5058 2017-11-24 Richard Biener <rguenther@suse.de>
5060 PR tree-optimization/83128
5061 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5062 (vn_reference_lookup_3): Likewise.
5064 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5067 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5068 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5071 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5072 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5073 check for the latter with gimple_call_builtin_p. Do not handle
5074 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5076 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5078 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5081 2017-11-23 Julia Koval <julia.koval@intel.com>
5083 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5084 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5085 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5086 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5087 _mm512_maskz_expandloadu_epi16): New intrinsics.
5088 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5089 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5090 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5091 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5092 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5093 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5094 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5095 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5096 _mm256_maskz_expandloadu_epi8): New intrinsics.
5097 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5098 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5099 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5100 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5101 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5102 * config/i386/sse.md (VI248_VLBW): New iterator.
5103 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5105 2017-11-23 Julia Koval <julia.koval@intel.com>
5107 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
5108 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
5109 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
5110 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
5111 _mm512_mask_compressstoreu_epi16): New.
5112 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
5113 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
5114 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
5115 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
5116 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
5117 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
5118 _mm256_mask_compressstoreu_epi8): New.
5119 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
5120 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
5121 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
5122 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
5123 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
5124 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
5125 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
5126 __builtin_ia32_compresshi128_mask,
5127 __builtin_ia32_compressstoreuqi512_mask,
5128 __builtin_ia32_compressstoreuhi512_mask,
5129 __builtin_ia32_compressstoreuqi256_mask,
5130 __builtin_ia32_compressstoreuqi128_mask,
5131 __builtin_ia32_compressstoreuhi256_mask,
5132 __builtin_ia32_compressstoreuhi128_mask): New builtins.
5133 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
5135 (ix86_expand_special_args_builtin): Handle new types.
5136 (s4fma_expand): Handle new builtin array.
5137 * config/i386/immintrin.h: Include new headers.
5138 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
5139 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
5141 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5143 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
5144 fixes. Declare temp and g variables at the top in order to avoid
5145 {} in most of the cases.
5147 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
5149 * match.pd (ptr-0): New transformation.
5151 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5153 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
5154 (struct addr_mode_cost_table): New.
5155 (struct tune_params): Add field addr_mode_costs.
5156 * config/arm/arm.c (generic_addr_mode_costs): New.
5157 (arm_slowmul_tune): Initialise addr_mode_costs field.
5158 (arm_fastmul_tune): Likewise.
5159 (arm_strongarm_tune): Likewise.
5160 (arm_xscale_tune): Likewise.
5161 (arm_9e_tune): Likewise.
5162 (arm_marvell_pj4_tune): Likewise.
5163 (arm_v6t2_tune): Likewise.
5164 (arm_cortex_tune): Likewise.
5165 (arm_cortex_a8_tune): Likewise.
5166 (arm_cortex_a7_tune): Likewise.
5167 (arm_cortex_a15_tune): Likewise.
5168 (arm_cortex_a35_tune): Likewise.
5169 (arm_cortex_a53_tune): Likewise.
5170 (arm_cortex_a57_tune): Likewise.
5171 (arm_exynosm1_tune): Likewise.
5172 (arm_xgene1_tune): Likewise.
5173 (arm_cortex_a5_tune): Likewise.
5174 (arm_cortex_a9_tune): Likewise.
5175 (arm_cortex_a12_tune): Likewise.
5176 (arm_cortex_a73_tune): Likewise.
5177 (arm_v7m_tune): Likewise.
5178 (arm_cortex_m7_tune): Likewise.
5179 (arm_v6m_tune): Likewise.
5180 (arm_fa726te_tune): Likewise.
5181 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
5183 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5185 * config/arm/arm.c (arm_mem_costs): New function.
5186 (arm_rtx_costs_internal): Use arm_mem_costs.
5188 2017-11-23 Mark Wielaard <mark@klomp.org>
5190 * dwarf2out.c (init_sections_and_labels): Use generation to create
5191 unique ranges_section_label and ranges_base_label. Return generation.
5192 (output_rnglists): Add generation argument. Use generation to create
5193 unique ranges labels.
5194 (dwarf2out_finish): Get generation from init_sections_and_labels
5195 and pass generation to output_rnglists.
5197 2017-11-23 Mike Stump <mikestump@comcast.net>
5198 Eric Botcazou <ebotcazou@adacore.com>
5200 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
5201 * cfgloop.h (struct loop): Add unroll field.
5202 * function.h (struct function): Add has_unroll bitfield.
5203 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
5204 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
5205 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
5206 (pass_rtl_unroll_loops::gate): Likewise.
5207 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
5208 for which loop->unroll==1.
5209 (decide_unroll_constant_iterations): Use note for consistency and
5210 take loop->unroll into account. Return early if loop->unroll is set.
5211 Fix thinko in existing test.
5212 (decide_unroll_runtime_iterations): Use note for consistency and
5213 take loop->unroll into account.
5214 (decide_unroll_stupid): Likewise.
5215 * lto-streamer-in.c (input_cfg): Read loop->unroll.
5216 * lto-streamer-out.c (output_cfg): Write loop->unroll.
5217 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
5219 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
5220 (print_loop): Print loop->unroll if set.
5221 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
5222 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
5223 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
5225 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
5226 loop->unroll is set and smaller than the trip count. Otherwise bypass
5227 entirely the heuristics if loop->unroll is set. Remove dead note.
5228 Fix off-by-one bug in other note.
5229 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
5230 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
5232 (tree_unroll_loops_completely): Make static.
5233 (pass_complete_unroll::execute): Use correct type for variable.
5234 (pass_complete_unrolli::execute): Fix formatting.
5235 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
5237 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
5239 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
5240 enable when TARGET_PREFER_AVX128 is set.
5242 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5244 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
5246 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
5247 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
5249 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5251 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
5252 frequencies. Fix estimation of aggregate parameters.
5254 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5256 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
5259 2017-11-23 Tom de Vries <tom@codesourcery.com>
5261 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
5263 2017-11-23 Tom de Vries <tom@codesourcery.com>
5265 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
5268 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
5271 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
5272 sibcall_value_pcrel_fdpic): Use local variable instead of
5274 (calli_tbr_rel): Add missing operand 2.
5275 (call_valuei_tbr_rel): Add missing operand 3.
5277 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5280 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
5281 bitpos/bitsize covering the whole destination, use store_expr only if
5282 the complex mode is the same. Otherwise, use expand_normal and if
5283 it returns CONCAT, subreg each part separately instead of trying to
5284 subreg the whole result.
5286 2017-11-23 Richard Biener <rguenther@suse.de>
5288 PR tree-optimization/23094
5289 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
5290 come here from walking over backedges in the first iteration.
5291 (vn_reference_lookup_3): Skip clobbers that store the same value.
5293 2017-11-23 Richard Biener <rguenther@suse.de>
5295 PR tree-optimization/81403
5296 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
5297 a block we need a leader relative to.
5298 (phi_translate_1): For nary processing require a leader from
5299 get_representative_for given we run expression simplification
5300 using match-and-simplify. Remove previous fix.
5302 2017-11-22 Jeff Law <law@redhat.com>
5304 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
5305 Use new method allocate_value_range rather than accessing the
5306 vrp_value_range_pool data member directly.
5307 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
5308 to use extract_range_from_stmt method to avoid need for
5309 extract_range_from_assignment method.
5310 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
5311 method rather than setting values_propgated data member directly.
5312 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
5313 and values propagated data members and extract_range_from_assignment
5314 method. Reorder private data members to conform to standards.
5315 Add new methods set_lattice_propagation_complete and
5316 allocate_value_range.
5318 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
5320 PR rtl-optimization/83030
5321 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
5322 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
5323 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
5324 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
5325 CROSSING_JUMP_P flag is already set before setting it.
5326 * cfgrtl.c (fixup_partition_crossing): Likewise.
5327 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
5330 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5332 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
5333 Handle the case where both arguments are using gen_const_vec_series.
5335 2017-11-22 David Malcolm <dmalcolm@redhat.com>
5338 * pretty-print.c (pp_format): Move quoting implementation to
5339 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
5340 to pass address of "quote" local.
5341 (pp_begin_quote): New function.
5342 (pp_end_quote): New function.
5343 * pretty-print.h (printer_fn): Convert penultimate param from bool
5345 (pp_begin_quote): New decl.
5346 (pp_end_quote): New decl.
5347 * tree-diagnostic.c (default_tree_printer): Convert penultimate
5348 param from bool to bool *.
5349 * tree-diagnostic.h (default_tree_printer): Likewise.
5351 2017-11-22 Jeff Law <law@redhat.com>
5353 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
5354 Initialize vr_values.
5355 (evrp_range_analyzer::try_find_new_range): Call methods attached to
5356 vr_values via vr_values class instance rather than delegators.
5357 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
5358 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5359 (evrp_range_analyzer::push_value_range): Likewise.
5360 (evrp_range_analyzer::pop_value_range): Likewise.
5361 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
5362 most delegators. Those remaining are exposed as public interfaces.
5363 Make vr_values a pointer and private.
5364 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
5366 (evrp_range_analyzer::get_vr_value): New method.
5367 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
5368 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
5369 (class evrp_dom_walker): Attach evrp_folder class, initialize
5370 it in the ctor. Remove temporary delegators.
5371 (evrp_dom_walker::before_dom_children): Call methods in attached
5372 evrp_range_analyzer class via class instance pointer. Use
5373 free value_range_constant_singleton to remove need for
5374 op_with_constant_singleton_value delegator method. Do not
5375 create a vrp_prop class instance for every call! Narrow
5376 scope of a couple variables.
5377 (evrp_dom_walker::cleanup): Call methods in attached
5378 evrp_range_analyzer class via class instance pointer.
5379 * vr-values.h (class vr_values): Privatize many methods and
5382 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5384 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
5385 * cfgexpand.c (expand_debug_expr): Remove handling for them.
5386 * expr.c (expand_expr_real_2): Likewise.
5387 * fold-const.c (const_unop): Likewise.
5388 * optabs-tree.c (optab_for_tree_code): Likewise.
5389 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
5390 * tree-inline.c (estimate_operator_cost): Likewise.
5391 * tree-pretty-print.c (dump_generic_node): Likewise.
5392 (op_code_prio): Likewise.
5393 (op_symbol_code): Likewise.
5394 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
5395 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
5396 * internal-fn.c (direct_internal_fn_optab): New function.
5397 (direct_internal_fn_array, direct_internal_fn_supported_p
5398 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
5399 * fold-const-call.c (fold_const_reduction): New function.
5400 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
5402 * tree-vect-loop.c: Include internal-fn.h.
5403 (reduction_code_for_scalar_code): Rename to...
5404 (reduction_fn_for_scalar_code): ...this and return an internal
5406 (vect_model_reduction_cost): Take an internal_fn rather than
5408 (vect_create_epilog_for_reduction): Likewise. Build calls rather
5410 (vectorizable_reduction): Use internal functions rather than tree
5411 codes for the reduction operation. Update calls to the functions
5413 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
5414 Use calls to internal functions rather than REDUC tree codes.
5415 * config/aarch64/aarch64-simd.md: Update comment accordingly.
5417 2017-11-22 Olivier Hainque <hainque@adacore.com>
5419 * config/vxworks.c (vxworks_override_options): Pick default
5420 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
5421 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
5422 DWARF_GNAT_ENCODINGS_DEFAULT.
5423 * config/vxworksae.h: Likewise.
5425 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
5427 PR tree-optimization/83104
5428 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
5431 2017-11-22 Marek Polacek <polacek@redhat.com>
5432 H.J. Lu <hongjiu.lu@intel.com>
5433 Jason Merrill <jason@redhat.com>
5438 * calls.c (initialize_argument_information): Call
5439 warn_parameter_passing_abi target hook.
5440 (store_one_arg): Use 0 for empty record size. Don't push 0 size
5441 argument onto stack.
5442 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
5443 * common.opt: Update -fabi-version description.
5444 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
5445 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
5447 (ix86_is_empty_record): New function.
5448 (ix86_warn_parameter_passing_abi): New function.
5449 (TARGET_EMPTY_RECORD_P): Redefine.
5450 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
5451 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
5452 * doc/tm.texi: Regenerated.
5453 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
5454 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
5455 * dwarf2out.c (get_ultimate_context): Move to tree.c.
5456 * explow.c (hard_function_value): Call arg_int_size_in_bytes
5457 instead of int_size_in_bytes.
5458 * expr.c (copy_blkmode_to_reg): Likewise.
5459 * function.c (aggregate_value_p): Return 0 for empty types.
5460 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
5462 (locate_and_pad_parm): Call arg size_in_bytes instead
5464 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
5465 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
5466 * target.def (empty_record_p, warn_parameter_passing_abi): New target
5468 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
5469 (std_gimplify_va_arg_expr): Skip empty records. Call
5470 arg_size_in_bytes instead size_in_bytes.
5471 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
5472 * tree-core.h (tree_type_common): Add empty_flag.
5473 (tree_decl_common): Update comments.
5474 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
5476 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5477 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
5479 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5480 * tree.c (default_is_empty_type): New function.
5481 (default_is_empty_record): New function.
5482 (arg_int_size_in_bytes): New function.
5483 (arg_size_in_bytes): New function.
5484 (get_ultimate_context): New function.
5485 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
5486 TRANSLATION_UNIT_WARN_EMPTY_P.
5487 (default_is_empty_record, arg_int_size_in_bytes,
5488 arg_size_in_bytes, get_ultimate_context): Declare.
5490 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5492 * config/arm/arm.c (cmse_clear_registers): New function.
5493 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
5494 code by call to cmse_clear_registers.
5495 (cmse_nonsecure_entry_clear_before_return): Likewise.
5497 2017-11-22 Tamar Christina <tamar.christina@arm.com>
5499 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
5500 (vdot_s32, vdotq_s32): New.
5501 (vdot_lane_u32, vdotq_lane_u32): New.
5502 (vdot_lane_s32, vdotq_lane_s32): New.
5505 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5508 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
5509 for unsigned values with fewer HWIs than the precision.
5510 (test_overflow): New function.
5511 (wide_int_cc_tests): Call it.
5513 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5515 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
5516 has the same mode class as Pmode.
5518 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5520 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
5521 padding_bits_to_clear_ptr.
5522 (cmse_nonsecure_entry_clear_before_return): Likewise.
5524 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5526 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
5527 auto_sbitap instead of integer bitfield to control register needing
5530 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5532 PR tree-optimization/83044
5533 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
5534 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
5535 that need the upper bound. Subtract offset from
5536 get_addr_base_and_unit_offset only if positive and subtract it
5537 before division by eltsize rather than after it.
5540 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
5541 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
5542 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
5546 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
5547 expand_binop with *mul_widen_optab, make sure at least one of the
5548 operands doesn't have VOIDmode.
5551 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
5553 PR rtl-optimization/82044
5554 PR tree-optimization/82042
5555 * dse.c (record_store): Check for overflow.
5556 (check_mem_read_rtx): Properly check for overflow if width == -1, call
5557 add_wild_read instead of clear_rhs_from_active_local_stores on
5558 overflow and log it into dump_file.
5560 2017-11-22 Richard Biener <rguenther@suse.de>
5562 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
5563 fake edges to exit when looking for a place to insert.
5564 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
5566 (insert_into_preds_of_block): Commit edge insertion immediately,
5567 assert that doesn't require new BBs.
5568 (fini_pre): Release expressions.
5569 (pass_pre::execute): Shuffle things around a bit, if the fn
5570 is too large do not compute AVAIL either as this is really the
5573 2017-11-22 Richard Biener <rguenther@suse.de>
5575 PR tree-optimization/83089
5576 * tree-if-conv.c (pass_if_conversion::execute): If anything
5577 changed reset SCEV and free the number of iteration estimates.
5579 2017-11-21 Martin Sebor <msebor@redhat.com>
5581 PR tree-optimization/82945
5582 * calls.h (warn_nonstring_bound): Remove unused function.
5584 2017-11-21 Martin Sebor <msebor@redhat.com>
5586 PR tree-optimization/82945
5587 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
5588 * calls.h (maybe_warn_nonstring_arg): Declare new function.
5589 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
5591 (initialize_argument_information): Call maybe_warn_nonstring_arg.
5592 * calls.h (get_attr_nonstring_decl): Declare new function.
5593 * doc/extend.texi (attribute nonstring): Update.
5594 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
5595 get_attr_nonstring_decl and handle it.
5596 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
5597 detection of nul-termination.
5598 (strlen_to_stridx): Change to a pointer.
5599 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
5600 (pass_strlen::execute): Same.
5602 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
5604 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
5605 for the new option -mprefer-vector-width=[none|128|256|512].
5606 * config/i386/i386.c (ix86_target_string): remove old style options
5607 -mprefer-avx256 and make -mprefer-avx128 as alias.
5608 (ix86_option_override_internal): Apply defaults for the
5609 -mprefer-vector-width=[128|256] option.
5610 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
5611 Implement macros to work with -mprefer-vector-width=.
5612 * config/i386/i386.opt: Implemented option
5613 -mprefer-vector-width=[none|128|256|512].
5614 * doc/invoke.texi: Documentation for
5615 -mprefer-vector-width=[none|128|256|512].
5617 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5619 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5621 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5623 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5625 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5627 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5629 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5630 * expr.c (expand_expr_real_2): Likewise.
5631 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5632 fold_binary_loc): Likewise.
5633 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5634 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5635 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5636 MINUS_EXPR transformations.
5637 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5638 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5639 * tree-inline.c (estimate_operator_cost): Likewise.
5640 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5641 op_symbol_code): Likewise.
5642 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5643 * vr-values.c (extract_range_from_binary_expr): Likewise.
5644 * varasm.c (initializer_constant_valid_p_1): Likewise.
5645 * tree.def: New tree code POINTER_DIFF_EXPR.
5647 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5649 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5650 integer suffix to movbe mnemonic.
5651 (*bswaphi2_movbe): Ditto.
5652 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5654 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5657 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5658 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5659 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5661 2017-11-21 Martin Liska <mliska@suse.cz>
5663 * tree-inline.c (expand_call_inline): Remove not needed
5666 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5667 Jakub Jelinek <jakub@redhat.com>
5670 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5671 Remove static keyword from f variable.
5673 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5675 PR tree-optimization/83086
5676 * gimple-ssa-store-merging.c
5677 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5678 rather than n.base_addr.
5680 2017-11-21 Martin Liska <mliska@suse.cz>
5682 PR rtl-optimization/82044
5683 PR tree-optimization/82042
5684 * dse.c (check_mem_read_rtx): Check for overflow.
5686 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5688 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5691 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5693 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5694 and its corresponding call arg location note.
5696 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5697 Andrew Burgess <andrew.burgess@embecosm.com>
5699 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5701 (arc_return_slot_offset): Likewise.
5702 (arc_eh_return_address_location): New declaration.
5703 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5704 (MUST_SAVE_REGISTER): Add exception handler case.
5705 (MUST_SAVE_RETURN_ADDR): Likewise.
5706 (arc_frame_pointer_required): Likewise.
5707 (arc_frame_pointer_needed): New function.
5708 (arc_compute_frame_size): Changed.
5709 (arc_expand_prologue): Likewise.
5710 (arc_expand_epilogue): Likewise.
5711 (arc_initial_elimination_offset): Likewise.
5712 (arc_return_slot_offset): Delete.
5713 (arc_eh_return_address_location): New function.
5714 (arc_builtin_setjmp_frame_value): Likewise.
5715 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5716 (EH_RETURN_STACKADJ_RTX): Define.
5717 (EH_RETURN_HANDLER_RTX): Likewise.
5718 * config/arc/arc.md (eh_return): Delete.
5720 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5722 * print-tree.h (debug_vec_tree): Remove prototype.
5723 * gdbinit.in (pvt): Remove macro.
5725 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5727 PR tree-optimization/83047
5728 * gimple-ssa-store-merging.c
5729 (imm_store_chain_info::output_merged_store): If the loads with the
5730 same vuse are in different basic blocks, for load_gsi pick a load
5731 location that is dominated by the other loads.
5734 * config/i386/i386.c (ix86_memmodel_check): Start
5735 -Winvalid-memory-model diagnostics with lowercase letter.
5738 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5739 set in_cold_section_p to the partition containing loc_list->first.
5740 When seeing loc_list->last_before_switch node, update secname and
5741 perform range_across_switch second partition handling only after that.
5744 * run-rtl-passes.c: Include debug.h.
5745 (run_rtl_passes): Call debug_hooks->assembly_start.
5746 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5750 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5751 OPTAB_DIRECT in calls to expand_simple_binop.
5753 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5756 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5758 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5761 * spellcheck.h (best_match::blithely_get_best_candidate): New
5764 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5766 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5767 with lower case letter.
5769 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5771 * config/i386/i386.md (bswaphi2): New expander.
5772 (*bswaphi2_movbe): New insn pattern.
5773 (bswaphi -> rorhi pepehole2): New peephole pattern.
5775 2017-11-20 Jeff Law <law@redhat.com>
5777 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5778 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5779 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5780 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5781 gimple-ssa-evrp-analyze.h.
5783 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5784 set BB_VISITED here.
5785 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5787 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5789 * tree-predcom.c: Add general comment on Store-Store chains.
5790 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5792 (get_chain_last_ref_at): Rename into...
5793 (get_chain_last_write_at): ...this.
5794 (get_chain_last_write_before_load): New function.
5795 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5796 CT_STORE_STORE when write reference is added.
5797 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5798 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5799 (initialize_root_vars_store_elim_1): Ditto.
5800 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5801 definition is created.
5802 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5803 chain by replacing it with dominant stored value.
5805 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5807 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5809 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5811 * vr-values.c (extract_range_from_binary_expr): Use a full range
5814 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5816 * config/arm/arm.md (R4_REGNUM): Define constant.
5817 (nonsecure_call_internal): Remove r4 clobber.
5818 (nonsecure_call_value_internal): Likewise.
5819 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5820 clobber and resequence match_operands.
5821 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5822 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5823 (nonsecure_call_value_reg_thumb2): Likewise.
5825 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5827 PR tree-optimization/78821
5828 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5829 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5830 to the address of the base rather than the base itself.
5831 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5832 (find_bswap_or_nop_finalize): New function.
5833 (find_bswap_or_nop): Use it.
5834 (bswap_replace): Return a tree rather than bool, change first
5835 argument from gimple * to gimple_stmt_iterator, allow inserting
5836 into an empty sequence, allow ins_stmt to be NULL - then emit
5837 all stmts into gsi. Fix up MEM_REF address gimplification.
5838 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5839 (struct store_immediate_info): Add N and INS_STMT non-static
5841 (store_immediate_info::store_immediate_info): Initialize them
5842 from newly added ctor args.
5843 (merged_store_group::apply_stores): Formatting fixes. Sort by
5845 (stmts_may_clobber_ref_p): For stores call also
5846 refs_anti_dependent_p.
5847 (gather_bswap_load_refs): New function.
5848 (imm_store_chain_info::try_coalesce_bswap): New method.
5849 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5850 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5851 (imm_store_chain_info::output_merged_store): Fail if number of
5852 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5853 and NOP_EXPR rhs_code.
5854 (pass_store_merging::process_store): Compute n and ins_stmt, if
5855 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5856 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5858 (pass_store_merging::execute): Calculate dominators.
5860 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5861 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5862 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
5863 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5864 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
5865 class pass_optimize_bswap, bswap_replace,
5866 pass_optimize_bswap::execute): Moved to ...
5867 * gimple-ssa-store-merging.c: ... this file.
5868 Include optabs-tree.h.
5869 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
5870 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5871 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
5872 anonymous namespace, remove static keywords.
5873 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
5874 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
5876 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
5879 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
5881 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
5883 * vec.h (debug_helper): New function.
5884 (DEFINE_DEBUG_VEC): New macro.
5885 * hash-set.h (debug_helper): New function.
5886 (DEFINE_DEBUG_HASH_SET): New macro.
5887 * cfg.c (debug_slim (edge)): New function.
5888 Call DEFINE_DEBUG_VEC for edges.
5889 Call DEFINE_DEBUG_HASH_SET for edges.
5890 * cfghooks.c (debug_slim (basic_block)): New function.
5891 Call DEFINE_DEBUG_VEC for basic blocks.
5892 Call DEFINE_DEBUG_HASH_SET for basic blocks.
5893 * print-tree.c (debug_slim): New function to handle trees.
5894 Call DEFINE_DEBUG_VEC for trees.
5895 Call DEFINE_DEBUG_HASH_SET for trees.
5896 (debug (vec<tree, va_gc>) &): Remove.
5897 (debug (<vec<tree, va_gc>) *): Remove.
5898 * print-rtl.c (debug_slim): New function to handle const_rtx.
5899 Call DEFINE_DEBUG_VEC for rtx_def.
5900 Call DEFINE_DEBUG_VEC for rtx_insn.
5901 Call DEFINE_DEBUG_HASH_SET for rtx_def.
5902 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
5903 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
5904 (debug (vec<rtx_insn *> *ptr): Remove.
5905 (debug_insn_vector): Remove.
5906 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
5908 2017-11-20 Tom de Vries <tom@codesourcery.com>
5910 PR rtl-optimization/82020
5911 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
5912 IF_THEN_ELSE condition.
5914 2017-11-19 Jeff Law <law@redhat.com>
5916 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
5917 of degenerates resulting from ignoring an edge.
5919 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5922 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
5924 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5927 * profile-count.c (profile_count::to_sreal_scale): Fix return value
5928 for uninitialied counts.
5930 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5933 * tree-inline.c (estimate_num_insns): Set to 1 at least.
5935 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5938 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
5940 2017-11-19 Tom de Vries <tom@codesourcery.com>
5942 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
5945 2017-11-19 Tom de Vries <tom@codesourcery.com>
5947 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
5950 2017-11-19 Tom de Vries <tom@codesourcery.com>
5952 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5955 2017-11-19 Tom de Vries <tom@codesourcery.com>
5957 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
5959 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
5961 2017-11-19 Tom de Vries <tom@codesourcery.com>
5963 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
5965 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
5966 semicolon after MCORE_EXPORT_NAME call.
5968 2017-11-19 Tom de Vries <tom@codesourcery.com>
5971 * vmsdbgout.c (vmsdbgout_early_finish): New function.
5972 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
5974 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
5976 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
5977 (cgraph_edge::clone): Cleanup updating of profile.
5978 * ipa-cp.c (update_profiling_info): Likewise.
5979 * ipa-inline-transform.c (inline_transform): Likewise.
5980 * ipa-inline.c (inline_small_functions): Add missing space to dump.
5981 * ipa-split.c (execute_split_functions): Do not split when function
5983 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
5984 * profile-count.c (profile_count::dump): Add global0.
5985 (profile_count::to_cgraph_frequency): Do not ICE when entry is
5987 (profile_count::to_sreal_scale): Likewise.
5988 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
5989 (profile_count::combine_with_ipa_count): New function.
5990 * profile-count.h (profile_guessed_global0adjusted): New.
5991 (profile_count::adjusted_zero): New.
5992 (profile_count::global0adjusted): New.
5993 (profile_count::combine_with_ipa_count): New.
5994 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
5995 correct profile of return block of split functions.
5996 (copy_cfg_body): Remove unused profile_count.
5997 (copy_body): Likewise.
5998 (expand_call_inline): Update.
5999 (tree_function_versioning): Update.
6001 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6003 * hash-set.h (hash_set::empty): New.
6004 * tree-ssa-threadbackward.h: Delete.
6005 * tree-ssa-threadbackward.c (class thread_jumps): New.
6006 Move max_threaded_paths into class.
6007 (fsm_find_thread_path): Remove arguments that are now in class.
6008 (profitable_jump_thread_path): Rename to...
6009 (thread_jumps::profitable_jump_thread_path): ...this.
6010 (convert_and_register_jump_thread_path): Rename to...
6011 (thread_jumps::convert_and_register_current_path): ...this.
6012 (check_subpath_and_update_thread_path): Rename to...
6013 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6014 (register_jump_thread_path_if_profitable): Rename to...
6015 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6016 (handle_phi): Rename to...
6017 (thread_jumps::handle_phi): ...this.
6018 (handle_assignment): Rename to...
6019 (thread_jumps::handle_assignment): ...this.
6020 (fsm_find_control_statement_thread_paths): Rename to...
6021 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6022 (find_jump_threads_backwards): Rename to...
6023 (thread_jumps::find_jump_threads_backwards): ...this.
6024 Initialize path local data.
6025 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6026 from within thread_jumps class.
6027 (pass_early_thread_jumps::execute): Same.
6029 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6031 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6033 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6035 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6036 to canonical location.
6038 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6041 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6043 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6045 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6047 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6048 rcrt1.o%s/grcrt1.o%s for -static-pie.
6050 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6052 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6053 ix86_shift_rotate_cost): Break out from ...
6054 (ix86_rtx_costs): ... here.
6055 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6058 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6060 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6061 when entry block was promoted unlikely.
6062 (estimate_bb_frequencies): Increase frequency scale.
6063 * profile-count.h (profile_count): Export precision info.
6065 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6067 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6068 disturbin profile of entry block.
6070 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6072 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6075 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6077 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6078 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6080 2017-11-17 Jeff Law <law@redhat.com>
6082 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6083 from evrp_dom_walker class. Various methods moved into new class.
6084 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6085 (evrp_range_analyzer::enter): New method.
6086 (evrp_range_analyzer::leave): New method.
6087 (evrp_dom_walker): Remove delegators no longer needed by this class.
6088 Replace vr_values data member with evrp_range_analyzer
6090 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6091 method extracted from evrp_dom_walker::before_dom_children.
6092 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6093 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6095 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6096 Add private copy constructor and move assignment operators.
6097 Privatize methods and class data where trivially possible.
6098 (evrp_dom_walker::cleanup): New function, extracted from
6099 execute_early_vrp. Simplify access to class data.
6101 * vr-values.h (get_output_for_vrp): Prototype.
6102 * vr-values.c (get_output_for_vrp): New function extracted from
6103 vrp_visit_assignment_or_call and extract_range_from_stmt.
6104 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
6106 2017-11-17 Luis Machado <luis.machado@linaro.org>
6108 * config/aarch64/aarch64.c
6109 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
6110 (qdf24xx_tunings) <autoprefetcher_model>: Set to
6111 tune_params::AUTOPREFETCHER_WEAK.
6113 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6116 * config/arm/arm.c (arm_valid_target_attribute_rec):
6117 Parse "arch=" and "+<ext>".
6118 (arm_valid_target_attribute_tree): Re-init global options.
6119 (arm_option_override): Make non-static.
6120 (arm_options_perform_arch_sanity_checks): Make errors fatal.
6121 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
6122 (__ARM_FEATURE_CRC32): Support undef.
6123 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
6124 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
6126 2017-11-17 David Malcolm <dmalcolm@redhat.com>
6128 * gdbinit.in (break-on-diagnostic): New command.
6130 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6132 * config/i386/i386.c (ix86_expand_epilogue): Change simple
6133 return to indirect jump for EH return if control-flow
6134 protection is enabled. Change explicit 'false' argument in
6135 pro_epilogue_adjust_stack with a value of
6137 * config/i386/i386.md (simple_return_indirect_internal):
6138 Remove SImode restriction to support 64-bit.
6140 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6142 * combine.c (added_notes_insn): New.
6143 (try_combine): Handle added_notes_insn like added_links_insn.
6144 Rewrite return value code.
6145 (distribute_notes): Set added_notes_insn to the earliest insn we added
6148 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6150 PR rtl-optimization/82621
6151 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
6152 dest of one of those SETs is unused.
6154 2017-11-17 Richard Biener <rguenther@suse.de>
6157 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
6158 * tree-pretty-print.c (dump_generic_node): Handle
6159 annot_expr_parallel_kind.
6160 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
6161 * gimplify.c (gimple_boolify): Likewise.
6163 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6165 * config.gcc (extra_headers): Add cet.h for x86 targets.
6166 * config/i386/cet.h: New file.
6167 * doc/install.texi: Add --enable-cet/--disable-cet.
6169 2017-11-17 Richard Biener <rguenther@suse.de>
6171 PR tree-optimization/83017
6172 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
6173 (gen_parallel_loop): Properly count iterations.
6174 (parallelize_loops): Handle loop->can_be_parallel independent
6175 of flag_loop_parallelize_all. Make static profitability test match
6177 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
6178 * invoke.texi (parloops-min-per-thread): Document.
6180 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
6182 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
6183 upstreaming review comments.
6185 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6187 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
6188 with fast unaligned access.
6189 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
6191 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
6193 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
6194 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
6196 2017-11-17 Richard Biener <rguenther@suse.de>
6198 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
6199 folding of references.
6201 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
6204 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
6205 of replacing call to strncmp with corresponding call to strcmp when
6208 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
6210 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
6211 option prefer-avx256 for skylake-avx512 configuration.
6212 * config/i386/i386.c (ix86_option_override_internal): Ditto.
6213 (get_builtin_code_for_version): Ditto.
6215 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6216 Monk Chiang <sh.chiang04@gmail.com>
6218 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
6219 (FIXED_REGISTERS): Reserve more register numbers.
6220 (CALL_USED_REGISTERS): Likewise.
6221 (REG_ALLOC_ORDER): Likewise.
6222 (REG_CLASS_CONTENTS): Likewise.
6223 (REGISTER_NAMES): Likewise.
6225 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6226 Kito Cheng <kito.cheng@gmail.com>
6228 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
6230 * config/nds32/iterators.md: Add vector mode iterators and attributes.
6232 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
6234 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
6235 parameter list for vec_splats.
6237 2017-11-16 Joseph Myers <joseph@codesourcery.com>
6239 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
6241 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
6243 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
6246 * doc/install.texi: Document incompatibility of Perl >=5.6.26
6247 with the required version of automake 1.11.6.
6249 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
6251 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
6253 (DU_C2_3_power9): Correct reservation combinations.
6254 (FP_DIV_power9, VEC_DIV_power9): New.
6255 (power9-alu): Split out rotate/shift...
6256 (power9-rot): ...to here, correct dispatch resource.
6257 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
6259 (power9-fp): Correct latency.
6260 (power9-sdiv): Add div/sqrt resource.
6261 (power9-ddiv): Correct latency, add div/sqrt resource.
6262 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
6263 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
6265 (power9-qpdiv, power9-qpmul): Adjust resource usage.
6267 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
6269 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
6270 switch statement mapping KF built-ins to TF built-ins if we don't
6271 have the proper ISA 3.0 assembler support.
6273 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6275 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
6276 (lower_emutls_function_body): Do not compute it.
6278 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6280 * ipa-split.c (split_bb_info): Turn time to sreal.
6281 (split_point): Likewise.
6282 (dump_split_point): Likewise.
6283 (fine_split_points): Likewise.
6284 (execute_split_functions): Only zero split_bbs; turn time to sreals.
6286 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6288 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
6291 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6293 * predict.c (combine_predictions_for_bb): Preserve zero predicted
6295 (expensive_function_p): Remove useless assert.
6296 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
6298 2017-11-16 Martin Sebor <msebor@redhat.com>
6300 PR tree-optimization/82588
6301 PR tree-optimization/82583
6302 * tree-vrp.c (check_array_ref): Handle flexible array members,
6303 string literals, and inner indices.
6304 (search_for_addr_array): Add detail to diagnostics.
6306 2017-11-16 Nathan Sidwell <nathan@acm.org>
6310 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
6311 (SET_DECL_ASSEMBLER_NAME): Forward to
6312 overwrite_decl_assembler_name.
6313 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
6314 (overwrite_decl_assembler_name): Declare.
6315 * tree.c (overwrite_decl_assembler_name): New.
6316 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
6317 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
6318 (LANG_HOOKS_INITIALIZER): Add it.
6319 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
6320 * langhooks.c (lhd_set_decl_assembler_name): Use
6321 SET_DECL_ASSEMBLER_NAME.
6322 (lhd_overwrite_decl_assembler_name): Default implementation.
6324 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6325 Jackson Woodruff <jackson.woodruff@arm.com>
6327 PR tree-optimization/71026
6328 * match.pd: Canonicalize constant multiplies in division.
6330 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6332 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
6333 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
6334 Remove OPT_fomit_frame_pointer entry.
6335 * common/config/arc/arc-common.c: Likewise.
6336 * common/config/arm/arm-common.c: Likewise.
6337 * common/config/avr/avr-common.c: Likewise.
6338 * common/config/c6x/c6x-common.c: Likewise.
6339 * common/config/cr16/cr16-common.c: Likewise.
6340 * common/config/cris/cris-common.c: Likewise.
6341 * common/config/epiphany/epiphany-common.c: Likewise.
6342 * common/config/fr30/fr30-common.c: Likewise.
6343 * common/config/frv/frv-common.c: Likewise.
6344 * common/config/ia64/ia64-common.c: Likewise.
6345 * common/config/iq2000/iq2000-common.c: Likewise.
6346 * common/config/lm32/lm32-common.c: Likewise.
6347 * common/config/m32r/m32r-common.c: Likewise.
6348 * common/config/mcore/mcore-common.c: Likewise.
6349 * common/config/microblaze/microblaze-common.c: Likewise.
6350 * common/config/mips/mips-common.c: Likewise.
6351 * common/config/mmix/mmix-common.c: Likewise.
6352 * common/config/mn10300/mn10300-common.c: Likewise.
6353 * common/config/nios2/nios2-common.c: Likewise.
6354 * common/config/pa/pa-common.c: Likewise.
6355 * common/config/pdp11/pdp11-common.c: Likewise.
6356 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
6357 * common/config/riscv/riscv-common.c: Likewise.
6358 * common/config/rs6000/rs6000-common.c: Likewise.
6359 * common/config/rx/rx-common.c: Likewise.
6360 * common/config/s390/s390-common.c: Likewise.
6361 * common/config/sh/sh-common.c: Likewise.
6362 * common/config/sparc/sparc-common.c: Likewise.
6363 * common/config/tilegx/tilegx-common.c: Likewise.
6364 * common/config/tilepro/tilepro-common.c: Likewise.
6365 * common/config/v850/v850-common.c: Likewise.
6366 * common/config/visium/visium-common.c: Likewise.
6367 * common/config/xstormy16/xstormy16-common.c: Likewise.
6368 * common/config/xtensa/xtensa-common.c: Likewise.
6369 * invoke.texi (-fomit-frame-pointer): Update documentation.
6371 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6373 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
6375 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6377 * cfg.c (scale_bbs_frequencies_int,
6378 cale_bbs_frequencies_gcov_type): Remove.
6379 * cfg.h (scale_bbs_frequencies_int,
6380 cale_bbs_frequencies_gcov_type): Remove.
6382 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6384 * tree-ssa-loop-manip.c
6385 (scale_dominated_blocks_in_loop): Update to profile counts.
6386 (tree_transform_and_unroll_loop): Likewise.
6388 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6390 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
6391 scale_bbs_frequencies_int.
6393 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6395 * final.c (compute_alignments): Use counts rather than frequencies.
6397 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6399 * cfgloopanal.c: Include sreal.h
6400 (average_num_loop_insns): Use counts and sreal for accounting.
6402 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6404 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
6407 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6408 Kito Cheng <kito.cheng@gmail.com>
6410 * config/nds32/constraints.md: Provide more constraints.
6411 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
6412 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
6413 support constraints usage.
6415 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6417 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
6419 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6420 Kito Cheng <kito.cheng@gmail.com>
6422 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
6423 * config/nds32/nds32.opt: Refine the layout.
6424 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
6425 TARGET_EXT_STRING): Support new options.
6426 * config/nds32/nds32.h: Likewise.
6427 * config/nds32/nds32.md: Likewise.
6428 * config/nds32/nds32-predicates.c: Likewise.
6429 * config/nds32/constraints.md: Likewise.
6430 * common/config/nds32/nds32-common.c: Likewise.
6432 2017-11-16 Julia Koval <julia.koval@intel.com>
6435 * config/i386/gfniintrin.h: Add sse check.
6436 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
6438 2017-11-16 Julia Koval <julia.koval@intel.com>
6440 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
6441 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
6442 (ix86_handle_option): Handle -mavx512vbmi2.
6443 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
6444 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
6445 * config/i386/i386-c.c (__AVX512VBMI2__): New.
6446 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
6447 (ix86_valid_target_attribute_inner_p): Ditto.
6448 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
6449 * config/i386/i386.opt (mavx512vbmi2): New option.
6450 * doc/invoke.texi: Add new option.
6452 2017-11-16 Julia Koval <julia.koval@intel.com>
6454 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
6455 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
6456 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
6457 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
6458 _mm512_gf2p8mul_epi8): New intrinsics.
6459 * config/i386/i386-builtin-types.def
6460 (V64QI_FTYPE_V64QI_V64QI): New type.
6461 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
6462 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
6463 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
6464 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
6465 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
6466 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
6468 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
6470 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
6471 explicitly as a stream of bytes.
6473 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6475 * config/rs6000/altivec.h (vec_xst_be): New #define.
6476 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
6477 and externalize from *altivec_vperm_<mode>_internal.
6478 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
6480 (XL_BE_V8HI): Likewise.
6481 (XL_BE_V4SI): Likewise.
6482 (XL_BE_V4SI): Likewise.
6483 (XL_BE_V2DI): Likewise.
6484 (XL_BE_V4SF): Likewise.
6485 (XL_BE_V2DF): Likewise.
6486 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
6487 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
6488 all array entries with these keys: VSX_BUILTIN_VEC_XL,
6489 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
6490 VSX_BUILTIN_VEC_XST_BE.
6491 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
6492 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
6494 (altivec_init_builtins): Replace conditional calls to def_builtin
6495 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
6496 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
6497 with unconditional calls. Remove calls to def_builtin for
6498 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
6499 __builtin_vec_xst_be.
6500 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
6501 to define_expand, and add alternate RTL generation for P8.
6502 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
6503 vsx_ld_elemrev_v8hi.
6504 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
6505 add alternate RTL generation for P8.
6506 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
6507 vsx_ld_elemrev_v16qi.
6508 (vsx_st_elemrev_v8hi): Convert define_insn
6509 to define_expand, and add alternate RTL generation for P8.
6510 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
6511 vsx_st_elemrev_v8hi.
6512 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
6513 add alternate RTL generation for P8.
6514 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
6515 vsx_st_elemrev_v16qi.
6517 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
6520 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
6521 TARGET_AVX512ER check.
6522 (ix86_option_override_internal): Set MASK_VZEROUPPER if
6523 neither -mzeroupper nor -mno-zeroupper is used and
6524 TARGET_EMIT_VZEROUPPER is set.
6525 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
6526 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
6528 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6530 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6531 folding of vector compares.
6532 (fold_build_vec_cmp): New helper function.
6533 (fold_compare_helper): New helper function.
6534 (builtin_function_type): Add compare builtins to the list of functions
6535 having unsigned arguments. Cosmetic updates to comment indentation.
6536 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
6537 the not+eq combination.
6539 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6541 PR tree-optimization/82726
6542 PR tree-optimization/70754
6543 * tree-predcom.c (order_drefs_by_pos): New function.
6544 (combine_chains): Move code setting has_max_use_after to...
6545 (try_combine_chains): ...here. New parameter. Sort combined chains
6546 according to position information.
6547 (tree_predictive_commoning_loop): Update call to above function.
6548 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
6550 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6552 PR tree-optimization/82726
6554 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6556 PR tree-optimization/70754
6557 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6558 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6559 combined stmt before it if not NULL.
6560 (combine_chains): Process refs reversely and compute dominance point
6564 2017-02-23 Bin Cheng <bin.cheng@arm.com>
6566 PR tree-optimization/79663
6567 * tree-predcom.c (combine_chains): Process refs in reverse order
6568 only for ZERO length chains, and add explaining comment.
6570 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6572 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
6573 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
6574 * doc/invoke.texi (ARM Options): Add armv8.3-a.
6576 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6578 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
6580 2017-11-15 Martin Liska <mliska@suse.cz>
6582 * tree-cfg.c (pass_warn_function_return::execute):
6583 Compare warn_return_type for greater than zero.
6585 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
6589 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
6590 to return true on Xeon and not on Xeon Phi.
6591 (ix86_check_avx256_register): Changed to ...
6592 (ix86_check_avx_upper_register): ... this. Add extra check for
6593 VALID_AVX512F_REG_OR_XI_MODE.
6594 (ix86_avx_u128_mode_needed): Changed
6595 ix86_check_avx256_register to ix86_check_avx_upper_register.
6596 (ix86_check_avx256_stores): Changed to ...
6597 (ix86_check_avx_upper_stores): ... this. Changed
6598 ix86_check_avx256_register to ix86_check_avx_upper_register.
6599 (ix86_avx_u128_mode_after): Changed
6600 avx_reg256_found to avx_upper_reg_found. Changed
6601 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
6602 (ix86_avx_u128_mode_entry): Changed
6603 ix86_check_avx256_register to ix86_check_avx_upper_register.
6604 (ix86_avx_u128_mode_exit): Ditto.
6605 * config/i386/i386.h: (host_detect_local_cpu): New define.
6607 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6609 * config/arm/xgene1.md (xgene1): Split into automatons
6610 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
6611 (xgene1_f_load): Adjust reservations and/or types.
6612 (xgene1_f_store): Likewise.
6613 (xgene1_load_pair): Likewise.
6614 (xgene1_store_pair): Likewise.
6615 (xgene1_fp_load1): Likewise.
6616 (xgene1_load1): Likewise.
6617 (xgene1_store1): Likewise.
6618 (xgene1_move): Likewise.
6619 (xgene1_alu): Likewise.
6620 (xgene1_simd): Likewise.
6621 (xgene1_bfm): Likewise.
6622 (xgene1_neon_load1): Likewise.
6623 (xgene1_neon_store1): Likewise.
6624 (xgene1_neon_logic): Likewise.
6625 (xgene1_neon_st1): Likewise.
6626 (xgene1_neon_ld1r): Likewise.
6627 (xgene1_alu_cond): Added.
6628 (xgene1_shift_reg): Likwise.
6629 (xgene1_bfx): Likewise.
6630 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6632 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6635 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6637 (can_widen_mult_without_libcall): New function.
6638 (expand_mul_overflow): If only checking unsigned mul overflow,
6639 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6640 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6641 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6642 (expand_DIVMOD): Formatting fix.
6643 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6644 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6645 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6647 PR tree-optimization/82977
6648 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6649 constructed temporary to strlen_to_stridx.put.
6651 2017-11-15 Martin Liska <mliska@suse.cz>
6653 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6654 * configure: Regenerate.
6656 2017-11-15 Martin Liska <mliska@suse.cz>
6659 * config/sh/sh-mem.cc: Use proper probability for
6662 2017-11-14 Jeff Law <law@redhat.com>
6664 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6665 the red zone for stack_clash_protection_final_dynamic_probe targets
6666 when the total dynamic stack size is zero bytes.
6668 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6669 blocks is post order.
6671 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6673 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6674 * final.c (rest_of_clean_state): Set it for the
6675 -fcompare-debug dump.
6676 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6677 class when TDF_COMPARE_DEBUG is set.
6679 * dwarf2out.c (gen_producer_string): Discard
6682 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6685 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6686 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6687 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6688 (__TGMATH_REAL_2_3): Remove macros.
6689 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6690 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6692 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6694 (remquo): Define using __TGMATH_REAL_3.
6696 2017-11-14 Jeff Law <law@redhat.com>
6698 * vr-values.c: New file with contents extracted from tree-vrp.c.
6699 * Makefile.in (OBJS): Add vr-values.o
6700 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6701 (set_value_range, set_and_canonicalize_value_range): Likewise.
6702 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6703 (value_range_constant_singleton, symbolic_range_p): Likewise.
6704 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6705 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6706 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6707 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6708 (find_case_label_range, find_case_label_index): Likewise.
6709 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6710 (range_int_cst_singleton_p, value_inside_range): Likewise.
6711 (get_single_symbol): Likewise.
6712 (switch_update): Move structure definition here.
6713 (to_remove_edges, to_update_switch_stmts): Provide externs.
6714 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6715 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6716 (vrp_val_is_min, set_value_range): Likewise.
6717 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6718 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6719 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6720 (range_is_nonnull, range_int_cst_p): Likewwise.
6721 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6722 (get_single_symbol, operand_less_p): Likewise
6723 (compare_values_warnv, compare_values): Likewise.
6724 (value_inside_range, value_range_constant_singleton): Likewise.
6725 (zero_nonzero_bitgs_from_vr): Likewise.
6726 (extract_range_from_binary_expr_1): Likewise.
6727 (overflow_comparison_p): Likewise.
6728 (to_remove_edges, to_update_switch_stmts): Likewise.
6729 (find_case_label-index, find_case_label_range): Likewise.
6730 (switch_update, set_value_range_to_nonnegative): Remove.
6731 (set_value_range_to_truthvalue): Likewise.
6732 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6733 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6734 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6735 (find_case_label_ranges, test_for_singularity): Likewise.
6736 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6737 (x_vr_values): Move to its remaining use site.
6739 2017-11-10 Jeff Law <law@redhat.com>
6741 * vr-values.h (VR_INITIALIZER): Move #define here.
6742 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6743 * Makefile.in (OBJS): Add tree-evrp.o
6744 * tree-vrp.h (assert_info): Move structure definition here.
6745 (set_value_range_to_varying): Prototype.
6746 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6747 (infer_value_range, register_edge_assert_for): Likewise.
6748 (stmt_interesting_for_vrp): Likewise.
6749 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6750 (set_value_range_to_varying): No longer static.
6751 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6752 (infer_value_range, register_edge_assert_for): Likewise.
6754 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6756 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6757 generate the XXBRD instruction.
6759 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6760 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6761 as being compatible if -mabi=ieeelongdouble.
6762 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6763 to setup float128 built-ins with hardware support.
6764 (BU_FLOAT128_HW_2): Likewise.
6765 (BU_FLOAT128_HW_3): Likewise.
6766 (BU_FLOAT128_HW_VSX_1): Likewise.
6767 (BU_FLOAT128_HW_VSX_2): Likewise.
6768 (scalar_extract_expq): Change float128 built-in functions to
6769 accommodate having both KFmode and TFmode functions. Use the
6770 KFmode variant as the default.
6771 (scalar_extract_sigq): Likewise.
6772 (scalar_test_neg_qp): Likewise.
6773 (scalar_insert_exp_q): Likewise.
6774 (scalar_insert_exp_qp): Likewise.
6775 (scalar_test_data_class_qp): Likewise.
6776 (sqrtf128_round_to_odd): Delete processing the round to odd
6777 built-in functions as special built-in functions, and define them
6778 as float128 built-ins. Use the KFmode variant as the default.
6779 (truncf128_round_to_odd): Likewise.
6780 (addf128_round_to_odd): Likewise.
6781 (subf128_round_to_odd): Likewise.
6782 (mulf128_round_to_odd): Likewise.
6783 (divf128_round_to_odd): Likewise.
6784 (fmaf128_round_to_odd): Likewise.
6785 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6786 support for KFmode and TFmode xststdcqp calls.
6787 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6788 point, switch the built-in handlers for the get/set float128
6789 exponent, get float128 mantissa, float128 test built-ins, and the
6790 float128 round to odd built-in functions. Eliminate creating the
6791 float128 round to odd built-in functions as special built-ins.
6792 (rs6000_init_builtins): Eliminate special creation of the float128
6793 round to odd built-in functions.
6794 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6795 function insns to support both TFmode and KFmode variants.
6796 (xsxsigqp_<mode>): Likewise.
6797 (xsiexpqpf_<mode>): Likewise.
6798 (xsiexpqp_<mode>): Likewise.
6799 (xststdcqp_<mode>): Likewise.
6800 (xststdcnegqp_<mode>): Likewise.
6801 (xststdcqp_<mode>): Likewise.
6803 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6805 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6806 unused path_in_freq_ptr parameter.
6807 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6809 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6811 * ipa-inline.c (edge_badness): Dump sreal frequency.
6812 (compute_inlined_call_time): Match natural implementaiton ...
6813 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6814 forgotten division by CGRAPH_FREQ_BASE.
6816 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6818 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6819 Solaris 11. Update comment.
6820 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6822 * config/sol2.h (STARTFILE_SPEC): Likewise.
6823 * configure: Regenerate.
6825 2017-11-14 Carl Love <cel@us.ibm.com>
6827 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6828 le_ and be_ prefixes to swap* variables. Remove
6829 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6832 2017-11-14 Jason Merrill <jason@redhat.com>
6834 Support GTY((cache)) on hash_map.
6835 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6836 (ggc_cache_remove): Override it instead of ggc_mx.
6837 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6838 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6839 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6840 (simple_cache_map_traits): Override maybe_mx.
6841 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6842 (hash_map): Friend gt_cleare_cache.
6843 (gt_cleare_cache): New.
6844 * tree.h (tree_cache_traits): New hash_map traits class.
6845 (tree_cache_map): New typedef.
6847 2017-11-14 Richard Biener <rguenther@suse.de>
6849 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6850 paramter and handling.
6851 (cleanup_control_flow_bb): Likewise.
6852 (cleanup_control_flow_pre): New helper performing a DFS walk
6853 to call cleanup_control_flow_bb in PRE order.
6854 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6855 via cleanup_control_flow_pre.
6857 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6859 * config/aarch64/aarch64-simd.md
6860 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6861 (*aarch64_simd_bsl<mode>_alt): Likewise.
6862 (aarch64_simd_bsldi_internal): New.
6863 (aarch64_simd_bsldi_alt): Likewise.
6865 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6867 * tracer.c (better_p): Do not compare frequencies.
6868 * reg-stack.c (better_edge): Likewise.
6869 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
6872 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6874 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
6875 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
6876 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
6877 * lto-streamer-in.c (input_function): Use update_max_bb_count.
6878 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
6879 * predict.c (maybe_hot_frequency_p): Inline to ...
6880 (maybe_hot_count_p): ... here; rewrite to counts.
6881 (counts_to_freqs): Rename to ...
6882 (update_max_bb_count): ... this one.
6883 (expensive_function_p): Use counts.
6884 (estimate_bb_frequencies): Update.
6885 (rebuild_frequencies): Update.
6886 * predict.h (counts_to_freqs): Rename to ...
6887 (update_max_bb_count): ... this one.
6888 * profile.c (compute_branch_probabilities): Add debug info
6889 * tree-inline.c (expand_call_inline): Update debug info.
6890 (optimize_inline_calls): Use update_max_bb_count..
6891 (tree_function_versioning): Use update_max_bb_count..
6892 * value-prof.c (gimple_value_profile_transformations):
6893 Do not use update_max_bb_count.
6895 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6897 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
6898 always use frequencies.
6900 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6902 * bb-reorder.c: Remove frequencies from comments.
6903 (better_edge_p): Use profile counts.
6904 (find_traces): Dump profile counts.
6905 (rotate_loop): Use profile counts.
6906 (find_traces_1_round): Likewise.
6907 (connect_better_edge_p): Use counts instead of probabilities for
6909 (copy_bb_p): Drop early check for non-0 frequency.
6910 (sanitize_hot_paths): Update comments.
6912 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6914 * ipa-split.c (struct split_point): Add count.
6915 (consider_split): Do not compute incoming frequency; compute incoming
6916 count and store it to split_point.
6917 (split_function): Set count of the call to split part correctly.
6919 2017-11-13 Carl Love <cel@us.ibm.com>
6921 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
6923 2017-11-13 Tom Tromey <tom@tromey.com>
6925 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
6927 2017-11-13 Carl Love <cel@us.ibm.com>
6929 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
6930 Add support for builtins:
6931 unsigned int vec_first_{,miss}_match_{,or_eos}index,
6932 vector {un,}signed {char,int,short},
6933 vector {un,}signed {char,int,short}) arguments.
6934 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
6935 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
6936 Add BU_P9V_AV_2 expansions for the builtins.
6937 * config/rs6000/altivec.h (vec_first_match_index,
6938 vec_first_mismatch_index, vec_first_match_or_eos_index,
6939 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
6940 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
6941 new extern declaration.
6942 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
6943 * config/rs6000/vsx.md (first_match_index_<mode>,
6944 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
6945 first_mismatch_or_eos_index_<mode>): Add define expand.
6946 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
6947 * doc/extend.texi: Update the built-in documenation file for the new
6950 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
6952 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
6953 into the min/max operations for _Float<N> and _Float<N>X types.
6955 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6958 * dwarf2out.c (do_eh_frame): New static variable.
6959 (dwarf2out_begin_prologue): Set it.
6960 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
6962 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6964 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
6966 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
6968 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6971 * debug.h (dwarf2out_do_eh_frame): Declare.
6972 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
6973 (dwarf2out_do_frame): Use it.
6974 (dwarf2out_do_cfi_asm): Likewise.
6975 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
6976 (dwarf2out_assembly_start): Likewise.
6977 (dwarf2out_begin_prologue): Fix comment.
6978 * toplev.c (compile_file): Always call dwarf2out_frame_finish
6979 if the target needs either debug or unwind DWARF2 info.
6980 * lto-opts.c (lto_write_options): Do not save -fexceptions,
6981 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
6982 -ftrapping-math, -ftrapv and -fwrapv.
6984 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6986 * cgraph.c (cgraph_edge::sreal_frequency): New function.
6987 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
6988 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
6989 (estimate_edge_size_and_time): Likewise.
6990 (ipa_merge_fn_summary_after_inlining): Likewise.
6991 * ipa-inline.c (cgraph_freq_base_rec): Remove.
6992 (compute_uninlined_call_time): Use sreal_frequency.
6993 (compute_inlined_call_time): Likewise.
6994 (ipa_inline): Do not initialize cgraph_freq_base_rec.
6995 * profile-count.c: Include sreal.h.
6996 (profile_count::to_sreal_scale): New.
6997 * profile-count.h: Forward declare sreal.
6998 (profile_count::to_sreal_scale): Declare.
7000 2017-11-13 Nathan Sidwell <nathan@acm.org>
7002 * diagnostic.c (maybe_line_and_column): New.
7003 (diagnostic_get_location_text): Use it.
7004 (diagnostic_report_current_module): Likewise.
7005 (test_diagnostic_get_location_text): Add tests.
7007 2017-11-13 Luis Machado <luis.machado@linaro.org>
7009 * doc/md.texi (Specifying processor pipeline description): Fix
7010 incorrect latency for the div instruction example.
7012 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7014 PR tree-optimization/78821
7015 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7016 that bit_not_p is the same.
7017 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7018 (split_group): Count precisely bit_not_p bits in each statement.
7019 (invert_op): New function.
7020 (imm_store_chain_info::output_merged_store): Use invert_op to
7021 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7022 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7024 2017-11-13 Martin Liska <mliska@suse.cz>
7026 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7027 (struct source_info): Likewise.
7028 (add_branch_counts): Likewise.
7029 (add_line_counts): Likewise.
7030 (function_summary): Likewise.
7031 (output_intermediate_line): Likewise.
7032 (generate_results): Likewise.
7034 2017-11-13 Martin Liska <mliska@suse.cz>
7036 * gcov.c (struct block_info): Remove typedef for block_t.
7037 (struct line_info): Likewise.
7038 (line_info::has_block): Likewise.
7039 (EXIT_BLOCK): Likewise.
7040 (unblock): Likewise.
7041 (circuit): Likewise.
7042 (get_cycles_count): Likewise.
7043 (process_file): Likewise.
7044 (read_graph_file): Likewise.
7045 (solve_flow_graph): Likewise.
7046 (find_exception_blocks): Likewise.
7047 (add_line_counts): Likewise.
7048 (accumulate_line_info): Likewise.
7049 (output_line_details): Likewise.
7051 2017-11-13 Martin Liska <mliska@suse.cz>
7053 * gcov.c (struct arc_info): Remove typedef for arc_t.
7054 (struct line_info): Likewise.
7055 (add_branch_counts): Likewise.
7056 (output_branch_count): Likewise.
7057 (function_info::~function_info): Likewise.
7058 (circuit): Likewise.
7059 (output_intermediate_line): Likewise.
7060 (read_graph_file): Likewise.
7061 (solve_flow_graph): Likewise.
7062 (find_exception_blocks): Likewise.
7063 (add_line_counts): Likewise.
7064 (accumulate_line_info): Likewise.
7065 (output_line_details): Likewise.
7066 (output_function_details): Likewise.
7068 2017-11-13 Martin Liska <mliska@suse.cz>
7070 * gcov.c (struct function_info): Remove typedef for function_t.
7071 (struct source_info): Likewise.
7072 (source_info::get_functions_at_location): Likewise.
7073 (solve_flow_graph): Likewise.
7074 (find_exception_blocks): Likewise.
7075 (add_line_counts): Likewise.
7076 (output_intermediate_file): Likewise.
7077 (process_file): Likewise.
7078 (generate_results): Likewise.
7079 (release_structures): Likewise.
7080 (read_graph_file): Likewise.
7081 (read_count_file): Likewise.
7082 (accumulate_line_counts): Likewise.
7083 (output_lines): Likewise.
7085 2017-11-13 Martin Liska <mliska@suse.cz>
7087 * gcov.c (function_info::function_info): Remove num_counts
7088 and add vector<gcov_type>.
7089 (function_info::~function_info): Use the vector.
7090 (process_file): Likewise.
7091 (read_graph_file): Likewise.
7092 (read_count_file): Likewise.
7093 (solve_flow_graph): Likewise.
7095 2017-11-13 Martin Liska <mliska@suse.cz>
7097 * gcov.c (function_info::is_artificial): New function.
7098 (process_file): Erase all artificial early.
7099 (generate_results): Skip as all artificial are already
7102 2017-11-13 Martin Liska <mliska@suse.cz>
7104 * gcov.c (read_graph_file): Store to global vector of functions.
7105 (read_count_file): Iterate the vector.
7106 (process_file): Likewise.
7107 (generate_results): Likewise.
7108 (release_structures): Likewise.
7110 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7112 PR tree-optimization/82954
7113 * gimple-ssa-store-merging.c
7114 (imm_store_chain_info::coalesce_immediate_stores): If
7115 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
7117 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
7119 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7120 Upddate call to ENDIAN_LANE_N.
7121 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
7122 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
7123 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
7124 and use aarch64_endian_lane_rtx.
7125 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
7127 2017-11-12 Tom de Vries <tom@codesourcery.com>
7129 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
7131 2017-11-12 Tom de Vries <tom@codesourcery.com>
7133 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
7135 2017-11-12 Tom de Vries <tom@codesourcery.com>
7137 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
7139 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7140 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7141 * defaults.h (ASM_OUTPUT_LABELREF): Same.
7143 2017-11-11 Martin Sebor <msebor@redhat.com>
7146 * doc/extend.texi (attribute nonstring): Remove spurious argument.
7149 * prefic.c (translate_name): Replace strncpy with memcpy to
7150 avoid -Wstringop-truncation.
7152 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7154 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
7156 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7158 * predict.c (maybe_hot_frequency_p): Do not use cfun.
7160 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7162 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
7165 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7167 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
7168 enable generating XXBRH if the value is in a vector register.
7169 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
7170 value is in a vector register.
7171 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
7172 register to register bswap64's instead of doing the GPR sequence
7173 used on previous machines.
7174 (bswapdi2_xxbrd): New insn.
7175 (bswapdi2_reg): Disallow on ISA 3.0.
7176 (register to register bswap64 splitter): Do not split the insn on
7177 ISA 3.0 systems that use XXBRD.
7179 2017-11-10 Martin Sebor <msebor@redhat.com>
7182 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
7184 (find_subframework_file): Same.
7186 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7188 * auto-profile.c (afdo_indirect_call): Drop frequency.
7189 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
7190 (cgraph_node::create_edge): Drop frequency argument.
7191 (cgraph_node::create_indirect_edge): Drop frequency argument.
7192 (cgraph_edge::make_speculative): Drop frequency arguments.
7193 (cgraph_edge::resolve_speculation): Do not update frequencies
7194 (cgraph_edge::dump_edge_flags): Do not dump frequency.
7195 (cgraph_node::dump): Check consistency in IPA mode.
7196 (cgraph_edge::maybe_hot_p): Use IPA counter.
7197 (cgraph_edge::verify_count_and_frequency): Rename to ...
7198 (cgraph_edge::verify_count): ... this one; drop frequency checking.
7199 (cgraph_node::verify_node): Update.
7200 * cgraph.h (struct cgraph_edge): Drop frequency.
7201 (cgraph_edge::frequency): New function.
7202 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
7204 (cgraph_edge::rebuild_edges): Likewise.
7205 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
7206 (duplicate_thunk_for_node): Do not pass frequency.
7207 (cgraph_node::create_clone): Scale only counts.
7208 (cgraph_node::create_virtual_clone): Do not pass frequency.
7209 (cgraph_node::create_edge_including_clones): Do not pass frequency.
7210 (cgraph_node::create_version_clone): Do not pass frequency.
7211 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
7212 (cgraph_node::expand_thunk): Do not pass frequency.
7213 (cgraph_node::create_wrapper): Do not pass frequency.
7214 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
7216 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
7217 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
7218 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
7219 (gather_caller_stats): Use frequency function.
7220 (ipcp_cloning_candidate_p): Use frequency function.
7221 (ipcp_propagate_stage): Use frequency function.
7222 (get_info_about_necessary_edges): Use frequency function.
7223 (update_profiling_info): Update only IPA profile.
7224 (update_specialized_profile): Use frequency functoin.
7225 (perhaps_add_new_callers): Update only IPA profile.
7226 * ipa-devirt.c (ipa_devirt): Use IPA profile.
7227 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
7228 (dump_ipa_call_summary): Use frequency function.
7229 (estimate_edge_size_and_time): Use frequency function.
7230 (ipa_merge_fn_summary_after_inlining): Use frequency function.
7231 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
7232 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
7233 (update_noncloned_counts): ... ths one; scale counts only.
7234 (clone_inlined_nodes): Do not scale frequency.
7235 (inline_call): Do not pass frequency.
7236 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
7237 (compute_inlined_call_time): Use IPA profile.
7238 (want_inline_small_function_p): Use IPA profile.
7239 (want_inline_self_recursive_call_p): Use IPA profile.
7240 (edge_badness): Use IPA profile.
7241 (lookup_recursive_calls): Use IPA profile.
7242 (recursive_inlining): Do not pass frequency.
7243 (resolve_noninline_speculation): Do not update frequency.
7244 (inline_small_functions): Collect max of IPA profile.
7245 (dump_overall_stats): Dump IPA porfile.
7246 (dump_inline_stats): Dump IPA porfile.
7247 (ipa_inline): Collect IPA stats.
7248 * ipa-inline.h (clone_inlined_nodes): Update prototype.
7249 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
7250 (ipa_propagate_frequency): Use frequency function.
7251 (ipa_profile): Cleanup.
7252 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
7253 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
7254 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
7255 (input_node): Do not stream frequency.
7256 (input_edge): Do not stream frequency.
7257 (merge_profile_summaries): Scale only IPA profiles.
7258 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
7259 * predict.c (drop_profile): Do not recompute frequency.
7260 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
7261 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
7262 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
7263 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
7264 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
7265 * tree-inline.c (copy_bb): Do not scale frequency.
7266 (expand_call_inline): Do not scale frequency.
7267 (tree_function_versioning): Do not scale frequency.
7268 * ubsan.c (ubsan_create_edge): Do not pass frequency.
7270 2017-11-10 Julia Koval <julia.koval@intel.com>
7272 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
7273 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
7274 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
7275 (_mm256_maskz_gf2p8affine_epi64_epi8)
7276 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
7277 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
7278 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
7279 (__builtin_ia32_vgf2p8affineqb_v32qi)
7280 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
7281 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
7283 2017-11-10 Tamar Christina <tamar.christina@arm.com>
7287 (arm_option_override): Refactor.
7288 (arm_option_reconfigure_globals): New.
7289 (arm_options_perform_arch_sanity_checks): New.
7290 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
7292 (arm_options_perform_arch_sanity_checks): Likewise
7294 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
7296 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
7297 (power9-qpmul): New.
7298 * rs6000/rs6000.md ("type" attr): Add qmul.
7299 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
7300 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
7301 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
7303 2017-11-10 Martin Sebor <msebor@redhat.com>
7306 * builtins.c (compute_objsize): Handle arrays that
7307 compute_builtin_object_size likes to fail for. Make extern.
7308 * builtins.h (compute_objsize): Declare.
7309 (check_strncpy_sizes): New function.
7310 (expand_builtin_strncpy): Call check_strncpy_sizes.
7311 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
7312 -Wstringop-truncation.
7313 (gimple_fold_builtin_strncat): Same.
7314 * gimple.c (gimple_build_call_from_tree): Set call location.
7315 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
7316 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
7317 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
7318 (handle_builtin_strlen): Use strlen_to_stridx.
7319 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
7321 Use strlen_to_stridx.
7322 (pass_strlen::execute): Release strlen_to_stridx.
7323 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
7324 (-Wstringop-truncation): Document new option.
7326 2017-11-10 Martin Liska <mliska@suse.cz>
7328 PR gcov-profile/82702
7329 * gcov.c (main): Handle intermediate files in a different
7331 (get_gcov_intermediate_filename): New function.
7332 (output_gcov_file): Remove support of intermediate files.
7333 (generate_results): Allocate intermediate file.
7334 (release_structures): Clean-up properly fn_end.
7335 (output_intermediate_file): Start iterating with line 1.
7337 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7339 PR tree-optimization/82929
7340 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7341 ops_swapped_p non-static data member.
7342 (store_immediate_info::store_immediate_info): Clear it.
7343 (imm_store_chain_info::coalesce_immediate_stores): If swapping
7344 ops set ops_swapped_p.
7345 (count_multiple_uses): Handle ops_swapped_p.
7347 2017-11-10 Martin Liska <mliska@suse.cz>
7349 * coverage.c (coverage_init): Stream information about
7350 support of has_unexecuted_blocks.
7351 * doc/gcov.texi: Document that.
7352 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
7353 * gcov.c (read_graph_file): Likewise.
7354 (output_line_beginning): Fix a small issue with
7357 2017-11-10 Bin Cheng <bin.cheng@arm.com>
7359 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
7360 reference of trivial component.
7362 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7365 * gimple-ssa-store-merging.c
7366 (pass_store_merging::terminate_all_aliasing_chains): For
7367 gimple_store_p stmts also call refs_output_dependent_p.
7369 PR rtl-optimization/82913
7370 * compare-elim.c (try_merge_compare): Punt if def_insn is not
7373 2017-11-09 Jeff Law <law@redhat.com>
7375 * vr-values.h: New file with vr_values class.
7376 * tree-vrp.c: Include vr-values.h
7377 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
7378 data objects into the vr_values class.
7379 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
7380 (get_value_range): Make it a member function within vr_values class.
7381 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
7382 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
7383 (op_with_constant_singleton_value_range): Likewise.
7384 (extract_range_for_var_from_comparison_expr): Likewise.
7385 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
7386 (extract_range_from_binary_expr): Likewise.
7387 (extract_range_from_unary_expr): Likewise.
7388 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
7389 (check_for_binary_op_overflow, extract_range_basic): Likewise.
7390 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
7391 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
7392 (compare_name_with_value, compare_names): Likewise.
7393 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
7394 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
7395 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
7396 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
7397 (extract_range_from_phi_node): Likewise.
7398 (simplify_truth_ops_using_ranges): Likewise.
7399 (simplify_div_or_mod_using_ranges): Likewise.
7400 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
7401 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
7402 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
7403 (simplify_float_conversion_using_ranges): Likewise.
7404 (simplify_internal_call_using_ranges): Likewise.
7405 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
7406 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
7407 poitner via x_vr_values for calls into gimple folder.
7408 (vrp_initialize_lattice): Make this the vr_values ctor.
7409 (vrp_free_lattice): Make this the vr_values dtor.
7410 (set_vr_value): New function.
7411 (class vrp_prop): Add vr_values data member. Add various member
7412 functions as well as member functions that delegate to vr_values.
7413 (check_array_ref): Make a member function within vrp_prop class.
7414 (search_for_addr_array, vrp_initialize): Likewise.
7415 (vrp_finalize): Likewise. Revamp to avoid direct access to
7416 vr_value, values_propagated, etc.
7417 (check_array_bounds): Extract vrp_prop class instance pointer from
7418 walk info structure. Use it to call member functions.
7419 (check_all_array_refs): Make a member function within vrp_prop class.
7420 Smuggle class instance pointer via walk info structure.
7421 (x_vr_values): New local static.
7422 (vrp_valueize): Use x_vr_values to get class instance.
7423 (vr_valueize_1): Likewise.
7424 (class vrp_folder): Add vr_values data member. Add various member
7425 functions as well as member functions that delegate to vr_values.
7426 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
7427 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
7428 class instance from vr_values. Use it to call member functions.
7429 (vrp_dom_walker): Add vr_values data member.
7430 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
7431 instance via x_vr_values.
7432 (identify_jump_threads): Accept vr_values as argument. Store
7433 it into the walker structure.
7434 (evrp_dom_walker): Add vr_values class data member. Add various
7436 (evrp_dom_walker::try_find_new_range): Use vr_values data
7437 member to access the memory allocator.
7438 (evrp_dom_walker::before_dom_children): Store vr_values class
7439 instance into the vrp_folder class.
7440 (evrp_dom_walker::push_value_range): Rework to avoid direct
7441 access to num_vr_values and vr_value.
7442 (evrp_dom_walker::pop_value_range): Likewise.
7443 (execute_early_vrp): Remove call to vrp_initialize_lattice.
7444 Use vr_values to get to dump_all_value_ranges member function.
7445 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
7446 and simplify_cond_using_ranges_2 via vrp_prop class instance.
7447 Pass vr_values class instance down to identify_jump_threads.
7448 Remove call to vrp_free_lattice.
7449 (debug_all_value_ranges): Remove.
7451 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
7452 (vrp_folder): Likewise.
7454 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
7455 Get it from the existing bitmap instead.
7456 (vrp_intersect_ranges_1): Likewise.
7458 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7460 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7462 (store_immediate_info::store_immediate_info): Add bitnotp argument,
7463 set bit_not_p to it.
7464 (imm_store_chain_info::coalesce_immediate_stores): Break group
7465 if bit_not_p is different.
7466 (count_multiple_uses, split_group,
7467 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
7468 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
7469 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
7470 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
7471 to store_immediate_info ctor.
7473 2017-11-09 Jim Wilson <jimw@sifive.com>
7475 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
7476 (scan_prog_file): Likewise.
7478 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7480 * bb-reorder.c (max_entry_frequency): Remove.
7481 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
7482 connect_traces, push_to_next_round_p): Remove prototypes.
7483 (find_traces_1_round): Use counts only.
7484 (push_to_next_round_p): Likewise.
7485 (find_traces): Likewise.
7486 (rotate_loop): Likewise.
7487 (find_traces_1_round): Likewise.
7488 (connect_traces): Likewise.
7489 (edge_order): Likewise.
7491 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
7493 * config/arm/arm.c (output_return_instruction): Add comments to
7494 indicate requirement for cmse_nonsecure_entry return to account
7495 for the size of clearing instruction output here.
7496 (thumb_exit): Likewise.
7497 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
7498 return in hardfloat mode.
7500 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
7502 * config/rs6000/rs6000.c (machine_function): Add a bool,
7503 "toc_is_wrapped_separately".
7504 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
7505 if it wasn't explicitly set or unset, we are optimizing for speed, and
7506 doing separate shrink-wrapping.
7507 (rs6000_get_separate_components): Enable the TOC component if
7508 saving the TOC register in the prologue.
7509 (rs6000_components_for_bb): Handle the TOC component.
7510 (rs6000_emit_prologue_components): Store the TOC register where needed.
7511 (rs6000_set_handled_components): Mark TOC as handled, if handled.
7512 (rs6000_emit_prologue): Don't save the TOC if that is already done.
7514 2017-11-09 Martin Jambor <mjambor@suse.cz>
7516 * ipa-param-manipulation.c: New file.
7517 * ipa-param-manipulation.h: Likewise.
7518 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
7519 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
7520 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
7521 (ipa_parm_adjustment): Likewise.
7522 (ipa_parm_adjustment_vec): Likewise.
7523 (ipa_get_vector_of_formal_parms): Moved declaration to
7524 ipa-param-manipulation.h.
7525 (ipa_get_vector_of_formal_parm_types): Likewise.
7526 (ipa_modify_formal_parameters): Likewise.
7527 (ipa_modify_call_arguments): Likewise.
7528 (ipa_combine_adjustments): Likewise.
7529 (ipa_dump_param_adjustments): Likewise.
7530 (ipa_modify_expr): Likewise.
7531 (ipa_get_adjustment_candidate): Likewise.
7532 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
7533 ipa-param-manipulation.c.
7534 (ipa_get_vector_of_formal_parm_types): Likewise.
7535 (ipa_modify_formal_parameters): Likewise.
7536 (ipa_modify_call_arguments): Likewise.
7537 (ipa_modify_expr): Likewise.
7538 (get_ssa_base_param): Likewise.
7539 (ipa_get_adjustment_candidate): Likewise.
7540 (index_in_adjustments_multiple_times_p): Likewise.
7541 (ipa_combine_adjustments): Likewise.
7542 (ipa_dump_param_adjustments): Likewise.
7543 * tree-sra.c: Also include ipa-param-manipulation.h
7544 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
7547 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7548 Alan Hayward <alan.hayward@arm.com>
7549 David Sherwood <david.sherwood@arm.com>
7551 * doc/sourcebuild.texi (vect_masked_store): Document.
7553 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7554 Alan Hayward <alan.hayward@arm.com>
7555 David Sherwood <david.sherwood@arm.com>
7557 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
7559 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7560 Alan Hayward <alan.hayward@arm.com>
7561 David Sherwood <david.sherwood@arm.com>
7563 * doc/sourcebuild.texi (vect_variable_length): Document.
7565 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7566 Alan Hayward <alan.hayward@arm.com>
7567 David Sherwood <david.sherwood@arm.com>
7569 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
7571 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7572 Alan Hayward <alan.hayward@arm.com>
7573 David Sherwood <david.sherwood@arm.com>
7575 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
7577 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7578 Alan Hayward <alan.hayward@arm.com>
7579 David Sherwood <david.sherwood@arm.com>
7581 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
7582 previously undocumented selectors.
7583 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
7585 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7587 * doc/rtl.texi (const_vector): Say that elements can be
7588 const_wide_ints too.
7589 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
7590 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
7591 (gen_vec_duplicate): Use it instead of CONSTANT_P.
7592 * optabs.c (expand_vector_broadcast): Likewise.
7594 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7595 Alan Hayward <alan.hayward@arm.com>
7596 David Sherwood <david.sherwood@arm.com>
7598 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
7599 scaled index even if the unscaled address was invalid.
7600 Don't increase the complexity of using a scale in that case.
7602 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7603 Alan Hayward <alan.hayward@arm.com>
7604 David Sherwood <david.sherwood@arm.com>
7606 * doc/rtl.texi: Rewrite the subreg rules so that they partition
7607 the inner register into REGMODE_NATURAL_SIZE bytes rather than
7608 UNITS_PER_WORD bytes.
7609 * emit-rtl.c (validate_subreg): Divide subregs into blocks
7610 based on REGMODE_NATURAL_SIZE of the inner mode.
7611 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
7612 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
7613 * expmed.c (lowpart_bit_field_p): Divide the value up into
7614 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
7615 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
7616 whether something is likely to occupy more than one register.
7618 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7621 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7622 profile_count::adjust_for_ipa_scaling.
7623 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7624 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7626 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7628 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7630 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7631 (split_group): Add total_orig and total_new arguments, estimate the
7632 number of statements related to the store group without store merging
7633 and with store merging.
7634 (imm_store_chain_info::output_merged_store): Adjust split_group
7635 callers, punt if estimated number of statements with store merging
7636 is not smaller than estimated number of statements without it.
7638 (handled_load): Remove has_single_use checks.
7639 (pass_store_merging::process_store): Likewise.
7641 2017-11-09 Richard Biener <rguenther@suse.de>
7643 PR tree-optimization/82902
7644 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7646 2017-11-09 Martin Liska <mliska@suse.cz>
7649 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7652 2017-11-09 Martin Liska <mliska@suse.cz>
7654 PR tree-optimization/82669
7655 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7657 2017-11-09 Martin Liska <mliska@suse.cz>
7659 PR gcov-profile/48463
7660 * coverage.c (coverage_begin_function): Output also end locus
7661 of a function and information whether the function is
7663 * gcov-dump.c (tag_function): Parse and print the information.
7664 * gcov.c (INCLUDE_MAP): Add include.
7665 (INCLUDE_SET): Likewise.
7666 (struct line_info): Move earlier in the source file because
7667 of vector<line_info> in function_info structure.
7668 (line_info::line_info): Likewise.
7669 (line_info::has_block): Likewise.
7670 (struct source_info): Add new member index.
7671 (source_info::get_functions_at_location): New function.
7672 (function_info::group_line_p): New function.
7673 (output_intermediate_line): New function.
7674 (output_intermediate_file): Use the mentioned function.
7675 (struct function_start): New.
7676 (struct function_start_pair_hash): Likewise.
7677 (process_file): Add code that identifies group functions.
7678 Assign lines either to global or function scope.
7679 (generate_results): Skip artificial functions.
7680 (find_source): Assign index for each source file.
7681 (read_graph_file): Read new flag artificial and end_line.
7682 (add_line_counts): Assign it either to global of function scope.
7683 (accumulate_line_counts): Isolate core of the function to
7684 accumulate_line_info and call it for both function and global
7686 (accumulate_line_info): New function.
7687 (output_line_beginning): Fix GNU coding style.
7688 (print_source_line): New function.
7689 (output_line_details): Likewise.
7690 (output_function_details): Likewise.
7691 (output_lines): Iterate both source (global) scope and function
7693 (struct function_line_start_cmp): New class.
7694 * doc/gcov.texi: Reflect changes in documentation.
7696 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7699 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7700 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7701 and similarly for not instead of neg.
7703 2017-11-08 Andi Kleen <ak@linux.intel.com>
7705 * config/i386/i386.opt: Add -mforce-indirect-call.
7706 * config/i386/predicates.md: Check for flag_force_indirect_call.
7707 * doc/invoke.texi: Document -mforce-indirect-call
7709 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7711 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7713 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7714 config/riscv/riscv.c (predict.h): New include.
7715 (riscv_slow_unaligned_access_p): No longer static.
7716 (riscv_block_move_straight): Add require.
7717 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7719 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7722 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7723 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7724 nonimmediate_operand predicate for operand 1 instead of
7727 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7729 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7731 * config/aarch64/constraints.md (Uml): New constraint.
7732 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7735 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7737 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7738 of two vec_duplicates into a vec_concat.
7740 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7742 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7743 Simplify vec_merge of vec_duplicate and vec_concat.
7744 * config/aarch64/constraints.md (Utq): New constraint.
7745 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7748 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7750 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7751 Simplify vec_merge of vec_duplicate and const_vector.
7752 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7754 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7755 mode iterator. Update predicate on operand 1 to
7756 handle non-const_vec constants. Delete constraints.
7757 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7759 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7761 PR tree-optimization/78821
7762 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7764 (store_operand_info::store_operand_info): Initialize it to false.
7765 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7766 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7767 store in the group, and if chain_info is non-NULL, to ignore altogether
7769 (compatible_load_p): Fail if bit_not_p does not match.
7770 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7771 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7772 (pass_store_merging::process_store): Adjust
7773 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7774 call terminate_all_aliasing_chains newly when adding a store into
7775 a chain with non-NULL chain_info.
7777 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7779 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7781 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7783 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7785 (aarch64_layout_frame): Initialise emit_frame_chain.
7786 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7787 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7789 2017-11-08 Martin Liska <mliska@suse.cz>
7791 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7792 of gimple_call_internal_p.
7794 2017-11-07 Tom de Vries <tom@codesourcery.com>
7796 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7798 2017-11-07 Tom de Vries <tom@codesourcery.com>
7800 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7803 2017-11-08 Martin Liska <mliska@suse.cz>
7806 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7808 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7810 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7812 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7815 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7816 (*cmp<mode>_ccz_1): New insn with $k alternative.
7819 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7820 RTX_COMM_COMPARE as commutative as well.
7821 (ix86_binary_operator_ok): Formatting fix.
7822 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7823 *<code><mode>3<mask_name><round_saeonly_name>,
7824 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7825 *<s>mul<mode>3_highpart<mask_name>,
7826 *vec_widen_umult_even_v16si<mask_name>,
7827 *vec_widen_umult_even_v8si<mask_name>,
7828 *vec_widen_umult_even_v4si<mask_name>,
7829 *vec_widen_smult_even_v16si<mask_name>,
7830 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7831 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7832 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7833 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7834 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7835 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7836 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7837 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7838 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7839 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7840 ix86_binary_operator_ok. Formatting fixes.
7841 (*<plusminus_insn><mode>3<mask_name><round_name>,
7842 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7845 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7847 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7848 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7849 eq and ne if TARGET_ISEL.
7850 (cmp): New code_iterator.
7851 (UNS, UNSU_, UNSIK): New code_attrs.
7852 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7853 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7855 ("eq<mode>3"): ... this.
7856 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7858 ("ne<mode>3"): ... this.
7860 2017-11-07 Julia Koval <julia.koval@intel.com>
7863 * common/config/i386/i386-common.c
7864 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
7865 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
7866 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
7867 * config/i386/i386.opt: Ditto.
7868 * config/i386/i386.c (ix86_target_string): Ditto.
7869 (ix86_option_override_internal): Ditto.
7870 (ix86_init_mpx_builtins): Move MPX to args2.
7871 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
7872 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7873 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7874 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7875 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7876 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7877 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
7879 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
7882 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
7883 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
7884 (zero-extendsidi peephole2): Remove peephole.
7886 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7889 * stor-layout.c: Include attribs.h.
7890 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
7891 explicit lookup of "aligned" attribute.
7893 2017-11-07 Andrew Waterman <andrew@sifive.com>
7895 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
7896 (riscv_expand_block_move): Likewise.
7897 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
7899 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
7900 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
7901 * config/riscv/riscv.c (riscv_block_move_straight): New function.
7902 (riscv_adjust_block_mem): Likewise.
7903 (riscv_block_move_loop): Likewise.
7904 (riscv_expand_block_move): Likewise.
7905 * config/riscv/riscv.md (movmemsi): New pattern.
7907 2017-11-07 Michael Clark <michaeljclark@mac.com>
7909 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
7910 (MUSL_DYNAMIC_LINKER): Likewise.
7912 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
7914 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
7915 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
7918 2017-11-07 Richard Biener <rguenther@suse.de>
7920 * match.pd: Fix build.
7922 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
7923 Jackson Woodruff <jackson.woodruff@arm.com>
7925 PR tree-optimization/71026
7926 * match.pd: Canonicalize negate in division.
7928 2017-11-07 Sudakshina Das <sudi.das@arm.com>
7931 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
7933 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7935 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
7936 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
7938 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7940 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
7941 non-scalar integral types.
7942 * match.pd (negate_expr_p): Handle MINUS_EXPR.
7943 (-(A-B), -(~A)): New transformations.
7945 2017-11-07 Tom de Vries <tom@codesourcery.com>
7947 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
7948 semicolon after "do {} while (0)".
7949 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7950 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7951 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7952 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7953 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7955 2017-11-07 Tom de Vries <tom@codesourcery.com>
7957 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
7958 after "do {} while (0)".
7959 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7960 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7961 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7962 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7963 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7965 2017-11-07 Tom de Vries <tom@codesourcery.com>
7968 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
7970 (arm_rtx_costs_internal): Add missing semicolon after
7971 HANDLE_NARROW_SHIFT_ARITH call.
7973 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7975 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7976 disable isel if it was not set explicitly.
7978 2017-11-06 James Bowman <james.bowman@ftdichip.com>
7980 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
7981 (add_type_attribute) likewise.
7983 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
7985 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
7986 of struct ix86_frame.
7987 (ix86_initial_elimination_offset): Likewise.
7988 (ix86_expand_split_stack_prologue): Likewise.
7990 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
7992 * tree-vrp.h (enum value_range_type): Update stale comment.
7994 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7995 Alan Hayward <alan.hayward@arm.com>
7996 David Sherwood <david.sherwood@arm.com>
7998 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
7999 (aarch64_expand_vec_perm_const): Take the number of units too.
8000 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8001 (aarch64_expand_vec_perm_const): Likewise.
8002 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8003 (vec_perm<mode>): Update accordingly.
8005 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8006 Alan Hayward <alan.hayward@arm.com>
8007 David Sherwood <david.sherwood@arm.com>
8009 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8010 Take the number of units too.
8011 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8012 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8013 but check for a vector mode before rather than after the call.
8014 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8015 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8016 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8017 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8018 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8019 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8020 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8021 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8022 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8023 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8024 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8025 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8026 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8027 (aarch64_sqdmull2_n<mode>): Update accordingly.
8029 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8030 Alan Hayward <alan.hayward@arm.com>
8031 David Sherwood <david.sherwood@arm.com>
8033 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8034 the number of units too.
8035 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8036 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8037 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
8038 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
8039 (vec_store_lanesxi<mode>): Update accordingly.
8041 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8042 Alan Hayward <alan.hayward@arm.com>
8043 David Sherwood <david.sherwood@arm.com>
8045 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8046 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8047 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8048 of units rather than the mode.
8049 * config/aarch64/iterators.md (nunits): New mode attribute.
8050 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8051 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8052 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8053 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8054 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8055 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8056 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8057 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8058 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8059 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8060 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8061 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8062 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8063 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8064 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8065 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8066 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8067 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8068 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8069 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8070 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8071 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8072 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8073 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8074 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8075 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8076 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8077 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8078 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8079 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8080 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8081 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8082 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8083 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8084 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8085 (aarch64_simd_vec_setv2di): Likewise.
8087 2017-11-06 Carl Love <cel@us.ibm.com>
8089 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8091 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8092 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8094 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8096 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8097 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8098 Add power 8 macro expansions.
8099 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8100 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8101 power 8 instructions. (VSX_XXBR): Add iterator.
8103 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8105 * config/arm/arm.md (predicable_short_it): Change default to "no",
8106 improve documentation, remove uses that are identical to the default.
8107 (enabled_for_depr_it): Rename to enabled_for_short_it.
8108 * gcc/config/arm/arm-fixed.md (predicable_short_it):
8109 Remove default uses.
8110 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
8111 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
8112 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
8113 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
8115 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
8118 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
8119 float128 helper macros, which are no longer used after deleting
8120 the old 'q' built-in functions, and moving the round to odd
8121 built-in functions to being special built-in functions.
8122 (BU_FLOAT128_2): Likewise.
8123 (BU_FLOAT128_1_HW): Likewise.
8124 (BU_FLOAT128_2_HW): Likewise.
8125 (BU_FLOAT128_3_HW): Likewise.
8126 (FABSQ): Delete old 'q' built-in functions.
8127 (COPYSIGNQ): Likewise.
8128 (SQRTF128_ODD): Move round to odd built-in functions to be
8129 special built-in functions, so that we can handle
8130 -mabi=ieeelongdouble.
8131 (TRUNCF128_ODD): Likewise.
8132 (ADDF128_ODD): Likewise.
8133 (SUBF128_ODD): Likewise.
8134 (MULF128_ODD): Likewise.
8135 (DIVF128_ODD): Likewise.
8136 (FMAF128_ODD): Likewise.
8137 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
8138 built-in names to 'f128'.
8139 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
8140 old 'q' built-in functions, as the machine independent code for
8141 'f128' built-in functions handles this.
8142 (rs6000_expand_builtin): Add expansion for float128 round to odd
8143 functions, keying off on -mabi=ieeelongdouble of whether to use
8144 the KFmode or TFmode variant.
8145 (rs6000_init_builtins): Initialize the _Float128 round to odd
8147 * doc/extend.texi (PowerPC Built-in Functions): Document the old
8148 _Float128 'q' built-in functions are now mapped into the new
8149 'f128' built-in functions.
8151 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
8153 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
8154 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
8156 2017-11-06 David Malcolm <dmalcolm@redhat.com>
8159 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
8160 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
8161 * toplev.c: Include "ipa-fnsummary.h".
8162 (toplev::finalize): Call ipa_fnsummary_c_finalize.
8164 2017-11-06 Jakub Jelinek <jakub@redhat.com>
8166 PR tree-optimization/82838
8167 * gimple-ssa-store-merging.c
8168 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
8169 on a separate gimple_seq which is then appended to seq.
8171 2017-11-06 Jeff Law <law@redhat.com>
8174 * config/i386/i386.c (PROBE_INTERVAL): Remove.
8175 (get_probe_interval): New functions.
8176 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
8177 (ix86_adjust_stack_and_probe): Likewise.
8178 (output_adjust_stack_and_probe): Likewise.
8179 (ix86_emit_probe_stack_range): Likewise.
8180 (ix86_expand_prologue): Likewise.
8182 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8184 PR tree-optimization/82816
8185 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
8186 if the modes of the two types are the same.
8187 (convert_plusminus_to_widen): Likewise.
8189 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8191 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
8193 (usadv16qi): New define_expand.
8194 (usadv8hi): New define_expand.
8196 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
8199 * ipa-inline-transform.c (update_noncloned_frequencies): Always
8201 (inline_transform): Likewise.
8202 * predict.c (counts_to_freqs): Remove useless conditional.
8203 * profile-count.h (profile_count::apply_scale): Move sanity check.
8204 * tree-inline.c (copy_bb): Always scale.
8205 (copy_cfg_body): Likewise.
8207 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
8210 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
8212 (*cmp_ite0): Add enabled_for_depr_it attribute.
8213 (*cmp_ite1): Likewise.
8215 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8217 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
8220 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8222 * tree-vrp.c (vrp_int_const_binop): Return true on success and
8223 return the value by pointer.
8224 (extract_range_from_multiplicative_op_1): Update accordingly.
8225 Return as soon as an operation fails.
8227 2017-11-05 Tom de Vries <tom@codesourcery.com>
8230 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
8231 (DEF_SANITIZER_BUILTIN): ... here.
8232 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
8233 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
8235 2017-11-05 Tom de Vries <tom@codesourcery.com>
8238 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
8240 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
8241 ASM_OUTPUT_BEFORE_CASE_LABEL call.
8242 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
8244 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8245 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8246 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8248 2017-11-05 Tom de Vries <tom@codesourcery.com>
8251 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
8254 2017-11-04 Michael Clark <michaeljclark@mac.com>
8256 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
8257 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
8259 (*addsi3_extended): Likewise.
8260 (*addsi3_extended2): Likewise.
8261 (<optab>si3): Likewise.
8262 (<optab>di3): Likewise.
8263 (<optab><mode>3): Likewise.
8264 (<*optabe>si3_internal): Likewise.
8265 (zero_extendqi<SUPERQI:mode>2): Likewise.
8266 (*add<mode>hi3): Likewise.
8267 (*xor<mode>hi3): Likewise.
8268 (<optab>di3): Likewise.
8269 (*<optab>si3_extend): Likewise.
8270 (*sge<u>_<X:mode><GPR:mode>): Likewise.
8271 (*slt<u>_<X:mode><GPR:mode>): Likewise.
8272 (*sle<u>_<X:mode><GPR:mode>): Likewise.
8274 2017-11-04 Andrew Waterman <andrew@sifive.com>
8276 * config/riscv/riscv.c (riscv_option_override): Conditionally set
8277 TARGET_STRICT_ALIGN based upon -mtune argument.
8279 2017-11-04 Andrew Waterman <andrew@sifive.com>
8281 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
8283 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
8285 * config/i386/i386.c (choose_basereg): Use optional scratch
8286 register and add assertion.
8287 (x86_emit_outlined_ms2sysv_save): Use scratch register when
8288 needed, and don't allocate stack.
8289 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
8290 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
8291 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
8293 2017-11-03 Jeff Law <law@redhat.com>
8295 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
8296 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
8297 to probe at the start of a noreturn function.
8299 2017-11-03 Jakub Jelinek <jakub@redhat.com>
8301 PR tree-optimization/78821
8302 * gimple-ssa-store-merging.c: Update the file comment.
8303 (MAX_STORE_ALIAS_CHECKS): Define.
8304 (struct store_operand_info): New type.
8305 (store_operand_info::store_operand_info): New constructor.
8306 (struct store_immediate_info): Add rhs_code and ops data members.
8307 (store_immediate_info::store_immediate_info): Add rhscode, op0r
8308 and op1r arguments to the ctor, initialize corresponding data members.
8309 (struct merged_store_group): Add load_align_base and load_align
8311 (merged_store_group::merged_store_group): Initialize them.
8312 (merged_store_group::do_merge): Update them.
8313 (merged_store_group::apply_stores): Pick the constant for
8314 encode_tree_to_bitpos from one of the two operands, or skip
8315 encode_tree_to_bitpos if neither operand is a constant.
8316 (class pass_store_merging): Add process_store method decl. Remove
8317 bool argument from terminate_all_aliasing_chains method decl.
8318 (pass_store_merging::terminate_all_aliasing_chains): Remove
8319 var_offset_p argument and corresponding handling.
8320 (stmts_may_clobber_ref_p): New function.
8321 (compatible_load_p): New function.
8322 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
8323 if there is overlap and rhs_code is not INTEGER_CST. For
8324 non-overlapping stores terminate group if rhs is not mergeable.
8325 (get_alias_type_for_stmts): Change first argument from
8326 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
8327 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
8328 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
8330 (get_location_for_stmts): Change first argument from
8331 auto_vec<gimple *> & to vec<gimple *> &.
8332 (struct split_store): Remove orig_stmts data member, add orig_stores.
8333 (split_store::split_store): Create orig_stores rather than orig_stmts.
8334 (find_constituent_stmts): Renamed to ...
8335 (find_constituent_stores): ... this. Change second argument from
8336 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
8337 to info structures rather than the statements.
8338 (split_group): Rename ALLOW_UNALIGNED argument to
8339 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
8340 it. Adjust find_constituent_stores caller.
8341 (imm_store_chain_info::output_merged_store): Handle rhs_code other
8342 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
8343 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
8344 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
8345 (mem_valid_for_store_merging): New function.
8346 (handled_load): New function.
8347 (pass_store_merging::process_store): New method.
8348 (pass_store_merging::execute): Use process_store method. Adjust
8349 terminate_all_aliasing_chains caller.
8351 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8353 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8354 Return true for more constants, symbols and label references.
8355 (aarch64_valid_floating_const): Remove unused function.
8357 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8360 * config/aarch64/aarch64.c (aarch64_layout_frame):
8361 Undo forcing of LR at bottom of frame.
8363 2017-11-03 Jeff Law <law@redhat.com>
8366 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
8367 for int_registers_saved.
8369 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
8370 extracted from tree-ssa-dom.c.
8371 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
8372 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
8373 (record_equivalences_from_incoming_edge): Add additional argument
8374 to single_pred_edge_ignoring_loop_edges call.
8375 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
8376 (uncprop_dom_walker::before_dom_children): Add additional argument
8377 to single_pred_edge_ignoring_loop_edges call.
8378 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
8379 single_pred_edge_ignoring_loop_edges rather than open coding.
8380 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
8382 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
8384 * match.pd (-(-A)): Rewrite.
8386 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8388 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
8389 (rs6000_emit_int_cmove): New declaration.
8390 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
8391 (rs6000_emit_sISEL): Delete.
8392 (rs6000_emit_int_cmove): Make non-static.
8393 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
8394 instead of rs6000_emit_sISEL.
8396 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
8398 * asan.c (create_cond_insert_point): Maintain profile.
8399 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
8401 * basic-block.h (struct basic_block_def): Remove frequency.
8402 (EDGE_FREQUENCY): Use to_frequency
8403 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
8405 (find_traces): Update to use to_frequency.
8406 (find_traces_1_round): Likewise; use only IPA counts.
8407 (bb_to_key): Likewise.
8408 (connect_traces): Use IPA counts only.
8409 (copy_bb_p): Update to use to_frequency.
8410 (fix_up_crossing_landing_pad): Likewise.
8411 (sanitize_hot_paths): Likewise.
8412 * bt-load.c (basic_block_freq): Likewise.
8413 * cfg.c (init_flow): Set count_max to uninitialized.
8414 (check_bb_profile): Remove frequencies; check counts.
8415 (dump_bb_info): Do not dump frequencies.
8416 (update_bb_profile_for_threading): Update counts only.
8417 (scale_bbs_frequencies_int): Likewise.
8418 (MAX_SAFE_MULTIPLIER): Remove.
8419 (scale_bbs_frequencies_gcov_type): Update counts only.
8420 (scale_bbs_frequencies_profile_count): Update counts only.
8421 (scale_bbs_frequencies): Update counts only.
8422 * cfg.h (struct control_flow_graph): Add count-max.
8423 (update_bb_profile_for_threading): Update prototype.
8424 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
8425 (find_many_sub_basic_blocks): Likewise.
8426 * cfgcleanup.c (try_forward_edges): Likewise.
8427 (try_crossjump_to_edge): Likewise.
8428 * cfgexpand.c (expand_gimple_cond): Likewise.
8429 (expand_gimple_tailcall): Likewise.
8430 (construct_init_block): Likewise.
8431 (construct_exit_block): Likewise.
8432 * cfghooks.c (verify_flow_info): Check consistency of counts.
8433 (dump_bb_for_graph): Do not dump frequencies.
8434 (split_block_1): Do not update frequencies.
8435 (split_edge): Do not update frequencies.
8436 (make_forwarder_block): Do not update frequencies.
8437 (duplicate_block): Do not update frequencies.
8438 (account_profile_record): Do not update frequencies.
8439 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
8440 for global heuristics.
8441 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
8442 (expected_loop_iterations_unbounded): Use counts only.
8443 * cfgloopmanip.c (scale_loop_profile): Simplify.
8444 (create_empty_loop_on_edge): Simplify
8446 (duplicate_loop_to_header_edge): Simplify
8447 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
8448 (update_br_prob_note): Take care of removing note when profile
8450 (relink_block_chain): Do not dump frequency.
8451 (rtl_account_profile_record): Use to_frequency.
8452 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
8453 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
8454 (cgraph_update_edges_for_call_stmt_node): Likewise.
8455 (cgraph_edge::verify_count_and_frequency): Update.
8456 (cgraph_node::verify_node): Temporarily disable frequency verification.
8457 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
8458 to_cgraph_frequency.
8459 (cgraph_edge::rebuild_edges): Convert to ipa counts.
8460 * cgraphunit.c (init_lowered_empty_function): Do not initialize
8462 (cgraph_node::expand_thunk): Update profile.
8463 * except.c (dw2_build_landing_pads): Do not update frequency.
8464 * final.c (compute_alignments): Use to_frequency.
8465 (dump_basic_block_info): Do not dump frequency.
8466 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
8467 (dump_gimple_bb_header): Do not dump frequency.
8468 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
8470 * gimple-streamer-in.c (input_bb): Do not stream frequency.
8471 * gimple-streamer-out.c (output_bb): Do not stream frequency.
8472 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
8473 (init_before_recovery): Do not update frequency.
8474 (sched_create_recovery_edges): Do not update frequency.
8475 * hsa-gen.c (convert_switch_statements): Do not update frequency.
8476 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
8477 (ipa_cp_c_finalize): Set max_count to uninitialized.
8478 * ipa-fnsummary.c (get_minimal_bb): Use counts.
8479 (param_change_prob): Use counts.
8480 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
8482 * ipa-split.c (consider_split): Use to_frequency.
8483 (split_function): Use to_frequency.
8484 * ira-build.c (loop_compare_func): Likewise.
8485 (mark_loops_for_removal): Likewise.
8486 (mark_all_loops_for_removal): Likewise.
8487 * loop-doloop.c (doloop_modify): Do not update frequency.
8488 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
8490 * lto-streamer-in.c (input_function): Update count_max.
8491 * omp-expand.c (expand_omp_taskreg): Update count_max.
8492 * omp-simd-clone.c (simd_clone_adjust): Update profile.
8493 * predict.c (maybe_hot_frequency_p): Use to_frequency.
8494 (maybe_hot_count_p): Use ipa counts only.
8495 (maybe_hot_bb_p): Simplify.
8496 (maybe_hot_edge_p): Simplify.
8497 (probably_never_executed): Do not take frequency argument.
8498 (probably_never_executed_bb_p): Do not pass frequency.
8499 (probably_never_executed_edge_p): Likewise.
8500 (combine_predictions_for_bb): Check that profile is nonzero.
8501 (propagate_freq): Do not set frequency.
8502 (drop_profile): Simplify.
8503 (counts_to_freqs): Simplify.
8504 (expensive_function_p): Use to_frequency.
8505 (propagate_unlikely_bbs_forward): Simplify.
8506 (determine_unlikely_bbs): Simplify.
8507 (estimate_bb_frequencies): Add hack to silence graphite issues.
8508 (compute_function_frequency): Use ipa counts.
8509 (pass_profile::execute): Update.
8510 (rebuild_frequencies): Use counts only.
8511 (force_edge_cold): Use counts only.
8512 * profile-count.c (profile_count::dump): Dump new count types.
8513 (profile_count::differs_from_p): Check compatiblity.
8514 (profile_count::to_frequency): New function.
8515 (profile_count::to_cgraph_frequency): New function.
8516 * profile-count.h (struct function): Declare.
8517 (enum profile_quality): Add profile_guessed_local and
8518 profile_guessed_global0.
8519 (class profile_proability): Decrease number of bits to 29;
8520 update from_reg_br_prob_note and to_reg_br_prob_note.
8521 (class profile_count: Update comment; decrease number of bits
8522 to 61. Check compatibility.
8523 (profile_count::compatible_p): New private member function.
8524 (profile_count::ipa_p): New member function.
8525 (profile_count::operator<): Handle global zero correctly.
8526 (profile_count::operator>): Handle global zero correctly.
8527 (profile_count::operator<=): Handle global zero correctly.
8528 (profile_count::operator>=): Handle global zero correctly.
8529 (profile_count::nonzero_p): New member function.
8530 (profile_count::force_nonzero): New member function.
8531 (profile_count::max): New member function.
8532 (profile_count::apply_scale): Handle IPA scalling.
8533 (profile_count::guessed_local): New member function.
8534 (profile_count::global0): New member function.
8535 (profile_count::ipa): New member function.
8536 (profile_count::to_frequency): Declare.
8537 (profile_count::to_cgraph_frequency): Declare.
8538 * profile.c (OVERLAP_BASE): Delete.
8539 (compute_frequency_overlap): Delete.
8540 (compute_branch_probabilities): Do not use compute_frequency_overlap.
8541 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
8542 * sched-ebb.c (rank): Use counts only.
8543 * shrink-wrap.c (handle_simple_exit): Use counts only.
8544 (try_shrink_wrapping): Use counts only.
8545 (place_prologue_for_one_component): Use counts only.
8546 * tracer.c (find_best_predecessor): Use to_frequency.
8547 (find_trace): Use to_frequency.
8548 (tail_duplicate): Use to_frequency.
8549 * trans-mem.c (expand_transaction): Do not update frequency.
8550 * tree-call-cdce.c: Do not update frequency.
8551 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
8552 (gimple_merge_blocks): Likewise.
8553 (gimple_split_edge): Likewise.
8554 (gimple_duplicate_sese_region): Likewise.
8555 (gimple_duplicate_sese_tail): Likewise.
8556 (move_sese_region_to_fn): Likewise.
8557 (gimple_account_profile_record): Likewise.
8558 (insert_cond_bb): Likewise.
8559 * tree-complex.c (expand_complex_div_wide): Likewise.
8560 * tree-eh.c (lower_resx): Update profile.
8561 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
8563 (initialize_cfun): Do not initialize frequencies
8564 (freqs_to_counts): Delete.
8565 (copy_cfg_body): Ignore count parameter.
8566 (copy_body): Update.
8567 (expand_call_inline): Update count_max.
8568 (optimize_inline_calls): Update count_max.
8569 (tree_function_versioning): Update count_max.
8570 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
8571 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
8573 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
8574 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
8575 (try_peel_loop): Likewise.
8576 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
8578 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
8579 (tree_transform_and_unroll_loop): Do not use frequencies
8580 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
8581 Use reliable prediction only.
8582 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
8583 * tree-ssa-sink.c (select_best_block): Use to_frequency.
8584 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
8585 probability scaling.
8586 * tree-ssa-threadupdate.c (create_block_for_threading): Do
8587 not update frequency
8588 (any_remaining_duplicated_blocks): Likewise.
8589 (update_profile): Likewise.
8590 (estimated_freqs_path): Delete.
8591 (freqs_to_counts_path): Delete.
8592 (clear_counts_path): Delete.
8593 (ssa_fix_duplicate_block_edges): Likewise.
8594 (duplicate_thread_path): Likewise.
8595 * tree-switch-conversion.c (gen_inbound_check): Use counts.
8596 * tree-tailcall.c (decrease_profile): Do not update frequency.
8597 (eliminate_tail_call): Likewise.
8598 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
8599 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
8600 (optimize_mask_stores): Likewise.
8601 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8602 * ubsan.c (ubsan_expand_null_ifn): Update profile.
8603 (ubsan_expand_ptr_ifn): Update profile.
8604 * value-prof.c (gimple_ic): Simplify.
8605 * value-prof.h (gimple_ic): Update prototype.
8606 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
8607 * ipa-inline.c (compute_uninlined_call_time): Be sure that
8609 (want_inline_self_recursive_call_p): Likewise.
8610 (resolve_noninline_speculation): Only cummulate defined counts.
8611 (inline_small_functions): Use nonzero_p.
8612 (ipa_inline): Do not access freed node.
8614 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8616 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8617 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8619 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8621 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8622 non-legitimate address.
8624 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8626 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8627 (*lt0_<mode>di, *lt0_<mode>si): New.
8629 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8631 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8632 TARGET_PAIRED_FLOAT.
8634 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8635 Jim Wilson <jim.wilson@linaro.org>
8637 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8638 * config/aarch64/aarch64-tune.md: Regenerated.
8639 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8640 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8642 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8644 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8645 (arc_expand_prologue): Restore blink for millicode.
8646 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8648 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8651 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8652 gen_vec_duplicate after forcing the scalar into a register.
8654 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8656 * combine (try_combine): Print the insns input to try_combine to the
8659 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8662 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8663 Remove second argument from aarch64_process_target_attr call.
8664 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8666 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8668 (aarch64_handle_attr_arch): Remove second argument.
8669 (aarch64_handle_attr_cpu): Ditto.
8670 (aarch64_handle_attr_tune): Ditto.
8671 (aarch64_handle_attr_isa_flags): Ditto.
8672 (aarch64_process_one_target_attr): Ditto.
8673 (aarch64_process_target_attr): Ditto.
8674 (aarch64_option_valid_attribute_p): Remove second argument.
8675 on aarch64_process_target_attr call.
8677 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8679 * diagnostic.c: Include "selftest-diagnostic.h".
8680 (selftest::assert_location_text): New function.
8681 (selftest::test_diagnostic_get_location_text): New function.
8682 (selftest::diagnostic_c_tests): Call it.
8684 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8686 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8687 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8688 (class selftest::test_diagnostic_context): Move to...
8689 * selftest-diagnostic.c: New file.
8690 * selftest-diagnostic.h: New file.
8692 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8694 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8695 offset range for FT32B.
8696 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8697 * config/ft32/ft32.md: Add TARGET_NOPM.
8698 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8699 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8701 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8703 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8705 2017-11-02 Jeff Law <law@redhat.com>
8707 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8708 virtual keyword on FINAL OVERRIDE members.
8710 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8711 virtuals together. Add virtual destructor.
8712 (substitute_and_fold_engine): Similarly.
8714 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8716 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8718 2017-11-02 Richard Biener <rguenther@suse.de>
8720 PR tree-optimization/82795
8721 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8723 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8725 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8727 (gcc_GAS_CHECK_FEATURE): Remove.
8728 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8729 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8730 * configure: Regenerate.
8732 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8734 * config/arc/arc.c (hwloop_optimize): Account for empty
8737 2017-11-02 Richard Biener <rguenther@suse.de>
8740 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8741 Truncate ARRAY_REF index and element size.
8743 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8745 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8747 2017-11-01 Jeff Law <law@redhat.com>
8749 * tree-ssa-ccp.c (ccp_folder): New class derived from
8750 substitute_and_fold_engine.
8751 (ccp_folder::get_value): New member function.
8752 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8753 (ccp_fold_stmt): Remove prototype.
8754 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8755 * tree-ssa-copy.c (copy_folder): New class derived from
8756 substitute_and_fold_engine.
8757 (copy_folder::get_value): Renamed from get_value.
8758 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8759 * tree-vrp.c (vrp_folder): New class derived from
8760 substitute_and_fold_engine.
8761 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8762 (vrp_folder::get_value): New member function.
8763 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8764 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8765 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8766 provide a class interface to folder/substitute routines.
8767 (ssa_prop_fold_stmt_fn): Remove typedef.
8768 (ssa_prop_get_value_fn): Likewise.
8769 (subsitute_and_fold): Remove prototype.
8770 (replace_uses_in): Likewise.
8771 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8772 Renamed from replace_uses_in. Call the virtual member function
8773 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8774 (substitute_and_fold_dom_walker): Remove initialization of
8775 data member entries for calbacks. Add substitute_and_fold_engine
8776 member and initialize it.
8777 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8778 member functions for get_value, replace_phi_args_in c
8779 replace_uses_in, and fold_stmt calls.
8780 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8781 substitute_and_fold. Remove assert. Update ctor call.
8783 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8784 (ssa_prop_visit_phi_fn): Likewise.
8785 (class ssa_propagation_engine): New class to provide an interface
8787 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8789 (ssa_prop_visit_phi): Likewise.
8790 (ssa_propagation_engine::simulate_stmt): Moved into class.
8791 Call visit_phi/visit_stmt from the class rather than via
8792 file scoped static variables.
8793 (ssa_propagation_engine::simulate_block): Moved into class.
8794 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8795 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8796 set file scoped statics for the visit_stmt/visit_phi callbacks.
8797 * tree-complex.c (complex_propagate): New class derived from
8798 ssa_propagation_engine.
8799 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8800 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8801 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8803 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8804 ssa_propagation_engine.
8805 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8806 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8807 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8808 * tree-ssa-copy.c (copy_prop): New class derived from
8809 ssa_propagation_engine.
8810 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8811 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8812 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8813 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8814 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8815 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8816 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8818 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8820 PR rtl-optimization/82778
8821 PR rtl-optimization/82597
8822 * compare-elim.c (struct comparison): Add in_a_setter field.
8823 (find_comparison_dom_walker::before_dom_children): Remove killed
8824 bitmap and df_simulate_find_defs call, instead walk the defs.
8825 Compute last_setter and initialize in_a_setter. Merge definitions
8826 with first initialization for a few variables.
8827 (try_validate_parallel): Use insn_invalid_p instead of
8828 recog_memoized. Return insn rather than just the pattern.
8829 (try_merge_compare): Fix up comment. Don't uselessly test if
8830 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8832 (execute_compare_elim_after_reload): Remove df_chain_add_problem
8835 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8836 Alan Hayward <alan.hayward@arm.com>
8837 David Sherwood <david.sherwood@arm.com>
8839 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
8840 aarch64_hard_regno_nregs to get the number of registers
8843 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8844 Alan Hayward <alan.hayward@arm.com>
8845 David Sherwood <david.sherwood@arm.com>
8847 * config/aarch64/constraints.md (Upl): Rename to...
8849 * config/aarch64/aarch64.md
8850 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
8853 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8854 Alan Hayward <alan.hayward@arm.com>
8855 David Sherwood <david.sherwood@arm.com>
8857 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
8858 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8861 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8862 Alan Hayward <alan.hayward@arm.com>
8863 David Sherwood <david.sherwood@arm.com>
8865 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
8866 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
8867 (aarch64_evpc_dup): Generate rtl direcly, rather than using
8869 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
8870 of a single element.
8871 * config/aarch64/iterators.md: Add a comment above the permute
8872 unspecs to say that they are generated directly by
8873 aarch64_expand_vec_perm_const.
8874 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
8876 2017-11-01 Nathan Sidwell <nathan@acm.org>
8878 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
8880 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8882 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
8883 and medany code models, and describe what they do.
8885 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8887 Revert accidental duplicate:
8889 * combine.c (can_change_dest_mode): Reject changes in
8890 REGMODE_NATURAL_SIZE.
8892 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
8894 PR rtl-optimization/64682
8895 PR rtl-optimization/69567
8896 PR rtl-optimization/69737
8897 PR rtl-optimization/82683
8898 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
8899 register mentioned in the note, drop the note, unless it came from I3,
8900 in which case it should go to I3 again.
8902 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8904 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
8905 and return false if not.
8906 (clear_bytes_written_by, live_bytes_read): Update accordingly.
8908 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8910 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
8911 range is known to be empty.
8913 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8914 Alan Hayward <alan.hayward@arm.com>
8915 David Sherwood <david.sherwood@arm.com>
8917 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
8918 and CONST_VECTOR_NUNITS instead of computing the number of units from
8919 the byte sizes of the vector and element.
8920 (simplify_binary_operation_1): Likewise.
8921 (simplify_const_binary_operation): Likewise.
8922 (simplify_ternary_operation): Likewise.
8924 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8925 Alan Hayward <alan.hayward@arm.com>
8926 David Sherwood <david.sherwood@arm.com>
8928 * var-tracking.c (INT_MEM_OFFSET): Replace with...
8929 (int_mem_offset): ...this new function.
8930 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
8931 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
8932 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
8935 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8936 Alan Hayward <alan.hayward@arm.com>
8937 David Sherwood <david.sherwood@arm.com>
8939 * lower-subreg.c (interesting_mode_p): New function.
8940 (compute_costs, find_decomposable_subregs, decompose_register)
8941 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
8942 (resolve_clobber, dump_choices): Use it.
8944 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8945 Alan Hayward <alan.hayward@arm.com>
8946 David Sherwood <david.sherwood@arm.com>
8948 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
8950 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8951 Alan Hayward <alan.hayward@arm.com>
8952 David Sherwood <david.sherwood@arm.com>
8954 * alias.c (find_base_value, find_base_term): Only process integer
8955 truncations. Check the precision rather than the size.
8957 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8958 Alan Hayward <alan.hayward@arm.com>
8959 David Sherwood <david.sherwood@arm.com>
8961 * machmode.h (is_narrower_int_mode): New function
8962 * optabs.c (expand_float, expand_fix): Use it.
8963 * dwarf2out.c (rotate_loc_descriptor): Likewise.
8965 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8966 Alan Hayward <alan.hayward@arm.com>
8967 David Sherwood <david.sherwood@arm.com>
8969 * rtl.h (narrower_subreg_mode): New function.
8970 * ira-color.c (update_costs_from_allocno): Use it.
8972 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8973 Alan Hayward <alan.hayward@arm.com>
8974 David Sherwood <david.sherwood@arm.com>
8976 * optabs-query.h (convert_optab_p): New function, split out from...
8977 (convert_optab_handler): ...here.
8978 (widening_optab_handler): Delete.
8979 (find_widening_optab_handler): Remove permit_non_widening parameter.
8980 (find_widening_optab_handler_and_mode): Likewise. Provide an
8981 override that operates on mode class wrappers.
8982 * optabs-query.c (widening_optab_handler): Delete.
8983 (find_widening_optab_handler_and_mode): Remove permit_non_widening
8984 parameter. Assert that the two modes are the same class and that
8985 the "from" mode is narrower than the "to" mode. Use
8986 convert_optab_handler instead of widening_optab_handler.
8987 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
8988 instead of widening_optab_handler.
8989 * expr.c (expand_expr_real_2): Update calls to
8990 find_widening_optab_handler.
8991 * optabs.c (expand_widen_pattern_expr): Likewise.
8992 (expand_binop_directly): Take the insn_code as a parameter.
8993 (expand_binop): Only call find_widening_optab_handler for
8994 conversion optabs; use optab_handler otherwise. Update calls
8995 to find_widening_optab_handler and expand_binop_directly.
8996 Use convert_optab_handler instead of widening_optab_handler.
8997 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
8998 find_widening_optab_handler and use scalar_mode rather than
9000 (convert_plusminus_to_widen): Likewise.
9002 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9003 Alan Hayward <alan.hayward@arm.com>
9004 David Sherwood <david.sherwood@arm.com>
9006 * machmode.h (fixed_size_mode): New class.
9007 * rtl.h (get_pool_mode): Return fixed_size_mode.
9008 * gengtype.c (main): Add fixed_size_mode.
9009 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9010 (get_raw_arg_mode): Likewise.
9011 * doc/tm.texi: Regenerate.
9012 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9013 * targhooks.c (default_get_reg_raw_mode): Likewise.
9014 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9015 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9016 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9017 (msp430_get_raw_result_mode): Likewise.
9018 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9019 * dbxout.c (dbxout_parms): Require fixed-size modes.
9020 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9021 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9022 * omp-low.c (lower_oacc_reductions): Likewise.
9023 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9024 (simplify_subreg): Update accordingly.
9025 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9026 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9027 that aren't fixed-size.
9028 (get_pool_mode): Return a fixed_size_mode.
9029 (output_constant_pool_2): Take a fixed_size_mode.
9031 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9032 Alan Hayward <alan.hayward@arm.com>
9033 David Sherwood <david.sherwood@arm.com>
9035 * doc/rtl.texi (vec_series): Document.
9036 (const): Say that the operand can be a vec_series.
9037 * rtl.def (VEC_SERIES): New rtx code.
9038 * rtl.h (const_vec_series_p_1): Declare.
9039 (const_vec_series_p): New function.
9040 * emit-rtl.h (gen_const_vec_series): Declare.
9041 (gen_vec_series): Likewise.
9042 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9043 (gen_vec_series): Likewise.
9044 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9045 * simplify-rtx.c (simplify_unary_operation): Handle negations
9047 (simplify_binary_operation_series): New function.
9048 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9049 (test_vector_ops_series): New function.
9050 (test_vector_ops): Call it.
9051 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9052 gen_const_vec_series.
9053 (altivec_lvsr): Likewise.
9054 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9056 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9057 Alan Hayward <alan.hayward@arm.com>
9058 David Sherwood <david.sherwood@arm.com>
9060 * doc/rtl.texi (const): Update description of address constants.
9061 Say that vector constants are allowed too.
9062 * common.md (E, F): Use CONSTANT_P instead of checking for
9064 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9065 checking for CONST_VECTOR.
9066 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9068 * expr.c (expand_expr_real_2): Check for vector modes instead
9069 of checking for CONST_VECTOR.
9070 * rtl.h (const_vec_p): New function.
9071 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9072 (unwrap_const_vec_duplicate): Handle them here too.
9074 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9075 David Malcolm <dmalcolm@redhat.com>
9076 Alan Hayward <alan.hayward@arm.com>
9077 David Sherwood <david.sherwood@arm.com>
9079 * rtl.h (vec_duplicate_p): New function.
9080 * selftest-rtl.c (assert_rtx_eq_at): New function.
9081 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9082 (assert_rtx_eq_at): Declare.
9083 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9084 * selftest-run-tests.c (selftest::run_tests): Call it.
9085 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9086 (simplify_unary_operation_1): Recursively handle vector duplicates.
9087 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9089 (simplify_subreg): Handle subregs of vector duplicates.
9090 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9091 (selftest::simplify_rtx_c_tests): New functions.
9093 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9094 Alan Hayward <alan.hayward@arm.com>
9095 David Sherwood <david.sherwood@arm.com>
9097 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9098 (gen_vec_duplicate): Likewise.
9099 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9101 (gen_const_vector): ...here.
9102 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9103 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9104 whose elements are all equal.
9105 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
9106 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
9107 (simplify_relational_operation): Likewise.
9108 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
9110 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
9111 (aarch64_expand_vector_init): Likewise.
9112 * config/arm/arm.c (neon_vdup_constant): Likewise.
9113 (neon_expand_vector_init): Likewise.
9114 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
9115 (arm_block_set_unaligned_vect): Likewise.
9116 (arm_block_set_aligned_vect): Likewise.
9117 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
9118 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
9119 (expand_vec_perm_even_odd_pack): Likewise.
9120 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
9121 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
9122 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
9123 gen_const_vec_duplicate.
9124 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
9125 * config/mips/mips.c (mips_gen_const_int_vector): Use
9126 gen_const_vec_duplicate.
9127 (mips_expand_vector_init): Use CONST0_RTX.
9128 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
9129 (define_split): Use gen_const_vec_duplicate.
9130 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
9131 (define_split): Use gen_const_vec_duplicate.
9132 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
9133 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
9134 * config/spu/spu.c (spu_const): Likewise.
9136 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9137 Alan Hayward <alan.hayward@arm.com>
9138 David Sherwood <david.sherwood@arm.com>
9140 * combine.c (can_change_dest_mode): Reject changes in
9141 REGMODE_NATURAL_SIZE.
9143 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
9145 * configure.ac (--enable-libssp): New.
9146 (gcc_cv_libc_provides_ssp): Check for explicit setting before
9147 trying to determine target-specific default. Adjust indentation.
9148 * configure: Regenerated.
9149 * doc/install.texi (Configuration): Expand --disable-libssp
9152 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
9154 config/i386/i386.c (ix86_expand_epilogue): Correct stack
9157 2017-10-31 Martin Jambor <mjambor@suse.cz>
9160 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
9162 2017-10-31 David Malcolm <dmalcolm@redhat.com>
9164 * auto-profile.c (autofdo_source_profile::read): Use
9165 UNKNOWN_LOCATION rather than 0.
9166 * diagnostic-core.h (warning_at_rich_loc): Rename to...
9167 (warning_at): ...this overload.
9168 (warning_at_rich_loc_n): Rename to...
9169 (warning_n): ...this overload.
9170 (error_at_rich_loc): Rename to...
9171 (error_at): ...this overload.
9172 (pedwarn_at_rich_loc): Rename to...
9173 (pedwarn): ...this overload.
9174 (permerror_at_rich_loc): Rename to...
9175 (permerror): ...this overload.
9176 (inform_at_rich_loc): Rename to...
9177 (inform): ...this overload.
9178 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
9179 (diagnostic_n_impl_richloc): Rename to...
9180 (diagnostic_n_impl): ...this rich_location *-based decl.
9181 (inform_at_rich_loc): Rename to...
9182 (inform): ...this, and add an assertion.
9183 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
9184 (warning_at_rich_loc): Rename to...
9185 (warning_at): ...this, and add an assertion.
9186 (warning_at_rich_loc_n): Rename to...
9187 (warning_n): ...this, and add an assertion.
9188 (warning_n): Update location_t-based implementation for removal of
9189 location_t-based diagnostic_n_impl.
9190 (pedwarn_at_rich_loc): Rename to...
9191 (pedwarn): ...this, and add an assertion.
9192 (permerror_at_rich_loc): Rename to...
9193 (permerror): ...this, and add an assertion.
9194 (error_n): Update for removal of location_t-based diagnostic_n_impl.
9195 (error_at_rich_loc): Rename to...
9196 (error_at): ...this, and add an assertion.
9197 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
9198 (driver::do_spec_on_infiles): Likewise.
9199 * substring-locations.c (format_warning_va): Update for renaming
9200 of inform_at_rich_loc.
9202 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
9204 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
9205 _Float<N>X built-in functions so that the variant without the
9206 "__builtin_" prefix is only enabled for the GNU C and Objective C
9207 languages when they are in non-strict ANSI/ISO mode.
9208 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
9209 * target.def (floatn_builtin_p): Add a target hook to control
9210 whether _Float<N> and _Float<N>X built-in functions without the
9211 "__builtin_" prefix are enabled, and return true for C and
9212 Objective C in the default hook. Include langhooks.h in
9214 * targhooks.h (default_floatn_builtin_p): Likewise.
9215 * targhooks.c (default_floatn_builtin_p): Likewise.
9216 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
9217 floatn_builtin_p target hook.
9218 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
9220 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
9221 Eric Botcazou <ebotcazou@adacore.com>
9223 PR rtl-optimization/81803
9224 * lra-constraints.c (curr_insn_transform): Also reload the whole
9225 register for a strict subreg no wider than a word if this is for
9226 a WORD_REGISTER_OPERATIONS target.
9228 2017-10-31 Jason Merrill <jason@redhat.com>
9230 * gdbinit.in: Skip over inlines from timevar.h.
9232 2017-10-31 Martin Liska <mliska@suse.cz>
9234 * doc/gcov.texi: Document new option.
9235 * gcov.c (print_usage): Likewise print it.
9236 (process_args): Support the argument.
9237 (format_count): New function.
9238 (format_gcov): Use the function.
9240 2017-10-31 Martin Liska <mliska@suse.cz>
9242 * gcov.c (struct name_map): do not use typedef.
9243 Define operator== and operator<.
9244 (name_search): Remove.
9245 (name_sort): Remove.
9246 (main): Do not allocate names.
9247 (process_file): Add vertical space.
9248 (generate_results): Use std::find.
9249 (release_structures): Do not release memory.
9250 (find_source): Use std::find.
9252 2017-10-31 Martin Liska <mliska@suse.cz>
9254 * gcov.c (struct line_info): Remove it's typedef.
9255 (line_info::line_info): Add proper ctor.
9256 (line_info::has_block): Do not use a typedef.
9257 (struct source_info): Do not use typedef.
9258 (circuit): Likewise.
9259 (get_cycles_count): Likewise.
9260 (output_intermediate_file): Iterate via vector iterator.
9261 (add_line_counts): Use std::vector methods.
9262 (accumulate_line_counts): Likewise.
9263 (output_lines): Likewise.
9265 2017-10-31 Martin Liska <mliska@suse.cz>
9267 * gcov.c (struct source_info): Remove typedef.
9268 (source_info::source_info): Add proper ctor.
9269 (accumulate_line_counts): Use struct, not it's typedef.
9270 (output_gcov_file): Likewise.
9271 (output_lines): Likewise.
9272 (main): Do not allocate an array.
9273 (output_intermediate_file): Use size of vector container.
9274 (process_file): Resize the vector.
9275 (generate_results): Do not preallocate, use newly added vector
9277 (release_structures): Do not release sources.
9278 (find_source): Use vector methods.
9279 (add_line_counts): Do not use typedef.
9281 2017-10-31 Martin Liska <mliska@suse.cz>
9283 * doc/gcov.texi: Document that.
9284 * gcov.c (add_line_counts): Mark lines with a non-executed
9286 (output_line_beginning): Handle such lines.
9287 (output_lines): Pass new argument.
9288 (output_intermediate_file): Print it in intermediate format.
9290 2017-10-31 Martin Liska <mliska@suse.cz>
9292 * color-macros.h: New file.
9293 * diagnostic-color.c: Factor out color related to macros to
9295 * doc/gcov.texi: Document -k option.
9296 * gcov.c (INCLUDE_STRING): Include string.h.
9297 (print_usage): Add -k option.
9298 (process_args): Parse it.
9299 (pad_count_string): New function.
9300 (output_line_beginning): Likewise.
9301 (DEFAULT_LINE_START): New macro.
9302 (output_lines): Support color output.
9304 2017-10-31 Martin Liska <mliska@suse.cz>
9306 PR gcov-profile/82633
9307 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
9308 their interaction with GCOV infrastructure.
9309 * configure.ac: Add -fkeep-{inline,static}-functions to
9311 * configure: Regenerate.
9313 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
9316 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
9318 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
9321 * config/rs6000/rs6000.md (allocate_stack): Force update interval
9322 into a register if it does not fit into an immediate offset field.
9324 2017-10-31 Olivier Hainque <hainque@adacore.com>
9326 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
9328 2017-10-31 Julia Koval <julia.koval@intel.com>
9330 * config.gcc: Add gfniintrin.h.
9331 * config/i386/gfniintrin.h: New.
9332 * config/i386/i386-builtin-types.def
9333 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
9334 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
9335 __builtin_ia32_vgf2p8affineinvqb_v32qi,
9336 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
9337 __builtin_ia32_vgf2p8affineinvqb_v16qi,
9338 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
9339 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
9340 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
9341 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
9342 V64QI_FTYPE_V64QI_V64QI_INT): New types.
9343 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
9344 * config/i386/immintrin.h: Include gfniintrin.h.
9345 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
9347 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
9349 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
9351 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9353 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
9354 (arm_ashldi3_1bit): Remove pattern.
9355 (ashrdi3): Remove shift by 1 expansion.
9356 (arm_ashrdi3_1bit): Remove pattern.
9357 (lshrdi3): Remove shift by 1 expansion.
9358 (arm_lshrdi3_1bit): Remove pattern.
9359 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
9360 cost of ashldi3 by 1.
9361 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
9362 (<shift>di3_neon): Likewise.
9364 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9366 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
9367 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
9368 and (*aarch64_simd_mov<VQ:mode>).
9369 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
9370 pattern alternative.
9371 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
9372 attributes to match pattern alternative.
9374 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
9376 * config.gcc (powerpc*-*-*): Add emmintrin.h.
9377 * config/rs6000/emmintrin.h: New file.
9378 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
9380 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9382 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
9383 * (movdi_vfp_cortexa8): Remove pattern.
9385 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9387 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
9389 (Specific, alpha*-dec-osf5.1): Remove.
9390 (Specific, mips-sgi-irix5): Remove.
9391 (Specific, mips-sgi-irix6): Remove.
9393 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9396 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
9397 arguments to clear_bit_region_be.
9399 2017-10-30 Jim Wilson <wilson@tuliptree.org>
9401 * gimplify.c: Include memmodel.h.
9403 2017-10-30 Martin Jambor <mjambor@suse.cz>
9405 * omp-grid.c (grid_attempt_target_gridification): Also insert a
9406 condition whether loop should be executed at all.
9408 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
9410 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9411 gimple folding of vec_madd() intrinsics.
9412 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
9413 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
9414 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
9416 2017-10-30 Richard Biener <rguenther@suse.de>
9418 PR tree-optimization/82762
9420 2017-10-23 Richard Biener <rguenther@suse.de>
9422 PR tree-optimization/82129
9424 2017-08-01 Richard Biener <rguenther@suse.de>
9426 PR tree-optimization/81181
9427 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9428 (compute_antic): ... end of iteration here.
9430 2017-10-30 Joseph Myers <joseph@codesourcery.com>
9432 * doc/invoke.texi (C Dialect Options): Document -std=c17,
9433 -std=iso9899:2017 and -std=gnu17.
9434 * doc/standards.texi (C Language): Document C17 support.
9435 * doc/cpp.texi (Overview): Mention -std=c17.
9436 (Standard Predefined Macros): Document C11 and C17 values of
9437 __STDC_VERSION__. Do not refer to C99 support as incomplete.
9438 * doc/extend.texi (Inline): Do not list individual options for
9439 standards newer than C99.
9440 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
9442 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
9445 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
9447 * asan.c (asan_finish_file): Align asan globals array by shadow
9450 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9453 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
9454 (struct store_immediate_info): Add bitregion_start and bitregion_end
9456 (store_immediate_info::store_immediate_info): Add brs and bre
9457 arguments and initialize bitregion_{start,end} from those.
9458 (struct merged_store_group): Add bitregion_start, bitregion_end,
9459 align_base and mask fields. Drop unnecessary struct keyword from
9460 struct store_immediate_info. Add do_merge method.
9461 (clear_bit_region_be): Use memset instead of loop storing zeros.
9462 (merged_store_group::do_merge): New method.
9463 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
9464 stores as long as the surrounding bitregions have no gaps.
9465 (merged_store_group::merge_overlapping): Use do_merge.
9466 (merged_store_group::apply_stores): Test that bitregion_{start,end}
9467 is byte aligned, rather than requiring that start and width are
9468 byte aligned. Drop unnecessary struct keyword from
9469 struct store_immediate_info. Allocate and populate also mask array.
9470 Make start of the arrays relative to bitregion_start rather than
9471 start and size them according to bitregion_{end,start} difference.
9472 (struct imm_store_chain_info): Drop unnecessary struct keyword from
9473 struct store_immediate_info.
9474 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
9475 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
9476 struct keyword from struct store_immediate_info.
9477 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
9478 between stores as long as the surrounding bitregions have no gaps.
9480 (struct split_store): Add orig non-static data member.
9481 (split_store::split_store): Initialize orig to false.
9482 (find_constituent_stmts): Return store_immediate_info *, non-NULL
9483 if there is exactly a single original stmt. Change stmts argument
9484 to pointer from reference, if NULL, don't push anything to it. Add
9485 first argument, use it to optimize skipping over orig stmts that
9486 are known to be before bitpos already. Simplify.
9487 (split_group): Return unsigned int count how many stores are or
9488 would be needed rather than a bool. Add allow_unaligned argument.
9489 Change split_stores argument from reference to pointer, if NULL,
9490 only do a dry run computing how many stores would be produced.
9491 Rewritten algorithm to use both alignment and misalign if
9492 !allow_unaligned and handle bitfield stores with gaps.
9493 (imm_store_chain_info::output_merged_store): Set start_byte_pos
9494 from bitregion_start instead of start. Compute allow_unaligned
9495 here, if true, do 2 split_group dry runs to compute which one
9496 produces fewer stores and prefer aligned if equal. Punt if
9497 new count is bigger or equal than original before emitting any
9498 statements, rather than during that. Remove no longer needed
9499 new_ssa_names tracking. Replace num_stmts with
9500 split_stores.length (). Use 32-bit stack allocated entries
9501 in split_stores auto_vec. Try to reuse original store lhs/rhs1
9502 if possible. Handle bitfields with gaps.
9503 (pass_store_merging::execute): Ignore bitsize == 0 stores.
9504 Compute bitregion_{start,end} for the stores and construct
9505 store_immediate_info with that. Formatting fixes.
9507 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
9510 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
9511 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
9513 2017-10-29 Jim Wilson <wilson@tuliptree.org>
9515 * gimplify.c: Include tm_p.h.
9517 * common.opt (gcoff): Re-add as ignored option.
9518 (gcoff1, gcoff2, gcoff3): Likewise.
9520 * Makefile.in (OBJS): Delete sdbout.o.
9521 (GTFILES): Delete $(srcdir)/sdbout.c.
9522 * debug.h: Delete sdb_debug_hooks.
9523 * final.c: Delete sdbout.h include.
9524 (final_scan_insn): Delete SDB_DEBUG check.
9525 (rest_of_clean_state): Likewise.
9526 * output.h: Delete sdb_begin_function_line.
9529 * toplev.c: Delete sdbout.h include.
9530 (process_options): Delete SDB_DEBUG check.
9531 * tree-core.h (tree_type_common): Delete pointer field of
9533 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
9534 TYPE_SYMTAB_POINTER.
9535 * tree.h (TYPE_SYMTAB_POINTER): Delete.
9536 (TYPE_SYMTAB_IS_POINTER): Delete.
9537 (TYPE_SYMTAB_IS_DIE): Renumber.
9538 * xcoffout.c: Refer to former sdbout.c file.
9539 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
9541 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
9542 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
9543 (Debugging Options): Delete -gcoff.
9544 (-gstabs): Delete SDB reference.
9546 (-gcoff@var{level}): Delete.
9547 * doc/passes.texi (Debugging information output): Delete SDB and
9548 sdbout.c references.
9549 * doc/tm.texi: Regenerate.
9550 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
9551 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
9553 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
9554 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
9555 (SDB_DEBUGGING_INFO): Delete.
9556 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
9557 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
9558 * target.def (output_source_filename): Delete COFF reference.
9560 * common.opt (gcoff): Delete.
9561 (gxcoff+): Update Negative chain.
9562 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
9564 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
9565 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
9566 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
9568 (expand_function_start): Change sdb reference to past tense.
9569 (expand_function_end): Change sdb reference to past tense.
9570 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
9571 * opts.c (debug_type_names): Delete coff entry.
9572 (common_handle_option): Delete OPT_gcoff case.
9573 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
9575 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
9576 * config/cris/cris.h: Delete SDB reference in comment.
9577 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
9578 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
9579 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
9580 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
9582 (ix86_expand_prologue): Likewise.
9583 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
9584 * config/ia64/ia64.h: Likewise.
9585 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
9586 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
9588 * config/mmix/mmix.h: Likewise.
9589 * config/nds32/nds32.c: Likewise.
9590 * config/stormy/storym16.h: Likewise.
9591 * config/visium/visium.h: Likewise.
9592 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
9594 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
9596 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
9597 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
9598 FRAME_POINTER_REGNUM point at high end of local var area.
9600 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
9602 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
9603 Move comment around. Do not reset best_edge for a copiable
9604 destination if the copy would cause a partition change.
9605 (better_edge_p): Remove redundant check.
9607 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9609 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
9611 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9613 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
9614 for math functions that have _Float<N> and _Float<N>X variants.
9615 (mathfn_built_in_2): Add support for math functions that have
9616 _Float<N> and _Float<N>X variants.
9617 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9618 (expand_builtin_mathfn_ternary): Add support for fma with
9619 _Float<N> and _Float<N>X variants.
9620 (expand_builtin): Likewise.
9621 (fold_builtin_3): Likewise.
9622 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9623 create math function _Float<N> and _Float<N>X variants as external
9625 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9626 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9627 the __builtin_ prefix and if not strict ansi, without the prefix.
9628 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9629 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9630 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9631 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9632 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9633 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9634 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9635 function signatures for fma _Float<N> and _Float<N>X variants.
9636 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9637 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9638 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9639 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9640 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9641 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9642 * gencfn-macros.c (print_case_cfn): Add support for math functions
9643 that have _Float<N> and _Float<N>X variants.
9644 (print_define_operator_list): Likewise.
9645 (fltfn_suffixes): Likewise.
9647 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9648 for math functions that have _Float<N> and _Float<N>X variants.
9649 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9650 and _Float<N>X variants.
9651 (COPYSIGN): Likewise.
9654 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9655 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9657 (integer_valued_read_call_p): Likewise.
9658 * fold-const-call.c (fold_const_call_ss): Likewise.
9659 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9660 _Float<N> and _Float<N>X variants.
9661 (fold_const_call_ssss): Add support for fma _Float<N> and
9662 _Float<N>X variants.
9663 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9664 support for copysign and fma _Float<N> and _Float<N>X variants.
9665 (backprop::process_builtin_call_use): Likewise.
9666 * tree-call-cdce.c (can_test_argument_range); Add support for
9667 sqrt _Float<N> and _Float<N>X variants.
9668 (edom_only_function): Likewise.
9669 (get_no_error_domain): Likewise.
9670 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9671 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9672 copysign _Float<N> and _Float<N>X variants.
9673 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9674 handled by machine independent code.
9675 (FMAF128): Likewise.
9676 * doc/cpp.texi (Common Predefined Macros): Document defining
9677 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9678 fma _Float<N> and _Float<N>X variants.
9680 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9683 * config/i386/i386-modes.def (CCFPU): Remove definition.
9684 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9685 (ix86_cc_modes_compatible): Ditto.
9686 (ix86_expand_carry_flag_compare): Ditto.
9687 (ix86_expand_int_movcc): Ditto.
9688 (ix86_expand_int_addcc): Ditto.
9689 (ix86_reverse_condition): Ditto.
9690 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9691 Return true/false for unordered/ordered fp comparisons.
9692 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9693 (ix86_prepare_fp_compare_args): Update for rename.
9694 (ix86_expand_fp_compare): Update for rename. Generate unordered
9695 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9696 (ix86_expand_sse_compare_and_jump): Ditto.
9697 * config/i386/predicates.md (fcmov_comparison_operator):
9698 Remove CCFPU mode handling.
9699 (ix86_comparison_operator): Ditto.
9700 (ix86_carry_flag_operator): Ditto.
9701 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9702 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9703 (*cmpu<mode>_cc_i387): Ditto.
9704 (FPCMP): Remove mode iterator.
9705 (unord): Remove mode attribute.
9706 (unord_subst): New define_subst transformation
9707 (unord): New define_subst attribute.
9709 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9710 (*cmpi<unord>xf_i387): Ditto.
9711 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9712 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9713 using unord_subst transformation.
9714 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9715 (round_saeonly): Also handle CCFP mode.
9716 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9717 Remove UNSPEC_SAHF unspec handling.
9719 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9721 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9723 2017-10-27 Jeff Law <law@redhat.com>
9725 * gimple-ssa-sprintf.c: Include domwalk.h.
9726 (class sprintf_dom_walker): New class, derived from dom_walker.
9727 (sprintf_dom_walker::before_dom_children): New function.
9728 (struct call_info): Moved into sprintf_dom_walker class
9729 (compute_formath_length, handle_gimple_call): Likewise.
9730 (sprintf_length::execute): Call the dominator walker rather
9731 than walking the statements.
9733 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9734 gimple statement locations.
9735 (check_array_bounds): Corresponding changes. Get the statement's
9736 location directly from wi->stmt.
9738 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9741 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9743 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9745 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9746 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9748 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9751 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9752 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9753 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9754 maybe_get_pool_constant.
9755 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9758 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9760 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9761 2.26 caveat. Update gas and gld versions.
9762 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9765 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9767 * cgraph.h (set_malloc_flag): Declare.
9768 * cgraph.c (set_malloc_flag_1): New function.
9769 (set_malloc_flag): Likewise.
9770 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9771 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9773 (read_ipa_call_summary): Add support for reading is_return_callee.
9774 (write_ipa_call_summary): Stream is_return_callee.
9775 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9776 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9777 ipa-prop.h, ipa-fnsummary.h.
9778 (pure_const_names): Change to static.
9779 (malloc_state_e): Define.
9780 (malloc_state_names): Define.
9781 (funct_state_d): Add field malloc_state.
9782 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9783 (check_retval_uses): New function.
9784 (malloc_candidate_p): Likewise.
9785 (analyze_function): Add support for malloc attribute.
9786 (pure_const_write_summary): Stream malloc_state.
9787 (pure_const_read_summary): Add support for reading malloc_state.
9788 (dump_malloc_lattice): New function.
9789 (propagate_malloc): New function.
9790 (warn_function_malloc): New function.
9791 (ipa_pure_const::execute): Call propagate_malloc and
9792 ipa_free_fn_summary.
9793 (pass_local_pure_const::execute): Add support for malloc attribute.
9794 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9795 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9797 2017-10-27 Martin Liska <mliska@suse.cz>
9799 PR gcov-profile/82457
9800 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9801 for fork-like functions to be properly instrumented.
9803 2017-10-27 Richard Biener <rguenther@suse.de>
9806 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9807 info when we redirected EH.
9809 2017-10-26 Michael Collison <michael.collison@arm.com>
9811 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9813 (<optab>_trunchf<GPI:mode>2: New pattern.
9814 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9815 * config/aarch64/iterators.md (wv): New mode attribute.
9816 (vf, VF): New mode attributes.
9817 (vgp, VGP): New mode attributes.
9818 (s): Update attribute with SImode and DImode prefixes.
9820 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9822 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9823 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9824 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9825 (nios2_option_overide): Initialize it. Don't allow R0-relative
9826 addressing with PIC.
9827 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9828 (nios2_symbolic_constant_p): Likewise.
9829 (nios2_legitimate_address_p): Likewise.
9830 (nios2_r0rel_section_name_p): New.
9831 (nios2_symbol_ref_in_r0rel_data_p): New.
9832 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9833 (r0rel_constant_p): New.
9834 (nios2_print_operand_address): Handle r0rel_constant_p.
9835 (nios2_cdx_narrow_form_p): Likewise.
9836 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9837 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9838 (Nios II Options): Document -mr0rel-sec.
9840 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9842 * config/nios2/nios2.c: Include xregex.h.
9843 (nios2_gprel_sec_regex): New.
9844 (nios2_option_overide): Initialize it. Don't allow GP-relative
9845 addressing with PIC.
9846 (nios2_small_section_name_p): Check for regex match.
9847 * config/nios2/nios2.opt (mgprel-sec=): New option.
9848 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9849 (Nios II Options): Document -mgprel-sec.
9851 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9853 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9855 2017-10-26 Tom de Vries <tom@codesourcery.com>
9857 PR tree-optimization/82707
9858 * gimple.c (gimple_copy): Fix unsharing of
9859 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9861 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
9863 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
9864 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
9865 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
9866 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
9867 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
9868 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
9869 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
9870 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
9871 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
9872 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
9873 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
9874 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
9875 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
9876 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
9877 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
9878 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
9879 _mm512_mask_cmpunord_ps_mask): New intrinsics.
9881 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9883 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
9885 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
9886 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
9887 warning to rs6000.c. Remove the Undocumented flag, since it has
9889 (-mabi=ibmlongdouble): Likewise.
9890 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
9891 already set, set the default format for long double.
9892 (rs6000_debug_reg_global): Print whether long double is IBM or
9894 (rs6000_option_override_internal): Rework setting long double
9895 format. Only warn if the user is changing the long double default
9896 and they did not use -Wno-psabi.
9897 * doc/invoke.texi (PowerPC options): Update the documentation for
9898 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
9900 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9901 Alan Hayward <alan.hayward@arm.com>
9902 David Sherwood <david.sherwood@arm.com>
9904 * rtl.h (wider_subreg_mode): New function.
9905 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
9906 rather than an unsigned int *.
9907 * ira-color.c (regno_max_ref_width): Replace with...
9908 (regno_max_ref_mode): ...this new variable.
9909 (coalesced_pseudo_reg_slot_compare): Update accordingly.
9910 Use wider_subreg_mode.
9911 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
9912 rather than an unsigned int *.
9913 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
9914 (process_alt_operands): Likewise.
9915 (invariant_p): Likewise.
9916 * lra-spills.c (assign_mem_slot): Likewise.
9917 (add_pseudo_to_slot): Likewise.
9918 * lra.c (collect_non_operand_hard_regs): Likewise.
9919 (add_regs_to_insn_regno_info): Likewise.
9920 * reload1.c (regno_max_ref_width): Replace with...
9921 (regno_max_ref_mode): ...this new variable.
9922 (reload): Update accordingly. Update call to
9923 ira_sort_regnos_for_alter_reg.
9924 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
9925 (init_eliminable_invariants): Update to use regno_max_ref_mode.
9926 (scan_paradoxical_subregs): Likewise.
9928 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9930 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
9931 (aarch64_frame): Add emit_frame_chain boolean.
9932 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9933 Move eh_return case to aarch64_layout_frame.
9934 (aarch64_layout_frame): Initialize emit_frame_chain.
9935 (aarch64_expand_prologue): Use emit_frame_chain.
9937 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9939 * config/aarch64/aarch64.c (aarch64_layout_frame):
9940 Ensure LR is always stored at the bottom of the callee-saves.
9941 Remove rarely used frame layout which saves callee-saves at top of
9942 frame, so the store of LR can be used as a valid probe in all cases.
9944 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9946 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
9947 Improve unaligned TImode/TFmode base/offset split.
9949 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9950 Alan Hayward <alan.hayward@arm.com>
9951 David Sherwood <david.sherwood@arm.com>
9953 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
9954 * combine.c (find_single_use_1): Likewise.
9955 (expand_field_assignment): Likewise.
9956 (move_deaths): Likewise.
9957 * lra-constraints.c (simplify_operand_subreg): Likewise.
9958 (curr_insn_transform): Likewise.
9959 * lra.c (collect_non_operand_hard_regs): Likewise.
9960 (add_regs_to_insn_regno_info): Likewise.
9961 * rtlanal.c (reg_referenced_p): Likewise.
9962 (covers_regno_no_parallel_p): Likewise.
9964 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9966 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
9967 Don't print any bits outside the precision of the value.
9968 * wide-int.cc (test_printing): Add some new tests.
9970 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9972 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
9973 supports -xbrace_comment option.
9974 * configure: Regenerate.
9975 * config.in: Regenerate.
9976 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
9977 (ASM_CPU_SPEC): Use it.
9979 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9981 * target.def (static_rtx_alignment): New hook.
9982 * targhooks.h (default_static_rtx_alignment): Declare.
9983 * targhooks.c (default_static_rtx_alignment): New function.
9984 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
9985 * doc/tm.texi: Regenerate.
9986 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
9987 instead of targetm.constant_alignment. Remove call to
9989 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9990 (cris_preferred_mininum_alignment): New function, split out from...
9991 (cris_constant_alignment): ...here.
9992 (cris_static_rtx_alignment): New function.
9993 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
9995 (ix86_constant_alignment): ...here.
9996 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9997 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9998 (mmix_static_rtx_alignment): New function.
9999 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10000 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10002 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10005 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10006 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10008 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10010 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10011 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10012 mark as a sign-extending load.
10013 (local_pic_load_u): Define.
10015 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10017 PR middle-end/82062
10018 * fold-const.c (operand_equal_for_comparison_p): Also return true
10019 if ARG0 is a simple variant of ARG1 with narrower precision.
10020 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10022 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10024 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10026 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10027 scatter_static, scatter_per_elt.
10028 * x86-tune-costs.h: Add new cost entries.
10030 2017-10-25 Richard Biener <rguenther@suse.de>
10032 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10033 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10034 class pass_fre): Move to ...
10035 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10036 class pass_fre): ... here and adjust for statistics.
10038 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10041 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10042 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10043 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10044 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10047 2017-10-25 Richard Biener <rguenther@suse.de>
10049 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10050 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10051 eliminate_push_avail, eliminate_insert): Move inside...
10052 (class eliminate_dom_walker): ... this class in preparation
10054 (fini_eliminate): Remove by merging with ...
10055 (eliminate): ... this function. Adjust for class changes.
10056 (pass_pre::execute): Remove fini_eliminate call.
10057 (pass_fre::execute): Likewise.
10059 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10062 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10063 (VPERMI2, VPERMI2I): New mode iterators.
10064 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10065 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10067 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10068 mode iterator. Remove 3 old define_insn patterns.
10069 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10070 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10071 VPERMI2 mode iterator, remove the other two expanders.
10072 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10073 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10074 instructions, remove the other two patterns.
10075 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10076 mode iterator, remove the other two patterns.
10077 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10078 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10079 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10080 and adjust argument order accordingly.
10081 (ix86_expand_vec_perm): Adjust caller.
10082 (expand_vec_perm_1): Likewise.
10083 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10084 (expand_vec_perm_vpermt2_vpshub2): ... this.
10085 (ix86_expand_vec_perm_const_1): Adjust caller.
10086 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10089 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10090 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10091 (vec_shl_<mode>): Remove unused expander.
10092 (avx512bw_<shift_insn><mode>3): New define_insn.
10093 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10094 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10096 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10099 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10102 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10104 PR rtl-optimization/82396
10105 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
10106 (autopref_multipass_init): Simplify initialization.
10107 (autopref_rank_data): Simplify sort order.
10108 * gcc/sched-int.h (autopref_multipass_data_): Remove
10109 multi_mem_insn_p, min_offset and max_offset.
10111 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10113 PR middle-end/60580
10114 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10115 Check special value of flag_omit_frame_pointer.
10116 (aarch64_can_eliminate): Likewise.
10117 (aarch64_override_options_after_change_1): Simplify handling of
10118 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
10120 2017-10-24 Richard Biener <rguenther@suse.de>
10122 PR tree-optimization/82697
10123 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
10124 zero for conditional load and unconditional store.
10126 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10128 * doc/install.texi: Document bootstrap-cet.
10130 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10133 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
10134 ENDBR instruction at function entrance if function is only
10137 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10140 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
10141 patterns to better describe from which operation the CF is computed.
10142 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
10143 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
10144 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
10145 is 0, use _0 suffixed expanders instead of emitting a comparison
10148 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
10150 * config/i386/i386.md(*movsf_internal, *movdf_internal):
10151 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
10153 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
10155 PR middle-end/82569
10156 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
10157 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
10158 * loop-iv.c (iv_get_reaching_def): Likewise.
10159 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
10160 variable is promoted and the partition contains undefined values.
10162 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10164 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
10166 (nios2_address_cost): Define.
10167 (nios2_legitimize_address): Recognize (exp + constant) directly.
10168 (TARGET_ADDRESS_COST): Define.
10170 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10172 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
10173 (nios2_symbolic_memory_operand_p): Declare.
10174 (nios2_split_large_constant): Declare.
10175 (nios2_split_symbolic_memory_operand): Declare.
10176 * config/nios2/nios2.c: Adjust includes.
10177 (nios2_symbolic_constant_allowed): New.
10178 (nios2_symbolic_constant_p): New.
10179 (nios2_plus_symbolic_constant_p): New.
10180 (nios2_valid_addr_expr_p): Recognize addresses involving
10181 symbolic constants.
10182 (nios2_legitimate_address_p): Likewise, also LO_SUM.
10183 (nios2_symbolic_memory_operand_p): New.
10184 (nios2_large_constant_p): New.
10185 (nios2_split_large_constant): New.
10186 (nios2_split_plus_large_constant): New.
10187 (nios2_split_symbolic_memory_operand): New.
10188 (nios2_legitimize_address): Code refactoring. Handle addresses
10189 involving symbolic constants.
10190 (nios2_emit_move_sequence): Likewise.
10191 (nios2_print_operand): Improve error output.
10192 (nios2_print_operand_address): Handle LO_SUM.
10193 (nios2_cdx_narrow_form_p): Likewise.
10194 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
10195 operands involving symbolic constants.
10196 (movhi_internal, movsi_internal): Likewise.
10197 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
10198 (extendhisi2, extendqi<mode>2): Likewise.
10200 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10202 * tree-pass.h (PROP_rtl_split_insns): Define.
10203 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
10205 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10207 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
10209 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10212 * target.def (const_not_ok_for_debug_p): Default to
10213 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
10214 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
10215 * targhooks.c (default_const_not_ok_for_debug_p): New function.
10216 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
10217 which targetm.const_not_ok_for_debug_p returned true.
10218 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
10220 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
10222 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
10223 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
10224 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
10226 (ix86_const_not_ok_for_debug_p): New function.
10227 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
10228 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
10230 2017-10-23 David Malcolm <dmalcolm@redhat.com>
10233 * system.h: Conditionally include "unique-ptr.h" if
10234 INCLUDE_UNIQUE_PTR is defined.
10235 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
10236 of defining INCLUDE_UNIQUE_PTR before including "system.h".
10238 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
10240 * config/rl78/rl78.md: New define_expand "subdi3".
10242 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
10245 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
10246 DF_REF_INSN if DF_REF_INSN_INFO is false.
10248 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
10250 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
10251 xmm_move instead of sse_move.
10252 (sse_store_index): New function.
10253 (ix86_register_move_cost): Be more sensible about mismatch stall;
10254 model AVX moves correctly; make difference between sse->integer and
10256 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
10257 moves; make difference between SSE and AVX.
10258 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
10259 and zmm_move. Increase size of sse load and store tables;
10260 add unaligned load and store tables; add ssemmx_to_integer.
10261 * x86-tune-costs.h: Update all entries according to real
10262 move latencies from Agner Fog's manual and chip documentation.
10264 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10267 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
10268 * config/i386/constraints.md (Wf): New constraint.
10269 * config/i386/i386.md (UNSPEC_SBB): New unspec.
10270 (cmp<dwi>_doubleword): Removed.
10271 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
10272 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
10273 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
10274 expand with cmp<dwi>_doubleword. For LTU and GEU use
10275 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
10277 * common.opt (gcolumn-info): Enable by default.
10278 * doc/invoke.texi (gcolumn-info): Document new default.
10280 2017-10-23 Richard Biener <rguenther@suse.de>
10282 PR tree-optimization/82672
10283 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
10284 Fold the stmt if we propagated into it.
10286 2017-10-23 Richard Biener <rguenther@suse.de>
10288 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
10289 (bitmap_remove_expr_from_set): ... this. All callers call this
10290 for non-constant values.
10291 (bitmap_set_subtract): Rename to...
10292 (bitmap_set_subtract_expressions): ... this. Adjust and
10294 (bitmap_set_contains_value): Remove superfluous check.
10295 (bitmap_set_replace_value): Inline into single caller ...
10296 (bitmap_value_replace_in_set): ... here and simplify.
10297 (dependent_clean): Merge into ...
10298 (clean): ... this using an overload. Adjust.
10299 (prune_clobbered_mems): Adjust.
10300 (compute_antic_aux): Likewise.
10301 (compute_partial_antic_aux): Likewise.
10303 2017-10-23 Richard Biener <rguenther@suse.de>
10305 PR tree-optimization/82129
10307 2017-08-01 Richard Biener <rguenther@suse.de>
10309 PR tree-optimization/81181
10310 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10311 (compute_antic): ... end of iteration here.
10313 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10315 * target.def (starting_frame_offset): New hook.
10316 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
10317 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
10318 * doc/tm.texi.in: Regenerate.
10319 * hooks.h (hook_hwi_void_0): Declare.
10320 * hooks.c (hook_hwi_void_0): New function.
10321 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
10322 STARTING_FRAME_OFFSET.
10323 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
10324 * reload1.c (reload): Likewise.
10325 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
10326 instead of STARTING_FRAME_OFFSET.
10327 * function.c (try_fit_stack_local): Likewise.
10328 (assign_stack_local_1): Likewise
10329 (instantiate_virtual_regs): Likewise.
10330 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
10331 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
10332 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
10333 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
10334 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
10335 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
10336 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
10337 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
10338 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
10339 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
10340 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
10341 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
10342 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
10343 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
10344 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
10345 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
10346 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
10347 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
10348 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
10349 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
10350 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
10351 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
10352 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
10353 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
10354 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
10355 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
10356 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
10357 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
10358 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
10359 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
10360 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
10361 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
10362 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
10363 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
10364 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
10365 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
10366 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
10367 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
10368 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
10369 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
10370 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
10371 * config/avr/avr.c (avr_starting_frame_offset): Make static and
10372 return a HOST_WIDE_INT.
10373 (avr_builtin_setjmp_frame_value): Use it instead of
10374 STARTING_FRAME_OFFSET.
10375 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10376 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
10377 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
10379 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10380 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
10381 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
10382 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10383 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
10384 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
10385 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10386 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
10387 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
10388 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10389 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
10390 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
10392 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10393 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
10394 * config/mips/mips.c (mips_compute_frame_info): Refer to
10395 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10396 (mips_starting_frame_offset): New function.
10397 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10398 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
10399 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
10400 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
10401 and return a HOST_WIDE_INT.
10402 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10403 (mmix_initial_elimination_offset): Refer to
10404 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10405 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
10406 * config/pa/pa.c (pa_starting_frame_offset): New function.
10407 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
10408 (pa_expand_prologue): Likewise.
10409 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10410 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
10411 !FRAME_GROWS_DOWNWARD handling to...
10412 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10413 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
10414 !FRAME_GROWS_DOWNWARD handling to...
10415 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10416 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
10417 !FRAME_GROWS_DOWNWARD handling to...
10418 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10419 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
10421 (rs6000_starting_frame_offset): New function.
10422 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
10423 !FRAME_GROWS_DOWNWARD handling to...
10424 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10425 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
10426 !FRAME_GROWS_DOWNWARD handling to...
10427 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10428 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
10429 !FRAME_GROWS_DOWNWARD handling to...
10430 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10431 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
10432 (rs6000_starting_frame_offset): New function.
10433 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
10434 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
10435 * config/vax/vax.c (vax_starting_frame_offset): New function.
10436 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
10437 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10438 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
10439 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
10440 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10441 * system.h (STARTING_FRAME_OFFSET): Poison.
10443 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10445 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10446 SCALAR_TYPE_MODE instead of TYPE_MODE.
10448 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10449 Alan Hayward <alan.hayward@arm.com>
10450 David Sherwood <david.sherwood@arm.com>
10452 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
10454 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10455 Alan Hayward <alan.hayward@arm.com>
10456 David Sherwood <david.sherwood@arm.com>
10458 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
10460 2017-10-23 Richard Biener <rguenther@suse.de>
10462 PR tree-optimization/82129
10463 * tree-ssa-pre.c (bitmap_set_and): Remove.
10464 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
10465 canonicalizing expressions in the set to those with lowest
10466 ID rather than taking that from the first edge.
10468 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10470 * combine.c (rtx_equal_for_field_assignment_p): Use
10471 byte_lowpart_offset.
10473 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10474 Alan Hayward <alan.hayward@arm.com>
10475 David Sherwood <david.sherwood@arm.com>
10477 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
10478 to a SUBREG_PROMOTED_VAR.
10480 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10481 Alan Hayward <alan.hayward@arm.com>
10482 David Sherwood <david.sherwood@arm.com>
10484 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
10485 (expand_debug_source_expr): Likewise.
10486 * combine.c (combine_simplify_rtx): Likewise.
10487 * cse.c (fold_rtx): Likewise.
10488 * optabs.c (expand_float): Likewise.
10489 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10490 (simplify_binary_operation_1): Likewise.
10492 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10493 Alan Hayward <alan.hayward@arm.com>
10494 David Sherwood <david.sherwood@arm.com>
10496 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
10497 (record_promoted_value): Likewise.
10498 * expr.c (expand_expr_real_2): Likewise.
10499 * ree.c (update_reg_equal_equiv_notes): Likewise.
10500 (combine_set_extension): Likewise.
10501 * rtlanal.c (low_bitmask_len): Likewise.
10502 * simplify-rtx.c (neg_const_int): Likewise.
10503 (simplify_binary_operation_1): Likewise.
10505 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10506 Alan Hayward <alan.hayward@arm.com>
10507 David Sherwood <david.sherwood@arm.com>
10509 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
10510 * regcprop.c (maybe_mode_change): Likewise.
10511 * reload1.c (alter_reg): Likewise.
10513 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10515 * inchash.h (inchash::hash::add_wide_int): New function.
10516 * lto-streamer-out.c (hash_tree): Use it.
10518 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10520 * inchash.h (inchash::hash::add_wide_int): Rename to...
10521 (inchash::hash::add_hwi): ...this.
10522 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
10523 (polymorphic_call_target_hasher::hash): Likewise.
10524 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
10525 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
10526 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
10527 * lto-streamer-out.c (hash_tree): Likewise.
10528 * optc-save-gen.awk: Likewise.
10529 * tree.c (add_expr): Likewise.
10531 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10534 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
10535 for ordered inequality comparisons even with TARGET_IEEE_FP.
10537 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10540 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
10541 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10542 Expand with cmp<dwi>_doubleword.
10544 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10546 * extend.texi: Add x86 specific to 'nocf_check' attribute.
10547 List CET intrinsics.
10548 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
10549 specific to -fcf-protection option.
10551 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10553 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
10554 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
10555 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
10556 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
10557 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
10558 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
10559 (extra_objs): Add cet.o for Linux/x86 targets.
10560 (tmake_file): Add i386/t-cet for Linux/x86 targets.
10561 * config/i386/cet.c: New file.
10562 * config/i386/cetintrin.h: Likewise.
10563 * config/i386/t-cet: Likewise.
10564 * config/i386/cpuid.h (bit_SHSTK): New.
10565 (bit_IBT): Likewise.
10566 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
10567 pass IBT and SHSTK bits.
10568 * config/i386/i386-builtin-types.def
10569 (VOID_FTYPE_UNSIGNED_PVOID): New.
10570 (VOID_FTYPE_UINT64_PVOID): Likewise.
10571 * config/i386/i386-builtin.def: Add CET intrinsics.
10572 * config/i386/i386-c.c (ix86_target_macros_internal): Add
10573 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
10574 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
10575 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
10577 * config/i386/i386.c (rest_of_insert_endbranch): New.
10578 (pass_data_insert_endbranch): Likewise.
10579 (pass_insert_endbranch): Likewise.
10580 (make_pass_insert_endbranch): Likewise.
10581 (ix86_notrack_prefixed_insn_p): Likewise.
10582 (ix86_target_string): Add -mibt, -mshstk flags.
10583 (ix86_option_override_internal): Add flag_cf_protection
10585 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
10586 (ix86_print_operand): Add 'notrack' prefix output.
10587 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
10588 (ix86_expand_builtin): Expand CET intrinsics.
10589 (x86_output_mi_thunk): Add 'endbranch' instruction.
10590 * config/i386/i386.h (TARGET_IBT): New.
10591 (TARGET_IBT_P): Likewise.
10592 (TARGET_SHSTK): Likewise.
10593 (TARGET_SHSTK_P): Likewise.
10594 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
10595 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
10596 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
10597 (builtin_setjmp_setup): New pattern.
10598 (builtin_longjmp): Likewise.
10599 (rdssp<mode>): Likewise.
10600 (incssp<mode>): Likewise.
10601 (saveprevssp): Likewise.
10602 (rstorssp): Likewise.
10603 (wrss<mode>): Likewise.
10604 (wruss<mode>): Likewise.
10605 (setssbsy): Likewise.
10606 (clrssbsy): Likewise.
10607 (nop_endbr): Likewise.
10608 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
10610 * config/i386/immintrin.h: Include <cetintrin.h>.
10611 * config/i386/linux-common.h
10612 (file_end_indicate_exec_stack_and_cet): New prototype.
10613 (TARGET_ASM_FILE_END): New.
10615 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10617 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10618 latencies instead of having separate table; make difference between
10619 integer and float costs.
10620 * i386.h (processor_costs): Remove scalar_stmt_cost,
10621 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10622 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10624 * x86-tune-costs.h: Remove entries which has been removed in
10625 procesor_costs from all tables; make cond_taken_branch_cost
10626 and cond_not_taken_branch_cost COST_N_INSNS based.
10628 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10630 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10632 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10634 * config/i386/i386.md (isa): Remove fma_avx512f.
10635 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10636 <avx512>_fmadd_<mode>_mask3<round_name>,
10637 <avx512>_fmsub_<mode>_mask<round_name>,
10638 <avx512>_fmsub_<mode>_mask3<round_name>,
10639 <avx512>_fnmadd_<mode>_mask<round_name>,
10640 <avx512>_fnmadd_<mode>_mask3<round_name>,
10641 <avx512>_fnmsub_<mode>_mask<round_name>,
10642 <avx512>_fnmsub_<mode>_mask3<round_name>,
10643 <avx512>_fmaddsub_<mode>_mask<round_name>,
10644 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10645 <avx512>_fmsubadd_<mode>_mask<round_name>,
10646 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10647 (*vec_widen_umult_even_v16si<mask_name>,
10648 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10649 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10651 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10653 * extend.texi: Add 'nocf_check' documentation.
10654 * gimple.texi: Add second parameter to
10655 gimple_build_call_from_tree.
10656 * invoke.texi: Add -fcf-protection documentation.
10657 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10659 2017-10-20 Richard Biener <rguenther@suse.de>
10661 PR tree-optimization/82473
10662 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10663 the largest input type.
10665 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10667 * c-attribs.c (handle_nocf_check_attribute): New function.
10668 (c_common_attribute_table): Add 'nocf_check' handling.
10669 * gimple-parser.c: Add second argument NULL to
10670 gimple_build_call_from_tree.
10671 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10672 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10674 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10675 * common.opt: Add fcf-protection flag.
10676 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10677 * flag-types.h: Add enum cf_protection_level.
10678 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10679 Add 'nocf_check' attribute propagation to gimple call.
10680 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10681 (gimple_build_call_from_tree): Update prototype.
10682 (gimple_call_nocf_check_p): New function.
10683 (gimple_call_set_nocf_check): Likewise.
10684 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10685 * ipa-icf.c: Add nocf_check attribute in statement hash.
10686 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10687 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10688 * toplev.c (process_options): Add flag_cf_protection handling.
10690 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10692 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10694 2017-10-20 Richard Biener <rguenther@suse.de>
10696 PR tree-optimization/82603
10697 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10698 remove false predicated stores.
10700 2017-10-20 Richard Biener <rguenther@suse.de>
10702 * graphite-isl-ast-to-gimple.c
10703 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10704 Remove return value and simplify, dump copied stmt after lhs
10706 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10707 Reduce dump verbosity.
10708 (gsi_insert_earliest): Likewise.
10709 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10710 * graphite.c (print_global_statistics): Adjust dumping.
10711 (print_graphite_scop_statistics): Likewise.
10712 (print_graphite_statistics): Do not dump loops here.
10713 (graphite_transform_loops): But here.
10715 2017-10-20 Nicolas Roche <roche@adacore.com>
10717 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10718 * configure: Regenerate.
10720 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10723 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10724 functions when optimizing replace GIMPLE_RETURN stmts with
10725 calls to __builtin_unreachable ().
10728 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10729 for -fsanitize=thread link of executables.
10730 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10731 link of executables.
10734 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10735 New mode iterators.
10736 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10737 define_insns for logical vector shifts to use VI248_AVX512BW
10738 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10739 condition, useless isa and prefix attributes. Change the first
10740 2 of these define_insns to ...
10741 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10742 define_insn for avx512vl.
10743 (<shift_insn><mode>3): ... and this, new define_insn without
10744 masking for non-avx512vl.
10747 * config/i386/sse.md (*andnot<mode>3,
10748 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10749 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10750 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10751 not applied use empty suffix even for TARGET_AVX512VL.
10752 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10753 is applied, supply evex,evex or evex,evex,evex instead of just
10756 2017-10-20 Julia Koval <julia.koval@intel.com>
10758 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10759 (OPTION_MASK_ISA_GFNI_UNSET): New.
10760 (ix86_handle_option): Handle OPT_mgfni.
10761 * config/i386/cpuid.h (bit_GFNI): New.
10762 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10763 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10764 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10765 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10766 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10767 * config/i386/i386.opt: Add mgfni.
10769 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10771 * config/msp430/msp430.c (msp430_option_override): Disable
10772 -fdelete-null-pointer-checks.
10773 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10775 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10777 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10778 of x87 and SSE instructions.
10780 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10782 * asan.c (create_cond_insert_point): Do not update edge count.
10783 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10784 (afdo_propagate_circuit): Likewise.
10785 (afdo_calculate_branch_prob): Likewise.
10786 (afdo_annotate_cfg): Likewise.
10787 * basic-block.h (struct edge_def): Remove count.
10788 (edge_def::count): New accessor.
10789 * bb-reorder.c (rotate_loop): Update.
10790 (find_traces_1_round): Update.
10791 (connect_traces): Update.
10792 (sanitize_hot_paths): Update.
10793 * cfg.c (unchecked_make_edge): Update.
10794 (make_single_succ_edge): Update.
10795 (check_bb_profile): Update.
10796 (dump_edge_info): Update.
10797 (update_bb_profile_for_threading): Update.
10798 (scale_bbs_frequencies_int): Update.
10799 (scale_bbs_frequencies_gcov_type): Update.
10800 (scale_bbs_frequencies_profile_count): Update.
10801 (scale_bbs_frequencies): Update.
10802 * cfganal.c (connect_infinite_loops_to_exit): Update.
10803 * cfgbuild.c (compute_outgoing_frequencies): Update.
10804 (find_many_sub_basic_blocks): Update.
10805 * cfgcleanup.c (try_forward_edges): Update.
10806 (try_crossjump_to_edge): Update
10807 * cfgexpand.c (expand_gimple_cond): Update
10808 (expand_gimple_tailcall): Update
10809 (construct_exit_block): Update
10810 * cfghooks.c (verify_flow_info): Update
10811 (redirect_edge_succ_nodup): Update
10812 (split_edge): Update
10813 (make_forwarder_block): Update
10814 (duplicate_block): Update
10815 (account_profile_record): Update
10816 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10817 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10818 * cfgloopmanip.c (scale_loop_profile): Update.
10820 (lv_adjust_loop_entry_edge): Update.
10821 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10822 (force_nonfallthru_and_redirect): Update.
10823 (purge_dead_edges): Update.
10824 (rtl_flow_call_edges_add): Update.
10825 * cgraphunit.c (init_lowered_empty_function): Update.
10826 (cgraph_node::expand_thunk): Update.
10827 * gimple-pretty-print.c (dump_probability): Update.
10828 (dump_edge_probability): Update.
10829 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10830 * haifa-sched.c (sched_create_recovery_edges): Update.
10831 * hsa-gen.c (convert_switch_statements): Update.
10832 * ifcvt.c (dead_or_predicable): Update.
10833 * ipa-inline-transform.c (inline_transform): Update.
10834 * ipa-split.c (split_function): Update.
10835 * ipa-utils.c (ipa_merge_profiles): Update.
10836 * loop-doloop.c (add_test): Update.
10837 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10838 * lto-streamer-in.c (input_cfg): Update.
10839 (input_function): Update.
10840 * lto-streamer-out.c (output_cfg): Update.
10841 * modulo-sched.c (sms_schedule): Update.
10842 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10843 * predict.c (maybe_hot_edge_p): Update.
10844 (unlikely_executed_edge_p): Update.
10845 (probably_never_executed_edge_p): Update.
10846 (dump_prediction): Update.
10847 (drop_profile): Update.
10848 (propagate_unlikely_bbs_forward): Update.
10849 (determine_unlikely_bbs): Update.
10850 (force_edge_cold): Update.
10851 * profile.c (compute_branch_probabilities): Update.
10852 * reg-stack.c (better_edge): Update.
10853 * shrink-wrap.c (handle_simple_exit): Update.
10854 * tracer.c (better_p): Update.
10855 * trans-mem.c (expand_transaction): Update.
10856 (split_bb_make_tm_edge): Update.
10857 * tree-call-cdce.c: Update.
10858 * tree-cfg.c (gimple_find_sub_bbs): Update.
10859 (gimple_split_edge): Update.
10860 (gimple_duplicate_sese_region): Update.
10861 (gimple_duplicate_sese_tail): Update.
10862 (gimple_flow_call_edges_add): Update.
10863 (insert_cond_bb): Update.
10864 (execute_fixup_cfg): Update.
10865 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
10866 * tree-complex.c (expand_complex_div_wide): Update.
10867 * tree-eh.c (lower_resx): Update.
10868 (unsplit_eh): Update.
10869 (cleanup_empty_eh_move_lp): Update.
10870 * tree-inline.c (copy_edges_for_bb): Update.
10871 (freqs_to_counts): Update.
10872 (copy_cfg_body): Update.
10873 * tree-ssa-dce.c (remove_dead_stmt): Update.
10874 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10875 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10876 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10877 (unloop_loops): Update.
10878 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10879 * tree-ssa-loop-split.c (connect_loops): Update.
10880 (split_loop): Update.
10881 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10882 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10883 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10884 * tree-ssa-reassoc.c (branch_fixup): Update.
10885 * tree-ssa-tail-merge.c (replace_block_by): Update.
10886 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10887 (compute_path_counts): Update.
10888 (update_profile): Update.
10889 (recompute_probabilities): Update.
10890 (update_joiner_offpath_counts): Update.
10891 (estimated_freqs_path): Update.
10892 (freqs_to_counts_path): Update.
10893 (clear_counts_path): Update.
10894 (ssa_fix_duplicate_block_edges): Update.
10895 (duplicate_thread_path): Update.
10896 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10897 (case_bit_test_cmp): Update.
10898 (collect_switch_conv_info): Update.
10899 (gen_inbound_check): Update.
10900 (do_jump_if_equal): Update.
10901 (emit_cmp_and_jump_insns): Update.
10902 * tree-tailcall.c (decrease_profile): Update.
10903 (eliminate_tail_call): Update.
10904 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10905 (vect_do_peeling): Update.
10906 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10907 * ubsan.c (ubsan_expand_null_ifn): Update.
10908 (ubsan_expand_ptr_ifn): Update.
10909 * value-prof.c (gimple_divmod_fixed_value): Update.
10910 (gimple_mod_pow2): Update.
10911 (gimple_mod_subtract): Update.
10912 (gimple_ic): Update.
10913 (gimple_stringop_fixed_value): Update.
10915 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10918 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
10920 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
10922 PR rtl-optimization/82395
10923 * ira-color.c (allocno_priority_compare_func): Fix comparison step
10924 based on non_spilled_static_chain_regno_p.
10926 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10928 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
10929 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
10930 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
10932 2017-10-19 Martin Sebor <msebor@redhat.com>
10934 PR tree-optimization/82596
10935 * tree.c (array_at_struct_end_p): Handle STRING_CST.
10937 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10939 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
10940 (get_mem_refs_of_builtin_call): Likewise.
10941 * builtins.c (expand_builtin_apply): Adjust call to
10942 allocate_dynamic_stack_space.
10943 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
10944 the third argument to allocate_dynamic_stack_space, otherwise -1.
10945 (expand_builtin): Deal with all alloca variants.
10946 (is_inexpensive_builtin): Likewise.
10947 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
10948 * calls.c (special_function_p): Deal with all alloca variants.
10949 (initialize_argument_information): Adjust call to
10950 allocate_dynamic_stack_space.
10951 (expand_call): Likewise.
10952 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
10953 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
10954 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
10955 use it for the stack usage computation.
10956 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
10957 * function.c (gimplify_parameters): Call build_alloca_call_expr.
10958 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
10959 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
10960 (in_loop_p): Remove first argument and useless check.
10961 (pass_walloca::execute): Remove useless test and adjust call to above.
10962 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
10963 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
10964 (gimplify_call_expr): Deal with all alloca variants.
10965 * hsa-gen.c (gen_hsa_alloca): Likewise.
10966 (gen_hsa_insns_for_call): Likewise.
10967 * ipa-pure-const.c (special_builtin_state): Likewise.
10968 * tree-chkp.c (chkp_build_returned_bound): Likewise.
10969 * tree-object-size.c (alloc_object_size): Likewise.
10970 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
10971 (call_may_clobber_ref_p_1): Likewise.
10972 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
10973 (ccp_fold_stmt): Likewise.
10974 (optimize_stack_restore): Likewise.
10975 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10976 (mark_all_reaching_defs_necessary_1): Likewise.
10977 (propagate_necessity): Likewise.
10978 (eliminate_unnecessary_stmts): Likewise.
10979 * tree.c (build_common_builtin_nodes): Build
10980 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
10981 (build_alloca_call_expr): New function.
10982 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
10983 (CASE_BUILT_IN_ALLOCA): Likewise.
10984 (build_alloca_call_expr): Declare.
10985 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
10987 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10990 * dwarf2out.c (new_die_raw): New static inline function.
10991 (new_die): Use it to create the DIE.
10992 (add_AT_external_die_ref): Likewise.
10993 (clone_die): Likewise.
10994 (clone_as_declaration): Likewise.
10995 (dwarf2out_vms_debug_main_pointer): Likewise.
10996 (base_type_die): Likewise. Remove early return for corner cases.
10997 Do not call add_pubtype on the DIE here.
10998 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
10999 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11000 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11001 native order exists for base types, attach the DIE manually and call
11002 add_pubtype on it. Do not equate a reverse order DIE to the type.
11004 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11006 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11007 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11009 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11011 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11013 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11014 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11015 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11016 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11017 * opts.c (sanitizer_opts): Add builtin.
11018 * ubsan.c (instrument_builtin): New function.
11019 (pass_ubsan::execute): Call it.
11020 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11021 * doc/invoke.texi: Document -fsanitize=builtin.
11023 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11024 builtins, store max (log2 (align), 0) into uchar field instead of
11025 align into uptr field.
11026 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11027 store uchar 0 field instead of uptr 0 field.
11028 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11029 instead of passing one address of struct with 2 locations pass
11030 two addresses of structs with 1 location each.
11031 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11032 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11033 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11034 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11035 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11036 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11037 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
11038 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
11040 2017-10-19 Martin Liska <mliska@suse.cz>
11043 * file-find.c (remove_prefix): Remove.
11044 * file-find.h (remove_prefix): Likewise.
11045 * gcc-ar.c: Remove smartness of lookup.
11047 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11049 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11050 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11051 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11053 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11056 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11057 (setcc + and to xor + setcc): New peephole2.
11059 2017-10-19 Tom de Vries <tom@codesourcery.com>
11061 * doc/sourcebuild.texi (Test Directives, Variants of
11062 dg-require-support): Add dg-require-stack-size.
11064 2017-10-19 Martin Liska <mliska@suse.cz>
11067 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11068 that have a large alignment.
11069 (gimplify_target_expr): Likewise.
11071 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11073 PR rtl-optimization/82602
11074 * ira.c (rtx_moveable_p): Return false for volatile asm.
11076 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11079 * config/i386/i386-modes.def (CCGZ): New CC mode.
11080 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11081 * config/i386/predicates.md (ix86_comparison_operator):
11083 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11084 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11085 with double-word subtraction.
11086 (put_condition_code): Handle CCGZmode.
11088 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11090 * wide-int.cc (debug (const wide_int &)): New.
11091 (debug (const wide_int *)): New.
11092 (debug (const widest_int &)): New.
11093 (debug (const widest_int *)): New.
11095 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11097 PR middle-end/82556
11098 * lra-constraints.c (curr_insn_transform): Use non-input operand
11099 instead of output one for matched reload.
11101 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11103 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11104 (tree-ssa-loop-ivopts.h): New header file.
11105 (struct builtin_info): New fields.
11106 (classify_builtin_1): Compute and record base and offset parts for
11107 memset builtin partition by calling strip_offset.
11108 (offset_cmp, fuse_memset_builtins): New functions.
11109 (finalize_partitions): Fuse adjacent memset partitions by calling
11111 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
11112 Expose the interface.
11113 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
11115 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11117 PR tree-optimization/82574
11118 * tree-loop-distribution.c (find_single_drs): New parameter. Check
11119 that data reference must be executed exactly once per iteration
11120 against the outermost loop in nest.
11121 (classify_partition): Update call to above function.
11123 2017-10-18 Richard Biener <rguenther@suse.de>
11125 PR tree-optimization/82591
11126 * graphite.c (graphite_transform_loops): Move code gen message
11128 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11129 Here. Handle scop_to_isl_ast failing.
11130 (scop_to_isl_ast): Limit the number of ISL operations.
11132 2017-10-18 Richard Biener <rguenther@suse.de>
11134 * graphite-isl-ast-to-gimple.c
11135 (translate_isl_ast_to_gimple::set_rename): Simplify.
11136 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
11137 (graphite_copy_stmts_from_block): ... here.
11138 (copy_bb_and_scalar_dependences): Simplify.
11139 (add_parameters_to_ivs_params): Canonicalize.
11140 (generate_entry_out_of_ssa_copies): Simplify.
11141 * graphite-sese-to-poly.c (extract_affine_name): Simplify
11142 by passing in ISL dimension.
11143 (parameter_index_in_region_1): Rename to ...
11144 (parameter_index_in_region): ... this.
11145 (extract_affine): Adjust assert, pass down parameter index.
11146 (add_param_constraints): Use range-info when available.
11147 (build_scop_context): Adjust.
11148 * sese.c (new_sese_info): Adjust.
11149 (free_sese_info): Likewise.
11150 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
11151 Remove unused typedefs.
11152 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
11154 2017-10-18 Martin Liska <mliska@suse.cz>
11156 * combine.c (simplify_compare_const): Add gcc_fallthrough.
11158 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11160 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
11161 (s390_sched_init): Do not reset s390_sched_state if we entered the
11162 current basic block via a fallthru edge and all others are unlikely.
11164 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11166 * config/s390/s390.c (NUM_SIDES): New variable.
11167 (LONGRUNNING_THRESHOLD): New variable.
11168 (LATENCY_FACTOR): New variable.
11169 (s390_sched_score): Decrease score for long-running instructions on
11171 (s390_sched_variable_issue): Perform bookkeeping for long-running
11174 2017-10-18 Richard Biener <rguenther@suse.de>
11176 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
11177 Simplify with removal of the parameter rename map.
11178 (set_rename): Likewise.
11179 (should_copy_to_new_region): Likewise.
11180 (graphite_copy_stmts_from_block): Likewise.
11181 (copy_bb_and_scalar_dependences): Remove initialization of
11182 unused copied_bb_map.
11183 (copy_def): Remove.
11184 (copy_internal_parameters): Likewise.
11185 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
11186 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
11187 Use INTEGRAL_TYPE_P.
11188 (parameter_index_in_region_1): Rename to ...
11189 (assign_parameter_index_in_region): ... this. Assert we have
11190 a parameter we handle.
11191 (scan_tree_for_params): Adjust.
11192 * sese.h (parameter_rename_map_t): Remove.
11193 (struct sese_info_t): Remove unused parameter_rename_map and
11194 copied_bb_map members.
11195 * sese.c (new_sese_info): Adjust.
11196 (free_sese_info): Likewise.
11198 2017-10-18 Martin Liska <mliska@suse.cz>
11201 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
11202 on an abnormal edge.
11204 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
11206 * doc/invoke.texi (ffunction-sections and fdata-sections):
11209 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11211 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
11212 the use statement can throw internally.
11214 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11216 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
11217 any RTX present on the RHS of a SET.
11218 * compare-elim.c (try_eliminate_compare): Restore comment.
11220 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11222 * langhooks.h (struct lang_hooks): Document that tree_size langhook
11223 may be also called on tcc_type nodes.
11224 * langhooks.c (lhd_tree_size): Likewise.
11226 2017-10-17 David Malcolm <dmalcolm@redhat.com>
11228 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
11229 format_warning_at_substring.
11230 (maybe_warn): Convert source_range * param to a location_t. Pass
11231 UNKNOWN_LOCATION rather than NULL to fmtwarn.
11232 (format_directive): Remove code to extract source_ranges and
11233 source_range * in favor of just a location_t.
11234 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
11236 * substring-locations.c (format_warning_va): Convert
11237 source_range * param to a location_t.
11238 (format_warning_at_substring): Likewise.
11239 * substring-locations.h (format_warning_va): Likewise.
11240 (format_warning_at_substring): Likewise.
11242 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
11244 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
11246 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
11247 and scatter/gather ops.
11249 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
11250 vec_gather_load and vec_scatter_store.
11251 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
11252 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
11253 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
11254 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
11255 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
11256 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
11258 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
11260 * reg-stack.c (compare_for_stack_reg): Add bool argument.
11261 Detect FTST instruction and handle its register pops. Only pop
11262 second operand if can_pop_second_op is true.
11263 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
11264 set can_pop_second_op to false in the compare_for_stack_reg call.
11266 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
11267 output_fp_compare for stack register operands.
11268 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
11269 instructions here. Do not emit stack register pops here. Assert
11270 that FCOMPP pops next to top stack register. Rewrite function.
11272 2017-10-17 Nathan Sidwell <nathan@acm.org>
11274 PR middle-end/82577
11275 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
11276 use DECL_ASSEMBLER_NAME_RAW.
11278 PR middle-end/82546
11279 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
11282 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
11283 Wilco Dijkstra <wilco.dijkstra@arm.com>
11285 * builtins.c (expand_builtin_update_setjmp_buf): Add a
11286 converstion to Pmode from the buf_addr.
11288 2017-10-17 Richard Biener <rguenther@suse.de>
11290 * graphite-dependences.c (scop_get_reads_and_writes): Change
11291 output parameters to references.
11293 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
11295 PR 71026/tree-optimization
11296 * fold-const.c (distribute_real_division): Removed.
11297 (fold_binary_loc): Remove calls to distribute_real_divison.
11299 2017-10-17 Richard Biener <rguenther@suse.de>
11301 * graphite-scop-detection.c
11302 (scop_detection::stmt_has_simple_data_refs_p): Always use
11303 the full nest as region.
11304 (try_generate_gimple_bb): Likewise.
11305 * sese.c (scalar_evolution_in_region): Simplify now that
11306 SCEV can handle instantiation in regions.
11307 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
11308 in the non-loop part of a function if requested.
11310 2017-10-17 Richard Biener <rguenther@suse.de>
11312 PR tree-optimization/82563
11313 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
11315 (graphite_regenerate_ast_isl): Call it.
11316 * graphite-scop-detection.c (build_scops): Remove entry edge split.
11318 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11320 PR tree-optimization/82549
11321 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
11322 Formatting fixes. Instead of calling make_bit_field_ref with negative
11325 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
11327 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
11328 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
11329 _mm_maskz_reduce_ss): New.
11330 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
11331 __builtin_ia32_reducess_mask): Ditto..
11332 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
11333 * config/i386/sse.md (reduces<mode>): Renamed to ...
11334 (reduces<mode><mask_scalar_name>): ... this.
11335 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
11337 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
11338 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
11340 2017-10-16 David Malcolm <dmalcolm@redhat.com>
11342 * Makefile.in (OBJS): Add unique-ptr-tests.o.
11343 * selftest-run-tests.c (selftest::run_tests): Call
11344 selftest::unique_ptr_tests_cc_tests.
11345 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
11346 * unique-ptr-tests.cc: New file.
11348 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
11351 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11353 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11354 (make_hard_regno_born, make_hard_regno_dead): Update
11355 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
11357 2017-10-16 Jeff Law <law@redhat.com>
11359 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11361 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
11363 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
11365 2017-10-16 Olivier Hainque <hainque@adacore.com>
11367 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
11368 with_cpu if we were configured for an e500v2 target cpu name.
11370 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
11372 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
11373 * doc/invoke.texi: Document +nodsp as a valid extension for
11376 2017-10-16 Martin Liska <mliska@suse.cz>
11378 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
11379 (test_set_range): Likewise.
11380 (test_range_functions): Rename to ...
11381 (test_bit_in_range): ... this.
11382 (sbitmap_c_tests): Add new test.
11384 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11386 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
11388 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
11389 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
11391 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11393 * config/aarch64/aarch64-builtins.c
11394 (aarch64_types_quadopu_lane_qualifiers): New.
11395 (TYPES_QUADOPU_LANE): New.
11396 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
11397 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
11398 (aarch64_<sur>dot_laneq<vsi2qi>): New.
11399 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
11400 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
11401 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
11402 (Vdottype, DOTPROD): New.
11403 (sur): Add SDOT and UDOT.
11405 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11407 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
11408 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
11409 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
11410 Add TARGET_DOTPROD.
11411 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
11412 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
11413 Enable TARGET_DOTPROD.
11414 (cortex-a75.cortex-a55): Likewise.
11415 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
11417 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11419 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
11420 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
11422 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
11424 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
11425 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
11426 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
11427 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
11428 * config/arm/types.md (neon_dot, neon_dot_q): New.
11429 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
11431 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11433 * config/arm/arm.h (TARGET_DOTPROD): New.
11434 * config/arm/arm.c (arm_arch_dotprod): New.
11435 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
11436 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
11437 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
11438 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
11439 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
11440 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
11441 * doc/invoke.texi (armv8.2-a): Document dotprod
11443 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11445 * i386.c (ix86_vec_cost): New function.
11446 (ix86_rtx_costs): Handle vector operations better.
11447 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
11448 * x86-tune-costs.h: Add new costs to all tables.
11450 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11452 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
11454 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
11455 divsd, sqrtss and sqrtsd
11456 * x86-tune-costs.h: Add new entries to all costs.
11457 (znver1_cost): Fix to match real instruction latencies.
11459 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11460 Michael Collison <michael.collison@arm.com>
11462 * compare-elim.c: Include emit-rtl.h.
11463 (can_merge_compare_into_arith): New function.
11464 (try_validate_parallel): Likewise.
11465 (try_merge_compare): Likewise.
11466 (try_eliminate_compare): Call the above when no previous clobber
11468 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
11471 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11473 PR middle-end/62263
11474 PR middle-end/82498
11475 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
11476 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
11478 PR middle-end/62263
11479 PR middle-end/82498
11480 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
11481 to be any operand_equal_p operands. For & (B - 1) require
11482 B to be power of 2. Recognize
11483 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
11485 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11488 * optabs.c (expand_memory_blockage): Fix call of
11489 targetm.have_memory_blockage.
11491 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11494 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
11495 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
11496 objects to extra_objs instead of overwriting it.
11498 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11500 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
11501 Use any_fp_register_operand as operand[3] predicate. Simplify
11502 equality test for operands[2] and operands[4] memory location.
11503 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
11504 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
11505 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
11506 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
11507 any_fp_register_operand as operand[1] predicate. Simplify
11508 equality test for operands[0] and operands[3] memory location.
11509 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
11510 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
11511 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
11513 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11515 * target-insns.def: Add memory_blockage.
11516 * optabs.c (expand_memory_blockage): New function.
11517 (expand_asm_memory_barrier): Rename ...
11518 (expand_asm_memory_blockage): ... to this.
11519 (expand_mem_thread_fence): Call expand_memory_blockage
11520 instead of expand_asm_memory_barrier.
11521 (expand_mem_singnal_fence): Ditto.
11522 (expand_atomic_load): Ditto.
11523 (expand_atomic_store): Ditto.
11524 * doc/md.texi (Standard Pattern Names For Generation):
11525 Document memory_blockage instruction pattern.
11527 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
11529 * config/rl78/rl78.c (rl78_emit_libcall): New function.
11530 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
11531 * config/rl78/rl78.md: New define_expand "adddi3".
11533 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11535 * cfghooks.c (verify_flow_info): Disable check that all probabilities
11538 2017-10-13 Jeff Law <law@redhat.com>
11540 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
11542 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11545 * internal-fn.c (expand_mul_overflow): If both operands have
11546 the same highpart of -1 or 0 and the topmost bit of lowpart
11547 is different, overflow is if res <= 0 rather than res < 0.
11549 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
11551 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
11552 TARGET_P9_VECTOR code for unaligned_load case.
11554 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11556 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
11558 2017-10-13 Nathan Sidwell <nathan@acm.org>
11560 * tree-core.h (tree_contains_struct): Make bool.
11561 * tree.c (tree_contains_struct): Likewise.
11562 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11563 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
11564 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
11565 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
11566 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
11568 2017-10-13 Richard Biener <rguenther@suse.de>
11570 * graphite-isl-ast-to-gimple.c
11571 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
11572 parameters and dominance check.
11573 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
11574 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
11575 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
11576 Do not update SSA form here or do intermediate IL verification.
11577 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
11578 (graphite_initialize): Remove check on the number of loops in
11579 the function and inline into graphite_transform_loops.
11580 (graphite_finalize): Inline into graphite_transform_loops.
11581 (graphite_transform_loops): Perform SSA update and IL verification
11583 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
11585 2017-10-13 Richard Biener <rguenther@suse.de>
11587 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
11588 graphite_expression_type_precision): Avoid global constructor
11590 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
11591 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
11592 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
11593 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
11594 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
11595 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
11597 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
11600 * config/i386/i386.h (ix86_red_zone_size): New.
11601 * config/i386/i386.md (push peephole2s): Replace
11602 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
11604 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11605 Alan Hayward <alan.hayward@arm.com>
11606 David Sherwood <david.sherwood@arm.com>
11608 * combine.c (can_change_dest_mode): Reject changes in
11609 REGMODE_NATURAL_SIZE.
11611 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11612 Alan Hayward <alan.hayward@arm.com>
11613 David Sherwood <david.sherwood@arm.com>
11615 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
11616 (expand_debug_source_expr): Likewise.
11617 * combine.c (combine_simplify_rtx): Likewise.
11618 * cse.c (fold_rtx): Likewise.
11619 * fwprop.c (canonicalize_address): Likewise.
11620 * targhooks.c (default_shift_truncation_mask): Likewise.
11622 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11623 Alan Hayward <alan.hayward@arm.com>
11624 David Sherwood <david.sherwood@arm.com>
11626 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11627 (widened_mode): Likewise.
11628 (expand_unop): Likewise.
11629 * ree.c (transform_ifelse): Likewise.
11630 (merge_def_and_ext): Likewise.
11631 (combine_reaching_defs): Likewise.
11632 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11634 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11635 Alan Hayward <alan.hayward@arm.com>
11636 David Sherwood <david.sherwood@arm.com>
11638 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11639 * combine.c (gen_lowpart_for_combine): Likewise.
11640 * dwarf2out.c (rtl_for_decl_location): Likewise.
11641 * final.c (alter_subreg): Likewise.
11642 * rtlhooks.c (gen_lowpart_general): Likewise.
11643 (gen_lowpart_if_possible): Likewise.
11645 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11646 Alan Hayward <alan.hayward@arm.com>
11647 David Sherwood <david.sherwood@arm.com>
11649 * calls.c (expand_call): Use subreg_lowpart_offset.
11650 * cse.c (cse_insn): Likewise.
11651 * regcprop.c (copy_value): Likewise.
11652 (copyprop_hardreg_forward_1): Likewise.
11654 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11657 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11658 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11659 =Q constraints to +Q and into insn condition add check
11660 that operands[0] and operands[1] are equal.
11661 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11662 =Q constraints to +Q and into insn condition add check
11663 that operands[0] is equal to either operands[1] or operands[2].
11666 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11667 instead of handling MINUS_EXPR twice (once for each argument),
11668 canonicalize operand order and handle just once, use rtype where
11669 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11672 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11673 any values of __C while still being pattern recognizable as a simple
11674 rotate instruction.
11676 2017-10-13 Richard Biener <rguenther@suse.de>
11678 PR tree-optimization/82451
11680 2017-10-02 Richard Biener <rguenther@suse.de>
11682 PR tree-optimization/82355
11683 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11684 a mapping for the enclosing loop but avoid generating one for
11685 the loop tree root.
11686 (copy_bb_and_scalar_dependences): Remove premature codegen
11687 error on PHIs in blocks duplicated into multiple places.
11688 * graphite-scop-detection.c
11689 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11690 in the region use it as loop and nest to analyze the DR in.
11691 (try_generate_gimple_bb): Likewise.
11692 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11693 (add_loop_constraints): For blocks in a loop not in the region
11694 create a dimension with a single iteration.
11695 * sese.h (gbb_loop_at_index): Remove assert.
11697 * cfgloop.c (loop_preheader_edge): For the loop tree root
11698 return the single successor of the entry block.
11699 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11700 Reset the SCEV hashtable and niters.
11701 * graphite-scop-detection.c
11702 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11703 assert that we only have POLYNOMIAL_CHREC that vary in loops
11704 contained in the region.
11705 (scop_detection::graphite_can_represent_expr): Adjust.
11706 (scop_detection::stmt_has_simple_data_refs_p): For loops
11707 not in the region set loop to NULL. The nest is now the
11708 entry edge to the region.
11709 (try_generate_gimple_bb): Likewise.
11710 * sese.c (scalar_evolution_in_region): Adjust for
11711 instantiate_scev change.
11712 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11713 Make nest parameter the edge into the region.
11714 (create_data_ref): Likewise.
11715 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11716 entry edge into a region and adjust instantiate_scev calls.
11717 (create_data_ref): Likewise.
11718 (graphite_find_data_references_in_stmt): Likewise.
11719 (find_data_references_in_stmt): Pass the loop preheader edge
11720 from the nest argument.
11721 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11722 parameter the edge into the region.
11723 (instantiate_parameters): Use the loop preheader edge as entry.
11724 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11726 (get_instantiated_value_entry): Make instantiate_below parameter
11727 the edge into the region.
11728 (instantiate_scev_name): Likewise. Adjust dominance checks,
11729 when we cannot use loop-based instantiation instantiate by
11730 walking use-def chains.
11731 (instantiate_scev_poly): Adjust.
11732 (instantiate_scev_binary): Likewise.
11733 (instantiate_scev_convert): Likewise.
11734 (instantiate_scev_not): Likewise.
11735 (instantiate_array_ref): Remove.
11736 (instantiate_scev_3): Likewise.
11737 (instantiate_scev_2): Likewise.
11738 (instantiate_scev_1): Likewise.
11739 (instantiate_scev_r): Do not blindly handle N-operand trees.
11740 Do not instantiate array-refs. Handle all constants and invariants.
11741 (instantiate_scev): Make instantiate_below parameter
11742 the edge into the region.
11743 (resolve_mixers): Use the loop preheader edge for the region
11744 parameter to instantiate_scev_r.
11745 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11747 2017-10-13 Richard Biener <rguenther@suse.de>
11749 PR tree-optimization/82525
11750 * graphite-isl-ast-to-gimple.c
11751 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11753 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11754 Fail code generation when we cannot represent the isl integer.
11755 (binary_op_to_tree): Elide modulo operations that are no-ops
11756 in the type we code generate. Remove now superfluous code
11759 2017-10-13 Richard Biener <rguenther@suse.de>
11761 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11762 (scop_detection::harmful_loop_in_region): Remove premature
11763 IV type restriction.
11764 (scop_detection::graphite_can_represent_scev): We can handle
11765 pointer IVs just fine.
11767 2017-10-13 Alan Modra <amodra@gmail.com>
11769 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11770 "Clobbers and Scratch Registers". Add paragraph on
11771 alternative to clobbers for scratch registers and OpenBLAS
11774 2017-10-13 Alan Modra <amodra@gmail.com>
11776 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11777 example of a memory input for a string of known length. Move
11778 commentary out of table. Add a number of new examples
11779 covering array memory inputs.
11781 2017-10-12 Martin Liska <mliska@suse.cz>
11783 PR tree-optimization/82493
11784 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11785 (test_range_functions): New function.
11786 (sbitmap_c_tests): Likewise.
11787 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11788 * selftest.h (sbitmap_c_tests): New function.
11790 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11792 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11794 * config/rs6000/amo.h: Fix spacing issue.
11796 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11799 * config/i386/i386.md (*ashl<mode>3_mask_1,
11800 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11801 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11804 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11806 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11807 (profile_probability): Set max_probability
11808 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11810 * profile-count.c (profile_probability::differs_from_p): Do not
11811 rely on max_probaiblity == 10000
11813 2017-10-12 Jeff Law <law@redhat.com>
11815 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11818 2017-10-12 Martin Sebor <msebor@redhat.com>
11822 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11823 (handle_alias_pairs): Call it.
11824 * common.opt (-Wattribute-alias): New option.
11825 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11826 * doc/invoke.texi (-Wattribute-alias): Document.
11828 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11831 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11833 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11835 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11836 (make_hard_regno_born, make_hard_regno_dead): Update
11837 bb_killed_pseudos and bb_gen_pseudos.
11839 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11841 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11843 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11845 * config/alpha/alpha.c (alpha_split_conditional_move):
11846 Use std::swap instead of manually swapping.
11847 (alpha_stdarg_optimize_hook): Ditto.
11848 (alpha_canonicalize_comparison): Ditto.
11850 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11852 * tree-loop-distribution.c (struct builtin_info): New struct.
11853 (struct partition): Refactor fields into struct builtin_info.
11854 (partition_free): Free struct builtin_info.
11855 (build_size_arg_loc, build_addr_arg_loc): Delete.
11856 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11857 information from struct builtin_info.
11858 (find_single_drs): New function refactored from classify_partition.
11859 Also moved builtin validity checks to this function.
11860 (compute_access_range, alloc_builtin): New functions.
11861 (classify_builtin_st, classify_builtin_ldst): New functions.
11862 (classify_partition): Refactor code into functions find_single_drs,
11863 classify_builtin_st and classify_builtin_ldst.
11864 (distribute_loop): Don't do runtime alias check when distributing
11866 (find_seed_stmts_for_distribution): New function.
11867 (pass_loop_distribution::execute): Refactor code finding seed
11868 stmts into above function. Support distribution for the innermost
11869 two-level loop nest. Adjust dump information.
11871 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11873 * tree-loop-distribution.c: Adjust the general comment.
11874 (NUM_PARTITION_THRESHOLD): New macro.
11875 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
11876 (classify_partition): Skip builtin pattern of loop nest's inner loop.
11877 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
11878 in call to build_partition_graph.
11879 (finalize_partitions): New parameter. Make loop distribution more
11880 conservative by fusing more partitions.
11881 (distribute_loop): Don't do runtime alias check in case of loop nest
11883 (find_seed_stmts_for_distribution): New function.
11884 (prepare_perfect_loop_nest): New function.
11885 (pass_loop_distribution::execute): Refactor code finding seed stmts
11886 and loop nest into above functions. Support loop nest distribution.
11887 Adjust dump information accordingly.
11889 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11891 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
11892 and set PTYPE_SEQUENTIAL for merged partition.
11894 2017-10-12 Richard Biener <rguenther@suse.de>
11896 PR tree-optimization/69728
11898 2017-09-19 Richard Biener <rguenther@suse.de>
11900 PR tree-optimization/69728
11901 * graphite-sese-to-poly.c (schedule_error): New global.
11902 (add_loop_schedule): Handle empty domain by failing the
11904 (build_original_schedule): Handle schedule_error.
11906 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
11907 domain by returning an unchanged schedule.
11909 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11911 * genrecog.c (validate_pattern): For VEC_SELECT verify that
11912 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
11914 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
11916 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
11919 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11922 * expr.c (store_field): Don't optimize away bitsize == 0 store
11923 from CALL_EXPR with addressable return type.
11925 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11927 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
11928 * config/rs6000/rs6000.md (sel): Delete mode attribute.
11929 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
11930 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
11931 TARGET_ISEL instead of TARGET_ISEL<sel>.
11933 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
11935 * config/rs6000/rs6000.c
11936 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
11938 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11940 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
11942 (reg_or_cint_operand): Fix comment.
11943 (reg_or_zero_operand): New predicate.
11944 * config/rs6000/rs6000-protos.h (output_isel): Delete.
11945 * config/rs6000/rs6000.c (output_isel): Delete.
11946 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
11947 instead of reg_or_cint_operand. Output instruction directly (not via
11949 (isel_unsigned_<mode>): Ditto.
11950 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
11951 gpc_reg_operand. Add an instruction alternative for this. Output
11952 instruction directly.
11953 (*isel_reversed_unsigned_<mode>): Ditto.
11955 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11957 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
11958 (TARGET_CANONICALIZE_COMPARISON): Define.
11960 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
11963 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11964 Check whether the dest is REG before adding REG_EQUIV note.
11966 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11969 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11971 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11972 (make_hard_regno_born, make_hard_regno_dead): Update
11973 bb_killed_pseudos and bb_gen_pseudos.
11975 2017-10-11 Nathan Sidwell <nathan@acm.org>
11977 * incpath.h (enum incpath_kind): Name enum, prefix values.
11978 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
11979 * incpath.c (heads, tails): Use INC_MAX.
11980 (add_env_var_paths, add_standard_paths): Use incpath_kind.
11981 (merge_include_chains, split_quote_chain,
11982 register_include_chains): Update incpath_kind names.
11983 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
11984 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
11986 (add_framework_path, darwin_register_objc_includes): Likewise.
11987 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
11989 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11991 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
11992 Do not use float_operator operator predicate.
11993 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
11994 * config/i386/predicates.md (float_operator): Remove predicate.
11996 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11998 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
11999 (*jccxf_i387): Ditto.
12000 (*jcc<mode>_i387): Ditto.
12001 (*jccu<mode>_i387): Ditto.
12002 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12003 (*jcc_*_i387 splitters): Remove.
12004 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12005 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12006 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12009 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12011 * profile-count.h (slow_safe_scale_64bit): New function.
12012 (safe_scale_64bit): New inline.
12013 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12014 * profile-count.c: Include wide-int.h
12015 (slow_safe_scale_64bit): New.
12017 2017-10-11 Nathan Sidwell <nathan@acm.org>
12019 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12020 HAS_DECL_ASSEMBLER_NAME_P.
12021 * gimple-expr.c (gimple_decl_printable_name: Check
12022 HAS_DECL_ASSEMBLER_NAME_P too.
12023 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12024 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12025 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12026 * passes.c (rest_of_decl_compilation): Check
12027 HAS_DECL_ASSEMBLER_NAME_P too.
12028 * recog.c (verify_changes): Likewise.
12029 * tree-pretty-print.c (dump_decl_name): Likewise.
12030 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12032 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12033 (SET_DECL_ASSEMBLER_NAME): Use it.
12034 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12035 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12036 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
12038 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12040 * config.gcc (i386, x86_64): Add extra objects.
12041 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12042 (ix86_min_insn_size): Declare.
12043 (ix86_issue_rate): Declare.
12044 (ix86_adjust_cost): Declare.
12045 (ia32_multipass_dfa_lookahead): Declare.
12046 (ix86_macro_fusion_p): Declare.
12047 (ix86_macro_fusion_pair_p): Declare.
12048 (ix86_bd_has_dispatch): Declare.
12049 (ix86_bd_do_dispatch): Declare.
12050 (ix86_core2i7_init_hooks): Declare.
12051 (ix86_atom_sched_reorder): Declare.
12052 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12053 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12054 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12055 (rip_relative_addr_p): Rename to ...
12056 (ix86_rip_relative_addr_p): ... this one; export.
12057 (memory_address_length): Update.
12058 (ix86_issue_rate): Move to x86-tune-sched.c.
12059 (ix86_flags_dependent): Move to x86-tune-sched.c.
12060 (ix86_agi_dependent): Move to x86-tune-sched.c.
12061 (exact_dependency_1): Move to x86-tune-sched.c.
12062 (exact_store_load_dependency): Move to x86-tune-sched.c.
12063 (ix86_adjust_cost): Move to x86-tune-sched.c.
12064 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12065 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12066 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12067 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12068 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12069 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12070 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12071 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12072 (min_insn_size): Rename to ...
12073 (ix86_min_insn_size): ... this one; export.
12074 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12075 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12076 (core2i7_first_cycle_multipass_backtrack): Move to
12077 x86-tune-sched-core.c.
12078 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12079 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12080 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12081 (ix86_avoid_jump_mispredicts): Update.
12082 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12083 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12084 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12085 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12086 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12087 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12088 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12089 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12090 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12091 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12092 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12093 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12094 (BIG): Move to ix86-tune-sched-bd.c.
12095 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12096 (enum insn_path): Move to ix86-tune-sched-bd.c.
12097 (get_mem_group): Move to ix86-tune-sched-bd.c.
12098 (is_cmp): Move to ix86-tune-sched-bd.c.
12099 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12100 (is_branch): Move to ix86-tune-sched-bd.c.
12101 (is_prefetch): Move to ix86-tune-sched-bd.c.
12102 (init_window): Move to ix86-tune-sched-bd.c.
12103 (allocate_window): Move to ix86-tune-sched-bd.c.
12104 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12105 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
12106 (process_end_window): Move to ix86-tune-sched-bd.c.
12107 (allocate_next_window): Move to ix86-tune-sched-bd.c.
12108 (find_constant): Move to ix86-tune-sched-bd.c.
12109 (get_num_immediates): Move to ix86-tune-sched-bd.c.
12110 (has_immediate): Move to ix86-tune-sched-bd.c.
12111 (get_insn_path): Move to ix86-tune-sched-bd.c.
12112 (get_insn_group): Move to ix86-tune-sched-bd.c.
12113 (count_num_restricted): Move to ix86-tune-sched-bd.c.
12114 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
12115 (add_insn_window): Move to ix86-tune-sched-bd.c.
12116 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
12117 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
12118 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
12119 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
12120 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
12121 (do_dispatch): Move to ix86-tune-sched-bd.c.
12122 (has_dispatch): Move to ix86-tune-sched-bd.c.
12123 * config/i386/t-i386: Add new object files.
12124 * config/i386/x86-tune-costs.h: New file.
12125 * config/i386/x86-tune-sched-atom.c: New file.
12126 * config/i386/x86-tune-sched-bd.c: New file.
12127 * config/i386/x86-tune-sched-core.c: New file.
12128 * config/i386/x86-tune-sched.c: New file.
12130 2017-10-11 Liu Hao <lh_mouse@126.com>
12132 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
12133 the generic version below instead.
12134 (should_colorize): Recognize Windows consoles as terminals
12136 * pretty-print.c [__MINGW32__] (write_all): New function.
12137 [__MINGW32__] (find_esc_head): Likewise.
12138 [__MINGW32__] (find_esc_terminator): Likewise.
12139 [__MINGW32__] (eat_esc_sequence): Likewise.
12140 [__MINGW32__] (mingw_ansi_fputs): New function that handles
12142 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
12145 2017-10-11 Richard Biener <rguenther@suse.de>
12147 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
12148 Properly call analyze_scalar_evolution with the loop of the stmt.
12150 2017-10-11 Richard Biener <rguenther@suse.de>
12152 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
12153 * tree-core.h (tree_base): Add chrec_var union member.
12154 * tree.h (CHREC_VAR): Remove.
12155 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
12156 * tree-chrec.h (build_polynomial_chrec): Adjust.
12157 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
12158 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
12160 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
12162 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
12163 * match.pd: ... here.
12164 ((T) X == (T) Y): Relax condition.
12166 2017-10-11 Bin Cheng <bin.cheng@arm.com>
12168 PR tree-optimization/82472
12169 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
12171 (break_alias_scc_partitions): Update postorder number.
12173 2017-10-11 Martin Liska <mliska@suse.cz>
12176 * opts.c (parse_no_sanitize_attribute): Do not use error_value
12178 * opts.h (parse_no_sanitize_attribute): Remove last argument.
12180 2017-10-11 Martin Liska <mliska@suse.cz>
12182 * print-rtl.c (print_insn): Move declaration of idbuf
12183 to same scope as name.
12185 2017-10-11 Martin Liska <mliska@suse.cz>
12190 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12191 volatile arguments.
12193 2017-10-11 Martin Liska <mliska@suse.cz>
12196 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12197 volatile arguments.
12199 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
12201 * config.gcc (default_gnu_indirect_function): Default to yes for
12202 arm*-*-linux* with glibc.
12204 2017-10-11 Richard Biener <rguenther@suse.de>
12206 * tree-scalar-evolution.c (get_scalar_evolution): Handle
12207 default-defs and types we do not want to analyze.
12208 (interpret_loop_phi): Replace unreachable code with an assert.
12209 (compute_scalar_evolution_in_loop): Remove and inline ...
12210 (analyze_scalar_evolution_1): ... here, replacing condition with
12211 what makes the intent clearer. Remove handling of cases
12212 get_scalar_evolution now handles.
12214 2017-10-10 Jim Wilson <wilson@tuliptree.org>
12216 PR rtl-optimization/81434
12217 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
12218 comment for main loop. In sched_group_found if, also add checks for
12219 pass and min_cost_group.
12221 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
12223 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
12224 (rs6000_insn_cost): New function.
12225 * config/rs6000/rs6000.md (cost): New attribute.
12227 2017-10-10 Jakub Jelinek <jakub@redhat.com>
12228 H.J. Lu <hongjiu.lu@intel.com>
12232 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
12233 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
12234 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
12235 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
12236 to OPTION_MASK_ISA_AVX512VL - builtins that have both
12237 OPTION_MASK_ISA_MMX and some other bit set require both
12238 mmx and the ISAs without the mmx bit.
12239 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
12240 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
12241 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
12242 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
12243 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
12244 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
12245 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
12246 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
12247 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
12248 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
12249 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
12250 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
12251 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
12252 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
12253 Add OPTION_MASK_ISA_MMX.
12255 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
12257 * config.gcc (armv7*-*-freebsd*): New target.
12258 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
12260 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
12262 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
12263 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
12266 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
12268 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
12269 a template parameter.
12270 (WIDE_INT_REF_FOR): Update accordingly.
12271 * tree.h (wi::int_traits <const_tree>): Delete.
12272 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
12273 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
12274 (wi::tree_to_wide_ref): New typedef.
12275 (wi::to_wide): New function.
12276 * calls.c (get_size_range): Use wi::to_wide when operating on
12277 trees as wide_ints.
12278 * cgraph.c (cgraph_node::create_thunk): Likewise.
12279 * config/i386/i386.c (ix86_data_alignment): Likewise.
12280 (ix86_local_alignment): Likewise.
12281 * dbxout.c (stabstr_O): Likewise.
12282 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
12283 * expr.c (const_vector_from_tree): Likewise.
12284 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
12285 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
12286 (fold_negate_expr_1, int_const_binop_1, const_binop)
12287 (fold_convert_const_int_from_real, optimize_bit_field_compare)
12288 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
12289 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
12290 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
12291 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
12292 (fold_not_const, round_up_loc): Likewise.
12293 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
12294 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
12295 (alloca_call_type): Likewise.
12296 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
12297 * godump.c (go_output_typedef): Likewise.
12298 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
12299 * internal-fn.c (get_min_precision): Likewise.
12300 * ipa-cp.c (ipcp_store_vr_results): Likewise.
12301 * ipa-polymorphic-call.c
12302 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
12303 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
12304 (ipa_modify_call_arguments): Likewise.
12305 * match.pd: Likewise.
12306 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
12307 * print-tree.c (print_node_brief, print_node): Likewise.
12308 * stmt.c (expand_case): Likewise.
12309 * stor-layout.c (layout_type): Likewise.
12310 * tree-affine.c (tree_to_aff_combination): Likewise.
12311 * tree-cfg.c (group_case_labels_stmt): Likewise.
12312 * tree-data-ref.c (dr_analyze_indices): Likewise.
12313 (prune_runtime_alias_test_list): Likewise.
12314 * tree-dump.c (dequeue_and_dump): Likewise.
12315 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
12316 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
12317 * tree-pretty-print.c (dump_generic_node): Likewise.
12318 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
12319 (simple_iv_with_niters): Likewise.
12320 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
12321 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
12322 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
12323 * tree-ssa-loop-niter.c (split_to_var_and_offset)
12324 (refine_value_range_using_guard, number_of_iterations_ne_max)
12325 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
12326 (get_cst_init_from_scev, record_nonwrapping_iv)
12327 (scev_var_range_cant_overflow): Likewise.
12328 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
12329 * tree-ssa-pre.c (compute_avail): Likewise.
12330 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
12331 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
12332 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
12333 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
12334 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
12335 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
12336 (dump_case_nodes, try_switch_expansion): Likewise.
12337 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
12338 (vect_do_peeling): Likewise.
12339 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
12340 * tree-vect-stmts.c (vectorizable_load): Likewise.
12341 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
12342 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
12343 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
12344 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
12345 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
12346 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
12347 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
12348 (evrp_dom_walker::before_dom_children): Likewise.
12349 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
12350 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
12351 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
12352 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
12353 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
12354 (get_range_pos_neg): Likewise.
12355 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
12356 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
12357 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
12358 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
12359 * config/avr/avr.c (avr_fold_builtin): Likewise.
12360 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
12361 * config/msp430/msp430.c (msp430_attr): Likewise.
12362 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
12363 * config/powerpcspe/powerpcspe-c.c
12364 (altivec_resolve_overloaded_builtin): Likewise.
12365 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
12366 (rs6000_expand_ternop_builtin): Likewise.
12367 * config/rs6000/rs6000-c.c
12368 (altivec_resolve_overloaded_builtin): Likewise.
12369 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
12370 (rs6000_expand_ternop_builtin): Likewise.
12371 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
12373 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12375 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
12376 when copying loop nest with only one inner loop.
12378 2017-10-10 Richard Biener <rguenther@suse.de>
12380 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
12381 blocks if SCEV is active.
12382 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
12384 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
12385 (scev_initialize): Assert we are not yet initialized.
12387 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12389 * tree-loop-distribution.c (generate_loops_for_partition): Remove
12390 inner loop's exit stmt by making it always exit the loop, otherwise
12391 we would generate an infinite empty loop.
12393 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12395 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
12396 renaming variables in new preheader if it's deleted.
12398 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12400 * tree-loop-distribution.c (struct partition): Remove unused field
12401 loops of the structure.
12402 (partition_alloc, partition_free): Ditto.
12403 (build_rdg_partition_for_vertex): Ditto.
12405 2017-10-09 Jeff Law <law@redhat.com>
12407 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
12408 return type to match prototype and documentation.
12410 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12412 * config/rs6000/rs6000.c (processor_costs): Move to ...
12413 * config/rs6000/rs6000.h: ... here.
12414 (rs6000_cost): Declare.
12416 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
12418 * except.c (setjmp_fn): New global variable.
12419 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
12420 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
12421 if DONT_USE_BUILTIN_SETJMP is defined.
12423 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12425 * target.def (insn_cost): New hook.
12426 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
12427 * doc/tm.texi: Regenerate.
12428 * rtlanal.c (insn_cost): Use the new hook.
12430 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12432 * combine.c (combine_validate_cost): Compute the new insn_cost,
12433 not just pattern_cost.
12434 (try_combine): Adjust comment.
12436 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12438 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
12440 * combine.c (uid_insn_cost): Adjust comment.
12441 (combine_validate_cost): Adjust comment. Use pattern_cost instead
12443 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
12444 * dse.c (find_shift_sequence): Ditto.
12445 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
12446 (bb_valid_for_noce_process_p): Use pattern_cost.
12447 * rtl.h (insn_rtx_cost): Delete.
12448 (pattern_cost): New prototype.
12449 (insn_cost): New prototype.
12450 * rtlanal.c (insn_rtx_cost): Rename to...
12451 (pattern_cost): ... this.
12454 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
12456 * config/i386/i386.md (*jcc_2): Remove insn pattern.
12457 (*jcc<mode>_0_r_i387): Ditto.
12458 (*jccxf_r_i387): Ditto.
12459 (*jcc<mode>_r_i387): Ditto.
12460 (*jccu<mode>_r_i387): Ditto.
12461 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
12462 (*jcc): Rename from *jcc_1.
12464 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12466 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
12467 deferred rescans after the lvx/stvx recombination pre-pass.
12469 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
12471 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
12472 memory operation instruction support.
12473 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
12474 (rs6000-ibm-aix[789]*): Likewise.
12475 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
12476 Document new functions.
12478 2017-10-09 Richard Biener <rguenther@suse.de>
12480 PR tree-optimization/82397
12481 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
12482 equality only for semantically equal trees.
12484 2017-10-09 Richard Biener <rguenther@suse.de>
12486 PR tree-optimization/82449
12487 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
12488 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
12489 allow constant addresses.
12490 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
12493 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12495 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
12498 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12501 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
12504 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12507 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
12509 2017-10-09 Jakub Jelinek <jakub@redhat.com>
12512 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
12513 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
12515 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
12517 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
12518 (WI_BINARY_PREDICATE_RESULT): Likewise.
12519 (wi::binary_traits::operator_result): New type.
12520 (wi::binary_traits::predicate_result): Likewise.
12521 (generic_wide_int::operator~, unary generic_wide_int::operator-)
12522 (generic_wide_int::operator==, generic_wide_int::operator!=)
12523 (generic_wide_int::operator&, generic_wide_int::and_not)
12524 (generic_wide_int::operator|, generic_wide_int::or_not)
12525 (generic_wide_int::operator^, generic_wide_int::operator+
12526 (binary generic_wide_int::operator-, generic_wide_int::operator*):
12528 (operator~, unary operator-, operator==, operator!=, operator&)
12529 (operator|, operator^, operator+, binary operator-, operator*): New
12531 * expr.c (get_inner_reference): Use wi::bit_and_not.
12532 * fold-const.c (fold_binary_loc): Likewise.
12533 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
12534 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
12535 (bit_value_binop): Likewise.
12536 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
12537 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
12538 (extract_range_from_binary_expr_1): Likewise.
12539 (masked_increment): Likewise.
12540 (simplify_bit_ops_using_ranges): Likewise.
12542 2017-10-09 Martin Jambor <mjambor@suse.cz>
12545 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
12546 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
12547 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
12549 (hsa_fixup_mov_insn_type): New function.
12550 (hsa_op_with_type::get_in_type): Use it.
12551 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
12552 immediates in an assert.
12553 (hsa_op_with_type::extend_int_to_32bit): New method.
12554 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
12555 types. Convert to dest type if necessary.
12556 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
12557 (reg_for_gimple_ssa): Pass false as min32int to
12558 hsa_type_for_scalar_tree_type.
12559 (gen_hsa_addr): Fixup type when creating addresable temporary.
12560 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
12561 (gen_hsa_unary_operation): Extend operands and convert to dest type if
12562 necessary. Call hsa_fixup_mov_insn_type.
12563 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
12564 extend operands and convert to dest type if necessary.
12565 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
12566 to dest type if necessary.
12567 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
12568 if conversion nt necessary and size matches.
12569 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
12570 to dest type if necessary.
12571 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
12572 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
12574 (gen_hsa_clrsb): Likewise.
12575 (gen_hsa_ffs): Likewise.
12576 (gen_hsa_divmod): Extend operands and convert to dest type if
12578 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
12580 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12582 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
12583 Remove empty default arguments. Use a brace block as output
12585 (conditional return): Ditto.
12587 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
12588 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
12590 (group_ending_nop): Ditto.
12591 (doloop_end): Ditto.
12592 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
12593 (splitters for those): Ditto.
12595 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12597 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
12598 a conditional jump (and the compare for it) so that pc_rtx is the
12600 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
12601 for the deleted and renamed ctr<mode>_internal[234] patterns.
12602 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
12603 Delete second conditional return pattern.
12604 (ctr<mode>_internal2): Delete this second bdnz pattern.
12605 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
12606 (ctr<mode>_internal4): Delete this second bdz pattern.
12608 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12610 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
12611 (always_initialized_rtx_for_ssa_name_p): New predicate.
12612 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
12613 (finish_out_of_ssa): Free new field of SA.
12614 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
12615 * tree-ssa-coalesce.c: Include tree-ssa.h.
12616 (get_parm_default_def_partitions): Remove extern keyword.
12617 (get_undefined_value_partitions): New function.
12618 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12619 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12620 uninitialized bits.
12621 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12623 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12625 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12627 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12629 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12630 for targets that preffer 128bit.
12632 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12634 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12636 2017-10-08 Olivier Hainque <hainque@adacore.com>
12638 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12639 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12641 (thumb_set_return_address): Likewise.
12643 2017-10-08 Olivier Hainque <hainque@adacore.com>
12645 * common/config/arm/arm-common.c (arm_except_unwind_info):
12646 Handle DWARF2_UNWIND_INFO.
12648 2017-10-07 Michael Collison <michael.collison@arm.com>
12650 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12653 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12655 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12656 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12657 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12658 defined, force the creation of a new block for a dispatch label.
12660 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12662 * invoke.texi (Wsuggest-attribute=cold): Document.
12663 * common.opt (Wsuggest-attribute=cold): New
12664 * ipa-pure-const.c (warn_function_cold): New function.
12665 * predict.c (compute_function_frequency): Use it.
12666 * predict.h (warn_function_cold): Declare.
12668 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12670 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12673 2017-10-06 Martin Liska <mliska@suse.cz>
12675 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12676 keyword for member functions.
12677 (struct sanopt_tree_couple): New struct.
12678 (struct sanopt_tree_couple_hash): New function.
12679 (struct sanopt_ctx): Add new hash_map.
12680 (has_dominating_ubsan_ptr_check): New function.
12681 (record_ubsan_ptr_check_stmt): Likewise.
12682 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12683 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12684 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12686 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12689 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12690 aarch64_simd_valid_immediate on CONST_VECTORs.
12691 (aarch64_reg_or_bic_imm): Likewise.
12693 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12695 PR rtl-optimization/82396
12696 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12698 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12700 * graphite-dependences.c (scop_get_reads): Move code to...
12701 (scop_get_must_writes): Move code to...
12702 (scop_get_may_writes): Move code to...
12703 (scop_get_reads_and_writes): ... here.
12704 (scop_get_dependences): Call scop_get_reads_and_writes.
12706 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12708 PR tree-optimization/82434
12709 * fold-const.h (can_native_encode_type_p,
12710 can_native_encode_string_p): Remove.
12711 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12712 don't encode anything, just return what would be otherwise returned.
12713 (native_encode_fixed, native_encode_complex, native_encode_vector):
12715 (native_encode_string): Likewise. Inline by hand
12716 can_native_encode_string_p.
12717 (can_native_encode_type_p): Remove.
12718 (can_native_encode_string_p): Remove.
12719 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12720 STRING_CSTs using can_native_encode_string_p, test all
12721 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12722 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12723 argument from native_encode_expr.
12724 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12725 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12728 2017-10-06 Richard Biener <rguenther@suse.de>
12730 PR tree-optimization/82397
12731 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12732 operand_equal_p but rely on data_ref_compare_tree for detecting
12734 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12735 to match up with dr_group_sort_cmp.
12737 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12740 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12742 * config/s390/s390-builtin-types.def: Regenerate.
12744 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12747 * config/s390/s390-builtin-types.def: Regenerate.
12748 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12749 Change flag from B_VXE to B_VX.
12750 (s390_vec_min_dbl): Remove B_VXE flag.
12752 2017-10-06 Richard Biener <rguenther@suse.de>
12754 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12755 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12756 translate_isl_ast_to_gimple::is_valid_rename,
12757 translate_isl_ast_to_gimple::get_rename,
12758 translate_isl_ast_to_gimple::get_def_bb_for_const,
12759 translate_isl_ast_to_gimple::get_new_name,
12760 translate_isl_ast_to_gimple::collect_all_ssa_names,
12761 translate_isl_ast_to_gimple::copy_loop_phi_args,
12762 translate_isl_ast_to_gimple::collect_all_ssa_names,
12763 translate_isl_ast_to_gimple::copy_loop_phi_args,
12764 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12765 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12766 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12767 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12768 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12769 translate_isl_ast_to_gimple::copy_cond_phi_args,
12770 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12771 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12772 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12773 translate_isl_ast_to_gimple::rename_uses,
12774 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12775 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12776 (set_rename_for_each_def): Likewise.
12777 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12778 here. Handle rewriting SCEV analyzable uses here.
12779 (copy_bb_and_scalar_dependences): Generate code for PHI
12781 (graphite_regenerate_ast_isl): Adjust.
12782 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12783 (add_write, add_read): New functions.
12784 (build_cross_bb_scalars_def): Use it and simplify.
12785 (build_cross_bb_scalars_use): Likewise.
12786 (graphite_find_cross_bb_scalar_vars): Inline into...
12787 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12788 simulating out-of-SSA. Compute liveout and add dependencies.
12789 (build_scops): Force an empty entry block.
12790 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12792 (sese_build_liveouts): Declare.
12793 (sese_trivially_empty_bb_p): Likewise.
12794 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12795 compute liveout and debug_liveout.
12796 (sese_bad_liveouts_use): Remove.
12797 (sese_reset_debug_liveouts_bb): Likewise.
12798 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12799 (sese_build_liveouts): Build liveout and debug_liveout and store
12801 (new_sese_info): Adjust.
12802 (free_sese_info): Likewise.
12803 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12804 do not build liveout here.
12805 (move_sese_in_condition): Adjust region entry.
12806 (scev_analyzable_p): Match up with chrec_apply requirements.
12807 (sese_trivially_empty_bb_p): New.
12808 * tree-into-ssa.c (get_reaching_def): Properly support generating
12809 default-defs for incremental rewrite of anonymous names.
12811 2017-10-06 Richard Biener <rguenther@suse.de>
12813 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12814 precision do not perform modulo reduction.
12816 2017-10-06 Richard Biener <rguenther@suse.de>
12818 PR tree-optimization/82436
12819 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12820 conservatively choose the vectorization factor when checking
12821 whether we can perform the required load permutation.
12822 (vect_transform_slp_perm_load): Assert when we may not fail.
12824 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12826 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12827 message for incompatible -msdata=* and -mcall-* options.
12829 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12831 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12832 rate for post-reload scheduling.
12834 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12836 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12838 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12840 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12841 to improve monte carlo in scimark.
12843 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12845 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12846 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12847 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12848 pentium4_cost, nocona_cost): Set reassociation width to 1.
12849 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12850 width to 2 for fp operations and 1 otherwise.
12851 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12853 (atom_cost): Set reassociation width to 2.
12854 (slm_cost, generic_cost): Set fp reassociation width
12855 to 2 and 1 otherwise.
12856 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12857 (core_cost): Set fp reassociation width to 4 and vector to 2.
12858 (ix86_reassociation_width): Rewrite using cost table; special case
12859 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12860 and TARGET_AVX128_OPTIMAL.
12861 * config/i386/i386.h (processor_costs): Add
12862 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
12863 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
12864 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
12865 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
12866 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
12867 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
12869 2017-10-05 Nathan Sidwell <nathan@acm.org>
12871 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
12873 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12875 * config/arm/arm.c (arm_test_fpu_data): New.
12876 (arm_run_selftests): Call arm_test_fpu_data.
12878 2017-10-04 Nathan Sidwell <nathan@acm.org>
12880 * toplev.c (toplev::main): Remove excess parens on pretty_printer
12882 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
12884 2017-10-04 Sudakshina Das <sudi.das@arm.com>
12886 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
12887 check type for aarch64_simd_valid_immediate.
12888 (aarch64_output_simd_mov_immediate): Update prototype.
12889 (aarch64_simd_valid_immediate): Update prototype.
12890 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
12891 support for ORR-immediate.
12892 (and<mode>3): modified pattern to add support for BIC-immediate.
12893 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
12894 now checks for valid immediate for BIC and ORR based on new enum
12896 (aarch64_output_simd_mov_immediate): Function now used to output
12897 BIC/ORR imm as well based on new enum argument.
12898 * config/aarch64/constraints.md (Do): New vector immediate constraint.
12900 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
12901 (aarch64_reg_or_bic_imm): Likewise.
12903 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12905 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
12906 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
12908 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12912 PR rtl-optimization/82396
12913 * haifa-sched.c (autopref_multipass_init): Simplify
12915 (autopref_rank_data): Simplify sort order.
12916 * sched-int.h (autopref_multipass_data_): Remove
12917 multi_mem_insn_p, min_offset and max_offset.
12919 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12921 * doc/sourcebuild.texi: Document vect_peeling_profitable.
12923 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12925 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
12926 vect_doubleint_cvt.
12928 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12930 * doc/sourcebuild.texi: Document vect_long_mult.
12932 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
12934 PR tree-optimization/82413
12935 * fold-const.c (build_range_check): Use widest_int when comparing
12936 the maximum ETYPE value with HIGH.
12938 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12940 PR rtl-optimization/82396
12941 * haifa-sched.c (autopref_multipass_init): Simplify
12943 (autopref_rank_data): Simplify sort order.
12944 * sched-int.h (autopref_multipass_data_): Remove
12945 multi_mem_insn_p, min_offset and max_offset.
12947 2017-10-04 Jakub Jelinek <jakub@redhat.com>
12949 PR tree-optimization/82381
12950 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
12951 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
12954 PR tree-optimization/82374
12955 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
12956 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
12957 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
12958 current_function_decl to the new decl.
12960 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
12962 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
12963 helper macro for IEEE float128 hardware built-in functions.
12964 (SQRTF128_ODD): Add built-in functions with the round-to-odd
12966 (TRUNCF128_ODD): Likewise.
12967 (ADDF128_ODD): Likewise.
12968 (SUBF128_ODD): Likewise.
12969 (MULF128_ODD): Likewise.
12970 (DIVF128_ODD): Likewise.
12971 (FMAF128_ODD): Likewise.
12972 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
12973 UNSPEC_TRUNC_ROUND_TO_ODD.
12974 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
12975 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
12976 floating point round to odd instructions.
12977 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
12978 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
12979 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
12980 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
12981 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
12982 (trunc<mode>sf2_hw): Change the truncate with round to odd
12983 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
12984 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
12985 to odd hardware instructions.
12986 (sub<mode>3_odd): Likewise.
12987 (mul<mode>3_odd): Likewise.
12988 (div<mode>3_odd): Likewise.
12989 (sqrt<mode>2_odd): Likewise.
12990 (fma<mode>4_odd): Likewise.
12991 (fms<mode>4_odd): Likewise.
12992 (nfma<mode>4_odd): Likewise.
12993 (nfms<mode>4_odd): Likewise.
12994 (trunc<mode>df2_odd): Change the truncate with round to odd
12995 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
12997 * doc/extend.texi (PowerPC built-in functions): Update documentation
12998 for existing IEEE float128-bit built-in functions. Add built-in
12999 functions that generate the IEEE 128-bit floating point round to
13002 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13004 PR rtl-optimization/77729
13005 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13006 to (X&(C1&~C2))|C2 transformations.
13008 2017-10-03 Martin Jambor <mjambor@suse.cz>
13010 PR tree-optimization/82363
13011 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13012 mismatch, mark lacc written regardless of racc.
13014 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13016 PR tree-optimization/82381
13017 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13018 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13019 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13020 fallthrough into reassoc_stmt_dominates_stmt_p.
13023 * combine.c (combine_instructions): Don't combine in unreachable
13026 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13029 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13030 function to not use the have_cpu variable. Do not set cpu_index,
13031 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13032 or the default cpu.
13033 (rs6000_valid_attribute_p): Remove duplicate initializations of
13034 old_optimize and func_optimize.
13035 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13036 (rs6000_activate_target_options): Make global.
13037 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
13040 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13042 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13043 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13044 Set *poffset to 0 and *psize and *pmax_size to -1 if
13045 *poffset + *psize overflows in HOST_WIDE_INT.
13047 PR tree-optimization/82387
13048 PR tree-optimization/82388
13049 PR tree-optimization/82389
13050 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13051 instead of live_bytes non-NULL.
13053 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13056 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13059 2017-10-02 Richard Biener <rguenther@suse.de>
13061 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13062 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13063 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13065 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13067 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13068 requested precision matches the type's.
13069 * calls.c (alloc_max_size): Calculate the new candidate size as
13070 a widest_int and use wi::to_widest when comparing it with the
13071 current candidate size.
13072 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13073 zero rather than integer_zero_node.
13074 * match.pd: Check for a no-op conversion before using wi::add
13075 rather than after. Use tree_to_uhwi when summing small shift
13076 counts into an unsigned int.
13078 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13079 Alan Hayward <alan.hayward@arm.com>
13080 David Sherwood <david.sherwood@arm.com>
13083 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13084 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13085 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13086 POINTER_AND_FP_REGS.
13088 2017-10-02 Richard Biener <rguenther@suse.de>
13090 PR tree-optimization/82355
13091 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13092 a mapping for the enclosing loop but avoid generating one for
13093 the loop tree root.
13094 (copy_bb_and_scalar_dependences): Remove premature codegen
13095 error on PHIs in blocks duplicated into multiple places.
13096 * graphite-scop-detection.c
13097 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13098 in the region use it as loop and nest to analyze the DR in.
13099 (try_generate_gimple_bb): Likewise.
13100 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13101 (add_loop_constraints): For blocks in a loop not in the region
13102 create a dimension with a single iteration.
13103 * sese.h (gbb_loop_at_index): Remove assert.
13105 2017-10-01 Kevin Buettner <kevinb@redhat.com>
13107 * omp-expand.c (adjust_context_scope): New function.
13108 (expand_parallel_call): Call adjust_context_scope.
13110 2017-10-01 Jeff Law <law@redhat.com>
13112 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
13113 dom_opt_dom_walker class with direct access to private members.
13114 Add comments. Call test_for_singularity.
13115 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
13116 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
13117 m_dummy_cond anymore.
13118 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
13120 (pass_dominator:execute): Build the dummy_cond here and pass it
13121 to the dom_opt_dom_walker ctor.
13122 (test_for_singularity): New function.
13124 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
13125 Maya Rashish <coypu@sdf.org>
13127 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
13128 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
13129 (alpha*-*-netbsd*) Use nbsd_tm_file.
13130 (arm*-*-netbsdelf*) Likewise.
13131 (i[34567]86-*-netbsdelf*) Likewise.
13132 (x86_64-*-netbsd*) Likewise.
13133 (mips*-*-netbsd*) Likewise.
13134 (powerpc-*-netbsd*) Likewise.
13135 (sh*-*-netbsd*) Likewise.
13136 (sparc-*-netbsdelf*) Likewise.
13137 (sparc64-*-netbsd*) Likewise.
13138 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
13140 (vax-*-netbsdelf*) Likewise.
13141 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
13142 (UINT_FAST8_TYPE) Likewise.
13143 (INT_FAST16_TYPE) Check CHAR_FAST16.
13144 (UINT_FAST16_TYPE) Likewise.
13146 2017-09-30 Jakub Jelinek <jakub@redhat.com>
13149 * config/i386/i386.md
13150 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
13151 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
13152 *divmodsi4_zext_2): New define_insn_and_split.
13153 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
13154 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
13155 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
13156 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
13157 New define_insn_and_split.
13158 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
13159 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
13160 operands[1] having DImode when mode is SImode.
13162 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
13163 always SImode for DIV and MOD in REG_EQUAL notes.
13165 2017-09-29 Yury Gribov <tetra2005@gmail.com>
13167 PR middle-end/82319
13168 * match.pd: Fix handling of NaNs in pattern.
13170 2017-09-29 Jeff Law <law@redhat.com>
13172 * sbitmap.c (bitmap_bit_in_range_p): New function.
13173 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
13174 * tree-ssa-dse.c (live_bytes_read): New function.
13175 (dse_classify_store): Ignore reads of dead bytes.
13177 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
13178 typos and whitespace errors.
13179 * config/i386/predicates.md (address_no_seg_operand): Likewise.
13180 * config/s390/s390.c (s390_emit_prologue): Likewise.
13182 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13185 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
13186 with a symbol for LRA.
13188 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13190 PR rtl-optimization/82338
13191 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
13193 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
13195 * genmodes.c (calc_wider_mode): Suppress qsort macro.
13196 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
13197 (qsort_chk): Declare.
13198 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
13199 (qsort_chk): New function.
13201 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13203 PR tree-optimization/82337
13204 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
13205 phi definition if the PHI result appears in an abnormal PHI.
13206 (find_basis_for_base_expr): Don't record a basis if the LHS of the
13207 basis appears in an abnormal PHI.
13209 2017-09-29 Richard Biener <rguenther@suse.de>
13211 * graphite-isl-ast-to-gimple.c
13212 (translate_isl_ast_to_gimple::set_codegen_error): New function.
13213 (binary_op_to_tree): Use it.
13214 (get_rename_from_scev): Likewise.
13215 (copy_loop_phi_nodes): Likewise.
13216 (copy_bb_and_scalar_dependences): Likewise.
13217 (translate_pending_phi_nodes): Likewise.
13219 2017-09-29 Jakub Jelinek <jakub@redhat.com>
13222 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
13223 for movabsq $(i32 << shift), r64.
13225 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
13227 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
13228 index when encoding %esp as %rsp to avoid 0x67 prefix.
13230 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
13232 * config/i386/i386.md (*movsf_internal, *movdf_internal):
13233 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
13235 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
13237 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
13238 Extensions with more than 16 double VFP registers.
13239 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
13240 to_clear_mask and all code related to it. Replace the remaining
13241 entry by a sbitmap and adapt code accordingly.
13243 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
13245 * brig-builtins.def: Change pure attributes to const.
13247 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13249 * config.gcc (default_gnu_indirect_function): Default to yes for
13250 sparc*-*-linux* with glibc.
13252 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13254 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
13255 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
13256 when creating .init_array and .fini_array sections with priority
13259 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
13262 * config/aarch64/aarch64.c
13263 (aarch64_builtin_support_vector_misalignment): Always return false
13264 when misalignment is unknown.
13266 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
13268 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
13269 this function to return false if the definition used by the swap
13270 instruction is artificial, or if the memory address from which the
13271 constant value is loaded is not represented by a base address held
13272 in a register or if the base address register is a frame or stack
13273 pointer. Additionally, return false if the base address of the
13274 loaded constant is a SYMBOL_REF but is not considered to be a
13276 (replace_swapped_load_constant): New function.
13277 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
13278 loaded constant vector with a load of a swapped constant vector.
13280 2017-09-27 Carl Love <cel@us.ibm.com>
13282 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
13283 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
13284 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
13287 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
13289 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
13290 first, always call autopref_rank_data otherwise.
13292 2017-09-27 Richard Biener <rguenther@suse.de>
13294 * graphite-scop-detection.c (find_scop_parameters): Move
13295 loop bound handling ...
13296 (gather_bbs::before_dom_children): ... here, avoiding the need
13297 to build scop_info->loop_nest.
13298 (record_loop_in_sese): Remove.
13299 * sese.h (sese_info_t::loop_nest): Remove.
13300 * sese.c (new_sese_info): Do not allocate loop_nest.
13301 (free_sese_info): Do not free loop_nest.
13303 2017-09-27 Jakub Jelinek <jakub@redhat.com>
13306 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
13307 lhs from calls if the lhs has addressable type.
13309 2017-09-27 Richard Biener <rguenther@suse.de>
13311 * graphite.h (scop::max_alias_set): New member.
13312 * graphite-scop-detection.c: Remove references to non-existing
13313 --param in comments.
13314 (build_alias_sets): Record the maximum alias set used for drs.
13315 (build_scops): Support zero as unlimited for
13316 --param graphite-max-arrays-per-scop.
13317 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
13318 and inline into ...
13319 (build_poly_sr_1): ... here. Compute alias set based on the
13320 maximum alias set used for drs rather than
13321 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
13323 2017-09-27 Richard Biener <rguenther@suse.de>
13325 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
13326 --param loop-block-tile-size=0 to disable tiling.
13328 2017-09-27 Richard Biener <rguenther@suse.de>
13330 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
13331 (graphite-max-nb-scop-params): Document special value zero.
13332 * domwalk.h (dom_walker::STOP): New symbolical constant.
13333 (dom_walker::dom_walker): Add optional parameter for bb to
13335 (dom_walker::~dom_walker): Declare.
13336 (dom_walker::before_dom_children): Document STOP return value.
13337 (dom_walker::m_user_bb_to_rpo): New member.
13338 (dom_walker::m_bb_to_rpo): Likewise.
13339 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
13340 mapping here if not provided by the user.
13341 (dom_walker::~dom_walker): Free bb to RPO mapping if not
13342 provided by the user.
13343 (dom_walker::STOP): Define.
13344 (dom_walker::walk): Do not compute bb to RPO mapping here.
13345 Support STOP return value from before_dom_children to stop
13347 * graphite-optimize-isl.c (optimize_isl): If the schedule
13348 is the same still generate code if -fgraphite-identity
13349 or -floop-parallelize-all are given.
13350 * graphite-scop-detection.c: Include cfganal.h.
13351 (gather_bbs::gather_bbs): Get and pass through bb to RPO
13353 (gather_bbs::before_dom_children): Return STOP for BBs
13355 (build_scops): Compute bb to RPO mapping and pass it to
13356 the domwalk. Treat --param graphite-max-nb-scop-params=0
13357 as not limiting the number of params.
13358 * graphite.c (graphite_initialize): Remove limit on the
13359 number of basic-blocks in a function.
13360 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
13361 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
13362 default value of 10.
13364 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13366 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
13367 Adjust code to eliminate needing to do the shift right 32-bits
13368 operation after XSCVDPSPN.
13370 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13372 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
13373 ((X / Y) != 0 -> X >= Y): Likewise.
13375 2017-09-26 Carl Love <cel@us.ibm.com>
13377 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
13378 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
13379 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
13380 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
13381 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
13382 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
13383 definitions and overloading.
13384 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
13385 statement for P9V_BUILTIN_XST_LEN_R.
13386 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
13387 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
13388 define_expand and define_insn for the instructions and builtins.
13389 * doc/extend.texi: Update the built-in documentation file for the new
13390 built-in functions.
13391 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
13392 define_insn for the instructions
13394 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
13397 * gcc/config/netbsd-protos.h: New file.
13398 * gcc/config/netbsd.c: New file.
13399 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
13400 * gcc/config/t-netbsd: New file.
13401 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
13402 (tmake_file) Add t-netbsd.
13403 (extra_objs) Add netbsd.o.
13405 2017-09-26 Janus Weil <janus@gcc.gnu.org>
13409 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
13411 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13413 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
13414 sign extension from a vector register to a GPR by doing a 32-bit
13415 direct move and then an EXTSW.
13416 (extendsi<mode>2 splitter): Likewise.
13417 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
13418 right or vector extract after doing XSCVDPSPN. Use
13419 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
13421 (movdi_from_sf_zero_ext): Likewise.
13422 (reload_gpr_from_vsxsf): Likewise.
13423 (p8_mfvsrd_4_disf): Delete, no longer used.
13424 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
13425 then needing to move the SFmode to a GPR to use the XSCVDPSP
13426 instruction instead of FRSP and XSCVDPSPN.
13427 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
13428 it is adjacent to the other XSCVSPDP insns.
13429 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
13430 SFmode to be in traditional Altivec registers.
13431 (vsx_xscvdpspn): Eliminate useless alternative constraint.
13432 (vsx_xscvspdpn): Likewise.
13433 (vsx_xscvspdpn_scalar): Likewise.
13435 2017-09-26 Martin Jambor <mjambor@suse.cz>
13437 * tree-sra.c (compare_access_positions): Put integral types first,
13438 stabilize sorting of integral types, remove conditions putting
13439 non-full-precision integers last.
13440 (sort_and_splice_var_accesses): Disable scalarization if a
13441 non-integert would be represented by a non-full-precision integer.
13443 2017-09-26 Joseph Myers <joseph@codesourcery.com>
13445 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
13446 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
13447 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
13448 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
13449 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
13450 conditionals inside the function instead of around it. Call
13451 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
13452 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
13454 2017-09-26 Richard Biener <rguenther@suse.de>
13456 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
13458 (scop_detection::build_scop_breadth): ... this. Removed.
13459 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
13460 (scop_detection::harmful_stmt_in_bb): Likewise.
13461 (scop_detection::graphite_can_represent_stmt): Likewise.
13462 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
13463 (scop_detection::can_represent_loop): Remove recursion, fold in ...
13464 (scop_detection::can_represent_loop_1): ... this. Removed.
13465 (scop_detection::harmful_loop_in_region): Simplify after inlining
13466 the above and remove more quadraticness.
13467 (build_scops): Adjust.
13468 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
13471 2017-09-26 Jakub Jelinek <jakub@redhat.com>
13474 * config/i386/i386.c (ix86_print_operand_address_as): Only test
13475 REGNO (base) == SP_REG if base is a REG.
13477 PR middle-end/35691
13478 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
13479 if it is different SSA_NAME.
13480 (optimize_range_tests_cmp_bitwise): New function.
13481 (optimize_range_tests): Call it.
13483 2017-09-26 Richard Biener <rguenther@suse.de>
13485 PR tree-optimization/82321
13486 * graphite.c (canonicalize_loop_closed_ssa): Properly check
13487 for the def being inside the loop.
13489 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13491 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
13493 * config/s390/s390-builtins.def: Fix constraint on op4.
13495 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13497 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
13498 independent expanders.
13499 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
13500 ("vec_ordered", "vec_unordered"): New expanders.
13502 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13504 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
13507 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13509 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
13510 vec_unpacks_lo_v16qi.
13511 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
13513 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13515 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
13516 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
13517 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
13519 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13521 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
13523 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
13525 ("*vec_slb<mode>"): New insn pattern.
13526 ("vec_shr_<mode>"): New expander.
13527 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
13528 and force the shift count operand to V16QImode.
13529 ("vec_srb<mode>"): Set shift count mode to V16QI.
13531 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13533 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
13534 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
13535 ("vec_widen_smult_hi_<mode>"): New expander definitions.
13537 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
13540 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
13542 2017-09-26 Richard Biener <rguenther@suse.de>
13544 PR tree-optimization/82320
13545 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
13548 2017-09-25 Jeff Law <law@redhat.com>
13550 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
13551 prototype for new argument.
13552 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
13553 mostly extracted from rs6000_emit_allocate_stack.
13554 (rs6000_emit_probe_stack_range_stack_clash): New function.
13555 (rs6000_emit_allocate_stack): Call
13556 rs6000_emit_probe_stack_range_stack_clash as needed.
13557 (rs6000_emit_probe_stack_range): Add additional argument
13558 to call to gen_probe_stack_range{si,di}.
13559 (output_probe_stack_range): New.
13560 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
13561 (output_probe_stack_range_stack_clash): New.
13562 (rs6000_emit_prologue): Emit notes into dump file as requested.
13563 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
13564 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
13565 Add additional operand and pass it to output_probe_stack_range.
13567 2017-09-25 Bin Cheng <bin.cheng@arm.com>
13569 PR tree-optimization/82163
13570 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
13571 (checking_verify_loop_closed_ssa): New parameter.
13572 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
13573 (check_loop_closed_ssa_stmt): Delete.
13574 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
13575 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
13576 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
13579 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
13581 * brig-builtins.def: Treat HSAIL barrier builtins as
13582 setjmp/longjump style functions.
13584 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13586 * target.def (constant_alignment): New hook.
13587 * defaults.h (CONSTANT_ALIGNMENT): Delete.
13588 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
13589 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
13590 * doc/tm.texi: Regenerate.
13591 * targhooks.h (default_constant_alignment): Declare.
13592 (constant_alignment_word_strings): Likewise.
13593 * targhooks.c (default_constant_alignment): New function.
13594 (constant_alignment_word_strings): Likewise.
13595 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
13596 instead of CONSTANT_ALIGNMENT.
13597 * varasm.c (align_variable, get_variable_align, build_constant_desc)
13598 (force_const_mem): Likewise.
13599 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
13600 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
13601 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
13602 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13603 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
13605 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
13606 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13607 constant_alignment_word_strings.
13608 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
13609 (CONSTANT_ALIGNMENT): Likewise.
13610 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13611 (arm_constant_alignment): New function.
13612 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
13613 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13614 constant_alignment_word_strings.
13615 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
13616 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13617 constant_alignment_word_strings.
13618 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13619 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13620 (cris_constant_alignment): New function.
13621 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13622 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13623 (epiphany_constant_alignment): New function.
13624 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13625 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13626 constant_alignment_word_strings.
13627 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13628 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13629 constant_alignment_word_strings.
13630 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13631 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13632 constant_alignment_word_strings.
13633 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13634 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13635 * config/i386/i386.c (ix86_constant_alignment): Make static.
13636 Use the same interface as the target hook.
13637 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13638 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13639 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13640 constant_alignment_word_strings.
13641 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13642 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13643 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13644 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13645 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13646 constant_alignment_word_strings.
13647 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13648 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13649 constant_alignment_word_strings.
13650 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13651 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13652 constant_alignment_word_strings.
13653 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13654 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13656 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13657 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13658 * config/mips/mips.c (mips_constant_alignment): New function.
13659 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13660 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13661 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13662 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13663 (mmix_constant_alignment): Make static. Use the same interface
13664 as the target hook.
13665 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13666 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13667 constant_alignment_word_strings.
13668 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13669 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13670 constant_alignment_word_strings.
13671 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13672 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13673 constant_alignment_word_strings.
13674 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13675 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13676 (rs6000_constant_alignment): New function.
13677 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13678 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13679 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13680 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13681 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13682 (rs6000_constant_alignment): New function.
13683 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13684 * config/s390/s390.c (s390_constant_alignment): New function.
13685 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13686 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13687 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13688 constant_alignment_word_strings.
13689 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13690 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13691 (sparc_constant_alignment): New function.
13692 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13693 * config/spu/spu.c (spu_constant_alignment): New function.
13694 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13695 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13696 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13697 constant_alignment_word_strings.
13698 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13699 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13700 constant_alignment_word_strings.
13701 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13702 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13703 constant_alignment_word_strings.
13704 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13705 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13706 (visium_constant_alignment): New function.
13707 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13708 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13709 (xtensa_constant_alignment): New function.
13710 * system.h (CONSTANT_ALIGNMENT): Poison.
13712 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13714 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13715 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13716 (rs6000_builtin_valid_without_lhs): New helper function.
13717 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13718 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13720 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13722 * target.h (vec_perm_indices): Use unsigned short rather than
13724 (auto_vec_perm_indices): Likewise.
13725 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13726 Use unsigned int rather than unsigned char.
13727 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13729 2017-09-25 Richard Biener <rguenther@suse.de>
13731 * cfgloop.h (sort_sibling_loops): Declare.
13732 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13733 (sort_sibling_loops): New function sorting the sibling loop list
13735 * graphite.c (graphite_transform_loops): Sort sibling loops.
13737 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13739 * target.def (vec_perm_const_ok): Change sel parameter to
13741 * optabs-query.c (can_vec_perm_p): Update accordingly.
13742 * doc/tm.texi: Regenerate.
13743 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13744 auto_vec_perm_indices and remove separate nelt field.
13745 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13746 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13747 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13748 (aarch64_expand_vec_perm_const): Update accordingly.
13749 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13750 to vec_perm_indices.
13751 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13752 auto_vec_perm_indices and remove separate nelt field.
13753 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13754 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13755 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13757 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13758 to vec_perm_indices.
13759 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13760 sel to vec_perm_indices.
13761 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13762 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13763 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13765 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13768 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13771 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13772 on the FUNCTION_DECL function context if it has a DIE that is a
13775 2017-09-25 Richard Biener <rguenther@suse.de>
13777 PR tree-optimization/82285
13778 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13781 2017-09-25 Tom de Vries <tom@codesourcery.com>
13785 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13788 2017-09-25 Richard Biener <rguenther@suse.de>
13790 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13791 ISL errors other than isl_error_quota happen. Dump if the
13792 schedule is the same.
13793 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13794 errors instead of aborting inside ISL.
13796 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13799 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13800 of libgcc_eh for m64.
13801 * config/i386/darwin64.h: Likewise.
13803 2017-09-25 Richard Biener <rguenther@suse.de>
13805 PR middle-end/82144
13806 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13807 attribute for incomplete types nor twice for complete ones.
13809 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13812 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13813 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13816 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13819 * config/i386/i386.opt (mprefer-avx256): Use
13820 ix86_target_flags variable.
13821 * config/i386/i386.c (ix86_target_string): Move
13822 -mprefer-avx256 to flag2_opts.
13824 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13826 PR middle-end/35691
13827 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13828 and x != -1 | y != -1 into (x & y) != -1.
13830 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13832 * config.gcc: Add new case statement to set
13833 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13834 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13835 s390x-*-linux* case statements. Added aarch64 to the list of
13836 supported architectures.
13838 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13840 PR tree-optimization/82289
13841 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13842 STMT_VINFO_RELEVANT_P.
13844 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13845 Alan Hayward <alan.hayward@arm.com>
13846 David Sherwood <david.sherwood@arm.com>
13848 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
13849 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13850 (extract_range_from_binary_expr_1): Don't call
13851 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
13853 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13854 Alan Hayward <alan.hayward@arm.com>
13855 David Sherwood <david.sherwood@arm.com>
13857 * target.def (preferred_vector_alignment): New hook.
13858 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13860 * doc/tm.texi: Regenerate.
13861 * targhooks.h (default_preferred_vector_alignment): Declare.
13862 * targhooks.c (default_preferred_vector_alignment): New function.
13863 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
13865 (DR_TARGET_ALIGNMENT): New macro.
13866 (aligned_access_p): Update commentary.
13867 (vect_known_alignment_in_bytes): New function.
13868 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
13870 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
13871 Calculate the misalignment based on the target alignment rather than
13873 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
13874 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
13875 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
13876 the target alignment, rather than masking the element misalignment
13877 with the number of elements in a vector. Also use the target
13878 alignment when calculating the maximum number of peels.
13879 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
13880 instead of TYPE_ALIGN_UNIT.
13881 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
13882 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
13883 (vect_create_addr_base_for_vector_ref): Update call accordingly.
13884 (vect_create_data_ref_ptr): Likewise.
13885 (vect_setup_realignment): Realign by ANDing with
13886 -DR_TARGET_MISALIGNMENT.
13887 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
13888 the number of peels based on DR_TARGET_ALIGNMENT.
13889 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
13890 with the guaranteed alignment boundary when deciding whether
13892 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
13893 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
13894 (ensure_base_align): Remove stmt_info parameter. Get the
13895 target base alignment from DR_TARGET_ALIGNMENT.
13896 (vectorizable_store): Update call accordingly. Interpret
13897 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
13899 (vectorizable_load): Likewise.
13901 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13902 Alan Hayward <alan.hayward@arm.com>
13903 David Sherwood <david.sherwood@arm.com>
13905 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
13906 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
13907 (vect_enhance_data_refs_alignment): Likewise.
13909 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13911 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13912 error. Only quit immediately if parsing is complete.
13913 (BEGIN): Initialize fatal_err and parse_done.
13914 (begin fpu, end fpu): Check number of arguments.
13915 (begin arch, end arch): Likewise.
13916 (begin cpu, end cpu): Likewise.
13917 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13918 (optalias): Likewise.
13920 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13922 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13923 * config/arm/arm-isa.h: Delete. Move definitions to ...
13924 * arm-cpus.in: ... here. Use new feature and fgroup values.
13925 * config/arm/arm.c (arm_option_override): Use lower case for feature
13927 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13928 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13929 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13931 (print_isa_bits_for): New function.
13932 (gen_isa): New function.
13933 (gen_comm_data): Use print_isa_bits_for.
13934 (define feature): New keyword.
13935 (define fgroup): New keyword.
13936 * config/arm/t-arm (TM_H): Remove.
13937 (GTM_H): Add arm-isa.h.
13938 (arm-isa.h): Add rule to generate file.
13939 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13940 case for feature bit names.
13942 2017-09-22 Richard Biener <rguenther@suse.de>
13944 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
13946 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
13947 print of no dependency loops ...
13948 * graphite.c (graphite_transform_loops): ... here.
13949 (canonicalize_loop_closed_ssa_form): Work from inner to outer
13951 (same_close_phi_node, remove_duplicate_close_phi,
13952 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
13953 (canonicalize_loop_closed_ssa): ... here and simplify.
13954 * graphite-optimize-isl.c: Include tree-vectorizer.h.
13955 (optimize_isl): Use dump_printf_loc to tell when we stopped
13956 optimizing because of an ISL timeout.
13958 2017-09-22 Richard Biener <rguenther@suse.de>
13960 PR tree-optimization/82291
13961 * tree-if-conv.c (predicate_mem_writes): Make sure to
13962 remove writes in blocks predicated with false.
13964 2017-09-22 Richard Biener <rguenther@suse.de>
13966 * sese.c: Include cfganal.h.
13967 (if_region_set_false_region): Remove.
13968 (create_if_region_on_edge): Likewise.
13969 (move_sese_in_condition): Re-implement without destroying
13972 2017-09-22 Richard Biener <rguenther@suse.de>
13974 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
13975 Verify both BBs contain loop PHI nodes before dispatching to
13976 copy_loop_phi_args.
13977 (graphite_regenerate_ast_isl): Do not recompute dominators,
13978 do not verify three times. Restructure for clarity.
13979 * graphite-scop-detection.c (same_close_phi_node,
13980 remove_duplicate_close_phi, make_close_phi_nodes_unique,
13981 defined_in_loop_p, canonicalize_loop_closed_ssa,
13982 canonicalize_loop_closed_ssa_form): Simplify, remove excess
13983 checking and SSA rewrite, move to ...
13984 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
13985 (graphite_initialize): Do not pass in ctx, do not reset the
13986 SCEV cache, compute only dominators.
13987 (graphite_transform_loops): Allocate ISL ctx after
13988 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
13989 Maintain post-dominators only around build_scops.
13990 * sese.c (if_region_set_false_region): Make static. Free
13991 and recompute dominators.
13992 (move_sese_in_condition): Assert we don't get called with
13993 post-dominators computed.
13994 * sese.h (if_region_set_false_region): Remove.
13996 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
13998 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
13999 mode attribute for TARGET_AVX512VL.
14001 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14003 * config/i386/i386.opt (mprefer-avx256): New option.
14004 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14006 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14007 for TARGET_PREFER_AVX256.
14008 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14010 2017-09-21 Jeff Law <law@redhat.com>
14012 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14013 Fix dump output if the only stack space is for pushed registers.
14015 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14017 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14020 2017-09-21 Martin Sebor <msebor@redhat.com>
14023 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14024 code (in C++) or code that triggers warnings.
14026 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14028 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14030 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14032 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14033 * sched-rgn.c: Ditto.
14034 * sel-sched-ir.c: Ditto.
14036 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
14038 * toplev.h (set_random_seed): Adjust return type.
14039 * toplev.c (init_local_tick): Move eager initialization of random_seed
14040 to get_random_seed. Adjust comment.
14041 (init_random_seed): Inline to get_random_seed, delete.
14042 (get_random_seed): Initialize random_seed lazily.
14043 (set_random_seed): Do not return previous value.
14044 (print_switch_value): Do not call get_random_seed.
14046 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14048 * cgraph.c (delete_function_version): New, broken out from...
14049 (cgraph_node::delete_function_version): ...here. Rename to
14050 cgraph_node::delete_function_version_by_decl. Update all uses.
14051 (cgraph_node::remove): Call delete_function_version.
14053 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14056 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14057 VAR_DECLs to which addressable non-volatile parameters are mapped
14058 and for id->retvar after the return value assignment. Clear
14059 id->retval and id->retbnd after inlining.
14061 2017-09-21 Richard Biener <rguenther@suse.de>
14063 PR tree-optimization/82276
14064 PR tree-optimization/82244
14065 * tree-vrp.c (build_assert_expr_for): Set
14066 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14068 (remove_range_assertions): Revert earlier change.
14070 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14073 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14075 2017-09-21 Richard Biener <rguenther@suse.de>
14077 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14078 Restore valid IL after code generation errors.
14079 * graphite.c (graphite_transform_loops): Diagnose code
14080 generation issues as MSG_MISSED_OPTIMIZATION and continue
14081 with processing SCOPs.
14083 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14084 Alan Hayward <alan.hayward@arm.com>
14085 David Sherwood <david.sherwood@arm.com>
14087 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14088 rather than choosing between plus_constant and gen_rtx_<CODE>.
14089 * expr.c (emit_push_insn): Likewise.
14090 (expand_expr_real_2): Likewise.
14092 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14093 Alan Hayward <alan.hayward@arm.com>
14094 David Sherwood <david.sherwood@arm.com>
14096 * loop-unroll.c (split_iv): Call copy_rtx on the step.
14098 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14099 Alan Hayward <alan.hayward@arm.com>
14100 David Sherwood <david.sherwood@arm.com>
14102 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14103 calling tree_to_uhwi.
14105 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14106 Alan Hayward <alan.hayward@arm.com>
14107 David Sherwood <david.sherwood@arm.com>
14109 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
14110 INTEGER_CST rather than a negative test for ADDR_EXPR.
14112 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14113 Alan Hayward <alan.hayward@arm.com>
14114 David Sherwood <david.sherwood@arm.com>
14116 * tree-vrp.c (extract_range_from_binary_expr_1): Check
14117 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
14119 2017-09-21 Richard Biener <rguenther@suse.de>
14121 PR tree-optimization/71351
14122 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
14123 graphite_create_new_loop_guard): Remove, fold remaining parts
14125 (translate_isl_ast_node_for): ... here and simplify.
14127 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14130 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
14131 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
14132 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
14133 alternative always use QI mode, for -Os imov (=R,R) alternative
14134 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
14137 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14138 Jeff Law <law@redhat.com>
14140 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
14141 (allocate_stack_space): New function, partially extracted from
14142 s390_emit_prologue.
14143 (s390_emit_prologue): Track offset to most recent stack probe.
14144 Code to allocate space moved into allocate_stack_space.
14145 Dump actions when no stack is allocated.
14146 (s390_prologue_plus_offset): New function.
14147 (s390_emit_stack_probe): Likewise.
14149 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
14151 * common.opt (Wa, Wl, Wp, g, gz=): Add
14153 (gno-column-info): Remove.
14154 (gcolumn-info): Drop RejectNegative.
14155 (gno-): New prefix.
14156 (gno-record-gcc-switches): Remove.
14157 (grecord-gcc-switches): Drop RejectNegative.
14158 (gno-split-dwarf): Remove.
14159 (gsplit-dwarf): Drop RejectNegative.
14160 (gno-strict-dwarf): Remove.
14161 (gstrict-dwarf): Drop RejectNegative.
14162 * config/darwin.opt (gfull, gused): Add RejectNegative.
14163 * dwarf2out.c (gen_producer_string): Drop
14164 gno-record-gcc-switches handler.
14165 * optc-gen.awk: Add g to prefixes with negative forms.
14166 * opts-common.c (remapping_prefix_p): New.
14167 (find_opt): Check it.
14168 (generate_canonical_option): Test g prefix.
14169 (option_map): Add -gno- mapping.
14170 (add_misspelling_candidates): Check remapping_prefix_p.
14172 2017-09-20 Jeff Law <law@redhat.com>
14174 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
14175 thinko in stack clash protection support.
14177 * explow.c (compute_stack_clash_protection_loop_data): Use
14178 CONST_INT_P instead of explicit test. Verify object is a
14179 CONST_INT_P before looking at INTVAL.
14180 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
14181 instead of explicit test.
14183 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
14186 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
14187 address instead of to r1 and r11.
14189 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
14191 * config.gcc: Support "knm".
14192 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
14193 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14195 * config/i386/i386.c (m_KNM): Define.
14196 (processor_target_table): Add "knm".
14198 (ix86_option_override_internal): Add "knm".
14199 (ix86_issue_rate): Add PROCESSOR_KNM.
14200 (ix86_adjust_cost): Ditto.
14201 (ia32_multipass_dfa_lookahead): Ditto.
14202 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
14203 (fold_builtin_cpu): Add M_INTEL_KNM.
14204 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
14205 (processor_type): Add PROCESSOR_KNM.
14206 * config/i386/x86-tune.def: Add m_KNM.
14207 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
14209 2017-09-20 Richard Biener <rguenther@suse.de>
14211 PR tree-optimization/80213
14212 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
14213 are allowed in empty BBs as well.
14214 (canonicalize_loop_closed_ssa): Also look for other complex
14216 (scop_detection::get_sese): Include the loop-closed PHI block
14218 (scop_detection::merge_sese): Remove code adding extra blocks.
14219 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
14220 (build_scops): Assert the final returned scop is invalid.
14222 2017-09-20 Richard Biener <rguenther@suse.de>
14224 PR tree-optimization/82264
14225 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
14227 (vn_phi_lookup): Likewise.
14228 (vn_phi_insert): Likewise.
14230 2017-09-20 Jakub Jelinek <jakub@redhat.com>
14232 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
14233 that fits into uhwi or shwi, add DW_AT_const_value regardless
14234 of early_dwarf without going through RTL, using add_AT_unsigned
14237 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
14238 (DEBUG_LTO_ABBREV_SECTION): Likewise.
14239 (DEBUG_LTO_MACINFO_SECTION): Likewise.
14240 (DEBUG_MACRO_SECTION): Likewise.
14241 (DEBUG_LTO_MACRO_SECTION): Likewise.
14242 (DEBUG_STR_DWO_SECTION): Likewise.
14243 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
14244 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
14245 (DEBUG_LTO_DWO_LINE_SECTION): Define.
14246 (DEBUG_LTO_LINE_STR_SECTION): Define.
14247 (init_sections_and_labels): Initialize debug_line_str_section
14248 variable. Initialize debug_loc_section for -gdwarf-5 to
14249 DEBUG_LOCLISTS_SECTION. Formatting fixes.
14251 2017-09-20 Richard Biener <rguenther@suse.de>
14253 * graphite-sese-to-poly.c (extract_affine): Properly handle
14254 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
14256 2017-09-20 Richard Biener <rguenther@suse.de>
14258 PR tree-optimization/81373
14259 * graphite-scop-detection.c (build_cross_bb_scalars_def):
14260 Force SESE live-out defs to be handled even if they are
14263 2017-09-19 Jeff Law <law@redhat.com>
14265 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
14266 nothing for stack adjustments with REG_STACK_CHECK.
14267 * sched-deps.c (parse_add_or_inc): Reject insns with
14268 REG_STACK_CHECK from dependency breaking.
14269 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
14270 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
14271 * reg-notes.def (STACK_CHECK): New note.
14273 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
14274 (ix86_expand_prologue): Dump stack clash info as needed.
14275 Call ix86_adjust_stack_and_probe_stack_clash as needed.
14277 * function.c (dump_stack_clash_frame_info): New function.
14278 * function.h (dump_stack_clash_frame_info): Prototype.
14279 (enum stack_clash_probes): New enum.
14281 * config/alpha/alpha.c (alpha_expand_prologue): Also check
14282 flag_stack_clash_protection.
14283 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
14284 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
14285 (arm_frame_pointer_required): Likewise.
14286 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
14287 (ia64_expand_prologue): Likewise.
14288 * config/mips/mips.c (mips_expand_prologue): Likewise.
14289 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
14290 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14291 (sparc_flat_expand_prologue): Likewise.
14292 * config/spu/spu.c (spu_expand_prologue): Likewise.
14294 * explow.c: Include "params.h".
14295 (anti_adjust_stack_and_probe_stack_clash): New function.
14296 (get_stack_check_protect): Likewise.
14297 (compute_stack_clash_protection_loop_data): Likewise.
14298 (emit_stack_clash_protection_loop_start): Likewise.
14299 (emit_stack_clash_protection_loop_end): Likewise.
14300 (allocate_dynamic_stack_space): Use get_stack_check_protect.
14301 Use anti_adjust_stack_and_probe_stack_clash.
14302 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
14303 (emit_stack_clash_protection_loop_start): Likewise.
14304 (emit_stack_clash_protection_loop_end): Likewise.
14305 * rtl.h (get_stack_check_protect): Prototype.
14306 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
14307 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
14308 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
14310 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
14312 * doc/tm.texi: Rebuilt.
14313 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
14314 get_stack_check_protect.
14315 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
14316 * config/arm/arm.c (arm_expand_prologue): Likewise.
14317 (arm_frame_pointer_required): Likewise.
14318 * config/i386/i386.c (ix86_expand_prologue): Likewise.
14319 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
14320 * config/mips/mips.c (mips_expand_prologue): Likewise.
14321 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
14322 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
14323 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14324 (sparc_flat_expand_prologue): Likewise.
14326 * common.opt (-fstack-clash-protection): New option.
14327 * flag-types.h (enum stack_check_type): Note difference between
14328 -fstack-check= and -fstack-clash-protection.
14329 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
14330 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
14331 * toplev.c (process_options): Issue warnings/errors for cases
14332 not handled with -fstack-clash-protection.
14333 * doc/invoke.texi (-fstack-clash-protection): Document new option.
14334 (-fstack-check): Note additional problem with -fstack-check=generic.
14335 Note that -fstack-check is primarily for Ada and refer users
14336 to -fstack-clash-protection for stack-clash-protection.
14337 Document new params for stack clash protection.
14339 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
14341 * config/i386/i386.c (ix86_split_long_move): Do not handle
14342 address used for LEA in a special way.
14344 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
14346 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
14348 2017-09-19 Martin Sebor <msebor@redhat.com>
14351 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
14352 of incompatible types.
14354 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
14356 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14357 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
14358 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14359 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
14361 2017-09-19 Richard Biener <rguenther@suse.de>
14363 PR tree-optimization/82244
14364 * tree-vrp.c (remove_range_assertions): Do not propagate
14365 a constant to abnormals but replace the assert with a copy.
14367 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
14369 PR rtl-optimization/57878
14370 PR rtl-optimization/68988
14371 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
14372 avoidance test involving non_reload_pseudos. Move frequency test
14373 below the general fragmentation avoidance test.
14375 2017-09-19 Richard Biener <rguenther@suse.de>
14377 PR tree-optimization/69728
14378 * graphite-sese-to-poly.c (schedule_error): New global.
14379 (add_loop_schedule): Handle empty domain by failing the
14381 (build_original_schedule): Handle schedule_error.
14383 2017-09-19 Richard Biener <rguenther@suse.de>
14385 * graphite-scop-detection.c (scop_detection::can_represent_loop):
14386 Do not iterate to sibling loops but only to siblings of inner
14389 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
14392 * config/m68k/m68k.md (moveq feeding equality comparison): Check
14393 that the registers are different.
14395 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
14397 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
14398 to processor_model and "amdfam17h" to arch_names_table.
14399 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
14401 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14404 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
14406 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14407 Alan Hayward <alan.hayward@arm.com>
14408 David Sherwood <david.sherwood@arm.com>
14410 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
14412 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
14414 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
14415 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
14416 vect_schedule_slp_instance.
14417 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
14418 Update call to vect_slp_analyze_node_operations. Simplify return
14420 (vect_slp_analyze_bb_1): Update call accordingly.
14421 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
14422 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
14423 (vect_schedule_slp): Update call accordingly.
14425 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14426 Alan Hayward <alan.hayward@arm.com>
14427 David Sherwood <david.sherwood@arm.com>
14429 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
14430 with types that aren't in fact scalar.
14432 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14434 * tree-vect-slp.c (vect_record_max_nunits): New function,
14436 (vect_build_slp_tree_1): ...here.
14437 (vect_build_slp_tree_2): Call it for phis too.
14439 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14441 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
14442 to vect_get_vec_def_for_operand when getting the mask operand.
14444 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14445 Alan Hayward <alan.hayward@arm.com>
14446 David Sherwood <david.sherwood@arm.com>
14448 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
14451 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14452 Alan Hayward <alan.hayward@arm.com>
14453 David Sherwood <david.sherwood@arm.com>
14455 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
14456 calculation for vector booleans.
14458 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14459 Alan Hayward <alan.hayward@arm.com>
14460 David Sherwood <david.sherwood@arm.com>
14462 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
14464 (vect_transform_stmt): ...here.
14465 (vect_analyze_stmt): Use it instead of calling
14466 vectorizable_live_operation directly.
14468 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
14470 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
14471 non-SIMT targets in acc vector loops.
14473 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
14475 * configure.ac: Add arc and check if assembler supports gdwarf2.
14476 * configure: Regenerate.
14478 2017-09-18 Richard Biener <rguenther@suse.de>
14480 PR tree-optimization/82220
14481 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
14482 epilogue niters from the min_profitable_iters compute.
14484 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14487 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
14488 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
14489 (ix86_init_pic_reg): Revert 2017-09-01 changes.
14491 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
14494 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
14495 switching to a new text section.
14497 2017-09-18 Richard Biener <rguenther@suse.de>
14499 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
14501 (build_alias_set): Reject aliases with no access function.
14503 2017-09-18 Richard Biener <rguenther@suse.de>
14505 PR tree-optimization/79622
14506 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
14508 (build_cross_bb_scalars_use): Likewise.
14510 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
14512 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
14514 2017-09-18 Alan Modra <amodra@gmail.com>
14517 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
14518 stack_pointer_rtx for count 0. Update comments. Break up
14519 large rtl expression.
14521 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
14523 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
14524 Increase to 20 bytes.
14525 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
14526 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
14527 or avx version of the stub.
14529 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
14532 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
14533 compute the minimum stack alignment. Also update preferred stack
14534 boundary for leaf functions.
14536 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
14538 PR tree-optimization/82228
14539 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
14542 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14544 * common/config/nds32/nds32-common.c
14545 (nds32_option_optimization_table): Refine formatting.
14546 (nds32_option_optimization_table): Use -fsched-pressure and
14547 -fomit-frame-pointer for specific optimization level.
14549 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14551 * config/nds32/nds32.c: Refine formatting and comments.
14552 * config/nds32/nds32.h: Likewise.
14553 * config/nds32/nds32.md: Likewise.
14554 * config/nds32/nds32-cost.c: Likewise.
14555 * config/nds32/nds32-isr.c: Likewise.
14556 * config/nds32/nds32-md-auxiliary.c: Likewise.
14557 * config/nds32/nds32-multiple.md: Likewise.
14558 * config/nds32/nds32-predicates.c: Likewise.
14560 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
14561 Jakub Jelinek <jakub@redhat.com>
14563 Add support for -std=c++2a.
14564 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
14566 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
14568 2017-09-15 Steve Ellcey <sellcey@cavium.com>
14571 * doc/extend.texi (Common Function Attributes): Add
14572 references to ARM, AArch64, and S/390 specific attributes.
14573 (Function Specific Option Pragmas): Add AArch64 and S/390
14574 to list of back ends that support the target pragma.
14576 2017-09-15 Nathan Sidwell <nathan@acm.org>
14578 * doc/standards.texi: Fix C++17 description. Update URLs for
14581 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14583 * common.opt (Wcast-align=strict): New warning option.
14584 * doc/invoke.texi: Document -Wcast-align=strict.
14586 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
14588 * cgraph.h (cgraph_thunk_info): Add comments.
14589 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
14590 assert for VIRTUAL_* arguments stricter.
14592 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
14594 PR tree-optimization/71026
14595 * match.pd: Move RDIV patterns from fold-const.c
14596 * fold-const.c (distribute_real_division): Removed.
14597 (fold_binary_loc): Remove calls to distribute_real_divison.
14599 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14601 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
14602 c++1z and gnu++1z as deprecated. Change other references to
14603 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
14604 Change -Wc++1z-compat to -Wc++17-compat.
14605 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
14606 * dwarf2out.c (highest_c_language): Handle C++17.
14607 (gen_compile_unit_die): Likewise.
14609 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14611 PR rtl-optimization/82192
14612 * combine.c (make_extraction): Don't look through non-paradoxical
14613 SUBREGs or TRUNCATE if pos + len is or might be bigger than
14616 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14617 Alan Hayward <alan.hayward@arm.com>
14618 David Sherwood <david.sherwood@arm.com>
14620 * target.def (function_arg_offset): New hook.
14621 * targhooks.h (default_function_arg_offset): Declare.
14622 * targhooks.c (default_function_arg_offset): New function.
14623 * function.c (locate_and_pad_parm): Use
14624 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14625 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14626 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14627 * doc/tm.texi: Regenerate.
14628 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14629 * config/spu/spu.c (spu_function_arg_offset): New function.
14630 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14631 * system.h (FUNCTION_ARG_OFFSET): Poison.
14633 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14634 Alan Hayard <alan.hayward@arm.com>
14635 David Sherwood <david.sherwood@arm.com>
14637 * target.def (truly_noop_truncation): New hook.
14638 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14639 than TRULY_NOOP_TRUNCATION.
14640 * hooks.h (hook_bool_uint_uint_true): Declare.
14641 * hooks.c (hook_bool_uint_uint_true): New function.
14642 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14643 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14644 * doc/tm.texi: Regenerate.
14645 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14646 rather than TRULY_NOOP_TRUNCATION in comments.
14647 (simplify_comparison): Likewise.
14648 (record_truncated_value): Likewise.
14649 * expmed.c (extract_bit_field_1): Likewise.
14650 (extract_split_bit_field): Likewise.
14651 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14652 instead of TRULY_NOOP_TRUNCATION.
14653 * function.c (assign_parm_setup_block): Likewise.
14654 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14655 * rtlhooks.c: Include target.h.
14656 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14657 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14658 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14659 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14660 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14661 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14662 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14663 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14664 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14665 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14666 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14667 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14668 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14669 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14670 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14671 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14672 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14673 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14674 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14675 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14676 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14677 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14678 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14679 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14680 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14681 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14682 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14683 rather than TRULY_NOOP_TRUNCATION in comments.
14684 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14685 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14686 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14687 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14688 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14689 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14690 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14691 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14692 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14693 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14694 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14695 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14696 rather than TRULY_NOOP_TRUNCATION in comments.
14697 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14698 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14699 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14700 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14701 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14702 TRULY_NOOP_TRUNCATION condition.
14703 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14704 (TRULY_NOOP_TRUNCATION): Delete.
14705 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14706 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14707 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14708 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14709 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14710 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14711 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14712 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14713 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14714 rather than TRULY_NOOP_TRUNCATION in comments.
14715 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14716 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14717 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14718 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14719 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14720 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14722 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14725 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14726 (*cmp_ior): Likewise.
14727 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14728 (*ior_scc_scc_cmp): Likewise.
14729 (*and_scc_scc): Likewise.
14730 (*and_scc_scc_cmp): Likewise.
14732 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14733 Alan Hayard <alan.hayward@arm.com>
14734 David Sherwood <david.sherwood@arm.com>
14736 * target.def (can_change_mode_class): New hook.
14737 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14738 (hard_regno_nregs): Likewise.
14739 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14740 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14741 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14742 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14743 (LOAD_EXTEND_OP): Update accordingly.
14744 * doc/tm.texi: Regenerate.
14745 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14746 CANNOT_CHANGE_MODE_CLASS.
14747 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14748 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14749 * combine.c (simplify_set): Update accordingly.
14750 * emit-rtl.c (validate_subreg): Likewise.
14751 * recog.c (general_operand): Likewise.
14752 * regcprop.c (mode_change_ok): Likewise.
14753 * reload1.c (choose_reload_regs): Likewise.
14754 (inherit_piecemeal_p): Likewise.
14755 * rtlanal.c (simplify_subreg_regno): Likewise.
14756 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14757 instead of CANNOT_CHANGE_MODE_CLASS.
14758 (reload_cse_simplify_operands): Likewise.
14759 * reload.c (push_reload): Use targetm.can_change_mode_class
14760 instead of CANNOT_CHANGE_MODE_CLASS.
14761 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14762 REG_CANNOT_CHANGE_MODE_P.
14763 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14764 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14765 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14766 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14767 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14768 (arm_can_change_mode_class): New function.
14769 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14770 than CANNOT_CHANGE_MODE_CLASS in comments.
14771 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14772 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14773 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14774 (ix86_can_change_mode_class): ...this new function, inverting the
14775 sense of the return value.
14776 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14777 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14778 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14779 (ia64_can_change_mode_class): New function.
14780 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14781 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14782 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14783 (m32c_can_change_mode_class): ...this new function, inverting the
14784 sense of the return value.
14785 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14786 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14787 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14788 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14789 (mips_can_change_mode_class): ...this new function, inverting the
14790 sense of the return value.
14791 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14792 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14793 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14794 (msp430_can_change_mode_class): New function.
14795 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14796 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14797 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14798 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14799 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14800 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14801 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14802 (pa_cannot_change_mode_class): Replace with...
14803 (pa_can_change_mode_class): ...this new function, inverting the
14804 sense of the return value.
14805 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14806 than CANNOT_CHANGE_MODE_CLASS in comments.
14807 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14808 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14809 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14810 (pdp11_cannot_change_mode_class): Replace with...
14811 (pdp11_can_change_mode_class): ...this new function, inverting the
14812 sense of the return value.
14813 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14814 * config/powerpcspe/powerpcspe-protos.h
14815 (rs6000_cannot_change_mode_class_ptr): Delete.
14816 * config/powerpcspe/powerpcspe.c
14817 (rs6000_cannot_change_mode_class_ptr): Delete.
14818 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14819 (rs6000_option_override_internal): Assign to
14820 targetm.can_change_mode_class instead of
14821 rs6000_cannot_change_mode_class_ptr.
14822 (rs6000_cannot_change_mode_class): Replace with...
14823 (rs6000_can_change_mode_class): ...this new function, inverting the
14824 sense of the return value.
14825 (rs6000_debug_cannot_change_mode_class): Replace with...
14826 (rs6000_debug_can_change_mode_class): ...this new function.
14827 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14828 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14829 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14830 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14831 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14833 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14834 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14835 (rs6000_option_override_internal): Assign to
14836 targetm.can_change_mode_class instead of
14837 rs6000_cannot_change_mode_class_ptr.
14838 (rs6000_cannot_change_mode_class): Replace with...
14839 (rs6000_can_change_mode_class): ...this new function, inverting the
14840 sense of the return value.
14841 (rs6000_debug_cannot_change_mode_class): Replace with...
14842 (rs6000_debug_can_change_mode_class): ...this new function.
14843 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14844 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14845 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14846 (s390_can_change_mode_class): ...this new function, inverting the
14847 sense of the return value.
14848 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14849 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14850 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14851 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14852 (sh_cannot_change_mode_class): Replace with...
14853 (sh_can_change_mode_class): ...this new function, inverting the
14854 sense of the return value.
14855 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14856 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14857 (sparc_can_change_mode_class): New function.
14858 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14859 * config/spu/spu.c (spu_can_change_mode_class): New function.
14860 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14861 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14862 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14863 (visium_can_change_mode_class): New function.
14864 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
14866 2017-09-15 Richard Biener <rguenther@suse.de>
14868 PR tree-optimization/82217
14869 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
14870 but not undefined case.
14872 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14875 * postreload.c (reload_cse_simplify_operands): Skip
14876 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
14878 2017-09-15 Richard Biener <rguenther@suse.de>
14880 PR tree-optimization/68823
14881 * graphite-scop-detection.c (build_alias_set): If we have a
14882 possible dependence check whether we can handle them by just
14883 looking at the DRs DR_ACCESS_FNs.
14884 (build_scops): If build_alias_set fails, fail the SCOP.
14886 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
14888 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
14889 to support float128 built-in functions that require the ISA 3.0
14891 (BU_FLOAT128_3_HW): Likewise.
14892 (SQRTF128): Add support for the IEEE 128-bit square root and fma
14893 built-in functions.
14894 (FMAF128): Likewise.
14896 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14897 support for built-in functions that need the ISA 3.0 IEEE 128-bit
14898 floating point instructions.
14899 (rs6000_invalid_builtin): Likewise.
14900 (rs6000_builtin_mask_names): Likewise.
14901 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
14902 (RS6000_BTM_FLOAT128_HW): Likewise.
14903 (RS6000_BTM_COMMON): Likewise.
14904 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
14906 * doc/extend.texi (RS/6000 built-in functions): Document the
14907 IEEE 128-bit floating point square root and fused multiply-add
14908 built-in functions.
14910 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
14912 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
14913 reg (r2) isn't in the set of registers defined in the prologue.
14915 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14916 Alan Hayward <alan.hayward@arm.com>
14917 David Sherwood <david.sherwood@arm.com>
14919 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
14920 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
14921 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
14922 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
14923 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
14925 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
14926 max_vectorization_factor.
14927 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
14929 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14930 Alan Hayward <alan.hayward@arm.com>
14931 David Sherwood <david.sherwood@arm.com>
14933 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
14934 (vect_worthwhile_without_simd_p): Declare.
14935 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
14936 (vectorizable_reduction): Use it.
14937 * tree-vect-stmts.c (vectorizable_shift): Likewise.
14938 (vectorizable_operation): Likewise.
14940 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14941 Alan Hayward <alan.hayward@arm.com>
14942 David Sherwood <david.sherwood@arm.com>
14944 * tree-vectorizer.h (vect_get_num_copies): New function.
14945 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
14946 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14947 (vectorizable_induction): Likewise.
14948 (vectorizable_live_operation): Likewise.
14949 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14950 (vectorizable_bswap): Likewise.
14951 (vectorizable_call): Likewise.
14952 (vectorizable_conversion): Likewise.
14953 (vectorizable_assignment): Likewise.
14954 (vectorizable_shift): Likewise.
14955 (vectorizable_operation): Likewise.
14956 (vectorizable_store): Likewise.
14957 (vectorizable_load): Likewise.
14958 (vectorizable_condition): Likewise.
14959 (vectorizable_comparison): Likewise.
14960 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
14962 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14963 Alan Hayward <alan.hayward@arm.com>
14964 David Sherwood <david.sherwood@arm.com>
14966 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
14967 of vect_init_vector.
14969 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14970 Alan Hayward <alan.hayward@arm.com>
14971 David Sherwood <david.sherwood@arm.com>
14973 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
14974 an inline wrapper that provides a location.
14975 (gimple_build_vector): Likewise.
14976 * gimple-fold.c (gimple_build_vector_from_val): New function.
14977 (gimple_build_vector): Likewise.
14978 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
14979 functions to build the initial value. Always return a gimple value.
14980 (get_initial_defs_for_reduction): Likewise. Only compute
14982 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
14983 vect_init_vector on the results from get_initial_def(s)_for_reduction.
14984 (vectorizable_induction): Use gimple_build_vector rather than
14987 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14988 Alan Hayward <alan.hayward@arm.com>
14989 David Sherwood <david.sherwood@arm.com>
14991 * target.h (vec_perm_indices): New typedef.
14992 (auto_vec_perm_indices): Likewise.
14993 * optabs-query.h: Include target.h
14994 (can_vec_perm_p): Take a vec_perm_indices *.
14995 * optabs-query.c (can_vec_perm_p): Likewise.
14996 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
14997 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14998 * tree-vect-generic.c (lower_vec_perm): Likewise.
14999 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15000 (vect_grouped_load_supported): Likewise.
15001 (vect_shift_permute_load_chain): Likewise.
15002 (vect_permute_store_chain): Use auto_vec_perm_indices.
15003 (vect_permute_load_chain): Likewise.
15004 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15005 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15006 Update uses of can_vec_perm_p.
15007 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15008 mode with a number of elements. Take a vec_perm_indices *.
15009 (vect_create_epilog_for_reduction): Update accordingly.
15010 Use auto_vec_perm_indices.
15011 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15012 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15013 (vect_transform_slp_perm_load): Likewise.
15014 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15015 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15016 (vect_gen_perm_mask_checked): Likewise.
15017 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15018 (vect_gen_perm_mask_checked): Likewise.
15019 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15020 (vectorizable_store): Likewise.
15021 (vectorizable_load): Likewise.
15022 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15023 (vectorizable_bswap): Likewise.
15025 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15026 Alan Hayward <alan.hayward@arm.com>
15027 David Sherwood <david.sherwood@arm.com>
15029 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15030 * tree.c (build_vector): Likewise.
15031 (build_vector_from_ctor): Update accordingly.
15032 (build_vector_from_val): Likewise.
15033 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15034 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15035 * tree-vect-generic.c (add_rshift): Likewise.
15036 (expand_vector_divmod): Likewise.
15037 (optimize_vector_constructor): Likewise.
15038 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15039 (vect_transform_slp_perm_load): Likewise.
15040 (vect_schedule_slp_instance): Likewise.
15041 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15042 (vectorizable_call): Likewise.
15043 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15044 * expmed.c (make_tree): Likewise.
15045 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15046 a vector passed to build_vector.
15047 (fold_convert_const): Likewise.
15048 (exact_inverse): Likewise.
15049 (fold_ternary_loc): Likewise.
15050 (fold_relational_const): Likewise.
15051 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15052 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15053 (const_unop): Likewise. Store the reduction accumulator in a
15054 variable rather than an array.
15055 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15056 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15057 the new vector, rather than constructing it after the input arrays.
15058 (native_interpret_vector): Use auto_vec<tree> when building
15059 a vector passed to build_vector. Add elements in order.
15060 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15061 auto_vec<tree> when building a vector passed to build_vector.
15062 (vect_create_epilog_for_reduction): Likewise.
15063 (vectorizable_induction): Likewise.
15064 (get_initial_def_for_reduction): Likewise. Fix indentation of
15066 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15068 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15069 passed to build_vector.
15071 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15072 Alan Hayward <alan.hayward@arm.com>
15073 David Sherwood <david.sherwood@arm.com>
15075 * tree-core.h (tree_base::u): Add an "nelts" field.
15076 (tree_vector): Use VECTOR_CST_NELTS as the length.
15077 * tree.c (tree_size): Likewise.
15078 (make_vector): Initialize VECTOR_CST_NELTS.
15079 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15080 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15081 TYPE_VECTOR_SUBPARTS.
15082 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15083 as the number of units, setting it from VECTOR_CST_NELTS.
15084 (const_vector_from_tree): Likewise.
15085 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15086 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15087 (fold_negate_expr_1): Likewise.
15088 (fold_convert_const): Likewise.
15089 (const_binop): Likewise. Differentiate the number of output and
15091 (const_unop): Likewise.
15092 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15093 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15094 in cases that did the opposite.
15096 2017-09-14 Richard Biener <rguenther@suse.de>
15098 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15101 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15103 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15105 2017-09-14 Jakub Jelinek <jakub@redhat.com>
15108 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
15109 if and where to split a bb, except for splitting before debug insn
15110 sequences followed by non-label real insn. Delete debug insns
15111 in between basic blocks.
15113 * combine.c (make_compound_operation_int): Formatting fixes.
15115 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
15116 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15117 * config/netbsd.h (LINK_EH_SPEC): Likewise.
15118 * config/sol2.h (LINK_EH_SPEC): Likewise.
15119 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15120 * config/s390/linux.h (LINK_SPEC): Likewise.
15121 * config/freebsd.h (LINK_EH_SPEC): Likewise.
15122 * config/openbsd.h (LINK_EH_SPEC): Likewise.
15123 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15124 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
15125 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
15126 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15127 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
15128 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
15130 2017-09-13 Jakub Jelinek <jakub@redhat.com>
15132 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
15134 (ENDFILE_LINUX_SPEC): Likewise.
15135 (LINK_EH_SPEC): Likewise.
15136 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
15137 (LINK_OS_LINUX_SPEC32): Likewise.
15138 (LINK_OS_LINUX_SPEC64): Likewise.
15139 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
15140 (LINK_OS_LINUX_SPEC): Likewise.
15142 2017-09-13 Martin Liska <mliska@suse.cz>
15144 PR middle-end/82154
15145 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
15146 CASE_HIGH is NULL_TREE.
15148 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15149 Alan Hayward <alan.hayward@arm.com>
15150 David Sherwood <david.sherwood@arm.com>
15152 * target.def (secondary_memory_needed): New hook.
15153 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
15154 instead of SECONDARY_MEMORY_NEEDED.
15155 (secondary_memory_needed_mode): Likewise.
15156 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
15157 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
15158 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
15159 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
15160 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
15161 * doc/tm.texi: Regenerate.
15162 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
15163 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
15164 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15165 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
15166 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
15167 * config/i386/i386.c (inline_secondary_memory_needed): Put the
15168 mode argument first and change the reg_class arguments to reg_class_t.
15169 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
15170 Make static. Update the call to inline_secondary_memory_needed.
15171 (ix86_register_move_cost): Update the call to
15172 inline_secondary_memory_needed.
15173 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15174 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
15176 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
15177 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15179 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
15180 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
15181 * config/mips/mips.c (mips_secondary_memory_needed): Make static
15182 and match hook interface. Add comment from mips.h.
15183 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15184 * config/mmix/mmix.md (truncdfsf2): Refer to
15185 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15187 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
15188 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
15189 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15190 (pa_secondary_memory_needed): New function.
15191 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
15192 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
15193 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15194 (pdp11_secondary_memory_needed): Make static and match hook interface.
15195 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
15196 * config/powerpcspe/powerpcspe-protos.h
15197 (rs6000_secondary_memory_needed_ptr): Delete.
15198 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
15200 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15201 (rs6000_option_override_internal): Assign to
15202 targetm.secondary_memory_needed rather than
15203 rs6000_secondary_memory_needed_ptr.
15204 (rs6000_secondary_memory_needed): Match hook interface.
15205 (rs6000_debug_secondary_memory_needed): Likewise.
15206 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
15207 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
15208 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
15209 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15210 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
15211 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
15213 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
15214 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15215 (rs6000_option_override_internal): Assign to
15216 targetm.secondary_memory_needed rather than
15217 rs6000_secondary_memory_needed_ptr.
15218 (rs6000_secondary_memory_needed): Match hook interface.
15219 (rs6000_debug_secondary_memory_needed): Likewise.
15220 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
15221 * config/s390/s390.c (s390_secondary_memory_needed): New function.
15222 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15223 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
15224 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15225 (sparc_secondary_memory_needed): New function.
15226 * lra-constraints.c (check_and_process_move): Refer to
15227 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15229 (curr_insn_transform): Likewise.
15230 (process_alt_operands): Use targetm.secondary_memory_needed
15231 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15232 (check_secondary_memory_needed_p): Likewise.
15233 (choose_split_class): Likewise.
15234 * reload.c: Unconditionally include code that was previously
15235 conditional on SECONDARY_MEMORY_NEEDED.
15236 (push_secondary_reload): Use targetm.secondary_memory_needed
15237 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15238 (push_reload): Likewise.
15239 * reload1.c: Unconditionally include code that was previously
15240 conditional on SECONDARY_MEMORY_NEEDED.
15241 (choose_reload_regs): Use targetm.secondary_memory_needed
15242 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15243 (gen_reload): Likewise.
15244 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
15246 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15247 Alan Hayward <alan.hayward@arm.com>
15248 David Sherwood <david.sherwood@arm.com>
15250 * target.def (secondary_memory_needed_mode): New hook:
15251 * targhooks.c (default_secondary_memory_needed_mode): Declare.
15252 * targhooks.h (default_secondary_memory_needed_mode): New function.
15253 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
15254 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
15255 * doc/tm.texi: Regenerate.
15256 * lra-constraints.c (check_and_process_move): Use
15257 targetm.secondary_memory_needed_mode instead of
15258 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
15259 (curr_insn_transform): Likewise.
15260 * reload.c (get_secondary_mem): Likewise.
15261 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15262 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
15264 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15265 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15266 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
15267 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15268 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
15270 * config/powerpcspe/powerpcspe-protos.h
15271 (rs6000_secondary_memory_needed_mode): Delete.
15272 * config/powerpcspe/powerpcspe.c
15273 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15274 (rs6000_secondary_memory_needed_mode): Make static.
15275 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15276 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
15278 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15280 (rs6000_secondary_memory_needed_mode): Make static.
15281 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15282 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
15283 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15284 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15285 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15287 (sparc_secondary_memory_needed_mode): New function.
15288 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
15290 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
15292 * config/aarch64/constraints.md (Umq): New constraint.
15293 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15295 (mov<mode>): Update condition.
15297 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15299 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
15300 when bitposition is the same.
15302 2017-09-13 Richard Biener <rguenther@suse.de>
15304 * dwarf2out.c (output_die_symbol): Remove.
15305 (output_die): Do not output a DIEs symbol.
15307 2017-09-13 Richard Biener <rguenther@suse.de>
15309 PR middle-end/82128
15310 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
15311 default-def to avoid breaking iterator update with the weird
15312 interaction with cgraph_update_edges_for_call_stmt_node.
15314 2017-09-13 Richard Biener <rguenther@suse.de>
15316 * tree-cfg.c (verify_gimple_assign_binary): Add verification
15317 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
15318 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
15319 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
15321 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
15323 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
15324 Disable pc relative literal load irrespective of
15325 TARGET_FIX_ERR_A53_84341 for default.
15327 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
15329 * config/sparc/sparc.c (output_return): Output the source location of
15330 the insn in the delay slot, if any.
15331 (output_sibcall): Likewise.
15333 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
15336 * common.opt (-static-pie): New alias.
15337 (shared): Negate static-pie.
15338 (-no-pie): Update help text.
15340 (static-pie): New option.
15341 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
15342 -static-pie support.
15343 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
15344 (LINK_EH_SPEC): Likewise.
15345 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15346 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15347 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15348 * gcc.c (LINK_COMMAND_SPEC): Likewise.
15349 (init_gcc_specs): Likewise.
15350 (init_spec): Likewise.
15351 (display_help): Update help message for -pie.
15352 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
15355 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
15357 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
15358 (movdi_aarch64): Likewise.
15359 (movti_aarch64): Likewise.
15361 2017-09-12 Simon Wright <simon@pushface.org>
15364 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
15365 calculation of the minor version, always output as 0.
15367 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15370 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15371 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
15372 on it early, rather than manual conversion late. For
15373 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
15374 instead of performing manual conversion.
15376 2017-09-12 Carl Love <cel@us.ibm.com>
15378 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
15379 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
15381 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15382 VMULOSW): Add definitions.
15383 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15384 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15385 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15386 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15387 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15389 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15391 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
15393 (movti_aarch64): Likewise.
15394 (movdf_aarch64): Likewise.
15395 (movtf_aarch64): Likewise.
15396 (load_pairdi): Likewise.
15397 (store_pairdi): Likewise.
15398 (load_pairdf): Likewise.
15399 (store_pairdf): Likewise.
15400 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
15401 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
15402 (ldr_got_small_<mode>): Likewise.
15403 (ldr_got_small_28k_<mode>): Likewise.
15404 (ldr_got_tiny): Likewise.
15405 * config/aarch64/iterators.md (ldst_sz): New.
15406 (ldpstp_sz): Likewise.
15407 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
15409 (thunderx_load): Split load_8 to load_16.
15410 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
15412 (thunderx2t99_storepair_basic): Split store_8 to store_16.
15413 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
15414 (xgene1_store_pair): Split store_8 to store_16.
15415 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
15416 (falkor_st_0_st_sd): Split store_8 to store_16.
15418 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15420 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
15421 and store1/2/3/4 to store_4/8/12/16.
15422 * config/aarch64/aarch64.md: Update for rename.
15423 * config/arm/arm.md: Likewise.: Likewise.
15424 * config/arm/arm.c: Likewise.
15425 * config/arm/thumb1.md: Likewise.
15426 * config/arm/thumb2.md: Likewise.
15427 * config/arm/vfp.md: Likewise.
15428 * config/arm/arm-generic.md: Likewise.
15429 * config/arm/arm1020e.md: Likewise.
15430 * config/arm/arm1026ejs.md: Likewise.
15431 * config/arm/arm1136jfs.md: Likewise.
15432 * config/arm/arm926ejs.md: Likewise.
15433 * config/arm/cortex-a15.md: Likewise.
15434 * config/arm/cortex-a17.md: Likewise.
15435 * config/arm/cortex-a5.md: Likewise.
15436 * config/arm/cortex-a53.md: Likewise.
15437 * config/arm/cortex-a57.md: Likewise.
15438 * config/arm/cortex-a7.md: Likewise.
15439 * config/arm/cortex-a8.md: Likewise.
15440 * config/arm/cortex-a9.md: Likewise.
15441 * config/arm/cortex-m4.md: Likewise.
15442 * config/arm/cortex-m7.md: Likewise.
15443 * config/arm/cortex-r4.md: Likewise.
15444 * config/arm/exynos-m1.md: Likewise.
15445 * config/arm/fa526.md: Likewise.
15446 * config/arm/fa606te.md: Likewise.
15447 * config/arm/fa626te.md: Likewise.
15448 * config/arm/fa726te.md: Likewise.
15449 * config/arm/fmp626.md: Likewise.
15450 * config/arm/iwmmxt.md: Likewise.
15451 * config/arm/ldmstm.md: Likewise.
15452 * config/arm/marvell-pj4.md: Likewise.
15453 * config/arm/xgene1.md: Likewise.
15454 * config/aarch64/thunderx.md: Likewise.
15455 * config/aarch64/thunderx2t99.md: Likewise.
15456 * config/aarch64/falkor.md: Likewise.
15458 2017-09-12 Martin Liska <mliska@suse.cz>
15460 * attribs.c (private_lookup_attribute): New function.
15461 * attribs.h (private_lookup_attribute): Declared here.
15462 (lookup_attribute): Called from this place.
15464 2017-09-12 Richard Biener <rguenther@suse.de>
15466 PR tree-optimization/82157
15467 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
15468 stmts with side-effects.
15470 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15471 Alan Hayward <alan.hayward@arm.com>
15472 David Sherwood <david.sherwood@arm.com>
15474 * target.def (hard_regno_nregs): New hook.
15475 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
15476 * targhooks.h (default_hard_regno_nregs): Declare.
15477 * targhooks.c (default_hard_regno_nregs): New function.
15478 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
15479 (TARGET_HARD_REGNO_NREGS): ...this hook.
15480 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
15481 (CLASS_MAX_NREGS): Likewise.
15482 * doc/tm.texi: Regenerate.
15483 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
15484 instead of HARD_REGNO_NREGS.
15485 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
15486 HARD_REGNO_NREGS in the comment.
15487 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
15488 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
15489 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
15490 Return an unsigned int.
15491 (TARGET_HARD_REGNO_NREGS): Redefine.
15492 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
15493 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
15494 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15495 (arc_hard_regno_nregs): New function.
15496 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
15497 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
15498 (arm_hard_regno_nregs): New function.
15499 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
15500 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
15501 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
15502 (TARGET_HARD_REGNO_NREGS): Redefine.
15503 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
15504 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
15505 (HARD_REGNO_NREGS): Delete.
15506 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
15507 (cr16_hard_regno_nregs): New function.
15508 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
15509 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
15510 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
15511 (cris_hard_regno_nregs): New function.
15512 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
15513 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
15514 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15515 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
15516 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
15517 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
15518 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
15519 (frv_hard_regno_nregs): Make static. Take and return an
15521 (frv_class_max_nregs): Remove outdated copy of documentation.
15522 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
15523 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
15524 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
15525 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
15526 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
15527 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
15528 (TARGET_HARD_REGNO_NREGS): Redefine.
15529 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
15530 (CLASS_MAX_NREGS): Update comment.
15531 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
15532 (ia64_hard_regno_nregs): New function.
15533 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
15534 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
15535 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
15536 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
15537 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
15539 (m32c_hard_regno_nregs): Likewise. Make static.
15540 (TARGET_HARD_REGNO_NREGS): Redefine.
15541 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
15542 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
15543 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
15544 (m68k_hard_regno_nregs): New function.
15545 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
15546 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
15547 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
15548 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
15549 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
15550 Take and return an unsigned int.
15551 (TARGET_HARD_REGNO_NREGS): Redefine.
15552 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
15553 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15554 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
15555 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
15556 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
15557 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
15558 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
15559 (msp430_hard_regno_nregs): Make static. Take and return an
15561 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
15562 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
15563 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
15564 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
15565 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
15566 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
15567 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
15568 (TARGET_HARD_REGNO_NREGS): Redefine.
15569 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
15570 (PA_HARD_REGNO_NREGS): ...this.
15571 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
15572 (PA_HARD_REGNO_NREGS): ...this.
15573 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15574 (pa_hard_regno_nregs): New function.
15575 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
15576 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
15577 (pdp11_hard_regno_nregs): New function.
15578 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
15579 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
15580 (rs6000_hard_regno_nregs_hook): New function.
15581 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
15582 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
15583 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
15584 Take and return an unsigned int. Move earlier in file.
15585 (TARGET_HARD_REGNO_NREGS): Redefine.
15586 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
15587 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
15588 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
15589 (rl78_hard_regno_nregs): Make static. Take and return an
15591 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
15592 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
15593 (rs6000_hard_regno_nregs_hook): New function.
15594 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
15595 * config/rx/rx.c (rx_hard_regno_nregs): New function.
15596 (TARGET_HARD_REGNO_NREGS): Redefine.
15597 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
15598 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
15599 instead of HARD_REGNO_NREGS.
15600 (s390_hard_regno_nregs): New function.
15601 (s390_hard_regno_mode_ok): Add comment from s390.h.
15602 (TARGET_HARD_REGNO_NREGS): Redefine.
15603 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
15604 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
15605 (sh_hard_regno_nregs): New function.
15606 (sh_pass_in_reg_p): Use it.
15607 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
15608 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15609 (sparc_hard_regno_nregs): New function.
15610 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
15611 * config/spu/spu.c (spu_hard_regno_nregs): New function.
15612 (spu_function_arg_advance): Use it, supplying a valid register number.
15613 (TARGET_HARD_REGNO_NREGS): Redefine.
15614 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
15615 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
15616 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
15617 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15618 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15619 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15620 (CLASS_MAX_NREGS): Remove copy of old documentation.
15621 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15622 (visium_hard_regno_nregs): New function.
15623 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15624 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15625 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15626 xtensa_hard_regno_nregs): New function.
15627 * system.h (HARD_REGNO_NREGS): Poison.
15629 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15631 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15632 hard_regno_nregs instead of HARD_REGNO_NREGS.
15633 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15634 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15635 (c6x_expand_epilogue): Likewise.
15636 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15637 (frv_read_iacc_argument): Likewise.
15638 * config/sh/sh.c: Include regs.h.
15639 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15640 (regs_used): Likewise.
15641 (output_stack_adjust): Likewise.
15642 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15643 * expmed.c: Include regs.h.
15644 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15645 * ree.c: Include regs.h.
15646 (combine_reaching_defs): Use hard_regno_nregs instead of
15648 (add_removable_extension): Likewise.
15650 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15652 * regs.h (hard_regno_nregs): Turn into a function.
15653 (end_hard_regno): Update accordingly.
15654 * caller-save.c (setup_save_areas): Likewise.
15655 (save_call_clobbered_regs): Likewise.
15656 (replace_reg_with_saved_mem): Likewise.
15657 (insert_restore): Likewise.
15658 (insert_save): Likewise.
15659 * combine.c (can_change_dest_mode): Likewise.
15660 (move_deaths): Likewise.
15661 (distribute_notes): Likewise.
15662 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15663 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15664 (rs6000_split_multireg_move): Likewise.
15665 (rs6000_register_move_cost): Likewise.
15666 (rs6000_memory_move_cost): Likewise.
15667 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15668 (rs6000_split_multireg_move): Likewise.
15669 (rs6000_register_move_cost): Likewise.
15670 (rs6000_memory_move_cost): Likewise.
15671 * cselib.c (cselib_reset_table): Likewise.
15672 (cselib_lookup_1): Likewise.
15673 * emit-rtl.c (set_mode_and_regno): Likewise.
15674 * function.c (aggregate_value_p): Likewise.
15675 * ira-color.c (setup_profitable_hard_regs): Likewise.
15676 (check_hard_reg_p): Likewise.
15677 (calculate_saved_nregs): Likewise.
15678 (assign_hard_reg): Likewise.
15679 (improve_allocation): Likewise.
15680 (calculate_spill_cost): Likewise.
15681 * ira-emit.c (modify_move_list): Likewise.
15682 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15683 (ira_hard_reg_in_set_p): Likewise.
15684 * ira.c (setup_reg_mode_hard_regset): Likewise.
15685 (clarify_prohibited_class_mode_regs): Likewise.
15686 (check_allocation): Likewise.
15687 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15688 (lra_setup_reg_renumber): Likewise.
15689 (setup_try_hard_regno_pseudos): Likewise.
15690 (spill_for): Likewise.
15691 (assign_hard_regno): Likewise.
15692 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15693 * lra-constraints.c (in_class_p): Likewise.
15694 (lra_constraint_offset): Likewise.
15695 (simplify_operand_subreg): Likewise.
15696 (lra_constraints): Likewise.
15697 (split_reg): Likewise.
15698 (split_if_necessary): Likewise.
15699 (invariant_p): Likewise.
15700 (inherit_in_ebb): Likewise.
15701 * lra-lives.c (process_bb_lives): Likewise.
15702 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15703 (get_hard_regs): Likewise.
15704 (do_remat): Likewise.
15705 * lra-spills.c (assign_spill_hard_regs): Likewise.
15706 * mode-switching.c (create_pre_exit): Likewise.
15707 * postreload.c (reload_combine_recognize_pattern): Likewise.
15708 * recog.c (peep2_find_free_register): Likewise.
15709 * regcprop.c (kill_value_regno): Likewise.
15710 (set_value_regno): Likewise.
15711 (copy_value): Likewise.
15712 (maybe_mode_change): Likewise.
15713 (find_oldest_value_reg): Likewise.
15714 (copyprop_hardreg_forward_1): Likewise.
15715 * regrename.c (check_new_reg_p): Likewise.
15716 (regrename_do_replace): Likewise.
15717 * reload.c (push_reload): Likewise.
15718 (combine_reloads): Likewise.
15719 (find_dummy_reload): Likewise.
15720 (operands_match_p): Likewise.
15721 (find_reloads): Likewise.
15722 (find_equiv_reg): Likewise.
15723 (reload_adjust_reg_for_mode): Likewise.
15724 * reload1.c (count_pseudo): Likewise.
15725 (count_spilled_pseudo): Likewise.
15726 (find_reg): Likewise.
15727 (clear_reload_reg_in_use): Likewise.
15728 (free_for_value_p): Likewise.
15729 (allocate_reload_reg): Likewise.
15730 (choose_reload_regs): Likewise.
15731 (reload_adjust_reg_for_temp): Likewise.
15732 (emit_reload_insns): Likewise.
15733 (delete_output_reload): Likewise.
15734 * rtlanal.c (subreg_get_info): Likewise.
15735 * sched-deps.c (sched_analyze_reg): Likewise.
15736 * sel-sched.c (init_regs_for_mode): Likewise.
15737 (mark_unavailable_hard_regs): Likewise.
15738 (choose_best_reg_1): Likewise.
15739 (verify_target_availability): Likewise.
15740 * valtrack.c (dead_debug_insert_temp): Likewise.
15741 * var-tracking.c (track_loc_p): Likewise.
15742 (emit_note_insn_var_location): Likewise.
15743 * varasm.c (make_decl_rtl): Likewise.
15744 * reginfo.c (choose_hard_reg_mode): Likewise.
15745 (init_reg_modes_target): Refer directly to
15746 this_target_regs->x_hard_regno_nregs.
15748 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15750 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15751 instead of hard_regno_nregs.
15753 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15755 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15756 end_hard_regno instead of hard_regno_nregs.
15757 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15758 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15759 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15760 * ira-color.c (improve_allocation): Likewise.
15761 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15762 * lra-lives.c (mark_regno_live): Likewise.
15763 (mark_regno_dead): Likewise.
15764 * lra-remat.c (operand_to_remat): Likewise.
15765 * lra.c (collect_non_operand_hard_regs): Likewise.
15766 * postreload.c (reload_combine_note_store): Likewise.
15767 (move2add_valid_value_p): Likewise.
15768 * reload.c (regno_clobbered_p): Likewise.
15770 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15772 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15774 * config/v850/v850.c (v850_reorg): Likewise.
15775 * reload.c (refers_to_regno_for_reload_p): Likewise.
15776 (find_equiv_reg): Likewise.
15777 * reload1.c (reload_reg_reaches_end_p): Likewise.
15779 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15781 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15783 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15784 * config/arm/arm.c (output_move_neon): Likewise.
15785 (arm_attr_length_move_neon): Likewise.
15786 (neon_split_vcombine): Likewise.
15787 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15788 (c6x_mark_reg_written): Likewise.
15789 (c6x_dwarf_register_span): Likewise.
15790 * config/i386/i386.c (ix86_save_reg): Likewise.
15791 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15792 (rws_access_reg): Likewise.
15793 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15794 * mode-switching.c (create_pre_exit): Likewise.
15795 * ree.c (combine_reaching_defs): Likewise.
15796 (add_removable_extension): Likewise.
15797 * regcprop.c (find_oldest_value_reg): Likewise.
15798 (copyprop_hardreg_forward_1): Likewise.
15799 * reload.c (reload_inner_reg_of_subreg): Likewise.
15800 (push_reload): Likewise.
15801 (combine_reloads): Likewise.
15802 (find_dummy_reload): Likewise.
15803 (reload_adjust_reg_for_mode): Likewise.
15804 * reload1.c (find_reload_regs): Likewise.
15805 (forget_old_reloads_1): Likewise.
15806 (reload_reg_free_for_value_p): Likewise.
15807 (reload_adjust_reg_for_temp): Likewise.
15808 (emit_reload_insns): Likewise.
15809 (delete_output_reload): Likewise.
15810 * sel-sched.c (choose_best_reg_1): Likewise.
15811 (choose_best_pseudo_reg): Likewise.
15813 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15814 Alan Hayward <alan.hayward@arm.com>
15815 David Sherwood <david.sherwood@arm.com>
15817 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15818 * target.def (slow_unaligned_access): New hook.
15819 * targhooks.h (default_slow_unaligned_access): Declare.
15820 * targhooks.c (default_slow_unaligned_access): New function.
15821 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15822 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15823 * doc/tm.texi: Regenerate.
15824 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15825 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15826 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15828 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15829 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15831 (rs6000_slow_unaligned_access): New function.
15832 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15833 (expand_block_compare): Likewise.
15834 (expand_strn_compare): Likewise.
15835 (rs6000_rtx_costs): Likewise.
15836 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15837 (riscv_slow_unaligned_access): Likewise.
15838 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15839 (riscv_slow_unaligned_access_p): ...this and make static.
15840 (riscv_option_override): Update accordingly.
15841 (riscv_slow_unaligned_access): New function.
15842 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15843 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15844 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15845 (rs6000_slow_unaligned_access): New function.
15846 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15847 (rs6000_rtx_costs): Likewise.
15848 * config/rs6000/rs6000-string.c (expand_block_compare)
15849 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15850 of SLOW_UNALIGNED_ACCESS.
15851 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15852 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15853 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15854 of SLOW_UNALIGNED_ACCESS.
15855 * expmed.c (simple_mem_bitfield_p): Likewise.
15856 * expr.c (alignment_for_piecewise_move): Likewise.
15857 (emit_group_load_1): Likewise.
15858 (emit_group_store): Likewise.
15859 (copy_blkmode_from_reg): Likewise.
15860 (emit_push_insn): Likewise.
15861 (expand_assignment): Likewise.
15862 (store_field): Likewise.
15863 (expand_expr_real_1): Likewise.
15864 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15865 * lra-constraints.c (simplify_operand_subreg): Likewise.
15866 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15867 * gimple-ssa-store-merging.c: Likewise in block comment at start
15869 * tree-ssa-strlen.c: Include target.h.
15870 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
15871 of SLOW_UNALIGNED_ACCESS.
15872 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
15874 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15876 PR rtl-optimization/82185
15877 * expmed.c (emit_store_flag_int): Only test tem if it has been
15880 2017-09-12 Richard Biener <rguenther@suse.de>
15882 PR middle-end/82149
15883 * match.pd ((FTYPE) N CMP CST): Fix typo.
15885 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
15887 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
15889 (mips_near_type_p): Add 'short_call' attribute as a synonym
15891 * doc/extend.texi (short_call): Document new function attribute.
15893 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15896 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
15897 assertion check that in the condition.
15898 (get_atomic_generic_size): Likewise. Before testing if parameter
15899 has pointer type, if it has array type, call for C++
15900 default_conversion to perform array-to-pointer conversion.
15902 2017-09-12 Richard Biener <rguenther@suse.de>
15904 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
15905 for operations we cannot scalarize.
15907 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
15909 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
15910 vectors heap vectors. Clean up comments.
15911 Make visited_bbs a reference.
15912 (profitable_jump_thread_path): Make GC
15913 vectors heap vectors. Clean up comments.
15915 (convert_and_register_jump_thread_path): Make GC vectors heap
15917 (check_subpath_and_update_thread_path): Same. Clean up comments.
15918 Make visited_bbs a reference.
15919 (handle_phi): Abstract common code to to
15920 register_jump_thread_path_if_profitable.
15921 Rename VAR_BB to DEF_BB.
15923 Make GC vectors heap vectors.
15924 Make visited_bbs a reference.
15925 (handle_assignment): Same.
15926 (register_jump_thread_path_if_profitable): New.
15927 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
15929 Make GC vectors heap vectors. Clean up comments.
15930 Make visited_bbs a reference.
15931 (find_jump_threads_backwards): Make visited_bbs live in the stack.
15932 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
15935 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
15938 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
15939 words of E_DImode object are reachable by xtensa_uimm8x4 access.
15941 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
15943 Revert r251800 and r251799.
15945 2017-09-11 Martin Jambor <mjambor@suse.cz>
15948 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
15949 arguments in advance.
15950 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
15951 use it to find predecessor edges.
15952 (naive_outof_ssa): Collect vector of predecessors.
15954 2017-09-08 Jason Merrill <jason@redhat.com>
15956 PR c++/70029 - ICE with ref-qualifier and -flto
15957 * langhooks.h (struct lang_hooks_for_types): Add
15958 copy_lang_qualifiers.
15959 * attribs.c (build_type_attribute_qual_variant): Use it.
15960 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
15962 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
15963 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
15965 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
15968 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
15969 (*mulsi3_sp64): New instruction.
15970 (mulsi3): New expander.
15972 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
15974 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
15976 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15978 * sancov.c: Include memmodel.h.
15980 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
15983 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
15986 2017-09-07 Carl Love <cel@us.ibm.com>
15988 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
15989 the sldi instruction.
15991 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
15993 * sancov.c: Include tm_p.h.
15995 2017-09-07 Jakub Jelinek <jakub@redhat.com>
15998 * output.h (switch_to_other_text_partition): New declaration.
15999 * varasm.c (switch_to_other_text_partition): New function.
16000 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16001 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16002 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16003 to the other text partition before emitting LCL label and switch back
16004 after emitting the word after it.
16006 2017-09-07 Richard Biener <rguenther@suse.de>
16008 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16009 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16010 critical edges here, after loop init.
16011 (pass_data_pre): Remove PROP_no_crit_edges flags.
16012 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16013 for valueization of call args to avoid leaking VN_TOP.
16014 (visit_use): Assert we do not visit default defs.
16015 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16016 Use error_mark_node to more easily detect leaking VN_TOP.
16017 All default-defs are varying, not VN_TOP. Mark them visited.
16018 (run_scc_vn): Make code match comment.
16020 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16022 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16023 OPTION_MASK_FLOAT128_KEYWORD.
16024 (POWERPC_MASKS): Likewise.
16025 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16026 support for the -mfloat128-type option, and make -mfloat128
16027 default on PowerPC Linux systems. Define or undefine
16028 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16029 Define __float128 to be __ieee128 if IEEE 128-bit support is
16030 enabled, or undefine it.
16031 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16032 Delete defining __FLOAT128_TYPE__.
16033 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16034 -mfloat128-type option and make -mfloat128 default on PowerPC
16036 (TARGET_FLOAT128_TYPE): Likewise.
16037 (-mfloat128-type): Likewise.
16038 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16039 Delete the -mfloat128-type option and make -mfloat128 default on
16040 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16041 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16042 128-bit floating point is enabled. Change tests from using
16043 -mfloat128-type to -mfloat128.
16044 (rs6000_mangle_type): Use the correct mangling for the __float128
16045 type even if normal long double is restricted to 64-bits.
16046 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16048 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16049 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16050 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16052 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16053 documentation for -mfloat128.
16055 2017-09-06 Olivier Hainque <hainque@adacore.com>
16057 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16059 2017-09-06 Wish Wu <wishwu007@gmail.com>
16060 Jakub Jelinek <jakub@redhat.com>
16062 * asan.c (initialize_sanitizer_builtins): Add
16063 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16064 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16065 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16066 BT_FN_VOID_UINT64_PTR variables.
16067 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16068 (BT_FN_VOID_UINT16_UINT16): Likewise.
16069 (BT_FN_VOID_UINT32_UINT32): Likewise.
16070 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16071 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16072 (BT_FN_VOID_UINT64_PTR): Likewise.
16073 * common.opt (flag_sanitize_coverage): New variable.
16074 (fsanitize-coverage=trace-pc): Remove.
16075 (fsanitize-coverage=): Add.
16076 * flag-types.h (enum sanitize_coverage_code): New enum.
16077 * fold-const.c (fold_range_test): Disable non-short-circuit
16078 optimization if flag_sanitize_coverage.
16079 (fold_truth_andor): Likewise.
16080 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16081 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16082 (coverage_sanitizer_opts): New array.
16083 (get_closest_sanitizer_option): Add OPTS argument, handle also
16084 OPT_fsanitize_coverage_.
16085 (parse_sanitizer_options): Adjusted to also handle
16086 OPT_fsanitize_coverage_.
16087 (common_handle_option): Add OPT_fsanitize_coverage_.
16088 * sancov.c (instrument_comparison, instrument_switch): New function.
16089 (sancov_pass): Add trace-cmp support.
16090 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16091 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16092 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16093 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16094 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16095 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16096 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16097 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16098 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16099 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16101 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16103 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16104 error. Only quit immediately if parsing is complete.
16105 (BEGIN): Initialize fatal_err and parse_done.
16106 (begin fpu, end fpu): Check number of arguments.
16107 (begin arch, end arch): Likewise.
16108 (begin cpu, end cpu): Likewise.
16109 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
16110 (optalias): Likewise.
16112 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16114 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
16115 * config/arm/arm-isa.h: Delete. Move definitions to ...
16116 * arm-cpus.in: ... here. Use new feature and fgroup values.
16117 * config/arm/arm.c (arm_option_override): Use lower case for feature
16119 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
16120 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
16121 * config/arm/parsecpu.awk (END): Add new command 'isa'.
16123 (print_isa_bits_for): New function.
16124 (gen_isa): New function.
16125 (gen_comm_data): Use print_isa_bits_for.
16126 (define feature): New keyword.
16127 (define fgroup): New keyword.
16128 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
16129 (arm-isa.h): Add rule to generate file.
16130 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
16131 case for feature bit names.
16133 2017-09-06 Richard Biener <rguenther@suse.de>
16135 * tree-ssa-pre.c (NECESSARY): Remove.
16136 (create_expression_by_pieces): Do not touch pass-local flags.
16137 (insert_into_preds_of_block): Likewise.
16138 (do_pre_regular_insertion): Likewise.
16139 (eliminate_insert): Likewise.
16140 (eliminate_dom_walker::before_dom_children): Likewise.
16141 (fini_eliminate): Do not look at inserted_exprs.
16142 (mark_operand_necessary): Remove.
16143 (remove_dead_inserted_code): Replace with simple work-list
16144 algorithm based on inserted_exprs and SSA uses.
16145 (pass_pre::execute): Re-order fini_eliminate and
16146 remove_dead_inserted_code.
16148 2017-09-06 Olivier Hainque <hainque@adacore.com>
16150 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
16151 for VxWorks 7. Adjust surrounding comments.
16153 2017-09-06 Richard Biener <rguenther@suse.de>
16155 * gimple-ssa-strength-reduction.c
16156 (find_candidates_dom_walker::before_dom_children): Also allow
16159 2017-09-06 Richard Biener <rguenther@suse.de>
16161 PR tree-optimization/82108
16162 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
16163 for gap in the non-permutation SLP case.
16165 2017-09-06 Martin Jambor <mjambor@suse.cz>
16167 PR tree-optimization/82078
16168 * tree-sra.c (sort_and_splice_var_accesses): Move call to
16169 add_access_to_work_queue...
16170 (build_accesses_from_assign): ...here.
16171 (propagate_all_subaccesses): Make sure racc is the group
16172 representative, if there is one.
16174 2017-09-06 Jakub Jelinek <jakub@redhat.com>
16176 PR middle-end/82095
16177 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
16180 2017-09-06 Richard Biener <rguenther@suse.de>
16182 * gimple-ssa-strength-reduction.c
16183 (find_candidates_dom_walker::before_doom_children): Use a
16184 type and not a mode check.
16186 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
16189 * config/arm/predicates.md (arm_general_adddi_operand): Create new
16191 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
16193 2017-09-05 Jeff Law <law@redhat.com>
16195 PR tree-optimization/64910
16196 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
16197 cases where we have 3 or more operands.
16199 2017-09-05 Jakub Jelinek <jakub@redhat.com>
16201 PR middle-end/81768
16202 * omp-low.c (lower_omp_for): Recompute tree invariant if
16203 gimple_omp_for_initial/final is ADDR_EXPR.
16205 PR middle-end/81768
16206 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
16207 into gimple val before gimplification fo the COND_EXPR.
16209 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
16211 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
16212 REGION_COPY argument.
16213 (thread_through_all_blocks): Remove unused argument to
16214 duplicate_thread_path.
16216 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16217 Alan Hayward <alan.hayward@arm.com>
16218 David Sherwood <david.sherwood@arm.com>
16220 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
16221 Take a scalar_mode rather than a machine_mode.
16222 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16223 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
16224 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16225 (aarch64_gen_adjusted_ldpstp): Likewise.
16226 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
16228 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16229 Alan Hayward <alan.hayward@arm.com>
16230 David Sherwood <david.sherwood@arm.com>
16232 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
16233 Take a scalar_int_mode instead of a machine_mode.
16234 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16235 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16236 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16237 (aarch64_simd_attr_length_rglist): Delete.
16238 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
16239 a scalar_int_mode instead of a machine_mode.
16240 (aarch64_add_offset): Likewise.
16241 (aarch64_internal_mov_immediate): Likewise
16242 (aarch64_add_constant_internal): Likewise.
16243 (aarch64_add_constant): Likewise.
16244 (aarch64_movw_imm): Likewise.
16245 (aarch64_rtx_arith_op_extract_p): Likewise.
16246 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16247 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16248 Remove assert that the mode isn't a vector.
16249 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16250 (aarch64_expand_mov_immediate): Update calls after above changes.
16251 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
16252 (aarch64_and_bitmask_imm): Check for scalar integer modes.
16253 (aarch64_move_imm): Likewise.
16254 (aarch64_can_const_movi_rtx_p): Likewise.
16255 (aarch64_strip_extend): Likewise.
16256 (aarch64_extr_rtx_p): Likewise.
16257 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
16258 a CONST_INT when the mode parameter is VOIDmode.
16259 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
16261 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16263 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
16264 * stor-layout.c (bitwise_mode_for_mode): Likewise.
16265 (bitwise_type_for_mode): Update accordingly.
16267 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16269 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
16270 * stor-layout.c (mode_for_size_tree): Likewise.
16271 (mode_for_array): Update accordingly.
16272 (layout_decl): Likewise.
16273 (compute_record_mode): Likewise. Only set the mode once.
16275 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16277 * target.def (get_mask_mode): Change return type to opt_mode.
16279 * doc/tm.texi: Regenerate.
16280 * targhooks.h (default_get_mask_mode): Return an opt_mode.
16281 * targhooks.c (default_get_mask_mode): Likewise.
16282 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16283 * optabs-query.c (can_vec_mask_load_store_p): Update use of
16284 targetm.get_mask_mode.
16285 * tree.c (build_truth_vector_type): Likewise.
16287 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16289 * machmode.h (mode_for_vector): Return an opt_mode.
16290 * stor-layout.c (mode_for_vector): Likewise.
16291 (mode_for_int_vector): Update accordingly.
16292 (layout_type): Likewise.
16293 * config/i386/i386.c (emit_memmov): Likewise.
16294 (ix86_expand_set_or_movmem): Likewise.
16295 (ix86_expand_vector_init): Likewise.
16296 (ix86_get_mask_mode): Likewise.
16297 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
16299 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
16300 * expmed.c (extract_bit_field_1): Likewise.
16301 * expr.c (expand_expr_real_2): Likewise.
16302 * optabs-query.c (can_vec_perm_p): Likewise.
16303 (can_vec_mask_load_store_p): Likewise.
16304 * optabs.c (expand_vec_perm): Likewise.
16305 * targhooks.c (default_get_mask_mode): Likewise.
16306 * tree-vect-stmts.c (vectorizable_store): Likewise.
16307 (vectorizable_load): Likewise.
16308 (get_vectype_for_scalar_type_and_size): Likewise.
16310 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16312 * machmode.h (mode_for_int_vector): New function.
16313 * stor-layout.c (mode_for_int_vector): Likewise.
16314 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
16315 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
16316 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16317 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16318 (s390_expand_vcond): Likewise.
16320 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16322 * machmode.h (opt_machine_mode): New type.
16323 (opt_mode<T>): Allow construction from anything that can be
16325 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
16326 (mode_for_size): Return an opt_machine_mode.
16327 * stor-layout.c (mode_for_size): Likewise.
16328 (mode_for_size_tree): Update call accordingly.
16329 (bitwise_mode_for_mode): Likewise.
16330 (make_fract_type): Likewise.
16331 (make_accum_type): Likewise.
16332 * caller-save.c (replace_reg_with_saved_mem): Update call
16334 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16335 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16336 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16337 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16338 * expmed.c (extract_bit_field_1): Likewise.
16339 * reload.c (get_secondary_mem): Likewise.
16340 * varasm.c (assemble_integer): Likewise.
16341 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
16344 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16346 * machmode.h (decimal_float_mode_for_size): New function.
16347 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
16348 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
16349 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
16350 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
16351 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
16352 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
16354 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16356 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
16357 (get_builtin_sync_mode): Likewise.
16358 (expand_ifn_atomic_compare_exchange): Likewise.
16359 (expand_builtin_atomic_clear): Likewise.
16360 (expand_builtin_atomic_test_and_set): Likewise.
16361 (fold_builtin_atomic_always_lock_free): Likewise.
16362 * calls.c (compute_argument_addresses): Likewise.
16363 (emit_library_call_value_1): Likewise.
16364 (store_one_arg): Likewise.
16365 * combine.c (combine_instructions): Likewise.
16366 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
16367 * config/arm/arm.c (arm_function_value): Likewise.
16368 (aapcs_allocate_return_reg): Likewise.
16369 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
16370 * config/i386/i386.c (construct_container): Likewise.
16371 (ix86_gimplify_va_arg): Likewise.
16372 (ix86_expand_sse_cmp): Likewise.
16373 (emit_memmov): Likewise.
16374 (emit_memset): Likewise.
16375 (expand_small_movmem_or_setmem): Likewise.
16376 (ix86_expand_pextr): Likewise.
16377 (ix86_expand_pinsr): Likewise.
16378 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
16379 * config/microblaze/microblaze.c (microblaze_block_move_straight):
16381 * config/mips/mips.c (mips_function_value_1) Likewise.
16382 (mips_block_move_straight): Likewise.
16383 (mips_expand_ins_as_unaligned_store): Likewise.
16384 * config/powerpcspe/powerpcspe.c
16385 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16386 (rs6000_darwin64_record_arg_flush): Likewise.
16387 * config/rs6000/rs6000.c
16388 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16389 (rs6000_darwin64_record_arg_flush): Likewise.
16390 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
16391 (sparc_function_value_1): Likewise.
16392 * config/spu/spu.c (adjust_operand): Likewise.
16393 (spu_emit_branch_or_set): Likewise.
16394 (arith_immediate_p): Likewise.
16395 * emit-rtl.c (gen_lowpart_common): Likewise.
16396 * expr.c (expand_expr_real_1): Likewise.
16397 * function.c (assign_parm_setup_block): Likewise.
16398 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
16399 * reload1.c (alter_reg): Likewise.
16400 * stor-layout.c (mode_for_vector): Likewise.
16401 (layout_type): Likewise.
16403 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16405 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
16406 (spu_convert_move): Likewise.
16407 * lower-subreg.c (resolve_simple_move): Likewise.
16409 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16412 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
16413 define_insn to a define_expand.
16414 (altivec_vsum2sws_direct): New define_insn.
16415 (altivec_vsumsws): Convert from a define_insn to a define_expand.
16417 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16419 * config/arm/arm.c (arm_option_params_internal): Improve setting of
16422 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
16427 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
16428 realign stack if stack alignment needed is less than incoming
16431 2017-09-05 Marek Polacek <polacek@redhat.com>
16434 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
16437 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16439 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
16441 2017-09-05 Richard Biener <rguenther@suse.de>
16443 PR tree-optimization/82084
16444 * fold-const.c (can_native_encode_string_p): Handle wide characters.
16446 2017-09-05 Richard Biener <rguenther@suse.de>
16448 PR tree-optimization/82102
16449 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
16451 2017-09-05 Martin Liska <mliska@suse.cz>
16453 PR tree-optimization/82032
16454 * tree-cfg.c (generate_range_test): New function.
16455 * tree-cfg.h (generate_range_test): Declared here.
16456 * tree-cfgcleanup.c (convert_single_case_switch): New function.
16457 (cleanup_control_expr_graph): Use it.
16458 * tree-switch-conversion.c (try_switch_expansion): Remove
16460 (emit_case_nodes): Use generate_range_test.
16462 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16465 * config/i386/i386.md (*<btsc><mode>_mask): Add
16466 TARGET_USE_BT to insn constraint.
16467 (*btr<mode>_mask): Ditto.
16469 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
16471 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
16472 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
16474 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
16477 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
16478 TARGET_NEON and TARGET_IWMMXT.
16479 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
16480 TARGET_NEON and TARGET_IWMMXT.
16481 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
16483 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16485 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
16486 (ix86_rewrite_tls_address): Ditto.
16487 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
16488 (ix86_rewrite_tls_address_1): Ditto.
16489 (ix86_rewrite_tls_address): Ditto.
16490 * config/i386/predicates.md (tls_address_pattern): New predicate.
16491 * config/i386/i386.md (TLS address splitter): New splitter.
16493 2017-09-04 Richard Biener <rguenther@suse.de>
16495 PR tree-optimization/82084
16496 * fold-const.h (can_native_encode_string_p): Declare.
16497 * fold-const.c (can_native_encode_string_p): Factor out from ...
16498 (native_encode_string): ... here.
16499 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
16500 vectorizing stores from constants we later cannot handle.
16502 2017-09-04 Marek Polacek <polacek@redhat.com>
16505 * doc/invoke.texi: Update -Wtautological-compare documentation.
16507 2017-09-04 Jeff Law <law@redhat.com>
16509 PR tree-optimization/64910
16510 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
16511 swap the first and last operand if the last is a constant.
16513 2017-09-04 Marek Polacek <polacek@redhat.com>
16516 * convert.c (do_narrow): When sanitizing signed integer overflows,
16517 bail out for signed types.
16518 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
16520 2017-09-04 Richard Biener <rguenther@suse.de>
16522 PR tree-optimization/82060
16523 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16524 Move devirtualization after stmt folding and before EH/AB/noreturn
16525 cleanup to get the stmt refs canonicalized. Use a bool instead
16526 of gimple_modified_p since that doesn't work for NOPs. Schedule
16527 NOPs generated by folding for removal.
16529 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16530 Alan Hayward <alan.hayward@arm.com>
16531 David Sherwood <david.sherwood@arm.com>
16533 * coretypes.h (pad_direction): New enum.
16534 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
16535 (FUNCTION_ARG_PADDING): Likewise.
16536 * target.def (function_arg_padding): New hook.
16537 * targhooks.h (default_function_arg_padding): Declare.
16538 * targhooks.c (default_function_arg_padding): New function.
16539 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
16540 (TARGET_FUNCTION_ARG_PADDING): ...this.
16541 * doc/tm.texi: Regenerate.
16542 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16543 instead of direction.
16544 (compute_argument_addresses): Likewise.
16545 (load_register_parameters): Likewise.
16546 (emit_library_call_value_1): Likewise.
16547 (store_one_arg): Use targetm.calls.function_arg_padding instead
16548 of FUNCTION_ARG_PADDING.
16549 (must_pass_in_stack_var_size_or_pad): Likewise.
16550 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
16551 (emit_group_store): Likewise.
16552 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
16553 instead of FUNCTION_ARG_PADDING.
16554 (emit_push_insn): Likewise, and propagate enum change throughout
16556 * function.h (direction): Delete.
16557 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
16559 * function.c (assign_parm_find_stack_rtl): Likewise.
16560 (assign_parm_setup_block_p): Likewise.
16561 (assign_parm_setup_block): Likewise.
16562 (gimplify_parameters): Likewise.
16563 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
16564 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
16566 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
16567 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16568 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
16569 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
16570 (aarch64_function_arg_padding): ...this new function.
16571 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
16572 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16573 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
16574 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16575 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
16576 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16577 (arm_pad_arg_upward): Replace with...
16578 (arm_function_arg_padding): ...this new function.
16579 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
16581 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
16582 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
16583 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16584 (ia64_hpux_function_arg_padding): Replace with...
16585 (ia64_function_arg_padding): ...this new function. Use pad_direction
16586 instead of direction. Check for TARGET_HPUX.
16587 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
16588 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16589 (iq2000_function_arg_padding): New function.
16590 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
16591 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
16592 (mips_function_arg_padding): ...this new function.
16593 (mips_pad_reg_upward): Update accordingly.
16594 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16595 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
16596 targetm.calls.function_arg_padding.
16597 (FUNCTION_ARG_PADDING): Delete.
16598 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16599 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
16600 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16601 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
16602 (nios2_block_reg_padding): Return pad_direction instead of direction.
16603 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
16604 instead of direction.
16605 (nios2_function_arg_padding): Likewise. Make static.
16606 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16607 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
16608 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
16609 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
16610 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16611 (pa_function_arg_padding): Make static. Return pad_direction instead
16613 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
16614 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16615 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
16616 instead of direction. Use targetm.calls.function_arg_padding.
16617 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16618 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16619 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16620 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16621 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16623 (function_arg_padding): Rename to...
16624 (rs6000_function_arg_padding): ...this. Make static. Return
16625 pad_direction instead of direction.
16626 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16627 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16628 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16629 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16630 instead of direction. Use targetm.calls.function_arg_padding.
16631 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16632 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16633 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16634 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16635 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16636 (function_arg_padding): Rename to...
16637 (rs6000_function_arg_padding): ...this. Make static. Return
16638 pad_direction instead of direction.
16639 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16640 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16641 * config/s390/s390.c (s390_function_arg_padding): New function.
16642 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16643 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16644 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16645 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16646 (function_arg_padding): Rename to...
16647 (sparc_function_arg_padding): ...this. Make static. Return
16648 pad_direction instead of direction.
16649 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16650 * config/spu/spu.c (spu_function_arg_padding): New function.
16651 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16652 * system.h (FUNCTION_ARG_PADDING): Poison.
16654 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16655 Alan Hayward <alan.hayward@arm.com>
16656 David Sherwood <david.sherwood@arm.com>
16658 * target.def (modes_tieable_p): New hook.
16659 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16660 (TARGET_MODES_TIEABLE_P): ...this.
16661 * doc/tm.texi.in: Regenerate.
16662 * hooks.h (hook_bool_mode_mode_true): Declare.
16663 * hooks.c (hook_bool_mode_mode_true): New function.
16664 * combine.c (subst): Use targetm.modes_tieable_p instead of
16666 * dse.c (find_shift_sequence): Likewise.
16667 * expmed.c (extract_low_bits): Likewise.
16668 * lower-subreg.c: Include target.h.
16669 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16671 * rtlanal.c (rtx_cost): Likewise.
16672 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16673 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16674 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16675 (TARGET_MODES_TIEABLE_P): Redefine.
16676 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16677 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16678 (TARGET_MODES_TIEABLE_P): Redefine.
16679 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16680 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16681 (arc_modes_tieable_p): New function.
16682 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16683 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16684 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16685 (arm_modes_tieable_p): Make static.
16686 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16687 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16688 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16689 (TARGET_MODES_TIEABLE_P): Redefine.
16690 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16691 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16692 (TARGET_MODES_TIEABLE_P): Redefine.
16693 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16694 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16695 (cr16_modes_tieable_p): New function.
16696 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16697 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16698 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16699 (TRULY_NOOP_TRUNCATION): Update comment.
16700 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16701 (TRULY_NOOP_TRUNCATION): Update comment.
16702 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16703 (frv_modes_tieable_p): New function.
16704 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16705 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16706 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16707 (TARGET_MODES_TIEABLE_P): Redefine.
16708 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16709 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16710 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16711 (TARGET_MODES_TIEABLE_P): Redefine.
16712 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16713 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16714 (ia64_modes_tieable_p): New function.
16715 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16716 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16717 (iq2000_modes_tieable_p): New function.
16718 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16719 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16720 (lm32_modes_tieable_p): New function.
16721 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16722 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16723 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16724 (TARGET_MODES_TIEABLE_P): Redefine.
16725 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16726 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16727 (m32r_modes_tieable_p): New function.
16728 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16729 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16730 (m68k_modes_tieable_p): New function.
16731 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16732 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16733 (mcore_modes_tieable_p): New function.
16734 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16735 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16737 (TARGET_MODES_TIEABLE_P): Redefine.
16738 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16739 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16740 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16741 (TARGET_MODES_TIEABLE_P): Redefine.
16742 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16743 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16744 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16745 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16746 (mn10300_modes_tieable_p): ...this and make static.
16747 (TARGET_MODES_TIEABLE_P): Redefine.
16748 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16749 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16750 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16751 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16752 (msp430_modes_tieable_p): Make static.
16753 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16754 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16755 (TARGET_MODES_TIEABLE_P): Redefine.
16756 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16757 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16758 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16759 (TARGET_MODES_TIEABLE_P): Redefine.
16760 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16761 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16762 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16763 (TARGET_MODES_TIEABLE_P): Redefine.
16764 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16765 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16766 (pdp11_modes_tieable_p): New function.
16767 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16768 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16769 (rs6000_modes_tieable_p): New function.
16770 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16771 * config/powerpcspe/powerpcspe.md: Update comment.
16772 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16773 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16774 (TARGET_MODES_TIEABLE_P): Redefine.
16775 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16776 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16777 (rl78_modes_tieable_p): New function.
16778 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16779 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16780 (rs6000_modes_tieable_p): New function.
16781 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16782 * config/rs6000/rs6000.md: Update comment.
16783 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16784 * config/rx/rx.c (rx_modes_tieable_p): New function.
16785 (TARGET_MODES_TIEABLE_P): Redefine.
16786 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16787 * config/s390/s390.c (s390_modes_tieable_p): New function.
16788 (TARGET_MODES_TIEABLE_P): Redefine.
16789 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16790 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16791 (sh_modes_tieable_p): New function.
16792 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16793 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16794 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16795 (sparc_modes_tieable_p): Make static.
16796 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16797 * config/spu/spu.c (spu_modes_tieable_p): New function.
16798 (TARGET_MODES_TIEABLE_P): Redefine.
16799 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16800 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16801 (TARGET_MODES_TIEABLE_P): Redefine.
16802 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16803 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16804 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16805 * config/v850/v850.c (v850_modes_tieable_p): New function.
16806 (TARGET_MODES_TIEABLE_P): Redefine.
16807 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16808 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16809 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16810 (visium_modes_tieable_p): New function.
16811 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16812 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16813 (xtensa_modes_tieable_p): New function.
16814 * system.h (MODES_TIEABLE_P): Poison.
16816 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16817 Alan Hayward <alan.hayward@arm.com>
16818 David Sherwood <david.sherwood@arm.com>
16820 * target.def (hard_regno_mode_ok): New hook.
16821 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16822 (TARGET_HARD_REGNO_MODE_OK): ...this.
16823 * doc/tm.texi.in: Regenerate.
16824 * hooks.h (hook_bool_uint_mode_true): Declare.
16825 * hooks.c (hook_bool_uint_mode_true): New function.
16826 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16827 HARD_REGNO_MODE_OK.
16828 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16829 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16830 instead of HARD_REGNO_MODE_OK.
16831 * caller-save.c: Include target.h.
16832 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16833 HARD_REGNO_MODE_OK.
16834 * combine.c (can_combine_p): Likewise.
16835 (combinable_i3pat): Likewise.
16836 (can_change_dest_mode): Likewise.
16837 * expr.c (init_expr_target): Likewise.
16838 (convert_move): Likewise.
16839 (convert_modes): Likewise.
16840 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16841 (setup_prohibited_mode_move_regs): Likewise.
16842 * ira.h (target_ira): Likewise.
16843 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16844 * lra-constraints.c (process_alt_operands): Likewise.
16845 (split_reg): Likewise.
16846 * recog.c (peep2_find_free_register): Likewise.
16847 * ree.c (combine_reaching_defs): Likewise.
16848 * regcprop.c (maybe_mode_change): Likewise.
16849 * reginfo.c (init_reg_sets_1): Likewise.
16850 (choose_hard_reg_mode): Likewise.
16851 (simplifiable_subregs): Likewise.
16852 * regrename.c (check_new_reg_p): Likewise.
16853 * reload.c (find_valid_class): Likewise.
16854 (find_valid_class_1): Likewise.
16855 (reload_inner_reg_of_subreg): Likewise.
16856 (push_reload): Likewise.
16857 (combine_reloads): Likewise.
16858 (find_dummy_reload): Likewise.
16859 (find_reloads): Likewise.
16860 * reload1.c (find_reg): Likewise.
16861 (set_reload_reg): Likewise.
16862 (allocate_reload_reg): Likewise.
16863 (choose_reload_regs): Likewise.
16864 (reload_adjust_reg_for_temp): Likewise.
16865 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
16866 (simplify_subreg_regno): Likewise.
16867 * sel-sched.c (init_regs_for_mode): Likewise.
16868 * varasm.c (make_decl_rtl): Likewise.
16869 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
16870 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
16871 HARD_REGNO_MODE_OK.
16872 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
16873 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
16874 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16875 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
16876 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
16877 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16878 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
16879 (arc_mode_class): Delete.
16880 (HARD_REGNO_MODE_OK): Delete.
16881 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16882 (arc_hard_regno_mode_ok): Rename old array to...
16883 (arc_hard_regno_mode_ok_modes): ...this.
16884 (arc_conditional_register_usage): Update accordingly.
16885 (arc_mode_class): Make static.
16886 (arc_hard_regno_mode_ok): New function.
16887 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
16888 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
16889 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16890 (arm_hard_regno_mode_ok): Make static.
16891 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
16892 HARD_REGNO_MODE_OK.
16893 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
16894 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
16895 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
16897 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16898 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
16899 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
16900 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
16901 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
16902 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16903 * config/bfin/predicates.md (valid_reg_operand): Use
16904 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
16905 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
16906 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
16907 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16908 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
16909 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
16910 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16911 (cr16_hard_regno_mode_ok): Make static and return a bool.
16912 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
16913 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16914 (cris_hard_regno_mode_ok): New function.
16915 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
16916 (epiphany_mode_class): Delete.
16917 (HARD_REGNO_MODE_OK): Delete.
16918 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
16919 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16920 (hard_regno_mode_ok): Rename to...
16921 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
16922 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
16923 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
16924 HARD_REGNO_MODE_OK.
16925 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
16926 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
16927 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16928 (frv_hard_regno_mode_ok): Make static and return a bool.
16929 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
16930 HARD_REGNO_MODE_OK.
16931 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
16932 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
16933 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
16934 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
16936 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16937 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
16938 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
16939 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
16941 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16942 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
16943 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16944 (ia64_hard_regno_mode_ok): New function.
16945 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
16946 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16947 (iq2000_hard_regno_mode_ok): New function.
16948 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
16949 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16950 (lm32_hard_regno_mode_ok): New function.
16951 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
16952 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
16953 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
16954 instead of HARD_REGNO_MODE_OK.
16955 (m32c_hard_regno_ok): Rename to...
16956 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
16957 (m32c_cannot_change_mode_class): Update accordingly.
16958 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16959 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
16960 (m32r_mode_class): Delete.
16961 (HARD_REGNO_MODE_OK): Delete.
16962 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16963 (m32r_hard_regno_mode_ok): Rename to...
16964 (m32r_hard_regno_modes): ...this.
16965 (m32r_mode_class): Make static.
16966 (m32r_hard_regno_mode_ok): New function.
16967 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
16968 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
16969 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16970 (m68k_hard_regno_mode_ok): Make static.
16971 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
16972 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16973 (mcore_hard_regno_mode_ok): New function.
16974 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
16975 (HARD_REGNO_MODE_OK): Delete.
16976 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
16978 (microblaze_hard_regno_mode_ok_p): ...this and make static.
16979 (microblaze_hard_regno_mode_ok): New function.
16980 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16981 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
16982 (mips_hard_regno_mode_ok): Delete.
16983 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
16984 (mips_hard_regno_mode_ok_p): ...this and make static.
16985 (mips_hard_regno_mode_ok_p): Rename to...
16986 (mips_hard_regno_mode_ok_uncached): ...this.
16987 (mips_hard_regno_mode_ok): New function.
16988 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
16989 of HARD_REGNO_MODE_OK.
16990 (mips_option_override): Update after above name changes.
16991 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16992 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
16993 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
16994 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
16995 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
16996 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16997 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
16998 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
16999 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17000 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17001 (msp430_hard_regno_mode_ok): Make static and return a bool.
17002 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17003 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17004 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17006 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17007 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17008 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17009 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17010 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17011 (PA_HARD_REGNO_MODE_OK): ...this
17012 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17013 (PA_HARD_REGNO_MODE_OK): ...this.
17014 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17015 (pa_hard_regno_mode_ok): New function.
17016 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17017 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17018 (pdp11_hard_regno_mode_ok): New function.
17019 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17020 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17022 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17024 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17025 (rs6000_hard_regno_mode_ok): Rename to...
17026 (rs6000_hard_regno_mode_ok_uncached): ...this.
17027 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17028 (rs6000_hard_regno_mode_ok): New function.
17029 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17030 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17031 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17032 (riscv_hard_regno_mode_ok): ...this and make static.
17033 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17034 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17035 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17036 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17037 (rl78_hard_regno_mode_ok): Make static and return bool.
17038 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
17039 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17041 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17042 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17043 (rs6000_hard_regno_mode_ok): Rename to...
17044 (rs6000_hard_regno_mode_ok_uncached): ...this.
17045 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17046 (rs6000_hard_regno_mode_ok): New function.
17047 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17048 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17049 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17050 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17051 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17052 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17053 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17054 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17055 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17056 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17057 (sh_hard_regno_mode_ok): Make static.
17058 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17059 instead of HARD_REGNO_MODE_OK.
17060 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17061 (sparc_mode_class): Delete.
17062 (HARD_REGNO_MODE_OK): Delete.
17063 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17064 (hard_regno_mode_classes): Make static.
17065 (sparc_mode_class): Likewise.
17066 (sparc_hard_regno_mode_ok): New function.
17067 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17068 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17069 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17071 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17072 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17073 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17074 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17075 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17076 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17077 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17078 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17079 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17080 (visium_hard_regno_mode_ok): New function.
17081 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17082 instead of HARD_REGNO_MODE_OK.
17083 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17084 (HARD_REGNO_MODE_OK): Delete.
17085 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17086 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17087 (xtensa_option_override): Update accordingly.
17088 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17089 (xtensa_hard_regno_mode_ok): New function.
17090 * system.h (HARD_REGNO_MODE_OK): Poison.
17092 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17093 Alan Hayward <alan.hayward@arm.com>
17094 David Sherwood <david.sherwood@arm.com>
17096 * target.def (hard_regno_call_part_clobbered): New hook.
17097 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17098 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17099 * doc/tm.texi: Regenerate.
17100 * hooks.h (hook_bool_uint_mode_false): Declare.
17101 * hooks.c (hook_bool_uint_mode_false): New function.
17102 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17103 * cselib.c (cselib_process_insn): Use
17104 targetm.hard_regno_call_part_clobbered instead of
17105 HARD_REGNO_CALL_PART_CLOBBERED.
17106 * ira-conflicts.c (ira_build_conflicts): Likewise.
17107 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17108 * lra-constraints.c (need_for_call_save_p): Likewise.
17109 * lra-lives.c: Include target.h.
17110 (check_pseudos_live_through_calls): Use
17111 targetm.hard_regno_call_part_clobbered instead of
17112 HARD_REGNO_CALL_PART_CLOBBERED.
17113 * regcprop.c: Include target.h.
17114 (copyprop_hardreg_forward_1): Use
17115 targetm.hard_regno_call_part_clobbered instead of
17116 HARD_REGNO_CALL_PART_CLOBBERED.
17117 * reginfo.c (choose_hard_reg_mode): Likewise.
17118 * regrename.c (check_new_reg_p): Likewise.
17119 * reload.c (find_equiv_reg): Likewise.
17120 * reload1.c (emit_reload_insns): Likewise.
17121 * sched-deps.c (deps_analyze_insn): Likewise.
17122 * sel-sched.c (init_regs_for_mode): Likewise.
17123 (mark_unavailable_hard_regs): Likewise.
17124 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
17125 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17126 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
17128 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17129 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17130 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
17132 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
17134 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17135 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17136 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
17138 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17139 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17140 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
17142 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17143 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
17145 * config/powerpcspe/powerpcspe.c
17146 (rs6000_hard_regno_call_part_clobbered): New function.
17147 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17148 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17149 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
17151 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17152 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17153 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
17155 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17156 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17157 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
17159 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17160 Alan Hayward <alan.hayward@arm.com>
17161 David Sherwood <david.sherwood@arm.com>
17163 * rtl.h (subreg_memory_offset): Declare.
17164 * emit-rtl.c (subreg_memory_offset): New function.
17165 * expmed.c (store_bit_field_1): Use it.
17166 * expr.c (undefined_operand_subword_p): Likewise.
17167 * simplify-rtx.c (simplify_subreg): Likewise.
17169 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
17171 PR rtl-optimization/57448
17174 * optabs.c (expand_atomic_load): Place compiler memory barriers if
17175 using atomic_load pattern.
17176 (expand_atomic_store): Likewise.
17178 2017-09-04 Jakub Jelinek <jakub@redhat.com>
17181 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
17182 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
17183 handling. Use replace_call_with_value with NULL instead of
17184 gsi_replace, unlink_stmt_vdef and release_defs.
17186 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
17189 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
17191 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17194 * rtl.h (emit_library_call_value_1): Declare.
17195 (emit_library_call): Replace declaration with a series of overloads.
17196 Remove the parameter count argument.
17197 (emit_library_call_value): Likewise.
17198 * calls.c (emit_library_call_value_1): Make global. Replace varargs
17199 with an "rtx_mode_t *".
17200 (emit_library_call_value): Delete.
17201 (emit_library_call): Likewise.
17202 * asan.c (asan_emit_stack_protection): Update calls accordingly.
17203 (asan_emit_allocas_unpoison): Likewise.
17204 * builtins.c (expand_builtin_powi): Likewise.
17205 (expand_asan_emit_allocas_unpoison): Likewise.
17206 * cfgexpand.c (expand_main_function): Likewise.
17207 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
17208 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
17209 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
17210 * config/arm/arm.c (arm_trampoline_init): Likewise.
17211 (arm_call_tls_get_addr): Likewise.
17212 (arm_expand_divmod_libfunc): Likewise.
17213 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
17214 (smulsi3_highpart): Likewise.
17215 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
17216 (c6x_expand_compare): Likewise.
17217 (c6x_expand_movmem): Likewise.
17218 * config/frv/frv.c (frv_trampoline_init): Likewise.
17219 * config/i386/i386.c (ix86_trampoline_init): Likewise.
17220 (ix86_expand_divmod_libfunc): Likewise.
17221 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
17222 (ia64_expand_compare): Likewise.
17223 (ia64_profile_hook): Likewise.
17224 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
17225 (nonlocal_goto): Likewise.
17226 (restore_stack_nonlocal): Likewise.
17227 * config/m32r/m32r.c (block_move_call): Likewise.
17228 (m32r_trampoline_init): Likewise.
17229 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
17230 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
17231 (m68k_call_m68k_read_tp): Likewise.
17232 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
17233 (microblaze_expand_divide): Likewise.
17234 * config/mips/mips.h (mips_args): Likewise.
17235 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
17236 (MIPS_ICACHE_SYNC): Likewise.
17237 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
17238 (nios2_trampoline_init): Likewise.
17239 * config/pa/pa.c (hppa_tls_call): Likewise.
17240 (pa_trampoline_init): Likewise.
17241 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
17242 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
17243 (expand_strn_compare): Likewise.
17244 (rs6000_generate_compare): Likewise.
17245 (rs6000_expand_float128_convert): Likewise.
17246 (output_profile_hook): Likewise.
17247 (rs6000_trampoline_init): Likewise.
17248 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
17249 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
17250 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
17251 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
17252 (rs6000_generate_compare): Likewise.
17253 (rs6000_expand_float128_convert): Likewise.
17254 (output_profile_hook): Likewise.
17255 (rs6000_trampoline_init): Likewise.
17256 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
17257 * config/sh/sh.c (sh_trampoline_init): Likewise.
17258 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
17259 (sparc_emit_float_lib_cmp): Likewise.
17260 (sparc32_initialize_trampoline): Likewise.
17261 (sparc64_initialize_trampoline): Likewise.
17262 (sparc_profile_hook): Likewise.
17263 * config/spu/spu.c (ea_load_store): Likewise.
17264 * config/spu/spu.md (floatunssidf2): Likewise.
17265 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
17266 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
17267 * config/visium/visium.c (expand_block_move_4): Likewise.
17268 (expand_block_move_2): Likewise.
17269 (expand_block_move_1): Likewise.
17270 (expand_block_set_4): Likewise.
17271 (expand_block_set_2): Likewise.
17272 (expand_block_set_1): Likewise.
17273 (visium_trampoline_init): Likewise.
17274 (visium_profile_hook): Likewise.
17275 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
17276 (xtensa_setup_frame_addresses): Likewise.
17277 (xtensa_trampoline_init): Likewise.
17278 * except.c (sjlj_emit_function_enter): Likewise.
17279 (sjlj_emit_function_exit): Likewise.
17280 * explow.c (allocate_dynamic_stack_space): Likewise.
17281 (probe_stack_range): Likewise.
17282 * expr.c (convert_mode_scalar): Likewise.
17283 * optabs.c (expand_binop): Likewise.
17284 (expand_twoval_binop_libfunc): Likewise.
17285 (expand_unop): Likewise.
17286 (prepare_cmp_insn): Likewise.
17287 (prepare_float_lib_cmp): Likewise.
17288 (expand_float): Likewise.
17289 (expand_fix): Likewise.
17290 (expand_fixed_convert): Likewise.
17291 (maybe_emit_sync_lock_test_and_set): Likewise.
17292 (expand_atomic_compare_and_swap): Likewise.
17293 (expand_mem_thread_fence): Likewise.
17294 (expand_atomic_fetch_op): Likewise.
17296 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
17298 * doc/generic.texi (OpenACC): Adjust URL.
17299 * doc/invoke.texi (C Dialect Options): Ditto.
17301 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
17303 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
17304 predicate for operand 1. Add (m,<S>) constraint.
17305 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
17306 Prevent memory operand 1 with register operand 2.
17308 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
17310 PR rtl-optimization/82024
17311 * combine.c (try_combine): If the combination result is a PARALLEL,
17312 and we only need to retain the SET in there that would be placed
17313 at I2, check that we can place that at I3 instead, before doing so.
17315 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17318 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
17320 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
17321 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
17324 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
17325 Jeff Law <law@redhat.com>
17327 * varasm.c (bss_initializer_p): Do not put constants into .bss
17328 (categorize_decl_for_section): Handle bss_initializer_p returning
17329 false when DECL_INITIAL is NULL.
17331 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17334 * config/s390/s390.c (s390_can_inline_p): New function.
17336 2017-09-01 Jeff Law <law@redhat.com>
17338 PR tree-optimization/82052
17339 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17340 Always initialize the returned slot after a hash table miss
17341 when INSERT is true.
17343 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
17345 * config/s390/s390.md (mem_signal_fence): Remove.
17346 * doc/md.texi (mem_signal_fence): Remove.
17347 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
17349 * target-insns.def (mem_signal_fence): Remove.
17351 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17354 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
17357 * asan.c (create_odr_indicator): Strip name encoding from assembler
17358 name before appending it after __odr_asan_.
17360 2017-09-01 Martin Liska <mliska@suse.cz>
17362 PR tree-optimization/82059
17363 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
17364 frequency only when an edge is redirected.
17366 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17368 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
17369 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
17370 (arc_conditional_register_usage): Remove ARC600 lp_count
17372 (arc_file_start): Emit Tag_ARC_CPU_variation.
17373 (arc_can_use_doloop_p): New conditions to use ZOLs.
17374 (hwloop_fail): New function.
17375 (hwloop_optimize): Likewise.
17376 (hwloop_pattern_reg): Likewise.
17377 (arc_doloop_hooks): New struct, to be used with reorg_loops.
17378 (arc_reorg_loops): New function, calls reorg_loops.
17379 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
17380 (arc600_corereg_hazard): Remove ZOL checking, case handled by
17382 (arc_loop_hazard): Remove function, functionality moved into
17384 (arc_hazard): Remove arc_loop_hazard call.
17385 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
17386 into hwloop_optimize.
17387 (arc_label_align): Remove ZOL handling.
17388 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
17389 * config/arc/arc.md (doloop_begin): Remove pattern.
17390 (doloop_begin_i): Likewise.
17391 (doloop_end_i): Likewise.
17392 (doloop_fallback): Likewise.
17393 (doloop_fallback_m): Likewise.
17394 (doloop_end): Reimplement expand.
17395 (arc_lp): New pattern for LP instruction.
17396 (loop_end): New pattern.
17397 (loop_fail): Likewise.
17398 (decrement_and_branch_until_zero): Likewise.
17399 * config/arc/arc.opt (mlpc-width): New option.
17400 * doc/invoke.texi (mlpc-width): Document option.
17402 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17404 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
17405 (arc_ccfsm_advance): Fix checking for delay slots.
17406 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
17408 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17410 * config/arc/arc.md (movqi_insn): Add stores to save constant long
17412 (movhi_insn): Update store instruction constraint which are saving
17413 6-bit short immediates.
17414 (movsi_insn): Consider also short scaled load operations.
17415 (zero_extendhisi2_i): Use Usd constraint instead of T.
17416 (extendhisi2_i): Add q constraint.
17417 (arc_clzsi2): Add type and length attributes.
17418 (arc_ctzsi2): Likewise.
17419 * config/arc/constraints.md (Usc): Update constraint, the
17420 assembler can parse two relocations for a single instruction.
17422 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17424 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
17425 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
17427 2017-08-31 Olivier Hainque <hainque@adacore.com>
17429 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
17430 match as powerpc-wrs-vxworks*.
17432 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
17434 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
17435 register constraint for by-element operand.
17436 (aarch64_mls_elt_merge<mode>): Likewise.
17438 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17440 * config/arc/arc.c (arc_can_follow_jump): Check for short
17443 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17445 * config.gcc: Use g.opt for arc.
17446 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
17447 functionality moved to ...
17448 (legitimate_scaled_address_p): New function, ...here.
17449 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
17450 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
17451 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
17453 (arc_override_options): Handle G option.
17454 (arc_output_pic_addr_const): Correct function definition.
17455 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
17456 (arc_decl_anon_ns_mem_p): Delete.
17457 (arc_in_small_data_p): Overhaul this function to take into
17458 consideration the value given via G option.
17459 (arc_rewrite_small_data_1): Renamed and corrected old
17460 arc_rewrite_small_data function.
17461 (arc_rewrite_small_data): New function.
17462 (small_data_pattern): Don't use pic_offset_table_rtx.
17463 * config/arc/arc.h (CC1_SPEC): Recognize G option.
17464 * config/arc/simdext.md (movmisalignv2hi): Use
17465 prepare_move_operands function.
17467 (movmisalign*): Likewise.
17468 * doc/invoke.texi (ARC options): Document -G option.
17470 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17472 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
17474 * config/arc/arc.c (arc_print_operand): Output scalled address for
17475 sdata whenever is possible.
17476 (arc_in_small_data_p): Allow sdata for 64bit datum when double
17477 load/stores are available.
17478 (compact_sda_memory_operand): Check for the alignment required by
17479 code density instructions.
17480 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
17482 * config/arc/constraints.md (Usd): Update constraint.
17483 (Us0): New constraint.
17484 (Usc): Update constraint.
17486 2017-08-31 Richard Biener <rguenther@suse.de>
17488 PR middle-end/82054
17489 * dwarf2out.c (dwarf2out_early_global_decl): Process each
17490 function only once.
17492 2017-08-31 Tamar Christina <tamar.christina@arm.com>
17494 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
17495 Resize type_signature.
17497 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17498 Alan Hayward <alan.hayward@arm.com>
17499 David Sherwood <david.sherwood@arm.com>
17501 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
17502 subregs whose inner modes can be stored in GPRs.
17503 (aarch64_classify_index): Likewise.
17505 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17506 Alan Hayward <alan.hayward@arm.com>
17507 David Sherwood <david.sherwood@arm.com>
17509 * config/aarch64/iterators.md (V_cmp_result): Rename to...
17510 (V_INT_EQUIV): ...this.
17511 (v_cmp_result): Rename to...
17512 (v_int_equiv): ...this.
17513 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
17514 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
17515 (copysign<mode>3): Likewise.
17516 (aarch64_simd_bsl<mode>_internal): Likewise.
17517 (aarch64_simd_bsl<mode>): Likewise.
17518 (vec_cmp<mode><mode>): Likewise.
17519 (vcond<mode><mode>): Likewise.
17520 (vcond<v_cmp_mixed><mode>): Likewise.
17521 (vcondu<mode><v_cmp_mixed>): Likewise.
17522 (aarch64_cm<optab><mode>): Likewise.
17523 (aarch64_cmtst<mode>): Likewise.
17524 (aarch64_fac<optab><mode>): Likewise.
17525 (vec_perm_const<mode>): Likewise.
17526 (vcond_mask_<mode><v_cmp_result>): Rename to...
17527 (vcond_mask_<mode><v_int_equiv>): ...this.
17528 (vec_cmp<mode><v_cmp_result>): Rename to...
17529 (vec_cmp<mode><v_int_equiv>): ...this.
17531 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17532 Alan Hayward <alan.hayward@arm.com>
17533 David Sherwood <david.sherwood@arm.com>
17535 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
17537 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
17538 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
17539 (UNSPEC_LD4_DREG): New unspecs.
17540 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
17541 (aarch64_ld2<mode>_dreg_be): Replace with...
17542 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
17544 (aarch64_ld3<mode>_dreg_le)
17545 (aarch64_ld3<mode>_dreg_be): Replace with...
17546 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
17548 (aarch64_ld4<mode>_dreg_le)
17549 (aarch64_ld4<mode>_dreg_be): Replace with...
17550 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
17553 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17555 PR tree-optimization/81987
17556 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
17557 insert an initializer in a location not dominated by the stride
17560 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17562 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
17563 on the entire header of the finally block in the fallthru case.
17565 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17567 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
17569 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
17571 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
17572 rs6000_emit_move_from_cr and call renamed function.
17573 (rs6000_emit_prologue): Call renamed functions.
17574 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
17575 movesi_from_cr, remove volatile CRs.
17577 2017-08-30 Jon Beniston <jon@beniston.com>
17578 Richard Biener <rguenther@suse.de>
17580 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
17581 of VECTOR_MODE_P check.
17582 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17583 element vector types.
17585 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17587 * df.h (df_read_modify_subreg_p): Remove in favor of...
17588 * rtl.h (read_modify_subreg_p): ...this new function. Take a
17589 const_rtx instead of an rtx.
17590 * cprop.c (local_cprop_find_used_regs): Update accordingly.
17591 * df-problems.c (df_word_lr_mark_ref): Likewise.
17592 * ira-lives.c (mark_pseudo_reg_live): Likewise.
17593 (mark_pseudo_reg_dead): Likewise.
17594 (mark_ref_dead): Likewise.
17595 * reginfo.c (init_subregs_of_mode): Likewise.
17596 * sched-deps.c (sched_analyze_1): Likewise.
17597 * df-scan.c (df_def_record_1): Likewise.
17598 (df_uses_record): Likewise.
17599 (df_read_modify_subreg_p): Remove in favor of...
17600 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
17601 const_rtx instead of an rtx.
17603 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17604 Alan Hayward <alan.hayward@arm.com>
17605 David Sherwood <david.sherwood@arm.com>
17607 * rtl.h (partial_subreg_p): New function.
17608 * caller-save.c (save_call_clobbered_regs): Use it.
17609 * calls.c (expand_call): Likewise.
17610 * combine.c (combinable_i3pat): Likewise.
17611 (simplify_set): Likewise.
17612 (make_extraction): Likewise.
17613 (make_compound_operation_int): Likewise.
17614 (gen_lowpart_or_truncate): Likewise.
17615 (force_to_mode): Likewise.
17616 (make_field_assignment): Likewise.
17617 (reg_truncated_to_mode): Likewise.
17618 (record_truncated_value): Likewise.
17619 (move_deaths): Likewise.
17620 * cse.c (record_jump_cond): Likewise.
17621 (cse_insn): Likewise.
17622 * cselib.c (cselib_lookup_1): Likewise.
17623 * expmed.c (extract_bit_field_using_extv): Likewise.
17624 * function.c (assign_parm_setup_reg): Likewise.
17625 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17626 * ira-build.c (create_insn_allocnos): Likewise.
17627 * lra-coalesce.c (merge_pseudos): Likewise.
17628 * lra-constraints.c (match_reload): Likewise.
17629 (simplify_operand_subreg): Likewise.
17630 (curr_insn_transform): Likewise.
17631 * lra-lives.c (process_bb_lives): Likewise.
17632 * lra.c (new_insn_reg): Likewise.
17633 (lra_substitute_pseudo): Likewise.
17634 * regcprop.c (mode_change_ok): Likewise.
17635 (maybe_mode_change): Likewise.
17636 (copyprop_hardreg_forward_1): Likewise.
17637 * reload.c (push_reload): Likewise.
17638 (find_reloads): Likewise.
17639 (find_reloads_subreg_address): Likewise.
17640 * reload1.c (alter_reg): Likewise.
17641 (eliminate_regs_1): Likewise.
17642 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17644 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17646 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17647 back to if statements, including unpack.
17649 2017-08-30 Martin Liska <mliska@suse.cz>
17651 PR inline-asm/82001
17652 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17654 (func_checker::compare_asm_inputs_outputs): ... this function.
17655 (func_checker::compare_gimple_asm): Use the function to compare
17656 also ASM constrains.
17657 * ipa-icf-gimple.h: Rename the function.
17659 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17660 Alan Hayward <alan.hayward@arm.com>
17661 David Sherwood <david.sherwood@arm.com>
17663 * coretypes.h (complex_mode): New type.
17664 * gdbhooks.py (build_pretty_printer): Handle it.
17665 * machmode.h (complex_mode): New class.
17666 (complex_mode::includes_p): New function.
17667 (is_complex_int_mode): Likewise.
17668 (is_complex_float_mode): Likewise.
17669 * genmodes.c (get_mode_class): Handle complex mode classes.
17670 * function.c (expand_function_end): Use is_complex_int_mode.
17672 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17673 Alan Hayward <alan.hayward@arm.com>
17674 David Sherwood <david.sherwood@arm.com>
17676 * coretypes.h (scalar_mode_pod): New typedef.
17677 * gdbhooks.py (build_pretty_printer): Handle it.
17678 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17679 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17680 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17681 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17682 as_a <scalar_mode>.
17684 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17685 Alan Hayward <alan.hayward@arm.com>
17686 David Sherwood <david.sherwood@arm.com>
17688 * machmode.h (mode_for_vector): Take a scalar_mode instead
17690 * stor-layout.c (mode_for_vector): Likewise.
17691 * explow.c (promote_mode): Use as_a <scalar_mode>.
17692 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
17694 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17695 Alan Hayward <alan.hayward@arm.com>
17696 David Sherwood <david.sherwood@arm.com>
17698 * target.def (preferred_simd_mode): Take a scalar_mode
17699 instead of a machine_mode.
17700 * targhooks.h (default_preferred_simd_mode): Likewise.
17701 * targhooks.c (default_preferred_simd_mode): Likewise.
17702 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17703 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17704 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17705 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17706 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17707 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17708 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17709 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17711 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17712 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17713 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17714 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17715 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17716 * doc/tm.texi: Regenerate.
17717 * optabs-query.c (can_vec_mask_load_store_p): Return false for
17720 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17721 Alan Hayward <alan.hayward@arm.com>
17722 David Sherwood <david.sherwood@arm.com>
17724 * target.def (scalar_mode_supported_p): Take a scalar_mode
17725 instead of a machine_mode.
17726 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17727 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17728 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17729 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17730 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17731 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17732 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17733 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17734 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17735 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17736 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17737 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17738 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17739 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17741 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17742 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17743 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17744 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17745 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17747 * doc/tm.texi: Regenerate.
17749 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17750 Alan Hayward <alan.hayward@arm.com>
17751 David Sherwood <david.sherwood@arm.com>
17753 * coretypes.h (opt_scalar_mode): New typedef.
17754 * gdbhooks.py (build_pretty_printers): Handle it.
17755 * machmode.h (mode_iterator::get_2xwider): Add overload for
17757 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17759 * expr.c (convert_mode_scalar): Likewise.
17760 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17761 * optabs.c (expand_float): Likewise.
17762 (expand_fix): Likewise.
17763 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17765 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17766 Alan Hayward <alan.hayward@arm.com>
17767 David Sherwood <david.sherwood@arm.com>
17769 * optabs.c (expand_float): Explicitly check for scalars before
17770 using a branching expansion.
17771 (expand_fix): Likewise.
17773 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17774 Alan Hayward <alan.hayward@arm.com>
17775 David Sherwood <david.sherwood@arm.com>
17777 * expr.c (convert_mode): Split scalar handling out into...
17778 (convert_mode_scalar): ...this new function. Treat the modes
17781 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17782 Alan Hayward <alan.hayward@arm.com>
17783 David Sherwood <david.sherwood@arm.com>
17785 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17787 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17789 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17790 Alan Hayward <alan.hayward@arm.com>
17791 David Sherwood <david.sherwood@arm.com>
17793 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17794 rather than a machine_mode.
17795 (fixed_from_string): Likewise.
17796 (fixed_convert): Likewise.
17797 (fixed_convert_from_int): Likewise.
17798 (fixed_convert_from_real): Likewise.
17799 (real_convert_from_fixed): Likewise.
17800 * fixed-value.c (fixed_from_double_int): Likewise.
17801 (fixed_from_string): Likewise.
17802 (fixed_convert): Likewise.
17803 (fixed_convert_from_int): Likewise.
17804 (fixed_convert_from_real): Likewise.
17805 (real_convert_from_fixed): Likewise.
17806 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17808 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17809 Alan Hayward <alan.hayward@arm.com>
17810 David Sherwood <david.sherwood@arm.com>
17812 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17813 of separate mode class checks. Do not allow vector modes here.
17814 (immed_wide_int_const): Use as_a <scalar_mode>.
17815 * explow.c (trunc_int_for_mode): Likewise.
17816 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17817 (wi::shwi): Likewise.
17818 (wi::min_value): Likewise.
17819 (wi::max_value): Likewise.
17820 * dwarf2out.c (loc_descriptor): Likewise.
17821 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17822 for CONST_WIDE_INT.
17824 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17825 Alan Hayward <alan.hayward@arm.com>
17826 David Sherwood <david.sherwood@arm.com>
17828 * tree.h (SCALAR_TYPE_MODE): New macro.
17829 * expr.c (expand_expr_addr_expr_1): Use it.
17830 (expand_expr_real_2): Likewise.
17831 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17832 (fold_convert_const_fixed_from_int): Likewise.
17833 (fold_convert_const_fixed_from_real): Likewise.
17834 (native_encode_fixed): Likewise
17835 (native_encode_complex): Likewise
17836 (native_encode_vector): Likewise.
17837 (native_interpret_fixed): Likewise.
17838 (native_interpret_real): Likewise.
17839 (native_interpret_complex): Likewise.
17840 (native_interpret_vector): Likewise.
17841 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17842 (simd_clone_adjust_argument_types): Likewise.
17843 (simd_clone_init_simd_arrays): Likewise.
17844 (simd_clone_adjust): Likewise.
17845 * stor-layout.c (layout_type): Likewise.
17846 * tree.c (build_minus_one_cst): Likewise.
17847 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17848 * tree-inline.c (estimate_move_cost): Likewise.
17849 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17850 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17851 (vectorizable_reduction): Likewise.
17852 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17853 (vect_recog_mixed_size_cond_pattern): Likewise.
17854 (check_bool_pattern): Likewise.
17855 (adjust_bool_pattern): Likewise.
17856 (search_type_for_mask_1): Likewise.
17857 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17858 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17859 (vectorizable_load): Likewise.
17860 (vectorizable_store): Likewise.
17861 * ubsan.c (ubsan_encode_value): Likewise.
17862 * varasm.c (output_constant): Likewise.
17864 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17865 Alan Hayward <alan.hayward@arm.com>
17866 David Sherwood <david.sherwood@arm.com>
17868 * coretypes.h (scalar_mode): New class.
17869 * machmode.h (scalar_mode): Likewise.
17870 (scalar_mode::includes_p): New function.
17871 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
17872 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
17873 * genmodes.c (get_mode_class): Handle remaining scalar modes.
17874 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
17875 * expmed.c (store_bit_field_1): Likewise.
17876 (extract_bit_field_1): Likewise.
17877 * expr.c (write_complex_part): Likewise.
17878 (read_complex_part): Likewise.
17879 (emit_move_complex_push): Likewise.
17880 (expand_expr_real_2): Likewise.
17881 * function.c (assign_parm_setup_reg): Likewise.
17882 (assign_parms_unsplit_complex): Likewise.
17883 * optabs.c (expand_binop): Likewise.
17884 * rtlanal.c (subreg_get_info): Likewise.
17885 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17886 * varasm.c (output_constant_pool_2): Likewise.
17888 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17889 Alan Hayward <alan.hayward@arm.com>
17890 David Sherwood <david.sherwood@arm.com>
17892 * expmed.c (extract_high_half): Use scalar_int_mode and remove
17894 (expmed_mult_highpart_optab): Likewise.
17895 (expmed_mult_highpart): Likewise.
17897 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17898 Alan Hayward <alan.hayward@arm.com>
17899 David Sherwood <david.sherwood@arm.com>
17901 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
17902 instead of a machine_mode.
17903 (builtin_memset_read_str): Likewise.
17904 * builtins.c (c_readstr): Likewise.
17905 (builtin_memcpy_read_str): Likewise.
17906 (builtin_strncpy_read_str): Likewise.
17907 (builtin_memset_read_str): Likewise.
17908 (builtin_memset_gen_str): Likewise.
17909 (expand_builtin_signbit): Use scalar_int_mode for local variables.
17910 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
17911 instead of a machine_mode.
17912 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
17914 (make_extraction): Likewise.
17915 (try_widen_shift_mode): Take and return scalar_int_modes instead
17917 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
17918 a scalar_int_mode instead of a machine_mode.
17919 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
17920 (avr_addr_space_pointer_mode): Likewise.
17921 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
17922 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
17923 (msp430_unwind_word_mode): Likewise.
17924 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
17925 (spu_addr_space_pointer_mode): Likewise.
17926 (spu_addr_space_address_mode): Likewise.
17927 (spu_libgcc_cmp_return_mode): Likewise.
17928 (spu_libgcc_shift_count_mode): Likewise.
17929 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
17930 (rl78_addr_space_pointer_mode): Likewise.
17931 (fl78_unwind_word_mode): Likewise.
17932 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
17934 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
17935 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
17936 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
17937 (mips_valid_pointer_mode): Likewise.
17938 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
17939 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
17940 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
17942 (ft32_addr_space_address_mode): Likewise.
17943 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
17944 scalar_int_mode instead of a machine_mode.
17945 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
17947 (m32c_addr_space_address_mode): Likewise.
17948 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
17949 (rs6000_eh_return_filter_mode): Likewise.
17950 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
17951 (rs6000_eh_return_filter_mode): Likewise.
17952 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
17953 (s390_libgcc_shift_count_mode): Likewise.
17954 (s390_unwind_word_mode): Likewise.
17955 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
17957 * target.def (mode_rep_extended): Likewise.
17958 (valid_pointer_mode): Likewise.
17959 (addr_space.valid_pointer_mode): Likewise.
17960 (eh_return_filter_mode): Return a scalar_int_mode rather than
17962 (libgcc_cmp_return_mode): Likewise.
17963 (libgcc_shift_count_mode): Likewise.
17964 (unwind_word_mode): Likewise.
17965 (addr_space.pointer_mode): Likewise.
17966 (addr_space.address_mode): Likewise.
17967 * doc/tm.texi: Regenerate.
17968 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
17970 (do_jump): Use scalar_int_mode for local variables.
17971 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
17972 rather than a machine_mode.
17973 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
17974 (scompare_loc_descriptor_wide): Likewise.
17975 (scompare_loc_descriptor_narrow): Likewise.
17976 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
17978 * except.c (sjlj_emit_dispatch_table): Likewise.
17979 (expand_builtin_eh_copy_values): Likewise.
17980 * explow.c (convert_memory_address_addr_space_1): Likewise.
17981 Take a scalar_int_mode rather than a machine_mode.
17982 (convert_memory_address_addr_space): Take a scalar_int_mode rather
17983 than a machine_mode.
17984 (memory_address_addr_space): Use scalar_int_mode for local variables.
17985 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
17986 rather than a machine_mode.
17987 * expmed.c (mask_rtx): Likewise.
17988 (init_expmed_one_conv): Likewise.
17989 (expand_mult_highpart_adjust): Likewise.
17990 (extract_high_half): Likewise.
17991 (expmed_mult_highpart_optab): Likewise.
17992 (expmed_mult_highpart): Likewise.
17993 (expand_smod_pow2): Likewise.
17994 (expand_sdiv_pow2): Likewise.
17995 (emit_store_flag_int): Likewise.
17996 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
17998 (extract_low_bits): Likewise.
17999 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18001 * expr.c (pieces_addr::adjust): Likewise.
18002 (can_store_by_pieces): Likewise.
18003 (store_by_pieces): Likewise.
18004 (clear_by_pieces_1): Likewise.
18005 (expand_expr_addr_expr_1): Likewise.
18006 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18007 (expand_expr_real_1): Likewise.
18008 (try_casesi): Likewise.
18009 * final.c (shorten_branches): Likewise.
18010 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18011 type of "mode" to machine_mode.
18012 * internal-fn.c (expand_arith_overflow_result_store): Take a
18013 scalar_int_mode rather than a machine_mode.
18014 (expand_mul_overflow): Use scalar_int_mode for local variables.
18015 * loop-doloop.c (doloop_modify): Likewise.
18016 (doloop_optimize): Likewise.
18017 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18018 than a machine_mode.
18019 (expand_doubleword_shift_condmove): Likewise.
18020 (expand_doubleword_shift): Likewise.
18021 (expand_doubleword_clz): Likewise.
18022 (expand_doubleword_popcount): Likewise.
18023 (expand_doubleword_parity): Likewise.
18024 (expand_absneg_bit): Use scalar_int_mode for local variables.
18025 (prepare_float_lib_cmp): Likewise.
18026 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18027 rather than a machine_mode.
18028 (convert_memory_address_addr_space): Likewise.
18029 (get_mode_bounds): Likewise.
18030 (get_address_mode): Return a scalar_int_mode rather than a
18032 * rtlanal.c (get_address_mode): Likewise.
18033 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18034 than a machine_mode.
18035 * targhooks.c (default_mode_rep_extended): Likewise.
18036 (default_valid_pointer_mode): Likewise.
18037 (default_addr_space_valid_pointer_mode): Likewise.
18038 (default_eh_return_filter_mode): Return a scalar_int_mode rather
18039 than a machine_mode.
18040 (default_libgcc_cmp_return_mode): Likewise.
18041 (default_libgcc_shift_count_mode): Likewise.
18042 (default_unwind_word_mode): Likewise.
18043 (default_addr_space_pointer_mode): Likewise.
18044 (default_addr_space_address_mode): Likewise.
18045 * targhooks.h (default_eh_return_filter_mode): Likewise.
18046 (default_libgcc_cmp_return_mode): Likewise.
18047 (default_libgcc_shift_count_mode): Likewise.
18048 (default_unwind_word_mode): Likewise.
18049 (default_addr_space_pointer_mode): Likewise.
18050 (default_addr_space_address_mode): Likewise.
18051 (default_mode_rep_extended): Take a scalar_int_mode rather than
18053 (default_valid_pointer_mode): Likewise.
18054 (default_addr_space_valid_pointer_mode): Likewise.
18055 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18057 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18058 rather than a machine_mode.
18059 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18060 for local variables.
18061 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18062 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18063 than a machine_mode.
18065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18066 Alan Hayward <alan.hayward@arm.com>
18067 David Sherwood <david.sherwood@arm.com>
18069 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18070 the mode argument to scalar_int_mode.
18071 (do_jump_by_parts_zero_rtx): Likewise.
18072 (do_jump_by_parts_equality_rtx): Likewise.
18073 (do_jump_by_parts_greater): Take a mode argument.
18074 (do_jump_by_parts_equality): Likewise.
18075 (do_jump_1): Update calls accordingly.
18077 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18078 Alan Hayward <alan.hayward@arm.com>
18079 David Sherwood <david.sherwood@arm.com>
18081 * is-a.h (safe_dyn_cast): New function.
18082 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18083 (jump_table_for_label): Likewise.
18084 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18085 instead of an rtx_insn *.
18086 (shorten_branches): Use dyn_cast instead of LABEL_P and
18087 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18088 rtx_jump_table_data::get_data_mode.
18089 (final_scan_insn): Likewise.
18091 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18092 Alan Hayward <alan.hayward@arm.com>
18093 David Sherwood <david.sherwood@arm.com>
18095 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18096 trying to combine a full-register integer set with a subreg
18099 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18100 Alan Hayward <alan.hayward@arm.com>
18101 David Sherwood <david.sherwood@arm.com>
18103 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18104 that is always either address_mode or pointer_mode.
18106 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18107 Alan Hayward <alan.hayward@arm.com>
18108 David Sherwood <david.sherwood@arm.com>
18110 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
18111 when the two are known to be equal.
18113 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18114 Alan Hayward <alan.hayward@arm.com>
18115 David Sherwood <david.sherwood@arm.com>
18117 * simplify-rtx.c (simplify_const_unary_operation): Use
18118 is_a <scalar_int_mode> instead of checking for a nonzero
18119 precision. Forcibly convert op_mode to a scalar_int_mode
18120 in that case. More clearly differentiate the operand and
18121 result modes and use the former when deciding what the value
18122 of a count-bits operation should be. Use is_int_mode instead
18123 of checking for a MODE_INT. Remove redundant check for whether
18124 this mode has a zero precision.
18126 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18127 Alan Hayward <alan.hayward@arm.com>
18128 David Sherwood <david.sherwood@arm.com>
18130 * optabs.c (widen_leading): Change the type of the mode argument
18131 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
18132 (widen_bswap): Likewise.
18133 (expand_parity): Likewise.
18134 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
18135 (expand_ffs): Likewise.
18136 (epand_unop): Check for scalar integer modes before calling the
18139 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18140 Alan Hayward <alan.hayward@arm.com>
18141 David Sherwood <david.sherwood@arm.com>
18143 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
18145 (expand_expr_real_1): Update call accordingly.
18147 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18148 Alan Hayward <alan.hayward@arm.com>
18149 David Sherwood <david.sherwood@arm.com>
18151 * expmed.c (store_bit_field_using_insv): Add op0_mode and
18152 value_mode arguments. Use scalar_int_mode internally.
18153 (store_bit_field_1): Rename the new integer mode from imode
18154 to op0_mode and use it instead of GET_MODE (op0). Update calls
18155 to store_split_bit_field, store_bit_field_using_insv and
18156 store_fixed_bit_field.
18157 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
18158 Use scalar_int_mode internally. Use a bit count rather than a mode
18159 when calculating the largest bit size for get_best_mode.
18160 Update calls to store_split_bit_field and store_fixed_bit_field_1.
18161 (store_fixed_bit_field_1): Add mode and value_mode arguments.
18162 Remove assertion that OP0 has a scalar integer mode.
18163 (store_split_bit_field): Add op0_mode and value_mode arguments.
18164 Update calls to extract_fixed_bit_field.
18165 (extract_bit_field_using_extv): Add an op0_mode argument.
18166 Use scalar_int_mode internally.
18167 (extract_bit_field_1): Rename the new integer mode from imode to
18168 op0_mode and use it instead of GET_MODE (op0). Update calls to
18169 extract_split_bit_field, extract_bit_field_using_extv and
18170 extract_fixed_bit_field.
18171 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
18172 to extract_split_bit_field and extract_fixed_bit_field_1.
18173 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
18174 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
18175 on the target mode.
18176 (extract_split_bit_field): Add an op0_mode argument. Update call
18177 to extract_fixed_bit_field.
18179 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18180 Alan Hayward <alan.hayward@arm.com>
18181 David Sherwood <david.sherwood@arm.com>
18183 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
18184 * explow.c (hard_function_value): Likewise.
18185 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
18186 convert_to_mode call outside the loop.
18187 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
18188 for the mode iterator. Require the mode specified by max_pieces
18190 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
18192 (copy_blkmode_to_reg): Likewise.
18193 (set_storage_via_setmem): Likewise.
18194 * optabs.c (prepare_cmp_insn): Likewise.
18195 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18196 * stor-layout.c (finish_bitfield_representative): Likewise.
18198 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18199 Alan Hayward <alan.hayward@arm.com>
18200 David Sherwood <david.sherwood@arm.com>
18202 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
18203 * expr.c (convert_move): Use them.
18204 (convert_modes): Likewise.
18205 (store_expr_with_bounds): Likewise.
18207 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18208 Alan Hayward <alan.hayward@arm.com>
18209 David Sherwood <david.sherwood@arm.com>
18211 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
18212 parameter for the mode of "x". Remove the "known_x", "known_mode"
18213 and "known_ret" arguments. Change the type of the mode argument
18214 to scalar_int_mode.
18215 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
18216 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
18217 (reg_num_sign_bit_copies_for_combine): Likewise.
18218 * rtlanal.c (nonzero_bits1): Likewise.
18219 (num_sign_bit_copies1): Likewise.
18220 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
18221 (reg_num_sign_bit_copies_general): Likewise.
18222 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
18223 (reg_nonzero_bits_general): Likewise.
18225 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18226 Alan Hayward <alan.hayward@arm.com>
18227 David Sherwood <david.sherwood@arm.com>
18229 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
18230 than in subroutines. Return 1 for non-integer modes.
18231 (cached_num_sign_bit_copies): Change the type of the mode parameter
18232 to scalar_int_mode.
18233 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
18234 classes. Handle CONST_INT_P first and then check whether X also
18235 has a scalar integer mode. Check the same thing for inner registers
18236 of a SUBREG and for values that are being extended or truncated.
18238 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18239 Alan Hayward <alan.hayward@arm.com>
18240 David Sherwood <david.sherwood@arm.com>
18242 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
18243 in subroutines. Return the mode mask for non-integer modes.
18244 (cached_nonzero_bits): Change the type of the mode parameter
18245 to scalar_int_mode.
18246 (nonzero_bits1): Likewise. Remove early exit for other mode
18247 classes. Handle CONST_INT_P first and then check whether X
18248 also has a scalar integer mode.
18250 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18251 Alan Hayward <alan.hayward@arm.com>
18252 David Sherwood <david.sherwood@arm.com>
18254 * expr.c (widest_int_mode_for_size): Make the comment match the code.
18255 Return a scalar_int_mode and assert that the size is greater than
18257 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
18258 (op_by_pieces_d::op_by_pieces_d): Likewise.
18259 (op_by_pieces_d::run): Likewise.
18260 (can_store_by_pieces): Likewise.
18262 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18263 Alan Hayward <alan.hayward@arm.com>
18264 David Sherwood <david.sherwood@arm.com>
18266 * combine.c (extract_left_shift): Add a mode argument and update
18268 (make_compound_operation_int): Change the type of the mode parameter
18269 to scalar_int_mode and update the call to extract_left_shift.
18271 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18272 Alan Hayward <alan.hayward@arm.com>
18273 David Sherwood <david.sherwood@arm.com>
18275 * combine.c (simplify_and_const_int): Change the type of the mode
18276 parameter to scalar_int_mode.
18277 (simplify_and_const_int_1): Likewise. Update recursive call.
18279 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18280 Alan Hayward <alan.hayward@arm.com>
18281 David Sherwood <david.sherwood@arm.com>
18283 * combine.c (simplify_compare_const): Check that the mode is a
18284 scalar_int_mode (rather than VOIDmode) before testing its
18286 (simplify_comparison): Move COMPARISON_P handling out of the
18287 loop and restrict the latter part of the loop to scalar_int_modes.
18288 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
18289 and when considering SUBREG_REGs. Use is_int_mode instead of
18290 checking GET_MODE_CLASS against MODE_INT.
18292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18293 Alan Hayward <alan.hayward@arm.com>
18294 David Sherwood <david.sherwood@arm.com>
18296 * combine.c (try_widen_shift_mode): Move check for equal modes to...
18297 (simplify_shift_const_1): ...here. Use scalar_int_mode for
18298 shift_unit_mode and for modes involved in scalar shifts.
18300 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18301 Alan Hayward <alan.hayward@arm.com>
18302 David Sherwood <david.sherwood@arm.com>
18304 * combine.c (force_int_to_mode): New function, split out from...
18305 (force_to_mode): ...here. Keep xmode up-to-date and use it
18306 instead of GET_MODE (x).
18308 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18309 Alan Hayward <alan.hayward@arm.com>
18310 David Sherwood <david.sherwood@arm.com>
18312 * optabs-query.h (extraction_insn::struct_mode): Change type to
18313 opt_scalar_int_mode and update comment.
18314 (extraction_insn::field_mode): Change type to scalar_int_mode.
18315 (extraction_insn::pos_mode): Likewise.
18316 * combine.c (make_extraction): Update accordingly.
18317 * optabs-query.c (get_traditional_extraction_insn): Likewise.
18318 (get_optab_extraction_insn): Likewise.
18319 * recog.c (simplify_while_replacing): Likewise.
18320 * expmed.c (narrow_bit_field_mem): Change the type of the mode
18321 parameter to opt_scalar_int_mode.
18323 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18324 Alan Hayward <alan.hayward@arm.com>
18325 David Sherwood <david.sherwood@arm.com>
18327 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
18328 to a scalar_int_mode instead of a machine_mode.
18329 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
18330 (get_best_mode): Return a boolean and use a pointer argument to store
18331 the selected mode. Replace the limit mode parameter with a bit limit.
18332 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
18333 for the values returned by bit_field_mode_iterator::next_mode.
18334 (store_bit_field): Update call to get_best_mode.
18335 (store_fixed_bit_field): Likewise.
18336 (extract_fixed_bit_field): Likewise.
18337 * expr.c (optimize_bitfield_assignment_op): Likewise.
18338 * fold-const.c (optimize_bit_field_compare): Likewise.
18339 (fold_truth_andor_1): Likewise.
18340 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
18341 Update for new type of m_mode.
18342 (get_best_mode): As above.
18344 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18345 Alan Hayward <alan.hayward@arm.com>
18346 David Sherwood <david.sherwood@arm.com>
18348 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
18349 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
18350 (store_bit_field): Check is_a <scalar_int_mode> before calling
18351 strict_volatile_bitfield_p.
18352 (extract_bit_field): Likewise.
18354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18355 Alan Hayward <alan.hayward@arm.com>
18356 David Sherwood <david.sherwood@arm.com>
18358 * target.def (cstore_mode): Return a scalar_int_mode.
18359 * doc/tm.texi: Regenerate.
18360 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
18361 * targhooks.h (default_cstore_mode): Likewise.
18362 * targhooks.c (default_cstore_mode): Likewise, using a forced
18364 * expmed.c (emit_cstore): Expect the target of the cstore to be
18367 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18368 Alan Hayward <alan.hayward@arm.com>
18369 David Sherwood <david.sherwood@arm.com>
18371 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
18373 (niter_desc): Likewise mode.
18374 (iv_analyze): Add a mode parameter.
18376 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
18377 and change its type to scalar_int_mode.
18378 * loop-iv.c: Update commentary at head of file.
18379 (iv_constant): Pass the mode paraeter before the rtx it describes
18380 and change its type to scalar_int_mode. Remove VOIDmode handling.
18381 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
18382 (iv_extend): Likewise.
18383 (shorten_into_mode): Likewise.
18384 (iv_add): Use scalar_int_mode.
18385 (iv_mult): Likewise.
18386 (iv_shift): Likewise.
18387 (canonicalize_iv_subregs): Likewise.
18388 (get_biv_step_1): Pass the outer_mode parameter before the rtx
18389 it describes and change its mode to scalar_int_mode. Also change
18390 the type of the returned inner_mode to scalar_int_mode.
18391 (get_biv_step): Likewise, turning outer_mode from a pointer
18392 into a direct parameter. Update call to get_biv_step_1.
18393 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
18394 iv_constant and get_biv_step.
18395 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
18396 and change its type to scalar_int_mode. Don't initialise iv->mode
18397 to VOIDmode and remove later checks for its still being VOIDmode.
18398 Update calls to iv_analyze_op and iv_analyze_expr. Check
18399 is_a <scalar_int_mode> when changing the mode under consideration.
18400 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
18401 Update call to iv_analyze_expr.
18402 (iv_analyze_op): Add a mode parameter. Reject subregs whose
18403 inner register is not also a scalar_int_mode. Update call to
18405 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
18406 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
18407 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
18408 separate mode class checks. Update calls to iv_analyze. Remove
18409 fix-up of VOIDmodes after iv_analyze_biv.
18410 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
18411 don't have a scalar_int_mode. Update call to biv_p.
18413 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18414 Alan Hayward <alan.hayward@arm.com>
18415 David Sherwood <david.sherwood@arm.com>
18417 * cfgexpand.c (convert_debug_memory_address): Use
18418 as_a <scalar_int_mode>.
18419 * combine.c (expand_compound_operation): Likewise.
18420 (make_extraction): Likewise.
18421 (change_zero_ext): Likewise.
18422 (simplify_comparison): Likewise.
18423 * cse.c (cse_insn): Likewise.
18424 * dwarf2out.c (minmax_loc_descriptor): Likewise.
18425 (mem_loc_descriptor): Likewise.
18426 (loc_descriptor): Likewise.
18427 * expmed.c (init_expmed_one_mode): Likewise.
18428 (synth_mult): Likewise.
18429 (emit_store_flag_1): Likewise.
18430 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
18431 of a comparison with size.
18432 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
18433 (reduce_to_bit_field_precision): Likewise.
18434 * function.c (expand_function_end): Likewise.
18435 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
18436 * loop-doloop.c (doloop_modify): Likewise.
18437 * optabs.c (expand_binop): Likewise.
18438 (expand_unop): Likewise.
18439 (expand_copysign_absneg): Likewise.
18440 (prepare_cmp_insn): Likewise.
18441 (maybe_legitimize_operand): Likewise.
18442 * recog.c (const_scalar_int_operand): Likewise.
18443 * rtlanal.c (get_address_mode): Likewise.
18444 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18445 (simplify_cond_clz_ctz): Likewise.
18446 * tree-nested.c (get_nl_goto_field): Likewise.
18447 * tree.c (build_vector_type_for_mode): Likewise.
18448 * var-tracking.c (use_narrower_mode): Likewise.
18450 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18451 Alan Hayward <alan.hayward@arm.com>
18452 David Sherwood <david.sherwood@arm.com>
18454 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
18455 * builtins.c (expand_builtin_signbit): Use it.
18456 * cfgexpand.c (expand_debug_expr): Likewise.
18457 * dojump.c (do_jump): Likewise.
18458 (do_compare_and_jump): Likewise.
18459 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
18460 * expmed.c (make_tree): Likewise.
18461 * expr.c (expand_expr_real_2): Likewise.
18462 (expand_expr_real_1): Likewise.
18463 (try_casesi): Likewise.
18464 * fold-const-call.c (fold_const_call_ss): Likewise.
18465 * fold-const.c (unextend): Likewise.
18466 (extract_muldiv_1): Likewise.
18467 (fold_single_bit_test): Likewise.
18468 (native_encode_int): Likewise.
18469 (native_encode_string): Likewise.
18470 (native_interpret_int): Likewise.
18471 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
18472 * internal-fn.c (expand_addsub_overflow): Likewise.
18473 (expand_neg_overflow): Likewise.
18474 (expand_mul_overflow): Likewise.
18475 (expand_arith_overflow): Likewise.
18476 * match.pd: Likewise.
18477 * stor-layout.c (layout_type): Likewise.
18478 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18479 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
18480 * tree-ssanames.c (get_range_info): Likewise.
18481 * tree-switch-conversion.c (array_value_type) Likewise.
18482 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
18483 (vect_recog_divmod_pattern): Likewise.
18484 (vect_recog_mixed_size_cond_pattern): Likewise.
18485 * tree-vrp.c (extract_range_basic): Likewise.
18486 (simplify_float_conversion_using_ranges): Likewise.
18487 * tree.c (int_fits_type_p): Likewise.
18488 * ubsan.c (instrument_bool_enum_load): Likewise.
18489 * varasm.c (mergeable_string_section): Likewise.
18490 (narrowing_initializer_constant_valid_p): Likewise.
18491 (output_constant): Likewise.
18493 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18494 Alan Hayward <alan.hayward@arm.com>
18495 David Sherwood <david.sherwood@arm.com>
18497 * machmode.h (NARROWEST_INT_MODE): New macro.
18498 * expr.c (alignment_for_piecewise_move): Use it instead of
18499 GET_CLASS_NARROWEST_MODE (MODE_INT).
18500 (push_block): Likewise.
18501 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
18503 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18505 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18506 Alan Hayward <alan.hayward@arm.com>
18507 David Sherwood <david.sherwood@arm.com>
18509 * postreload.c (move2add_valid_value_p): Change the type of the
18510 mode parameter to scalar_int_mode.
18511 (move2add_use_add2_insn): Add a mode parameter and use it instead
18513 (move2add_use_add3_insn): Likewise.
18514 (reload_cse_move2add): Update accordingly.
18516 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18517 Alan Hayward <alan.hayward@arm.com>
18518 David Sherwood <david.sherwood@arm.com>
18520 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
18522 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
18523 * optabs.c (expand_unop): Likewise.
18525 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18526 Alan Hayward <alan.hayward@arm.com>
18527 David Sherwood <david.sherwood@arm.com>
18529 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
18530 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
18531 (popcount_loc_descriptor): Likewise.
18532 (bswap_loc_descriptor): Likewise.
18533 (rotate_loc_descriptor): Likewise.
18534 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
18535 calling the functions above.
18537 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18538 Alan Hayward <alan.hayward@arm.com>
18539 David Sherwood <david.sherwood@arm.com>
18541 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
18543 (try_combine): Likewise.
18544 (simplify_if_then_else): Likewise.
18545 * cse.c (cse_insn): Likewise.
18546 * dwarf2out.c (mem_loc_descriptor): Likewise.
18547 * emit-rtl.c (gen_lowpart_common): Likewise.
18548 * simplify-rtx.c (simplify_truncation): Likewise.
18549 (simplify_binary_operation_1): Likewise.
18550 (simplify_const_relational_operation): Likewise.
18551 (simplify_ternary_operation): Likewise.
18552 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
18554 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18555 Alan Hayward <alan.hayward@arm.com>
18556 David Sherwood <david.sherwood@arm.com>
18558 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
18559 * reload.c (push_reload): Likewise.
18560 (find_reloads): Likewise.
18562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18563 Alan Hayward <alan.hayward@arm.com>
18564 David Sherwood <david.sherwood@arm.com>
18566 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
18567 (make_compound_operation_int): Likewise.
18568 (change_zero_ext): Likewise.
18569 * expr.c (convert_move): Likewise.
18570 (convert_modes): Likewise.
18571 * fwprop.c (forward_propagate_subreg): Likewise.
18572 * loop-iv.c (get_biv_step_1): Likewise.
18573 * optabs.c (widen_operand): Likewise.
18574 * postreload.c (move2add_valid_value_p): Likewise.
18575 * recog.c (simplify_while_replacing): Likewise.
18576 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18577 (simplify_binary_operation_1): Likewise. Remove redundant
18578 mode equality check.
18580 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18581 Alan Hayward <alan.hayward@arm.com>
18582 David Sherwood <david.sherwood@arm.com>
18584 * combine.c (combine_simplify_rtx): Add checks for
18585 is_a <scalar_int_mode>.
18586 (simplify_if_then_else): Likewise.
18587 (make_field_assignment): Likewise.
18588 (simplify_comparison): Likewise.
18589 * ifcvt.c (noce_try_bitop): Likewise.
18590 * loop-invariant.c (canonicalize_address_mult): Likewise.
18591 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18594 Alan Hayward <alan.hayward@arm.com>
18595 David Sherwood <david.sherwood@arm.com>
18597 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
18598 is_a <scalar_int_mode> instead of != BLKmode.
18600 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18601 Alan Hayward <alan.hayward@arm.com>
18602 David Sherwood <david.sherwood@arm.com>
18604 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
18605 instead of != VOIDmode.
18606 * combine.c (if_then_else_cond): Likewise.
18607 (change_zero_ext): Likewise.
18608 * dwarf2out.c (mem_loc_descriptor): Likewise.
18609 (loc_descriptor): Likewise.
18610 * rtlanal.c (canonicalize_condition): Likewise.
18611 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
18613 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18614 Alan Hayward <alan.hayward@arm.com>
18615 David Sherwood <david.sherwood@arm.com>
18617 * simplify-rtx.c (simplify_binary_operation_1): Use
18618 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18621 Alan Hayward <alan.hayward@arm.com>
18622 David Sherwood <david.sherwood@arm.com>
18624 * wide-int.h (int_traits<unsigned char>) New class.
18625 (int_traits<unsigned short>) Likewise.
18626 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18627 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18629 * combine.c (set_nonzero_bits_and_sign_copies): Use
18630 is_a <scalar_int_mode>.
18631 (find_split_point): Likewise.
18632 (combine_simplify_rtx): Likewise.
18633 (simplify_logical): Likewise.
18634 (expand_compound_operation): Likewise.
18635 (expand_field_assignment): Likewise.
18636 (make_compound_operation): Likewise.
18637 (extended_count): Likewise.
18638 (change_zero_ext): Likewise.
18639 (simplify_comparison): Likewise.
18640 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18641 (ucompare_loc_descriptor): Likewise.
18642 (minmax_loc_descriptor): Likewise.
18643 (mem_loc_descriptor): Likewise.
18644 (loc_descriptor): Likewise.
18645 * expmed.c (init_expmed_one_mode): Likewise.
18646 * lra-constraints.c (lra_constraint_offset): Likewise.
18647 * optabs.c (prepare_libcall_arg): Likewise.
18648 * postreload.c (move2add_note_store): Likewise.
18649 * reload.c (operands_match_p): Likewise.
18650 * rtl.h (load_extend_op): Likewise.
18651 * rtlhooks.c (gen_lowpart_general): Likewise.
18652 * simplify-rtx.c (simplify_truncation): Likewise.
18653 (simplify_unary_operation_1): Likewise.
18654 (simplify_binary_operation_1): Likewise.
18655 (simplify_const_binary_operation): Likewise.
18656 (simplify_const_relational_operation): Likewise.
18657 (simplify_subreg): Likewise.
18658 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18659 * var-tracking.c (adjust_mems): Likewise.
18660 (prepare_call_arguments): Likewise.
18662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18663 Alan Hayward <alan.hayward@arm.com>
18664 David Sherwood <david.sherwood@arm.com>
18666 * machmode.h (is_int_mode): New fuction.
18667 * combine.c (find_split_point): Use it.
18668 (combine_simplify_rtx): Likewise.
18669 (simplify_if_then_else): Likewise.
18670 (simplify_set): Likewise.
18671 (simplify_shift_const_1): Likewise.
18672 (simplify_comparison): Likewise.
18673 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18674 * cse.c (notreg_cost): Likewise.
18675 (cse_insn): Likewise.
18676 * cselib.c (cselib_lookup_1): Likewise.
18677 * dojump.c (do_jump_1): Likewise.
18678 (do_compare_rtx_and_jump): Likewise.
18679 * dse.c (get_call_args): Likewise.
18680 * dwarf2out.c (rtl_for_decl_init): Likewise.
18681 (native_encode_initializer): Likewise.
18682 * expmed.c (emit_store_flag_1): Likewise.
18683 (emit_store_flag): Likewise.
18684 * expr.c (convert_modes): Likewise.
18685 (store_field): Likewise.
18686 (expand_expr_real_1): Likewise.
18687 * fold-const.c (fold_read_from_constant_string): Likewise.
18688 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18689 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18690 * optabs.c (expand_binop): Likewise.
18691 (expand_unop): Likewise.
18692 (expand_abs_nojump): Likewise.
18693 (expand_one_cmpl_abs_nojump): Likewise.
18694 * simplify-rtx.c (mode_signbit_p): Likewise.
18695 (val_signbit_p): Likewise.
18696 (val_signbit_known_set_p): Likewise.
18697 (val_signbit_known_clear_p): Likewise.
18698 (simplify_relational_operation_1): Likewise.
18699 * tree.c (vector_type_mode): Likewise.
18701 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18702 Alan Hayward <alan.hayward@arm.com>
18703 David Sherwood <david.sherwood@arm.com>
18705 * machmode.h (smallest_mode_for_size): Fix formatting.
18706 (smallest_int_mode_for_size): New function.
18707 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18708 instead of smallest_mode_for_size.
18709 * combine.c (make_extraction): Likewise.
18710 * config/arc/arc.c (arc_expand_movmem): Likewise.
18711 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18712 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18713 * config/s390/s390.c (s390_expand_insv): Likewise.
18714 * config/sparc/sparc.c (assign_int_registers): Likewise.
18715 * config/spu/spu.c (spu_function_value): Likewise.
18716 (spu_function_arg): Likewise.
18717 * coverage.c (get_gcov_type): Likewise.
18718 (get_gcov_unsigned_t): Likewise.
18719 * dse.c (find_shift_sequence): Likewise.
18720 * expmed.c (store_bit_field_1): Likewise.
18721 * expr.c (convert_move): Likewise.
18722 (store_field): Likewise.
18723 * internal-fn.c (expand_arith_overflow): Likewise.
18724 * optabs-query.c (get_best_extraction_insn): Likewise.
18725 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18726 * stor-layout.c (layout_type): Likewise.
18727 (initialize_sizetypes): Likewise.
18728 * targhooks.c (default_get_mask_mode): Likewise.
18729 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18731 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18732 Alan Hayward <alan.hayward@arm.com>
18733 David Sherwood <david.sherwood@arm.com>
18735 * machmode.h (opt_mode::else_blk): New function.
18736 (int_mode_for_mode): Declare.
18737 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18738 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18740 * cfgexpand.c (expand_debug_expr): Likewise.
18741 * combine.c (gen_lowpart_or_truncate): Likewise.
18742 (gen_lowpart_for_combine): Likewise.
18743 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18744 * config/avr/avr.c (avr_to_int_mode): Likewise.
18745 (avr_out_plus_1): Likewise.
18746 (avr_out_plus): Likewise.
18747 (avr_out_round): Likewise.
18748 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18749 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18750 (s390_expand_vcond): Likewise.
18751 * config/spu/spu.c (spu_split_immediate): Likewise.
18752 (spu_expand_mov): Likewise.
18753 * dse.c (get_stored_val): Likewise.
18754 * expmed.c (store_bit_field_1): Likewise.
18755 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18757 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18758 (extract_low_bits): Likewise.
18759 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18760 handling rather than repeating the check.
18761 (emit_group_store): Likewise.
18762 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18763 * optabs.c (expand_absneg_bit): Likewise.
18764 (expand_copysign_absneg): Likewise.
18765 (expand_copysign_bit): Likewise.
18766 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18767 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18768 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18769 * var-tracking.c (prepare_call_arguments): Likewise.
18770 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18771 int_mode_for_mode instead of mode_for_size.
18772 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18774 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18775 Alan Hayward <alan.hayward@arm.com>
18776 David Sherwood <david.sherwood@arm.com>
18778 * machmode.h (int_mode_for_size): New function.
18779 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18780 instead of mode_for_size.
18781 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18783 * combine.c (expand_field_assignment): Use int_mode_for_size
18784 instead of mode_for_size.
18785 (make_extraction): Likewise.
18786 (simplify_shift_const_1): Likewise.
18787 (simplify_comparison): Likewise.
18788 * dojump.c (do_jump): Likewise.
18789 * dwarf2out.c (mem_loc_descriptor): Likewise.
18790 * emit-rtl.c (init_derived_machine_modes): Likewise.
18791 * expmed.c (flip_storage_order): Likewise.
18792 (convert_extracted_bit_field): Likewise.
18793 * expr.c (copy_blkmode_from_reg): Likewise.
18794 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18795 * internal-fn.c (expand_mul_overflow): Likewise.
18796 * lower-subreg.c (simple_move): Likewise.
18797 * optabs-libfuncs.c (init_optabs): Likewise.
18798 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18799 * tree.c (vector_type_mode): Likewise.
18800 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18801 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18802 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18803 * tree-vect-stmts.c (vectorizable_load): Likewise.
18804 (vectorizable_store): Likewise.
18806 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18807 Alan Hayward <alan.hayward@arm.com>
18808 David Sherwood <david.sherwood@arm.com>
18810 * coretypes.h (pod_mode): New type.
18811 (scalar_int_mode_pod): New typedef.
18812 * machmode.h (pod_mode): New class.
18813 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18814 * genmodes.c (emit_mode_int_n): Update accordingly.
18815 * lower-subreg.h (target_lower_subreg): Change type to
18816 scalar_int_mode_pod.
18817 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18818 scalar_int_mode_pod.
18820 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18821 Alan Hayward <alan.hayward@arm.com>
18822 David Sherwood <david.sherwood@arm.com>
18824 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18825 machine_mode to scalar_int_mode.
18826 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18827 (rs6000_option_override_internal): Remove cast to int.
18828 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18829 machine_mode to scalar_int_mode.
18830 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18831 (rs6000_option_override_internal): Remove cast to int.
18832 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18833 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18835 * config/s390/s390.c (s390_expand_builtin): Likewise.
18836 * coretypes.h (scalar_int_mode): New type.
18837 (opt_scalar_int_mode): New typedef.
18838 * machmode.h (scalar_int_mode): New class.
18839 (scalar_int_mode::includes_p): New function.
18840 (byte_mode): Change type to scalar_int_mode.
18841 (word_mode): Likewise.
18842 (ptr_mode): Likewise.
18843 * emit-rtl.c (byte_mode): Likewise.
18844 (word_mode): Likewise.
18845 (ptr_mode): Likewise.
18846 (init_derived_machine_modes): Update accordingly.
18847 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18848 and MODE_PARTIAL_INT.
18849 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18850 opt_scalar_int_mode.
18852 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18853 Alan Hayward <alan.hayward@arm.com>
18854 David Sherwood <david.sherwood@arm.com>
18856 * target.def (libgcc_floating_mode_supported_p): Take a
18858 * doc/tm.texi: Regenerate.
18859 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18861 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18862 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
18865 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18866 Alan Hayward <alan.hayward@arm.com>
18867 David Sherwood <david.sherwood@arm.com>
18869 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
18870 * doc/tm.texi: Regenerate.
18871 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
18872 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
18873 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
18874 * targhooks.h (default_floatn_mode): Likewise.
18875 * targhooks.c (default_floatn_mode): Likewise.
18876 * tree.c (build_common_tree_nodes): Update accordingly.
18878 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18879 Alan Hayward <alan.hayward@arm.com>
18880 David Sherwood <david.sherwood@arm.com>
18882 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
18883 (mode_iterator::iterate_p): Likewise.
18884 (mode_iterator::get_wider): Likewise.
18885 * expr.c (init_expr_target): Use opt_scalar_float_mode.
18887 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18888 Alan Hayward <alan.hayward@arm.com>
18889 David Sherwood <david.sherwood@arm.com>
18891 * coretypes.h (opt_scalar_float_mode): New typedef.
18892 * machmode.h (float_mode_for_size): New function.
18893 * emit-rtl.c (double_mode): Delete.
18894 (init_emit_once): Use float_mode_for_size.
18895 * stor-layout.c (layout_type): Likewise.
18896 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
18898 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18899 Alan Hayward <alan.hayward@arm.com>
18900 David Sherwood <david.sherwood@arm.com>
18902 * output.h (assemble_real): Take a scalar_float_mode.
18903 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
18904 * config/arm/arm.md (consttable_4): Likewise.
18905 (consttable_8): Likewise.
18906 (consttable_16): Likewise.
18907 * config/mips/mips.md (consttable_float): Likewise.
18908 * config/s390/s390.c (s390_output_pool_entry): Likewise.
18909 * varasm.c (assemble_real): Take a scalar_float_mode.
18910 (output_constant_pool_2): Update accordingly.
18911 (output_constant): Likewise.
18913 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18914 Alan Hayward <alan.hayward@arm.com>
18915 David Sherwood <david.sherwood@arm.com>
18917 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
18918 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
18919 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
18920 (native_encode_real): Likewise.
18921 (native_interpret_real): Likewise.
18922 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
18923 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18926 Alan Hayward <alan.hayward@arm.com>
18927 David Sherwood <david.sherwood@arm.com>
18929 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
18930 <scalar_float_mode>. Simplify.
18931 (gen_extend_conv_libfunc): Likewise.
18933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18934 Alan Hayward <alan.hayward@arm.com>
18935 David Sherwood <david.sherwood@arm.com>
18937 * coretypes.h (scalar_float_mode): New type.
18938 * machmode.h (mode_traits::from_int): Use machine_mode if
18939 USE_ENUM_MODES is defined.
18940 (is_a): New function.
18942 (dyn_cast): Likewise.
18943 (scalar_float_mode): New class.
18944 (scalar_float_mode::includes_p): New function.
18945 (is_float_mode): Likewise.
18946 * gdbhooks.py (MachineModePrinter): New class.
18947 (build_pretty_printer): Use it for scalar_float_mode.
18948 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
18949 (format_helper::format_helper): Turn into a template.
18950 * genmodes.c (get_mode_class): New function.
18951 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
18952 or machine_mode if none.
18953 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
18954 as_a <scalar_float_mode>.
18955 * dwarf2out.c (mem_loc_descriptor): Likewise.
18956 (insert_float): Likewise.
18957 (add_const_value_attribute): Likewise.
18958 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18959 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
18960 (expand_unop): Update accordingly.
18961 (expand_abs_nojump): Likewise.
18962 (expand_copysign_absneg): Take a scalar_float_mode.
18963 (expand_copysign_bit): Likewise.
18964 (expand_copysign): Update accordingly.
18966 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18967 Alan Hayward <alan.hayward@arm.com>
18968 David Sherwood <david.sherwood@arm.com>
18970 * coretypes.h (opt_mode): New class.
18971 * machmode.h (opt_mode): Likewise.
18972 (opt_mode::else_void): New function.
18973 (opt_mode::require): Likewise.
18974 (opt_mode::exists): Likewise.
18975 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
18976 (GET_MODE_2XWIDER_MODE): Likewise.
18977 (mode_iterator::get_wider): Update accordingly.
18978 (mode_iterator::get_2xwider): Likewise.
18979 (mode_iterator::get_known_wider): Likewise, turning into a template.
18980 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
18981 forcing a wider mode to exist.
18982 * config/cr16/cr16.h (LONG_REG_P): Likewise.
18983 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18984 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
18985 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
18986 * lower-subreg.c (init_lower_subreg): Likewise.
18987 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
18988 on the final iteration.
18989 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
18990 a wider mode exists before asking for a move pattern.
18991 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
18992 forcing a wider mode to exist.
18993 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
18994 returning false if no such mode exists.
18995 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
18996 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
18997 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
18998 Avoid checking for a MODE_INT if we already know the mode is not a
19000 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19001 forcing a wider mode to exist.
19002 (expmed_mult_highpart_optab): Likewise.
19003 (expmed_mult_highpart): Likewise.
19004 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19006 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19007 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19008 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19009 * internal-fn.c (expand_mul_overflow): Update use of
19010 GET_MODE_2XWIDER_MODE.
19011 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19012 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19013 GET_MODE_WIDER_MODE.
19014 (convert_plusminus_to_widen): Likewise.
19015 * tree-switch-conversion.c (array_value_type): Likewise.
19016 * var-tracking.c (emit_note_insn_var_location): Likewise.
19017 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19018 Return false inside rather than outside the loop if no wider mode
19020 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19021 and GET_MODE_2XWIDER_MODE
19022 (can_compare_p): Use else_void.
19023 * gdbhooks.py (OptMachineModePrinter): New class.
19024 (build_pretty_printer): Use it for opt_mode.
19026 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19027 Alan Hayward <alan.hayward@arm.com>
19028 David Sherwood <david.sherwood@arm.com>
19030 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19031 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19033 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19034 Alan Hayward <alan.hayward@arm.com>
19035 David Sherwood <david.sherwood@arm.com>
19037 * machmode.h (mode_traits): New structure.
19038 (get_narrowest_mode): New function.
19039 (mode_iterator::start): Likewise.
19040 (mode_iterator::iterate_p): Likewise.
19041 (mode_iterator::get_wider): Likewise.
19042 (mode_iterator::get_known_wider): Likewise.
19043 (mode_iterator::get_2xwider): Likewise.
19044 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19045 (FOR_EACH_MODE): Likewise.
19046 (FOR_EACH_MODE_FROM): Likewise.
19047 (FOR_EACH_MODE_UNTIL): Likewise.
19048 (FOR_EACH_WIDER_MODE): Likewise.
19049 (FOR_EACH_2XWIDER_MODE): Likewise.
19050 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19051 * combine.c (simplify_comparison): Likewise
19052 * config/i386/i386.c (type_natural_mode): Likewise.
19053 * cse.c (cse_insn): Likewise.
19054 * dse.c (find_shift_sequence): Likewise.
19055 * emit-rtl.c (init_derived_machine_modes): Likewise.
19056 (init_emit_once): Likewise.
19057 * explow.c (hard_function_value): Likewise.
19058 * expmed.c (extract_fixed_bit_field_1): Likewise.
19059 (extract_bit_field_1): Likewise.
19060 (expand_divmod): Likewise.
19061 (emit_store_flag_1): Likewise.
19062 * expr.c (init_expr_target): Likewise.
19063 (convert_move): Likewise.
19064 (alignment_for_piecewise_move): Likewise.
19065 (widest_int_mode_for_size): Likewise.
19066 (emit_block_move_via_movmem): Likewise.
19067 (copy_blkmode_to_reg): Likewise.
19068 (set_storage_via_setmem): Likewise.
19069 (compress_float_constant): Likewise.
19070 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19071 * optabs-query.c (get_best_extraction_insn): Likewise.
19072 * optabs.c (expand_binop): Likewise.
19073 (expand_twoval_unop): Likewise.
19074 (expand_twoval_binop): Likewise.
19075 (widen_leading): Likewise.
19076 (widen_bswap): Likewise.
19077 (expand_parity): Likewise.
19078 (expand_unop): Likewise.
19079 (prepare_cmp_insn): Likewise.
19080 (prepare_float_lib_cmp): Likewise.
19081 (expand_float): Likewise.
19082 (expand_fix): Likewise.
19083 (expand_sfix_optab): Likewise.
19084 * postreload.c (move2add_use_add2_insn): Likewise.
19085 * reg-stack.c (reg_to_stack): Likewise.
19086 * reginfo.c (choose_hard_reg_mode): Likewise.
19087 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19088 * stor-layout.c (mode_for_size): Likewise.
19089 (smallest_mode_for_size): Likewise.
19090 (mode_for_vector): Likewise.
19091 (finish_bitfield_representative): Likewise.
19092 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19093 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19094 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19095 * var-tracking.c (prepare_call_arguments): Likewise.
19097 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19098 Alan Hayward <alan.hayward@arm.com>
19099 David Sherwood <david.sherwood@arm.com>
19101 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19102 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19103 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19104 * machmode.h (mode_size): Move earlier in file.
19105 (mode_precision): Likewise.
19106 (mode_inner): Likewise.
19107 (mode_nunits): Likewise.
19108 (mode_unit_size): Likewise.
19109 (unit_unit_precision): Likewise.
19110 (mode_wider): Likewise.
19111 (mode_2xwider): Likewise.
19112 (machine_mode): New class.
19113 (mode_to_bytes): New function.
19114 (mode_to_bits): Likewise.
19115 (mode_to_precision): Likewise.
19116 (mode_to_inner): Likewise.
19117 (mode_to_unit_size): Likewise.
19118 (mode_to_unit_precision): Likewise.
19119 (mode_to_nunits): Likewise.
19120 (GET_MODE_SIZE): Use mode_to_bytes.
19121 (GET_MODE_BITSIZE): Use mode_to_bits.
19122 (GET_MODE_PRECISION): Use mode_to_precision.
19123 (GET_MODE_INNER): Use mode_to_inner.
19124 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
19125 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
19126 (GET_MODE_NUNITS): Use mode_to_nunits.
19127 * system.h (ALWAYS_INLINE): New macro.
19128 * config/powerpcspe/powerpcspe-c.c
19129 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
19130 int for arg1_mode and arg2_mode.
19132 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19133 Alan Hayward <alan.hayward@arm.com>
19134 David Sherwood <david.sherwood@arm.com>
19136 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19137 Prefix mode names with E_ in case statements.
19138 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19139 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
19140 (aarch64_split_simd_move): Likewise.
19141 (aarch64_gen_storewb_pair): Likewise.
19142 (aarch64_gen_loadwb_pair): Likewise.
19143 (aarch64_gen_store_pair): Likewise.
19144 (aarch64_gen_load_pair): Likewise.
19145 (aarch64_get_condition_code_1): Likewise.
19146 (aarch64_constant_pool_reload_icode): Likewise.
19147 (get_rsqrte_type): Likewise.
19148 (get_rsqrts_type): Likewise.
19149 (get_recpe_type): Likewise.
19150 (get_recps_type): Likewise.
19151 (aarch64_gimplify_va_arg_expr): Likewise.
19152 (aarch64_simd_container_mode): Likewise.
19153 (aarch64_emit_load_exclusive): Likewise.
19154 (aarch64_emit_store_exclusive): Likewise.
19155 (aarch64_expand_compare_and_swap): Likewise.
19156 (aarch64_gen_atomic_cas): Likewise.
19157 (aarch64_emit_bic): Likewise.
19158 (aarch64_emit_atomic_swap): Likewise.
19159 (aarch64_emit_atomic_load_op): Likewise.
19160 (aarch64_evpc_trn): Likewise.
19161 (aarch64_evpc_uzp): Likewise.
19162 (aarch64_evpc_zip): Likewise.
19163 (aarch64_evpc_ext): Likewise.
19164 (aarch64_evpc_rev): Likewise.
19165 (aarch64_evpc_dup): Likewise.
19166 (aarch64_gen_ccmp_first): Likewise.
19167 (aarch64_gen_ccmp_next): Likewise.
19168 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
19169 (alpha_emit_xfloating_libcall): Likewise.
19170 (emit_insxl): Likewise.
19171 (alpha_arg_type): Likewise.
19172 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
19173 (arc_preferred_simd_mode): Likewise.
19174 (arc_secondary_reload): Likewise.
19175 (get_arc_condition_code): Likewise.
19176 (arc_print_operand): Likewise.
19177 (arc_legitimate_constant_p): Likewise.
19178 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19179 * config/arc/arc.md (casesi_load): Likewise.
19180 (casesi_compact_jump): Likewise.
19181 * config/arc/predicates.md (proper_comparison_operator): Likewise.
19182 (cc_use_register): Likewise.
19183 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19184 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19185 (arm_init_iwmmxt_builtins): Likewise.
19186 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
19187 (neon_expand_vector_init): Likewise.
19188 (arm_attr_length_move_neon): Likewise.
19189 (maybe_get_arm_condition_code): Likewise.
19190 (arm_emit_vector_const): Likewise.
19191 (arm_preferred_simd_mode): Likewise.
19192 (arm_output_iwmmxt_tinsr): Likewise.
19193 (thumb1_output_casesi): Likewise.
19194 (thumb2_output_casesi): Likewise.
19195 (arm_emit_load_exclusive): Likewise.
19196 (arm_emit_store_exclusive): Likewise.
19197 (arm_expand_compare_and_swap): Likewise.
19198 (arm_evpc_neon_vuzp): Likewise.
19199 (arm_evpc_neon_vzip): Likewise.
19200 (arm_evpc_neon_vrev): Likewise.
19201 (arm_evpc_neon_vtrn): Likewise.
19202 (arm_evpc_neon_vext): Likewise.
19203 (arm_validize_comparison): Likewise.
19204 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
19205 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
19206 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
19207 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
19208 (c6x_preferred_simd_mode): Likewise.
19209 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
19210 (epiphany_rtx_costs): Likewise.
19211 * config/epiphany/predicates.md (proper_comparison_operator):
19213 * config/frv/frv.c (condexec_memory_operand): Likewise.
19214 (frv_emit_move): Likewise.
19215 (output_move_single): Likewise.
19216 (output_condmove_single): Likewise.
19217 (frv_hard_regno_mode_ok): Likewise.
19218 (frv_matching_accg_mode): Likewise.
19219 * config/h8300/h8300.c (split_adds_subs): Likewise.
19220 (h8300_rtx_costs): Likewise.
19221 (h8300_print_operand): Likewise.
19222 (compute_mov_length): Likewise.
19223 (output_logical_op): Likewise.
19224 (compute_logical_op_length): Likewise.
19225 (compute_logical_op_cc): Likewise.
19226 (h8300_shift_needs_scratch_p): Likewise.
19227 (output_a_shift): Likewise.
19228 (compute_a_shift_length): Likewise.
19229 (compute_a_shift_cc): Likewise.
19230 (expand_a_rotate): Likewise.
19231 (output_a_rotate): Likewise.
19232 * config/i386/i386.c (classify_argument): Likewise.
19233 (function_arg_advance_32): Likewise.
19234 (function_arg_32): Likewise.
19235 (function_arg_64): Likewise.
19236 (function_value_64): Likewise.
19237 (ix86_gimplify_va_arg): Likewise.
19238 (ix86_legitimate_constant_p): Likewise.
19239 (put_condition_code): Likewise.
19240 (split_double_mode): Likewise.
19241 (ix86_avx256_split_vector_move_misalign): Likewise.
19242 (ix86_expand_vector_logical_operator): Likewise.
19243 (ix86_split_idivmod): Likewise.
19244 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
19245 (ix86_build_const_vector): Likewise.
19246 (ix86_build_signbit_mask): Likewise.
19247 (ix86_match_ccmode): Likewise.
19248 (ix86_cc_modes_compatible): Likewise.
19249 (ix86_expand_branch): Likewise.
19250 (ix86_expand_sse_cmp): Likewise.
19251 (ix86_expand_sse_movcc): Likewise.
19252 (ix86_expand_int_sse_cmp): Likewise.
19253 (ix86_expand_vec_perm_vpermi2): Likewise.
19254 (ix86_expand_vec_perm): Likewise.
19255 (ix86_expand_sse_unpack): Likewise.
19256 (ix86_expand_int_addcc): Likewise.
19257 (ix86_split_to_parts): Likewise.
19258 (ix86_vectorize_builtin_gather): Likewise.
19259 (ix86_vectorize_builtin_scatter): Likewise.
19260 (avx_vpermilp_parallel): Likewise.
19261 (inline_memory_move_cost): Likewise.
19262 (ix86_tieable_integer_mode_p): Likewise.
19263 (x86_maybe_negate_const_int): Likewise.
19264 (ix86_expand_vector_init_duplicate): Likewise.
19265 (ix86_expand_vector_init_one_nonzero): Likewise.
19266 (ix86_expand_vector_init_one_var): Likewise.
19267 (ix86_expand_vector_init_concat): Likewise.
19268 (ix86_expand_vector_init_interleave): Likewise.
19269 (ix86_expand_vector_init_general): Likewise.
19270 (ix86_expand_vector_set): Likewise.
19271 (ix86_expand_vector_extract): Likewise.
19272 (emit_reduc_half): Likewise.
19273 (ix86_emit_i387_round): Likewise.
19274 (ix86_mangle_type): Likewise.
19275 (ix86_expand_round_sse4): Likewise.
19276 (expand_vec_perm_blend): Likewise.
19277 (canonicalize_vector_int_perm): Likewise.
19278 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
19279 (expand_vec_perm_1): Likewise.
19280 (expand_vec_perm_interleave3): Likewise.
19281 (expand_vec_perm_even_odd_pack): Likewise.
19282 (expand_vec_perm_even_odd_1): Likewise.
19283 (expand_vec_perm_broadcast_1): Likewise.
19284 (ix86_vectorize_vec_perm_const_ok): Likewise.
19285 (ix86_expand_vecop_qihi): Likewise.
19286 (ix86_expand_mul_widen_hilo): Likewise.
19287 (ix86_expand_sse2_abs): Likewise.
19288 (ix86_expand_pextr): Likewise.
19289 (ix86_expand_pinsr): Likewise.
19290 (ix86_preferred_simd_mode): Likewise.
19291 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
19292 * config/i386/sse.md (*andnot<mode>3): Likewise.
19293 (<mask_codefor><code><mode>3<mask_name>): Likewise.
19294 (*<code><mode>3): Likewise.
19295 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
19296 (ia64_expand_atomic_op): Likewise.
19297 (ia64_arg_type): Likewise.
19298 (ia64_mode_to_int): Likewise.
19299 (ia64_scalar_mode_supported_p): Likewise.
19300 (ia64_vector_mode_supported_p): Likewise.
19301 (expand_vec_perm_broadcast): Likewise.
19302 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
19303 (iq2000_function_arg_advance): Likewise.
19304 (iq2000_function_arg): Likewise.
19305 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
19306 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
19307 (m68k_libcall_value): Likewise.
19308 (m68k_function_value): Likewise.
19309 (sched_attr_op_type): Likewise.
19310 * config/mcore/mcore.c (mcore_output_move): Likewise.
19311 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
19313 (microblaze_function_arg): Likewise.
19314 * config/mips/mips.c (mips16_build_call_stub): Likewise.
19315 (mips_print_operand): Likewise.
19316 (mips_mode_ok_for_mov_fmt_p): Likewise.
19317 (mips_vector_mode_supported_p): Likewise.
19318 (mips_preferred_simd_mode): Likewise.
19319 (mips_expand_vpc_loongson_even_odd): Likewise.
19320 (mips_expand_vec_unpack): Likewise.
19321 (mips_expand_vi_broadcast): Likewise.
19322 (mips_expand_vector_init): Likewise.
19323 (mips_expand_vec_reduc): Likewise.
19324 (mips_expand_msa_cmp): Likewise.
19325 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
19326 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
19327 (cc_flags_for_mode): Likewise.
19328 * config/msp430/msp430.c (msp430_print_operand): Likewise.
19329 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
19330 (nds32_output_casesi_pc_relative): Likewise.
19331 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19332 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
19333 (nvptx_gen_unpack): Likewise.
19334 (nvptx_gen_pack): Likewise.
19335 (nvptx_gen_shuffle): Likewise.
19336 (nvptx_gen_wcast): Likewise.
19337 (nvptx_preferred_simd_mode): Likewise.
19338 * config/pa/pa.c (pa_secondary_reload): Likewise.
19339 * config/pa/predicates.md (base14_operand): Likewise.
19340 * config/powerpcspe/powerpcspe-c.c
19341 (altivec_resolve_overloaded_builtin): Likewise.
19342 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
19344 (rs6000_preferred_simd_mode): Likewise.
19345 (output_vec_const_move): Likewise.
19346 (rs6000_expand_vector_extract): Likewise.
19347 (rs6000_split_vec_extract_var): Likewise.
19348 (reg_offset_addressing_ok_p): Likewise.
19349 (rs6000_legitimate_offset_address_p): Likewise.
19350 (rs6000_legitimize_address): Likewise.
19351 (rs6000_emit_set_const): Likewise.
19352 (rs6000_const_vec): Likewise.
19353 (rs6000_emit_move): Likewise.
19354 (spe_build_register_parallel): Likewise.
19355 (rs6000_darwin64_record_arg_recurse): Likewise.
19356 (swap_selector_for_mode): Likewise.
19357 (spe_init_builtins): Likewise.
19358 (paired_init_builtins): Likewise.
19359 (altivec_init_builtins): Likewise.
19360 (do_load_for_compare): Likewise.
19361 (rs6000_generate_compare): Likewise.
19362 (rs6000_expand_float128_convert): Likewise.
19363 (emit_load_locked): Likewise.
19364 (emit_store_conditional): Likewise.
19365 (rs6000_output_function_epilogue): Likewise.
19366 (rs6000_handle_altivec_attribute): Likewise.
19367 (rs6000_function_value): Likewise.
19368 (emit_fusion_gpr_load): Likewise.
19369 (emit_fusion_p9_load): Likewise.
19370 (emit_fusion_p9_store): Likewise.
19371 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
19372 (fusion_gpr_mem_load): Likewise.
19373 (fusion_addis_mem_combo_load): Likewise.
19374 (fusion_addis_mem_combo_store): Likewise.
19375 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
19376 (fusion_gpr_mem_load): Likewise.
19377 (fusion_addis_mem_combo_load): Likewise.
19378 (fusion_addis_mem_combo_store): Likewise.
19379 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19381 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
19382 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19383 (rs6000_preferred_simd_mode): Likewise.
19384 (output_vec_const_move): Likewise.
19385 (rs6000_expand_vector_extract): Likewise.
19386 (rs6000_split_vec_extract_var): Likewise.
19387 (reg_offset_addressing_ok_p): Likewise.
19388 (rs6000_legitimate_offset_address_p): Likewise.
19389 (rs6000_legitimize_address): Likewise.
19390 (rs6000_emit_set_const): Likewise.
19391 (rs6000_const_vec): Likewise.
19392 (rs6000_emit_move): Likewise.
19393 (rs6000_darwin64_record_arg_recurse): Likewise.
19394 (swap_selector_for_mode): Likewise.
19395 (paired_init_builtins): Likewise.
19396 (altivec_init_builtins): Likewise.
19397 (rs6000_expand_float128_convert): Likewise.
19398 (emit_load_locked): Likewise.
19399 (emit_store_conditional): Likewise.
19400 (rs6000_output_function_epilogue): Likewise.
19401 (rs6000_handle_altivec_attribute): Likewise.
19402 (rs6000_function_value): Likewise.
19403 (emit_fusion_gpr_load): Likewise.
19404 (emit_fusion_p9_load): Likewise.
19405 (emit_fusion_p9_store): Likewise.
19406 * config/rx/rx.c (rx_gen_move_template): Likewise.
19407 (flags_from_mode): Likewise.
19408 * config/s390/predicates.md (s390_alc_comparison): Likewise.
19409 (s390_slb_comparison): Likewise.
19410 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
19411 (s390_vector_mode_supported_p): Likewise.
19412 (s390_cc_modes_compatible): Likewise.
19413 (s390_match_ccmode_set): Likewise.
19414 (s390_canonicalize_comparison): Likewise.
19415 (s390_emit_compare_and_swap): Likewise.
19416 (s390_branch_condition_mask): Likewise.
19417 (s390_rtx_costs): Likewise.
19418 (s390_secondary_reload): Likewise.
19419 (__SECONDARY_RELOAD_CASE): Likewise.
19420 (s390_expand_cs): Likewise.
19421 (s390_preferred_simd_mode): Likewise.
19422 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
19423 * config/sh/sh.c (sh_print_operand): Likewise.
19424 (dump_table): Likewise.
19425 (sh_secondary_reload): Likewise.
19426 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19427 * config/sh/sh.md (casesi_worker_1): Likewise.
19428 (casesi_worker_2): Likewise.
19429 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
19430 (fcc_comparison_operator): Likewise.
19431 * config/sparc/sparc.c (sparc_expand_move): Likewise.
19432 (emit_soft_tfmode_cvt): Likewise.
19433 (sparc_preferred_simd_mode): Likewise.
19434 (output_cbranch): Likewise.
19435 (sparc_print_operand): Likewise.
19436 (sparc_expand_vec_perm_bmask): Likewise.
19437 (vector_init_bshuffle): Likewise.
19438 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
19439 (spu_vector_mode_supported_p): Likewise.
19440 (spu_expand_insv): Likewise.
19441 (spu_emit_branch_or_set): Likewise.
19442 (spu_handle_vector_attribute): Likewise.
19443 (spu_builtin_splats): Likewise.
19444 (spu_builtin_extract): Likewise.
19445 (spu_builtin_promote): Likewise.
19446 (spu_expand_sign_extend): Likewise.
19447 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
19448 (tilegx_simd_int): Likewise.
19449 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
19450 (tilepro_simd_int): Likewise.
19451 * config/v850/v850.c (const_double_split): Likewise.
19452 (v850_print_operand): Likewise.
19453 (ep_memory_offset): Likewise.
19454 * config/vax/vax.c (vax_rtx_costs): Likewise.
19455 (vax_output_int_move): Likewise.
19456 (vax_output_int_add): Likewise.
19457 (vax_output_int_subtract): Likewise.
19458 * config/visium/predicates.md (visium_branch_operator): Likewise.
19459 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
19460 (visium_print_operand_address): Likewise.
19461 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19462 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
19463 (xtensa_expand_conditional_branch): Likewise.
19464 (xtensa_copy_incoming_a7): Likewise.
19465 (xtensa_output_literal): Likewise.
19466 * dfp.c (decimal_real_maxval): Likewise.
19467 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19469 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19470 Alan Hayward <alan.hayward@arm.com>
19471 David Sherwood <david.sherwood@arm.com>
19473 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
19474 (mode_nunits_inline): Likewise.
19475 (mode_inner_inline): Likewise.
19476 (mode_unit_size_inline): Likewise.
19477 (mode_unit_precision_inline): Likewise.
19478 (emit_insn_modes_h): Likewise. Also emit a #define of the
19480 (emit_mode_wider): Add an E_ prefix to mode names.
19481 (emit_mode_complex): Likewise.
19482 (emit_mode_inner): Likewise.
19483 (emit_mode_adjustments): Likewise.
19484 (emit_mode_int_n): Likewise.
19485 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
19486 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
19487 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
19488 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
19489 (CRC32_BUILTIN, ENTRY): Likewise.
19490 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
19491 (aarch64_pop_regs): Likewise.
19492 (aarch64_process_components): Likewise.
19493 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
19494 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
19495 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
19496 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
19497 * config/arm/arm.c (arm_init_libfuncs): Likewise.
19498 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
19500 * config/i386/i386-builtin.def (pcmpestr): Likewise.
19501 (pcmpistr): Likewise.
19502 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
19503 * config/mmix/mmix.c (mmix_output_condition): Likewise.
19504 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19506 * config/rl78/rl78.c (mduc_regs): Likewise.
19507 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19508 (htm_expand_builtin): Likewise.
19509 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
19510 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
19511 * config/xtensa/xtensa.c (print_operand): Likewise.
19512 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
19513 (NUM_MODE_VECTOR_INT): Likewise.
19514 * genoutput.c (null_operand): Likewise.
19515 (output_operand_data): Likewise.
19516 * genrecog.c (print_parameter_value): Likewise.
19517 * lra.c (debug_operand_data): Likewise.
19519 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19521 * dwarf2out.c (scompare_loc_descriptor_wide)
19522 (scompare_loc_descriptor_narrow): New functions, split out from...
19523 (scompare_loc_descriptor): ...here.
19524 * expmed.c (emit_store_flag_int): New function, split out from...
19525 (emit_store_flag): ...here.
19527 2017-08-30 Richard Biener <rguenther@suse.de>
19529 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
19530 (dwarf2out_early_finish): Move setting of AT_pubnames from
19531 early debug output to early finish.
19533 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19535 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
19536 and -mdata-region to the assembler.
19538 2017-08-30 Richard Biener <rguenther@suse.de>
19540 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
19542 (gen_subprogram_die): Add DW_AT_object_pointer only early.
19543 (dwarf2out_early_global_decl): Only generate a DIE for the
19544 abstract origin if it doesn't already exist or is a declaration DIE.
19545 (resolve_addr): Do not add the linkage name twice when
19546 generating a stub DIE for the DW_TAG_GNU_call_site target.
19548 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19550 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19551 Use machine_mode rather than int for arg1_mode.
19553 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19556 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
19557 that the second argument of the built-in functions to unpack
19558 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
19559 switch statement instead a lot of if statements.
19560 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
19561 Allow 64-bit values to be in Altivec registers as well as
19562 traditional floating point registers.
19563 (pack<mode>, FMOVE128_VSX iterator): Likewise.
19565 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
19567 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
19568 MAX_REGS_PER_ADDRESS == 1.
19570 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
19572 * config/i386/i386.opt (flag_fentry): Do not init to -1.
19573 * config/i386/i386.c (ix86_option_override_internal): Simplify
19574 setting of opts->x_flag_entry.
19576 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19577 Jakub Jelinek <jakub@redhat.com>
19578 Richard Biener <rguenther@suse.de>
19580 PR tree-optimization/81503
19581 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
19582 folded constant fits in the target type; reorder tests for clarity.
19584 2017-08-29 Martin Liska <mliska@suse.cz>
19586 * passes.def: Include pass_lower_switch.
19587 * stmt.c (dump_case_nodes): Remove and move to
19588 tree-switch-conversion.
19589 (case_values_threshold): Likewise.
19590 (expand_switch_as_decision_tree_p): Likewise.
19591 (emit_case_decision_tree): Likewise.
19592 (expand_case): Likewise.
19593 (balance_case_nodes): Likewise.
19594 (node_has_low_bound): Likewise.
19595 (node_has_high_bound): Likewise.
19596 (node_is_bounded): Likewise.
19597 (emit_case_nodes): Likewise.
19598 (struct simple_case_node): New struct.
19599 (add_case_node): Remove.
19600 (emit_case_dispatch_table): Use vector instead of case_list.
19601 (reset_out_edges_aux): Remove.
19602 (compute_cases_per_edge): Likewise.
19603 (expand_case): Build list of simple_case_node.
19604 (expand_sjlj_dispatch_table): Use it.
19605 * tree-switch-conversion.c (struct case_node): Moved from
19606 stmt.c and adjusted.
19607 (emit_case_nodes): Likewise.
19608 (node_has_low_bound): Likewise.
19609 (node_has_high_bound): Likewise.
19610 (node_is_bounded): Likewise.
19611 (case_values_threshold): Likewise.
19612 (reset_out_edges_aux): Likewise.
19613 (compute_cases_per_edge): Likewise.
19614 (add_case_node): Likewise.
19615 (dump_case_nodes): Likewise.
19616 (balance_case_nodes): Likewise.
19617 (expand_switch_as_decision_tree_p): Likewise.
19618 (emit_jump): Likewise.
19619 (emit_case_decision_tree): Likewise.
19620 (try_switch_expansion): Likewise.
19621 (do_jump_if_equal): Likewise.
19622 (emit_cmp_and_jump_insns): Likewise.
19623 (fix_phi_operands_for_edge): New function.
19624 (record_phi_operand_mapping): Likewise.
19625 (class pass_lower_switch): New pass.
19626 (pass_lower_switch::execute): New function.
19627 (make_pass_lower_switch): Likewise.
19628 (conditional_probability):
19629 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19630 * tree-pass.h: Add make_pass_lower_switch.
19632 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19635 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19638 2017-08-29 Richard Biener <rguenther@suse.de>
19640 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19641 we do not add a DW_AT_inline attribute twice.
19642 (gen_subprogram_die): Remove code setting DW_AT_inline on
19643 DECL_ABSTRACT_P nodes.
19645 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19647 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19648 calls to internal functions.
19649 (gimplify_modify_expr): Likewise.
19650 * tree-call-cdce.c (use_internal_fn): Likewise.
19651 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19652 (convert_to_divmod): Set the nothrow flag.
19653 * tree-if-conv.c (predicate_mem_writes): Likewise.
19654 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19655 (vectorizable_call): Likewise.
19656 (vectorizable_store): Likewise.
19657 (vectorizable_load): Likewise.
19658 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19659 (vect_recog_mask_conversion_pattern): Likewise.
19661 2017-08-29 Martin Liska <mliska@suse.cz>
19664 * gcc.c (driver_handle_option): Add new argument.
19665 * opts-common.c (handle_option): Pass
19666 target_option_override_hook.
19667 * opts-global.c (lang_handle_option): Add new option.
19668 (set_default_handlers): Add new argument.
19669 (decode_options): Likewise.
19670 * opts.c (target_handle_option): Likewise.
19671 (common_handle_option): Call target_option_override_hook.
19672 * opts.h (struct cl_option_handler_func): Add hook for
19673 target option override.
19674 (struct cl_option_handlers): Likewise.
19675 (set_default_handlers): Add new argument.
19676 (decode_options): Likewise.
19677 (common_handle_option): Likewise.
19678 (target_handle_option): Likewise.
19679 * toplev.c (toplev::main): Pass targetm.target_option.override
19682 2017-08-29 Richard Biener <rguenther@suse.de>
19683 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19685 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19686 life to the active subtree.
19688 2017-08-28 Jeff Law <law@redhat.com>
19690 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19691 derive_equivalences.
19692 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19693 Code moved into....
19694 (edge_info::derive_equivalences): New private member function
19696 * tree-ssa-dom.c (class edge_info): Changed from a struct
19697 to a class. Add ctor/dtor, methods and data members.
19698 (edge_info::edge_info): Renamed from allocate_edge_info.
19699 Initialize additional members.
19700 (edge_info::~edge_info): New.
19701 (free_dom_edge_info): Delete the edge info.
19702 (record_edge_info): Use new class & associated member functions.
19703 Tighten forms for testing for edge equivalences.
19704 (record_temporary_equivalences): Iterate over the simple
19705 equivalences rather than assuming there's only one per edge.
19706 (cprop_into_successor_phis): Iterate over the simple
19707 equivalences rather than assuming there's only one per edge.
19708 (optimize_stmt): Use operand_equal_p rather than pointer
19709 equality for mini-DSE code.
19711 2017-08-28 Nathan Sidwell <nathan@acm.org>
19713 * gcc.c (execute): Fold SIGPIPE handling into switch
19714 statement. Adjust internal error message.
19716 2017-08-28 Richard Biener <rguenther@suse.de>
19719 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19720 Do nothing for removed DIEs.
19722 2017-08-28 Richard Biener <rguenther@suse.de>
19724 PR tree-optimization/81977
19725 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19728 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19731 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19732 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19733 using targetm.gen_mem_thread_fence.
19735 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19738 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19739 predicate to register_operand. Reorder operands.
19740 (*btr<mode>): Ditto.
19741 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19742 (*btr<mode>_mask): Ditto.
19744 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19746 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19747 * config/rs6000/xmmintrin.h: New file.
19748 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19750 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19753 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19754 parameter and_insn and return it.
19755 (recombine_lvx_pattern): Insert a copy to ensure availability of
19756 the base register of the copied masking operation at the point of
19757 the instruction replacement.
19758 (recombine_stvx_pattern): Likewise.
19760 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19762 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19763 undocumented switches.
19764 (-mpower9-dform-vector): Likewise.
19765 (-mpower9-dform): Likewise.
19766 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19767 comments to delete references to -mpower9-dform* switches.
19768 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19769 Delete reference to -mpower9-dform* switches, test for
19770 -mpower9-vector instead.
19771 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19772 (OTHER_P9_VECTOR_MASKS): Likewise.
19773 (POWERPC_MASKS): Likewise.
19774 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19775 tests against -mpower9-dform* to -mpower9-vector. Delete code
19776 that checked for -mpower9-dform* consistancy with other options.
19777 Add test for -mpower9-misc to enable other power9 switches.
19778 (rs6000_init_hard_regno_mode_ok): Likewise.
19779 (rs6000_option_override_internal): Likewise.
19780 (rs6000_emit_prologue): Likewise.
19781 (rs6000_emit_epilogue): Likewise.
19782 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19783 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19784 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19786 (emit_fusion_p9_store): Likewise.
19787 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19788 resetting these macros if the assembler does not support ISA 3.0
19790 (TARGET_P9_DFORM_VECTOR): Likewise.
19791 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19792 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19794 2017-08-25 Alan Modra <amodra@gmail.com>
19796 PR rtl-optimization/81747
19797 * cse.c (cse_extended_basic_block): Don't attempt to record
19798 equivalences for degenerate conditional branches that branch
19799 to their fall-through.
19801 2017-08-24 Martin Sebor <msebor@redhat.com>
19803 PR middle-end/81908
19804 * gimple-fold.c (size_must_be_zero_p): New function.
19805 (gimple_fold_builtin_memory_op): Call it.
19807 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19809 * config/rs6000/mm_malloc.h: New file.
19811 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19813 PR tree-optimization/81913
19814 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19815 analysis when either IVs in condition can wrap.
19817 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19819 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19820 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19822 2017-08-24 Richard Biener <rguenther@suse.de>
19825 * targhooks.c (default_target_can_inline_p): Properly
19826 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19827 is present and always compare.
19828 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19829 infer -mfpmath=sse from TARGET_SSE_P.
19830 (ix86_can_inline_p): Properly use target_option_default_node when
19831 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19833 2017-08-24 Richard Biener <rguenther@suse.de>
19836 * dwarf2out.c (output_die): Handle flag_generate_offload like
19838 (output_comp_unit): Likewise.
19839 (gen_array_type_die): Likewise.
19840 (dwarf2out_early_finish): Likewise.
19841 (note_variable_value_in_expr): Likewise.
19842 (dwarf2out_finish): Likewise. Adjust assert.
19843 * cgraphunit.c (symbol_table::compile): Move setting of
19844 flag_generate_offload earlier ...
19845 (symbol_table::finalize_compilation_unit): ... here, before
19846 early debug finalization.
19848 2017-08-24 Richard Biener <rguenther@suse.de>
19850 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19851 and ipa-fnsummary.h.
19852 (ix86_can_inline_p): When ix86_fpmath flags do not match
19853 check whether the callee uses FP math at all.
19855 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19857 PR middle-end/81931
19858 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19859 instead of TYPE_PRECISION.
19861 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19862 Alan Hayward <alan.hayward@arm.com>
19863 David Sherwood <david.sherwood@arm.com>
19865 * combine.c (make_extraction): Use subreg_offset_from_lsb.
19867 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
19869 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
19871 (ix86_frame::stack_realign_allocate): New field.
19872 (struct machine_frame_state): Modify comments.
19873 (machine_frame_state::sp_realigned_fp_end): New field.
19874 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
19875 layout calculation.
19876 (sp_valid_at): Add assertion to assure no attempt to access invalid
19877 offset of a realigned stack.
19878 (fp_valid_at): Likewise.
19879 (choose_baseaddr): Modify comments.
19880 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
19881 ix86_expand_prologue.
19882 (ix86_expand_prologue): Modify stack realignment and allocation.
19883 (ix86_expand_epilogue): Modify comments.
19884 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
19885 avx2_runtime, avx512f, and avx512f_runtime.
19887 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
19889 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
19890 (mstackrealign): Do not init to -1.
19891 * config/i386/i386.c (ix86_option_override_internal):
19892 Check opts_set, not opts when setting default value of
19893 opts->x_ix86_force_align_arg_pointer.
19895 2017-08-23 Richard Biener <rguenther@suse.de>
19897 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
19898 lang_hooks.decl_printable_name.
19899 * print-rtl-function.c (print_rtx_function): Likewise.
19900 * tree-pretty-print.c (dump_function_header): Likewise.
19902 2017-08-23 Richard Biener <rguenther@suse.de>
19905 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
19906 -g0 at compile-time.
19908 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19910 PR middle-end/19706
19911 * doc/sourcebuild.texi (Other hardware attributes):
19914 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19916 PR middle-end/19706
19917 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
19918 Fix single-use check.
19920 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
19922 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
19924 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19926 * doc/install.texi: Modify to add more details on running selected
19929 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19931 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
19932 is combined with -mabi=ms.
19933 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
19936 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19938 PR tree-optimization/81488
19939 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
19940 and cached_basis fields.
19941 (MAX_SPREAD): New constant.
19942 (alloc_cand_and_find_basis): Initialize new fields.
19943 (clear_visited): New function.
19944 (create_phi_basis_1): Rename from create_phi_basis, set visited
19945 and cached_basis fields.
19946 (create_phi_basis): New wrapper function.
19947 (phi_add_costs_1): Rename from phi_add_costs, add spread
19948 parameter, set visited field, short-circuit when limits reached.
19949 (phi_add_costs): New wrapper function.
19950 (record_phi_increments_1): Rename from record_phi_increments, set
19952 (record_phi_increments): New wrapper function.
19953 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
19954 (phi_incr_cost): New wrapper function.
19955 (all_phi_incrs_profitable_1): Rename from
19956 all_phi_incrs_profitable, set visited field.
19957 (all_phi_incrs_profitable): New wrapper function.
19959 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19960 Alan Hayward <alan.hayward@arm.com>
19961 David Sherwood <david.sherwood@arm.com>
19963 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
19964 that takes the outer and inner modes.
19965 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
19966 comparison as the canonical test for a paradoxical subreg.
19967 * combine.c (simplify_set): Use paradoxical_subreg_p.
19968 (make_extraction): Likewise.
19969 (force_to_mode): Likewise.
19970 (rtx_equal_for_field_assignment_p): Likewise.
19971 (gen_lowpart_for_combine): Likewise.
19972 (simplify_comparison): Likewise.
19973 * cse.c (equiv_constant): Likewise.
19974 * expmed.c (store_bit_field_1): Likewise.
19975 * final.c (alter_subreg): Likewise.
19976 * fwprop.c (propagate_rtx): Likewise.
19977 (forward_propagate_subreg): Likewise.
19978 * ira-conflicts.c (ira_build_conflicts): Likewise.
19979 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
19980 * lra-constraints.c (curr_insn_transform): Likewise.
19981 (split_reg): Likewise.
19982 * lra-eliminations.c (move_plus_up): Likewise.
19983 (lra_eliminate_regs_1): Likewise.
19984 * recog.c (general_operand): Likewise.
19985 * ree.c (combine_reaching_defs): Likewise.
19986 * reload.c (push_reload): Likewise.
19987 (find_reloads): Likewise.
19988 * reload1.c (elimination_effects): Likewise.
19989 (compute_reload_subreg_offset): Likewise.
19990 (choose_reload_regs): Likewise.
19991 * rtlanal.c (subreg_lsb_1): Likewise.
19992 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19993 (simplify_subreg): Likewise.
19994 * var-tracking.c (track_loc_p): Likewise.
19995 * emit-rtl.c (byte_lowpart_offset): Likewise.
19996 (paradoxical_subreg_p): Delete out-of-line definition.
19998 2017-08-22 Jeff Law <law@redhat.com>
20000 PR tree-optimization/81741
20001 PR tree-optimization/71947
20002 * tree-ssa-dom.c: Include tree-inline.h.
20003 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20004 equivalences if one is more expensive to compute than the other.
20005 * tree-ssa-scopedtables.h (class const_or_copies): Make
20006 record_const_or_copy_raw method private.
20007 (class avail_exprs_stack): New method simplify_binary_operation.
20008 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20009 avail_exprs_stack::simplify_binary_operation as needed.
20010 (avail_exprs_stack::simplify_binary_operation): New function.
20012 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20014 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20015 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20016 (DOT_SYMBOLS): Likewise.
20017 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20018 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20019 (RS6000_ABI_NAME): Likewise.
20020 (TARGET_CMODEL): Likewise.
20021 (TOC_SECTION_ASM_OP): Likewise.
20022 (SET_CMODEL): New macro.
20023 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20025 2017-08-22 Richard Biener <rguenther@suse.de>
20027 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20028 to allow for free-lang-data replacements similar to verify_type_variant.
20030 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20031 Alan Hayward <alan.hayward@arm.com>
20032 David Sherwood <david.sherwood@arm.com>
20034 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20035 VOIDmode for the LABEL_REF.
20037 2017-08-22 Richard Biener <rguenther@suse.de>
20039 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20041 2017-08-22 Richard Biener <rguenther@suse.de>
20043 * common.opt (feliminate-dwarf2-dups): Ignore.
20044 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20045 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20046 same_die_p_wrap, compute_section_prefix,
20047 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20048 (comdat_symbol_id, comdat_symbol_number): Likewise.
20049 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20051 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20052 (output_die): Mark unreachable path unreachable.
20053 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20054 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20055 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20056 (dwarf2out_early_finish): Likewise.
20058 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20060 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20062 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20065 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20066 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20067 (avr_attribute_table) <io, io_low, address>: Initialize
20068 .decl_required with true.
20070 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20072 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20073 undocumented debugging options.
20074 (-mvsx-scalar-double): Likewise.
20075 (-mallow-df-permute): Likewise.
20076 (-mvectorize-builtins): Likewise.
20077 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20078 (rs6000_builtin_vectorized_function): Likewise.
20079 (rs6000_builtin_md_vectorized_function): Likewise.
20080 (rs6000_opt_vars): Likewise.
20082 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20085 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20086 (*btrq_imm): Rename from *btrq.
20087 (*btcq_imm): Rename from *btcq.
20088 (btsc): New code attribute.
20089 (*<btsc><mode>): New insn pattern.
20090 (*btr<mode>): Ditto.
20091 (*<btsc><mode>_mask): New insn_and_split pattern.
20092 (*btr<mode>_mask): Ditto.
20094 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20095 Alan Hayward <alan.hayward@arm.com>
20096 David Sherwood <david.sherwood@arm.com>
20098 * function.c (pad_below): Simplify padding calculation.
20100 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20101 Alan Hayward <alan.hayward@arm.com>
20102 David Sherwood <david.sherwood@arm.com>
20104 * target.def (function_prologue): Remove frame size argument.
20105 (function_epilogue): Likewise.
20106 * doc/tm.texi: Regenerate.
20107 * final.c (final_start_function): Update call to function_prologue.
20108 (final_end_function): Update call to function_epilogue.
20109 (default_function_pro_epilogue): Remove frame size argument.
20110 * output.h (default_function_pro_epilogue): Likewise.
20111 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
20112 (arm_output_function_prologue): Likewise.
20113 * config/frv/frv.c (frv_function_prologue): Likewise.
20114 (frv_function_epilogue): Likewise.
20115 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
20116 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
20117 (ia64_output_function_epilogue): Likewise.
20118 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
20119 (m32r_output_function_epilogue): Likewise.
20120 * config/microblaze/microblaze.c (microblaze_function_prologue)
20121 (microblaze_function_epilogue): Likewise.
20122 * config/mips/mips.c (mips_output_function_prologue): Likewise.
20123 (mips_output_function_epilogue): Likewise.
20124 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
20125 (mmix_target_asm_function_epilogue): Likewise.
20126 * config/msp430/msp430.c (msp430_start_function): Likewise.
20127 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
20128 (nds32_asm_function_epilogue): Likewise.
20129 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
20130 * config/pa/pa.c (pa_output_function_prologue): Likewise.
20131 (pa_output_function_epilogue): Likewise.
20132 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
20133 (rs6000_output_function_epilogue): Likewise.
20134 * config/rl78/rl78.c (rl78_start_function): Likewise.
20135 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
20136 (rs6000_output_function_epilogue): Likewise.
20137 * config/rx/rx.c (rx_output_function_prologue): Likewise.
20138 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
20139 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
20140 (sparc_asm_function_epilogue): Likewise.
20142 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20144 * tree.h (type_has_mode_precision_p): New function.
20145 * convert.c (convert_to_integer_1): Use it.
20146 * expr.c (expand_expr_real_2): Likewise.
20147 (expand_expr_real_1): Likewise.
20148 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
20149 * match.pd: Likewise.
20150 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
20151 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
20152 * tree-tailcall.c (process_assignment): Likewise.
20153 * tree-vect-loop.c (vectorizable_reduction): Likewise.
20154 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
20155 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
20156 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
20157 (vectorizable_assignment): Likewise.
20158 (vectorizable_shift): Likewise.
20159 (vectorizable_operation): Likewise.
20160 * tree-vrp.c (register_edge_assert_for_2): Likewise.
20162 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
20164 * match.pd: Add pow (C, x) simplification.
20166 2017-08-21 Richard Biener <rguenther@suse.de>
20168 PR tree-optimization/81900
20169 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
20170 for blocks with abnormal predecessors.
20171 (compute_antic): Do not set visited flag prematurely.
20173 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
20176 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
20178 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20180 * stor-layout.h (vector_type_mode): Move to...
20181 * tree.h (vector_type_mode): ...here.
20182 * stor-layout.c (vector_type_mode): Move to...
20183 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
20185 2017-08-21 Richard Biener <rguenther@suse.de>
20187 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
20188 register_external_die hooks.
20189 (debug_false_tree_charstarstar_uhwistar): Declare.
20190 (debug_nothing_tree_charstar_uhwi): Likewise.
20191 * debug.c (do_nothing_debug_hooks): Adjust.
20192 (debug_false_tree_charstarstar_uhwistar): New do nothing.
20193 (debug_nothing_tree_charstar_uhwi): Likewise.
20194 * dbxout.c (dbx_debug_hooks): Adjust.
20195 (xcoff_debug_hooks): Likewise.
20196 * sdbout.c (sdb_debug_hooks): Likewise.
20197 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20198 * dwarf2out.c (macinfo_label_base): New global.
20199 (dwarf2out_register_external_die): New function for the
20200 register_external_die hook.
20201 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
20202 (dwarf2_debug_hooks): Use them.
20203 (dwarf2_lineno_debug_hooks): Adjust.
20204 (struct die_struct): Add with_offset flag.
20205 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
20206 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
20207 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
20208 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
20209 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
20210 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
20211 defining section names for the early LTO debug variants.
20212 (reset_indirect_string): New helper.
20213 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
20214 (print_dw_val): Add support for offsetted symbol references.
20215 (get_ultimate_context): Split out from is_cxx.
20216 (is_cxx): Use get_ultimate_context.
20217 (is_fortran): Add decl overload.
20218 (compute_comp_unit_symbol): Split out worker from
20219 compute_section_prefix.
20220 (compute_section_prefix): Call compute_comp_unit_symbol and
20221 set comdat_type_p here.
20222 (output_die): Skip DIE symbol output for the LTO added one.
20223 Handle DIE symbol references with offset.
20224 (output_comp_unit): Guard section name mangling properly.
20225 For LTO debug sections emit a symbol at the section beginning
20226 which we use to refer to its DIEs.
20227 (add_abstract_origin_attribute): For DIEs registered via
20228 dwarf2out_register_external_die directly refer to the early
20229 DIE rather than indirectly through the shadow one we created.
20230 Remove obsolete call to dwarf2out_abstract_function for
20231 non-function/block origins.
20232 (gen_array_type_die): When generating early LTO debug do
20233 not emit DW_AT_string_length.
20234 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
20235 late when in LTO. As suggested place a gcc_unreachable for
20236 the DECL_ABSTRACT_P case.
20237 (gen_subprogram_die): Avoid another specification DIE
20238 for early built declarations/definitions for the late LTO case.
20239 (gen_variable_die): Add type references for late duplicated VLA dies
20241 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
20242 we have the abstract instance already.
20243 (process_scope_var): Adjust decl DIE contexts in LTO which
20244 first puts them in limbo.
20245 (gen_decl_die): Do not generate type DIEs late apart from
20246 types for VLAs or for decls we do not yet have a DIE. Do not
20247 call dwarf2out_abstract_function late.
20248 (dwarf2out_early_global_decl): Make sure to create DIEs
20249 for abstract instances of a decl first.
20250 (dwarf2out_late_global_decl): Adjust comment.
20251 (output_macinfo_op): With multiple macro sections use
20252 macinfo_label_base to distinguish labels.
20253 (output_macinfo): Likewise. Update macinfo_label_base.
20254 Pass in the line info label.
20255 (note_variable_value_in_expr): When generating LTO resolve
20256 all variable values here by generating DIEs as needed.
20257 (init_sections_and_labels): Add early LTO debug flag parameter
20258 and generate different sections and names if set. Add generation
20259 counter for the labels so we can have multiple of them.
20260 (reset_dies): Helper to allow DIEs to be output multiple times.
20261 (dwarf2out_finish): When outputting DIEs to the fat part of an
20262 LTO object first reset DIEs.
20263 (dwarf2out_early_finish): Output early DIEs when generating LTO.
20264 (modified_type_die): Check for decl_ultimate_origin being self
20266 (gen_type_die_with_usage): Likewise.
20267 (gen_typedef_die): Allow decl_ultimate_origin being self.
20268 (set_decl_abstract_flags): Remove.
20269 (set_block_abstract_flags): Likewise.
20270 (dwarf2out_abstract_function): Treat the early generated DIEs
20271 as the abstract copy and only add DW_AT_inline and
20272 DW_AT_artificial here and call set_decl_origin_self.
20273 If the DIE has an abstract origin don't do anything.
20274 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
20275 if we have none yet (Go fails to build one, PR78628).
20276 (variably_modified_type_p): Prevent endless recursion for Ada
20277 cyclic pointer types.
20278 * lto-streamer-in.c: Include debug.h.
20279 (dref_queue): New global.
20280 (lto_read_tree_1): Stream in DIE references.
20281 (lto_input_tree): Register DIE references.
20282 (input_function): Stream DECL_DEBUG_ARGS.
20283 * lto-streamer-out.c: Include debug.h.
20284 (lto_write_tree_1): Output DIE references.
20285 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
20286 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
20287 (output_function): Stream DECL_DEBUG_ARGS.
20288 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
20289 Stream DECL_ABSTRACT_ORIGIN.
20290 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
20291 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
20292 DECL_CONTEXT for file-scope decls.
20293 * lto-streamer.h (struct dref_entry): Declare.
20294 (dref_queue): Likewise.
20295 * cfgexpand.c (pass_expand::execute): Do not call the
20296 outlining_inline_function hook here.
20297 * lto-wrapper.c (debug_obj): New global.
20298 (tool_cleanup): Unlink it if required.
20299 (debug_objcopy): New function.
20300 (run_gcc): Handle early debug sections in the IL files by
20301 extracting them to separate files, partially linkin them and
20302 feeding the result back as result to the linker.
20303 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
20304 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
20305 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
20306 sections into a separate segment.
20307 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
20309 (darwin_asm_dwarf_section): Likewise.
20310 (darwin_asm_output_dwarf_offset): Likewise.
20311 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
20313 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20314 Alan Hayward <alan.hayward@arm.com>
20315 David Sherwood <david.sherwood@arm.com>
20317 * read-md.h (md_reader::record_potential_iterator_use): Replace
20318 pointer argument with an rtx and an index.
20319 * read-rtl.c (iterator_group::apply_iterator): Likewise.
20320 (apply_mode_iterator): Likewise.
20321 (apply_code_iterator): Likewise.
20322 (apply_int_iterator): Likewise.
20323 (apply_subst_iterator): Likewise.
20324 (record_iterator_use): Likewise.
20325 (record_attribute_use): Likewise.
20326 (md_reader::record_potential_iterator_use): Likewise. Update calls
20327 to record_iterator_use and apply_iterator.
20328 (iterator_use): Replace ptr with x and index.
20329 (attribute_use): Likewise.
20330 (apply_attribute_uses): Update calls to apply_iterator.
20331 (apply_iterators): Likewise. Update initialization of iterator_use.
20332 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
20333 and record_potential_iterator_use.
20334 (rtx_reader::read_rtx_operand): Likewise.
20336 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20337 Alan Hayward <alan.hayward@arm.com>
20338 David Sherwood <david.sherwood@arm.com>
20340 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
20343 2017-08-21 Richard Biener <rguenther@suse.de>
20345 PR middle-end/81884
20346 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
20347 at struct end conservatively when comparing common bases.
20349 2017-08-21 Richard Biener <rguenther@suse.de>
20351 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
20352 (mem_ref_in_stmt): Remove.
20353 (determine_max_movement): Use ref index to get at the reference.
20354 (invariantness_dom_walker::before_dom_children): Deal with
20355 lim data already initialized.
20356 (gather_mem_refs_stmt): Initialize lim data and record ref index.
20358 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20360 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
20361 (TARGET_ISA_ROUND): Ditto.
20362 (TARGET_ROUND): Ditto.
20363 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
20364 * config/i386/i386.md: Ditto.
20365 * config/i386/sse.md: Ditto.
20366 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
20367 with OPTION_MASK_ISA_SSE4_1.
20369 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20372 * doc/extend.texi (x86 Built-in Functions): Correct the name of
20373 __builtin_ia32_lzcnt_u16.
20375 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
20378 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
20379 (rs6000_set_current_function): Rewrite function to use it.
20381 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
20384 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
20385 and TYPE_WARN_IF_NOT_ALIGN.
20386 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
20387 (handle_warn_if_not_align): New.
20388 (place_union_field): Call handle_warn_if_not_align.
20389 (place_field): Call handle_warn_if_not_align.
20390 Copy TYPE_WARN_IF_NOT_ALIGN.
20391 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
20392 (layout_type): Likewise.
20393 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
20395 (tree_decl_common): Add warn_if_not_align.
20396 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
20397 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
20398 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
20399 (DECL_WARN_IF_NOT_ALIGN): Likewise.
20400 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
20401 * doc/extend.texi: Document warn_if_not_aligned attribute.
20402 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
20404 2017-08-17 Martin Liska <mliska@suse.cz>
20407 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
20408 (get_data_dependence): Use it as pointer type.
20409 (distribute_loop): Likewise.
20411 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20413 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
20414 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
20415 (p8_vmrgow_<mode>_direct): New define_insn.
20416 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
20417 handle endianness for vmrgew and vmrgow permute patterns.
20419 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20421 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
20422 * config/rs6000/rs6000-cpus.def: Remove comment.
20423 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
20424 (POWERPC_MASKS): Likewise.
20425 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
20426 use of TARGET_VSX_TIMODE.
20427 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
20428 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
20429 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
20430 (rs6000_option_override_internal): Remove dead code.
20431 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
20432 (rs6000_legitimize_reload_address): Likewise.
20433 (rs6000_legitimate_address_p): Likewise.
20434 (rs6000_opt_masks): Delete "vsx-timode".
20435 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
20436 from function comment.
20437 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
20438 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
20439 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
20441 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
20442 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
20443 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
20444 (VSX_TI): Likewise.
20446 (define_peephole2): Likewise.
20448 2017-08-17 Martin Sebor <msebor@redhat.com>
20451 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
20452 past the end of an array.
20453 (test_pp_format): Add test cases.
20455 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
20457 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
20458 missing ECF_NOTHROW flags.
20460 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20463 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
20464 operands residing in integer registers.
20465 (*vsx_le_perm_load_<mode>): Likewise.
20466 (*vsx_le_perm_store_<mode>): Likewise.
20467 (define_peephole2): Add peepholes to optimize the above.
20469 2017-08-17 Marek Polacek <polacek@redhat.com>
20471 PR middle-end/81814
20472 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
20473 to mimic what shorten_compare did. Change the return type to bool.
20474 (fold_cond_expr_with_comparison): Update call to
20475 operand_equal_for_comparison_p.
20476 (fold_ternary_loc): Likewise.
20478 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
20480 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
20482 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
20484 2017-08-17 Richard Biener <rguenther@suse.de>
20486 * tree-ssa-structalias.c (solve_graph): When propagating
20487 to successors update the graphs succ edges and avoid duplicate work.
20489 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
20492 * config/i386/i386.c (ix86_option_override_internal): Save target
20493 specific options after ix86_stack_protector_guard_reg was changed.
20495 2017-08-17 Richard Biener <rguenther@suse.de>
20497 PR tree-optimization/81827
20498 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
20499 (new_var_info): Initialize it conservatively.
20500 (get_call_vi): Mark register vars.
20501 (new_scalar_tmp_constraint_exp): Likewise.
20502 (handle_rhs_call): Likewise.
20503 (handle_const_call): Likewise.
20504 (create_function_info_for): Likewise.
20505 (solve_constraints): Sort varinfos to separate register from
20506 non-register vars to pack points-to solution bitmaps during
20509 2017-08-17 Marek Polacek <polacek@redhat.com>
20511 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
20513 2017-08-17 Richard Biener <rguenther@suse.de>
20515 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
20516 to true when overflow is undefined and we saturated the result.
20518 2017-08-17 Alan Modra <amodra@gmail.com>
20521 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
20522 Don't use store multiple if only one reg needs saving.
20523 (interesting_frame_related_regno): New function.
20524 (rs6000_frame_related): Don't emit frame info for regs that
20526 (rs6000_emit_epilogue): Likewise.
20528 2017-08-16 Nathan Sidwell <nathan@acm.org>
20530 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
20531 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
20532 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
20533 * tree.c (free_lang_data_in_type): Use else-if chain. Always
20534 clear TYPE_LANG_1. Remove obsolete member-function stripping.
20535 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
20536 (verify_type): Adjust for TYPE_BINFO move.
20537 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
20538 process TYPE_BINFO directly.
20539 (hash_tree): Likewise.
20540 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20542 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20545 2017-08-16 David Malcolm <dmalcolm@redhat.com>
20547 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
20549 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
20552 * config/i386/i386.md (*anddi_1_btr): Change predicates of
20553 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
20554 Add ix86_binary_operator_ok to insn constraint.
20555 (*iordi_1_bts): Ditto.
20556 (*xordi_1_btc): Ditto.
20557 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
20558 Update corresponding peephole2 pattern.
20562 2017-08-16 Bin Cheng <bin.cheng@arm.com>
20564 PR tree-optimization/81832
20565 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
20566 copy loop header which has IFN_LOOP_DIST_ALIAS call.
20568 2017-08-16 Marek Polacek <polacek@redhat.com>
20571 * fold-const.c (fold_indirect_ref_1): Restore original behavior
20572 regarding size_zero_node.
20574 2017-08-16 Martin Liska <mliska@suse.cz>
20577 * config.gcc: Respect previously set extra_objs in case
20580 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
20582 PR tree-optimization/81835
20583 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
20584 the phi SSA_NAME. Check that the condition in a COND_EXPR does
20585 not depend on the phi.
20587 2017-08-16 Alan Modra <amodra@gmail.com>
20589 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
20592 2017-08-16 Alan Modra <amodra@gmail.com>
20594 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
20595 (save_reg_p): ..into this. Update all callers.
20596 (first_reg_to_save): Simplify.
20598 2017-08-16 Alan Modra <amodra@gmail.com>
20600 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
20603 2017-08-15 Joseph Myers <joseph@codesourcery.com>
20607 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
20608 constant count if that count is less than 32.
20610 2017-08-15 Nathan Sidwell <nathan@acm.org>
20612 * gcc.c (execute): Emit friendlier message if inferior is killed
20613 by an external cause.
20615 2017-08-15 Richard Biener <rguenther@suse.de>
20617 PR tree-optimization/81790
20618 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20619 CONSTRUCTORs from simplifying and VN.
20621 2017-08-14 Martin Sebor <msebor@redhat.com>
20623 * builtin-attrs.def: Add comments.
20625 2017-08-14 Martin Sebor <msebor@redhat.com>
20628 * doc/extend.texi (attribute nonstring): Document new attribute.
20630 2017-08-14 Martin Sebor <msebor@redhat.com>
20633 * tree-diagnostic.c (default_tree_printer): Handle %G.
20634 * gimple-pretty-print.h (percent_G_format): Declare new function.
20635 * gimple-pretty-print.c (percent_G_format): Define.
20636 * tree-pretty-print.c (percent_K_format): Add argument.
20638 2017-08-14 Martin Sebor <msebor@redhat.com>
20640 PR translation/79998
20641 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20642 Remove a stray space.
20644 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20647 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20648 (*iordi_1_bts): Ditto.
20649 (*xordi_1_btc): Ditto.
20651 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20654 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20655 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20657 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20658 quoted strings, and make more translator-friendly.
20659 (darwin_rs6000_override_options): Likewise.
20660 (rs6000_option_override_internal): Likewise.
20661 (rs6000_return_in_memory): Fix overlong line.
20662 (init_cmulative_args): Use quoted strings, and make more
20663 translator-friendly.
20664 (rs6000_pass_by_reference): Fix overlong line.
20665 (def_builtin): Use quoted strings.
20666 (altivec_expand_predicate_builtin): Use quoted strings, and make
20667 more translator-friendly.
20668 (htm_expand_builtin): Use quoted strings.
20669 (cpu_expand_builtin): Use quoted strings, and make more
20670 translator-friendly.
20671 (altivec_expand_builtin): Likewise.
20672 (paired_expand_predicate_builtin): Likewise.
20673 (rs6000_invalid_builtin): Likewise.
20674 (builtin_function_type): Use quoted strings.
20675 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20676 more translator-friendly.
20677 (rs6000_trampoline_init): Likewise.
20678 (rs6000_handle_altivec_attribute): Likewise.
20679 (rs6000_inner_target_options): Use quoted strings.
20680 (rs6000_disable_incompatible_switches): Likewise.
20681 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20682 strings, and make more translator-friendly.
20683 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20685 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20687 PR tree-optimization/81799
20688 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20689 cond_expr to simple gimple operand.
20691 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20693 PR middle-end/46932
20694 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20696 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20700 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20701 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20702 TARGET_GASISR_PROLOGUES.
20703 * config/avr/avr.c (avr_option_override): Same.
20704 (avr_pass_pre_proep::execute): Same.
20706 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20709 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20710 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20712 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20714 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20715 define_insn_and_split. Split to a memory load from 0 in
20716 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20717 using PTR mode iterator.
20718 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20719 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20720 (*add_tp_<mode>): Redefine as define_insn_and_split.
20721 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20722 address space. Merge with *add_tp_x32 using PTR mode iterator.
20723 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20724 Split to an add with a memory load from 0 in
20725 DEFAULT_TLS_SEG_REG address space.
20727 2017-08-12 Andrew Pinski <apinski@cavium.com>
20729 * config/aarch64/aarch64-option-extensions.def (rdma):
20730 Fix feature string to what Linux prints out in /proc/cpuinfo.
20732 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20735 * dwarf2out.c (modified_type_die): For C typedef types that have
20736 an ultimate origin, process the ultimate origin instead of the
20738 (gen_typedef_die): Assert that input DECLs have no ultimate
20740 (gen_type_die_with_usage): For typedef variants that have an
20741 ultimate origin, just call gen_decl_die on the original DECL.
20742 (process_scope_var): Avoid creating DIEs for local typedefs and
20743 concrete static variables.
20745 2017-08-12 Alan Modra <amodra@gmail.com>
20749 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20750 match gnu-user.h startfile.
20751 (ENDFILE_LINUX_SPEC): Similarly.
20753 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20756 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20758 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20760 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20761 * config/aarch64/aarch64.md (mov<mode>): Change.
20762 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20763 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20764 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20766 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20768 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20769 for storage order barriers.
20771 2017-08-11 Martin Liska <mliska@suse.cz>
20774 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20775 variables of void type.
20777 2017-08-11 Martin Liska <mliska@suse.cz>
20779 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20780 TARGET_SUPPORTS_ALIASES.
20781 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20782 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20783 (optimize_weakref): Likewise.
20784 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20785 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20786 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20788 2017-08-11 Martin Liska <mliska@suse.cz>
20791 * config/i386/i386.c (make_resolver_func): Do complete
20792 refactoring of the function.
20794 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20797 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20798 * config/i386/i386.c (ix86_stack_protect_guard): Use
20799 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20800 * doc/invoke.texi (x86 Options): Document
20801 -mstack-protector-guard-symbol= option.
20803 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20805 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20806 * config/i386/i386.c (ix86_split_stack_guard): New function.
20807 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20808 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20809 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20810 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20811 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20812 (split_stack_space_check): Call ix86_split_stack_guard.
20814 2017-08-10 Martin Sebor <msebor@redhat.com>
20816 * print-tree.c (print_node): Print location using the established
20818 Replace spaces with colons.
20819 (debug_raw, debug): Ditto.
20821 2017-08-10 Martin Sebor <msebor@redhat.com>
20824 * pretty-print.c (pp_format): Correct the handling of %s precision.
20826 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20829 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20831 (ix86_finalize_stack_frame_flags): This. Also clear
20832 frame_pointer_needed if -fno-omit-frame-pointer is used without
20834 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20835 with ix86_finalize_stack_frame_flags.
20836 (ix86_expand_epilogue): Likewise.
20837 (ix86_expand_split_stack_prologue): Likewise.
20838 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20840 2017-08-10 Martin Liska <mliska@suse.cz>
20843 * c-attribs.c (handle_target_attribute):
20844 Report warning for an empty string argument of target attribute.
20846 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20849 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20851 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20852 or DECL_NONLOCAL labels.
20853 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20854 or DECL_NONLOCAL labels here.
20856 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20858 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20859 to indicate when early gimple folding has been disabled.
20860 (rs6000_gimple_fold_builtin): Add debug content.
20861 (rs6000_invalid_builtin): Fix whitespace.
20862 (rs6000_expand_builtin): Fix whitespace.
20863 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
20865 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
20868 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
20869 SAVE_MULTIPLE if not all the registers that saves, should be saved.
20871 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
20873 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
20874 (qdf24xx): Likewise.
20875 * config/aarch64/aarch64.md: Include falkor.md.
20876 * config/aarch64/falkor.md: New.
20878 2017-08-09 Marek Polacek <polacek@redhat.com>
20881 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
20882 * diagnostic.c (emit_diagnostic): Add a comment.
20883 (emit_diagnostic_valist): New function.
20885 2017-08-09 Marek Polacek <polacek@redhat.com>
20888 * input.c (make_location): New overload.
20889 * input.h (make_location): Declare.
20891 2017-08-08 Alan Modra <amodra@gmail.com>
20892 H.J. Lu <hongjiu.lu@intel.com>
20895 * gcc.c (NO_PIE_SPEC): Delete.
20896 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
20898 (LINK_PIE_SPEC): ..to here.
20899 (LINK_COMMAND_SPEC): Support -no-pie.
20900 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
20901 chain of crtbegin*.o selection, update for PIE_SPEC changes and
20903 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
20904 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
20905 (ENDFILE_CRTEND_SPEC): Similarly.
20907 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
20910 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
20911 (mstack-protector-guard-offset=): Ditto.
20912 * config/i386/i386.c (ix86_option_override): Handle
20913 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
20915 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
20916 ix86_stack_protect_guard_offset variables.
20917 (TARGET_STACK_PROTECT_GUARD): Always define.
20918 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
20919 and -mstack-protector-guard-offset= options.
20921 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20923 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
20924 boundary case for the last candidate.
20926 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20928 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
20929 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
20931 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20933 PR middle-end/19706
20934 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
20935 * config/aarch64/aarch64-builtins.c
20936 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
20937 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
20938 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
20940 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20941 Andrew Pinski <pinskia@gmail.com>
20943 PR middle-end/19706
20944 * internal-fn.def (XORSIGN): New.
20945 * optabs.def (xorsign_optab): New.
20946 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
20947 (convert_expand_mult_copysign): New.
20948 (pass_optimize_widening_mul::execute): Call
20949 convert_expand_mult_copysign.
20951 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20953 PR tree-optimization/81354
20954 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
20955 Insert on edges rather than explicitly creating landing pads.
20956 (analyze_candidates_and_replace): Commit edge inserts.
20958 2017-08-08 Richard Biener <rguenther@suse.de>
20960 PR middle-end/81719
20961 * tree-ssa-loop-niter.c: Include tree-dfa.h.
20962 (expand_simple_operations): Also look through ADDR_EXPRs with
20963 MEM_REF bases treating them as POINTER_PLUS_EXPR.
20965 2017-08-08 Richard Biener <rguenther@suse.de>
20967 PR tree-optimization/81723
20968 * tree-vect-slp.c (struct bst_traits): New hash traits.
20969 (bst_fail): New global.
20970 (vect_build_slp_tree_2): New worker, split out from ...
20971 (vect_build_slp_tree): ... this now wrapping it with using
20972 bst_fail set to cache SLP tree build fails. Properly handle
20974 (vect_analyze_slp_instance): Allocate and free bst_fail.
20976 2017-08-08 Martin Liska <mliska@suse.cz>
20979 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
20980 LABEL_DECLs that can be from a different function.
20982 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20984 PR tree-optimization/81744
20985 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
20986 loop's number of iterations.
20988 2017-08-08 Martin Liska <mliska@suse.cz>
20990 * asan.c: Include header files.
20991 * attribs.c (build_decl_attribute_variant): New function moved
20993 (build_type_attribute_qual_variant): Likewise.
20994 (cmp_attrib_identifiers): Likewise.
20995 (simple_cst_list_equal): Likewise.
20996 (omp_declare_simd_clauses_equal): Likewise.
20997 (attribute_value_equal): Likewise.
20998 (comp_type_attributes): Likewise.
20999 (build_type_attribute_variant): Likewise.
21000 (lookup_ident_attribute): Likewise.
21001 (remove_attribute): Likewise.
21002 (merge_attributes): Likewise.
21003 (merge_type_attributes): Likewise.
21004 (merge_decl_attributes): Likewise.
21005 (merge_dllimport_decl_attributes): Likewise.
21006 (handle_dll_attribute): Likewise.
21007 (attribute_list_equal): Likewise.
21008 (attribute_list_contained): Likewise.
21009 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21010 (lookup_attribute_by_prefix): Likewise.
21011 * bb-reorder.c: Include header files.
21012 * builtins.c: Likewise.
21013 * calls.c: Likewise.
21014 * cfgexpand.c: Likewise.
21015 * cgraph.c: Likewise.
21016 * cgraphunit.c: Likewise.
21017 * convert.c: Likewise.
21018 * dwarf2out.c: Likewise.
21019 * final.c: Likewise.
21020 * fold-const.c: Likewise.
21021 * function.c: Likewise.
21022 * gimple-expr.c: Likewise.
21023 * gimple-fold.c: Likewise.
21024 * gimple-pretty-print.c: Likewise.
21025 * gimple.c: Likewise.
21026 * gimplify.c: Likewise.
21027 * hsa-common.c: Likewise.
21028 * hsa-gen.c: Likewise.
21029 * internal-fn.c: Likewise.
21030 * ipa-chkp.c: Likewise.
21031 * ipa-cp.c: Likewise.
21032 * ipa-devirt.c: Likewise.
21033 * ipa-fnsummary.c: Likewise.
21034 * ipa-inline.c: Likewise.
21035 * ipa-visibility.c: Likewise.
21037 * lto-cgraph.c: Likewise.
21038 * omp-expand.c: Likewise.
21039 * omp-general.c: Likewise.
21040 * omp-low.c: Likewise.
21041 * omp-offload.c: Likewise.
21042 * omp-simd-clone.c: Likewise.
21043 * opts-global.c: Likewise.
21044 * passes.c: Likewise.
21045 * predict.c: Likewise.
21046 * sancov.c: Likewise.
21047 * sanopt.c: Likewise.
21048 * symtab.c: Likewise.
21049 * toplev.c: Likewise.
21050 * trans-mem.c: Likewise.
21051 * tree-chkp.c: Likewise.
21052 * tree-eh.c: Likewise.
21053 * tree-into-ssa.c: Likewise.
21054 * tree-object-size.c: Likewise.
21055 * tree-parloops.c: Likewise.
21056 * tree-profile.c: Likewise.
21057 * tree-ssa-ccp.c: Likewise.
21058 * tree-ssa-live.c: Likewise.
21059 * tree-ssa-loop.c: Likewise.
21060 * tree-ssa-sccvn.c: Likewise.
21061 * tree-ssa-structalias.c: Likewise.
21062 * tree-ssa.c: Likewise.
21063 * tree-streamer-in.c: Likewise.
21064 * tree-vectorizer.c: Likewise.
21065 * tree-vrp.c: Likewise.
21066 * tsan.c: Likewise.
21067 * ubsan.c: Likewise.
21068 * varasm.c: Likewise.
21069 * varpool.c: Likewise.
21070 * tree.c: Remove functions moved to attribs.[ch].
21071 * tree.h: Likewise.
21072 * config/aarch64/aarch64.c: Add attrs.h header file.
21073 * config/alpha/alpha.c: Likewise.
21074 * config/arc/arc.c: Likewise.
21075 * config/arm/arm.c: Likewise.
21076 * config/avr/avr.c: Likewise.
21077 * config/bfin/bfin.c: Likewise.
21078 * config/c6x/c6x.c: Likewise.
21079 * config/cr16/cr16.c: Likewise.
21080 * config/cris/cris.c: Likewise.
21081 * config/darwin.c: Likewise.
21082 * config/epiphany/epiphany.c: Likewise.
21083 * config/fr30/fr30.c: Likewise.
21084 * config/frv/frv.c: Likewise.
21085 * config/ft32/ft32.c: Likewise.
21086 * config/h8300/h8300.c: Likewise.
21087 * config/i386/winnt.c: Likewise.
21088 * config/ia64/ia64.c: Likewise.
21089 * config/iq2000/iq2000.c: Likewise.
21090 * config/lm32/lm32.c: Likewise.
21091 * config/m32c/m32c.c: Likewise.
21092 * config/m32r/m32r.c: Likewise.
21093 * config/m68k/m68k.c: Likewise.
21094 * config/mcore/mcore.c: Likewise.
21095 * config/microblaze/microblaze.c: Likewise.
21096 * config/mips/mips.c: Likewise.
21097 * config/mmix/mmix.c: Likewise.
21098 * config/mn10300/mn10300.c: Likewise.
21099 * config/moxie/moxie.c: Likewise.
21100 * config/msp430/msp430.c: Likewise.
21101 * config/nds32/nds32-isr.c: Likewise.
21102 * config/nds32/nds32.c: Likewise.
21103 * config/nios2/nios2.c: Likewise.
21104 * config/nvptx/nvptx.c: Likewise.
21105 * config/pa/pa.c: Likewise.
21106 * config/pdp11/pdp11.c: Likewise.
21107 * config/powerpcspe/powerpcspe.c: Likewise.
21108 * config/riscv/riscv.c: Likewise.
21109 * config/rl78/rl78.c: Likewise.
21110 * config/rx/rx.c: Likewise.
21111 * config/s390/s390.c: Likewise.
21112 * config/sh/sh.c: Likewise.
21113 * config/sol2.c: Likewise.
21114 * config/sparc/sparc.c: Likewise.
21115 * config/spu/spu.c: Likewise.
21116 * config/stormy16/stormy16.c: Likewise.
21117 * config/tilegx/tilegx.c: Likewise.
21118 * config/tilepro/tilepro.c: Likewise.
21119 * config/v850/v850.c: Likewise.
21120 * config/vax/vax.c: Likewise.
21121 * config/visium/visium.c: Likewise.
21122 * config/xtensa/xtensa.c: Likewise.
21124 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21127 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
21128 constraints since the -mupper-regs-* switches have been
21130 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
21131 into a vector from a double word element that was extracted from
21132 another vector, and eliminate extra XXPERMDI instructions.
21133 (vsx_concat_<mode>_2): Likewise.
21134 (vsx_concat_<mode>_3): Likewise.
21135 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
21136 concat to allow optimizing inserts from previous extracts.
21138 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
21140 * config/i386/i386.c (ix86_stack_protect_guard): Generate
21141 memory reference to a SSP offset in TLS address space.
21142 (ix86_print_operand) <case '@'>: Remove.
21143 (ix86_print_operand_punct_valid_p): Remove '@' code.
21144 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
21145 UNSPEC_SP_TLS_TEST.
21146 (stack_tls_protect_set_<mode>): Remove.
21147 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
21148 (stack_tls_protect_test_<mode>): Remove.
21149 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
21151 2017-08-07 Olivier Hainque <hainque@adacore.com>
21154 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
21156 2017-08-07 Douglas Rupp <rupp@adacore.com>
21158 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
21159 variable was referenced as multidir in command.
21161 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21164 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
21167 2017-08-07 Martin Liska <mliska@suse.cz>
21169 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
21170 * config/rl78/rl78.c: Add include of attribs.h.
21171 * config/sh/sh.c: Likewise.
21172 * config/v850/v850.c: Likewise.
21174 2017-08-07 Tom de Vries <tom@codesourcery.com>
21176 PR middle-end/78266
21177 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
21179 2017-08-07 Martin Liska <mliska@suse.cz>
21181 * config/mips/mips.c: Include attribs.h.
21183 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
21186 * doc/invoke.texi: Document change in behvaior for -Ofast for
21189 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
21191 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
21193 (aarch64_pop_regs): Likewise.
21194 (aarch64_gen_load_pair): Likewise.
21195 (aarch64_save_callee_saves): Likewise.
21196 (aarch64_restore_callee_saves): Likewise.
21198 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21200 * config/i386/i386.c: Revert the last change.
21202 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21205 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21207 (ix86_finalize_stack_frame_flags): This. Also clear
21208 frame_pointer_needed if -fno-omit-frame-pointer is used without
21210 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21211 with ix86_finalize_stack_frame_flags.
21212 (ix86_expand_epilogue): Likewise.
21213 (ix86_expand_split_stack_prologue): Likewise.
21215 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21218 * config/i386/i386.c (get_builtin_code_for_version): Set priority
21219 to P_AES for Westmere.
21221 2017-08-07 Jonathan Yong <10walls@gmail.com>
21223 * config/i386/mingw.opt (fset-stack-executable): Removed.
21224 * config/i386/cygming.opt (fset-stack-executable): Moved
21226 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
21228 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
21230 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
21232 2017-08-07 Marek Polacek <polacek@redhat.com>
21234 PR middle-end/81737
21235 * fold-const.c (fold_indirect_ref_1): Check type_domain.
21237 2017-08-07 Martin Liska <mliska@suse.cz>
21239 * attribs.h (canonicalize_attr_name): New function.
21240 (cmp_attribs): Move from c-format.c and adjusted.
21241 (is_attribute_p): Moved from tree.h.
21242 * tree-inline.c: Add new includes.
21243 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
21244 (private_is_attribute_p): Remove.
21245 (private_lookup_attribute): Likewise.
21246 (private_lookup_attribute_by_prefix): Simplify.
21247 (remove_attribute): Use is_attribute_p.
21248 * tree.h: Remove removed declarations.
21250 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21252 PR middle-end/81698
21253 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
21254 instead of computing it in the function. Formatting fix.
21255 (expand_case): Don't rely on default_edge being the first edge,
21256 clear it if removing it, pass default_edge to
21257 emit_case_dispatch_table.
21258 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
21261 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
21263 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
21264 insn in the function, emit NOP after the insn.
21266 2017-08-06 Tom de Vries <tom@codesourcery.com>
21268 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
21271 2017-08-06 Tom de Vries <tom@codesourcery.com>
21273 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
21276 2017-08-04 Yury Gribov <tetra2005@gmail.com>
21278 PR tree-optimization/57371
21279 * match.pd: New pattern.
21281 2017-08-04 Marek Polacek <polacek@redhat.com>
21283 PR middle-end/81695
21284 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
21285 perform the computation in offset_int.
21287 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21289 PR tree-optimization/81136
21290 * tree-vectorizer.h: Include tree-hash-traits.h.
21291 (vec_base_alignments): New typedef.
21292 (vec_info): Add a base_alignments field.
21293 (vect_record_base_alignments): Declare.
21294 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
21296 (DR_IS_CONDITIONAL_IN_STMT): New macro.
21297 (create_data_ref): Add an is_conditional_in_stmt argument.
21298 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
21299 the is_conditional_in_stmt field.
21300 (data_ref_loc): Add an is_conditional_in_stmt field.
21301 (get_references_in_stmt): Set the is_conditional_in_stmt field.
21302 (find_data_references_in_stmt): Update call to create_data_ref.
21303 (graphite_find_data_references_in_stmt): Likewise.
21304 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
21305 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
21306 (vect_record_base_alignment): New function.
21307 (vect_record_base_alignments): Likewise.
21308 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
21309 for nested statements even if we fail to compute a misalignment.
21310 Use pooled base alignments for unconditional references.
21311 (vect_find_same_alignment_drs): Compare base addresses instead
21313 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
21314 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
21316 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21318 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
21319 Add an explicit name for the enum. Use auto_vec for slp_instances
21320 and grouped_stores.
21321 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
21323 (_bb_vec_info): Add a constructor and destructor.
21324 (vinfo_for_stmt): Return NULL for uids of -1 as well.
21325 (destroy_loop_vec_info): Delete.
21326 (vect_destroy_datarefs): Likewise.
21327 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
21328 (vec_info::vec_info): New function.
21329 (vec_info::~vec_info): Likewise.
21330 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
21331 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
21332 destroy_loop_vec_info.
21333 * tree-vect-loop.c (new_loop_vec_info): Replace with...
21334 (_loop_vec_info::_loop_vec_info): ...this.
21335 (destroy_loop_vec_info): Replace with...
21336 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
21337 the stmt_vec_infos. Leave handling of vec_info information to its
21338 destructor. Remove explicit vector releases.
21339 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
21340 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
21341 * tree-vect-slp.c (new_bb_vec_info): Replace with...
21342 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
21343 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
21344 (destroy_bb_vec_info): Replace with...
21345 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
21346 information to its destructor.
21347 (vect_slp_analyze_bb_1): Use new and delete instead of
21348 new_bb_vec_info and destroy_bb_vec_info.
21349 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
21352 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21354 * tree-data-ref.h (subscript): Add access_fn field.
21355 (data_dependence_relation): Add could_be_independent_p.
21356 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
21357 (same_access_functions): Move to tree-data-ref.c.
21358 * tree-data-ref.c (ref_contains_union_access_p): New function.
21359 (access_fn_component_p): Likewise.
21360 (access_fn_components_comparable_p): Likewise.
21361 (dr_analyze_indices): Add a reference to access_fn_component_p.
21362 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
21364 (constant_access_functions): Likewise.
21365 (add_other_self_distances): Likewise.
21366 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
21367 (initialize_data_dependence_relation): Use XCNEW and remove
21368 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
21369 of access functions that have the same type. Allow the
21370 subsequence to end with different bases in some circumstances.
21371 Record the chosen access functions in SUB_ACCESS_FN.
21372 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
21373 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
21374 (subscript_dependence_tester_1): Likewise dra and drb.
21375 (build_classic_dist_vector): Update calls accordingly.
21376 (subscript_dependence_tester): Likewise.
21377 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
21378 DDR_COULD_BE_INDEPENDENT_P.
21379 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
21380 comp_alias_ddrs instead of may_alias_ddrs.
21381 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
21383 (vect_analyze_data_ref_dependence): Use it if
21384 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
21385 distance vectors if that fails.
21386 (dependence_distance_ge_vf): New function.
21387 (vect_prune_runtime_alias_test_list): Use it. Don't clear
21388 LOOP_VINFO_MAY_ALIAS_DDRS.
21390 2017-08-04 Richard Biener <rguenther@suse.de>
21392 PR middle-end/81705
21393 * fold-const.c (fold_binary_loc): Properly restrict
21394 minus_var0 && minus_var1 case when associating undefined overflow
21397 2017-08-04 Tom de Vries <tom@codesourcery.com>
21399 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
21401 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21403 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21404 Don't start diagnostic messages with a capital letter.
21405 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21407 (rs6000_invalid_builtin): Likewise.
21408 (rs6000_trampoline_init): Likewise.
21410 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21413 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
21414 after setting changeable df flags.
21416 2017-08-03 Richard Biener <rguenther@suse.de>
21418 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
21419 up if the use is in USE - X.
21421 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
21423 * toplev.c (dumpfile.h): New include.
21424 (internal_error_reentered): New static function. Use it...
21425 (internal_error_function): ...here to handle reentered internal_error.
21427 2017-08-03 Richard Biener <rguenther@suse.de>
21429 PR middle-end/81148
21430 * fold-const.c (split_tree): Add minus_var and minus_con
21431 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
21432 here but always use minus_*.
21433 (associate_trees): Assert we never associate with MINUS_EXPR
21434 and NULL first operand. Do not recurse for PLUS_EXPR operands
21435 when associating as MINUS_EXPR either.
21436 (fold_binary_loc): Track minus_var and minus_con.
21438 2017-08-03 Tom de Vries <tom@codesourcery.com>
21441 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
21442 ACCEL_COMPILER, apply finish_options on
21443 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
21445 2017-08-03 Tom de Vries <tom@codesourcery.com>
21448 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
21449 function_entry_patch_area_size > 0.
21451 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21454 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
21455 instead of 10??LU, perform unit multiplication in wide_int,
21456 don't change alloc_object_size_limit if the limit is larger
21459 PR tree-optimization/81655
21460 PR tree-optimization/81588
21461 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
21462 the case when ranges[i].low and high are 1 for unsigned type with
21465 PR middle-end/81052
21466 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
21467 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
21469 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21471 * tree-vrp.h: Add include guard.
21473 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
21476 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
21477 (ud2): New insn pattern.
21478 * config/i386/i386.c (ix86_expand_epilogue):
21479 For naked functions, generate ud2 instead of trap insn.
21481 2017-08-02 Marek Polacek <polacek@redhat.com>
21484 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
21486 2017-08-02 Tom de Vries <tom@codesourcery.com>
21487 Cesar Philippidis <cesar@codesourcery.com>
21489 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
21490 Add missing edge probabilities.
21492 2017-08-02 Tamar Christina <tamar.christina@arm.com>
21494 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
21495 Correct endianness.
21497 2017-08-02 Jakub Jelinek <jakub@redhat.com>
21499 PR middle-end/79499
21500 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
21501 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
21502 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
21503 prologue_seq sequences - if any.
21505 2017-08-02 Richard Biener <rguenther@suse.de>
21507 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
21508 via vectors if supported, integer extracts via punning if supported
21509 or otherwise vector extracts.
21511 2017-08-02 Richard Biener <rguenther@suse.de>
21513 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
21515 (bitmap_insert_into_set): ... this.
21517 2017-08-02 Richard Biener <rguenther@suse.de>
21519 PR tree-optimization/81633
21521 2015-08-17 Alan Hayward <alan.hayward@arm.com>
21523 PR tree-optimization/71752
21524 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
21526 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
21528 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
21529 (machine_function::call_ms2sysv_pad_out): Remove field.
21530 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
21531 (ix86_compute_frame_layout): Likewise.
21533 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
21536 * config/i386/i386.c (ix86_set_func_type): Disallow naked
21537 attribute with interrupt attribute.
21539 2017-08-01 Andrew Pinski <apinski@cavium.com>
21541 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
21542 BIT_INSERT_EXPR's operand 1
21543 to see if the types precision matches.
21545 2017-08-01 Martin Liska <mliska@suse.cz>
21547 PR middle-end/70140
21548 * builtins.c (expand_builtin_memcpy_args): Remove.
21549 (expand_builtin_memcpy): Call newly added function
21550 expand_builtin_memory_copy_args.
21551 (expand_builtin_memcpy_with_bounds): Likewise.
21552 (expand_builtin_mempcpy): Remove last argument.
21553 (expand_builtin_mempcpy_with_bounds): Likewise.
21554 (expand_builtin_memory_copy_args): New function created from
21555 expand_builtin_mempcpy_args with small modifications.
21556 (expand_builtin_mempcpy_args): Remove.
21557 (expand_builtin_stpcpy): Remove unused argument.
21558 (expand_builtin): Likewise.
21559 (expand_builtin_with_bounds): Likewise.
21561 2017-08-01 Martin Liska <mliska@suse.cz>
21564 Make mempcpy more optimal (PR middle-end/70140).
21566 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21569 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
21570 __builtin_vec_cmpne verify both arguments are compatible vectors
21571 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
21572 verify arg1_type is a pointer or array type. For __builtin_vec_st,
21573 move computation of aligned to after checking the argument types.
21577 * config/rs6000/vsx.md (vextract_fp_from_shorth,
21578 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
21581 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
21582 Doug Rupp <rupp@adacore.com>
21583 Olivier Hainque <hainque@adacore.com>
21585 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
21586 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
21588 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
21589 for TARGET_OS_CPP_BUILTIN.
21590 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
21591 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
21593 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
21594 passing required abi options to the assembler for EABI configurations.
21595 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
21596 of .text.hot and .text.unlikely sections for kernel modules when
21597 using ARM style exceptions.
21598 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
21599 options. Add EXTRA_CC1_SPEC.
21600 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
21602 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
21604 (ARM_TARGET2_DWARF_FORMAT): Define.
21605 * config/arm/t-vxworks: Adjust multilib control to removal of the
21606 Diab command line options.
21608 2017-08-01 Martin Liska <mliska@suse.cz>
21610 PR gcov-profile/81561
21611 * gcov.c (unblock): Make unblocking safe as we need to preserve
21612 index correspondence of blocks and block_lists.
21614 2017-08-01 Richard Biener <rguenther@suse.de>
21616 PR tree-optimization/81181
21617 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21618 (compute_antic): ... end of iteration here.
21620 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21622 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21623 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21624 (ftree-slp-vectorize): Likewise.
21625 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21626 can no longer be set independent of flag_tree_loop_vectorize.
21627 * omp-general.c (emp_max_vf): Likewise.
21628 * opts.c (enable_fdo_optimizations): Remove references to
21629 flag_tree_vectorize, these are now implicit.
21630 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21631 and leave it for the options machinery.
21633 2017-08-01 Martin Liska <mliska@suse.cz>
21635 PR middle-end/70140
21636 * builtins.c (expand_builtin_memcpy_args): Remove.
21637 (expand_builtin_memcpy): Call newly added function
21638 expand_builtin_memory_copy_args.
21639 (expand_builtin_memcpy_with_bounds): Likewise.
21640 (expand_builtin_mempcpy): Remove last argument.
21641 (expand_builtin_mempcpy_with_bounds): Likewise.
21642 (expand_builtin_memory_copy_args): New function created from
21643 expand_builtin_mempcpy_args with small modifications.
21644 (expand_builtin_mempcpy_args): Remove.
21645 (expand_builtin_stpcpy): Remove unused argument.
21646 (expand_builtin): Likewise.
21647 (expand_builtin_with_bounds): Likewise.
21649 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21652 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21653 print "ds:" only for immediates in generic address space.
21655 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21658 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21659 (ix86_function_ok_for_sibcall): Return false for naked functions.
21661 2017-08-01 Richard Biener <rguenther@suse.de>
21663 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21664 (compute_antic): Seed worklist with exit block predecessors.
21665 * cfganal.c (dfs_find_deadend): For a cycle return the source
21666 of the edge closing it.
21668 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21670 * config/aarch64/aarch64.c
21671 (aarch64_can_const_movi_rtx_p): Move 0 check.
21673 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21675 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21676 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21678 * match.pd: Ditto in address comparison pattern.
21680 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21682 PR tree-optimization/81627
21683 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21684 closed ssa form for store-store chain.
21686 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21688 PR tree-optimization/81620
21689 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21690 for store-store chain.
21692 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21694 PR tree-optimization/81588
21695 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21696 ranges[i].in_p, invert comparison code ccode. For >/>=,
21697 swap rhs1 and rhs2 and comparison code unconditionally,
21698 for </<= don't do that. Don't swap rhs1/rhs2 again if
21699 ranges[i].in_p, instead invert comparison code ccode if
21700 opcode or oe->rank is BIT_IOR_EXPR.
21703 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21704 a direct optab to conversion optab.
21705 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21706 with GET_MODE_INNER as last argument instead of optab_handler.
21707 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21708 vector extraction if possible and optab is available.
21709 * expr.c (store_constructor): Use convert_optab_handler instead
21710 of optab_handler. Use vector initialization from smaller
21711 vectors if possible and optab is available.
21712 * tree-vect-stmts.c (vectorizable_load): Likewise.
21713 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21714 now have two modes.
21715 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21716 of vec_init from half-sized vectors with the same element mode.
21717 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21718 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21719 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21720 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21721 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21722 after mode in gen_vec_extract* calls.
21723 (vec_extract<mode>): Renamed to ...
21724 (vec_extract<mode><ssescalarmodelower>): ... this.
21725 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21726 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21727 element mode after mode in gen_vec_init* calls.
21728 (VEC_INIT_HALF_MODE): New mode iterator.
21729 (vec_init<mode>): Renamed to ...
21730 (vec_init<mode><ssescalarmodelower>): ... this.
21731 (vec_init<mode><ssehalfvecmodelower>): New expander.
21732 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21733 (vec_extractv2sfsf): ... this.
21734 (vec_initv2sf): Renamed to ...
21735 (vec_initv2sfsf): ... this.
21736 (vec_extractv2si): Renamed to ...
21737 (vec_extractv2sisi): ... this.
21738 (vec_initv2si): Renamed to ...
21739 (vec_initv2sisi): ... this.
21740 (vec_extractv4hi): Renamed to ...
21741 (vec_extractv4hihi): ... this.
21742 (vec_initv4hi): Renamed to ...
21743 (vec_initv4hihi): ... this.
21744 (vec_extractv8qi): Renamed to ...
21745 (vec_extractv8qiqi): ... this.
21746 (vec_initv8qi): Renamed to ...
21747 (vec_initv8qiqi): ... this.
21748 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21749 (vec_init<mode>): Renamed to ...
21750 (vec_init<mode><VEC_base_l>): ... this.
21751 (vec_extract<mode>): Renamed to ...
21752 (vec_extract<mode><VEC_base_l>): ... this.
21753 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21754 (vec_initv2sfsf): ... this.
21755 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21756 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21757 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21758 element mode after mode in gen_vec_init* calls.
21759 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21760 (vec_init<mode><Vel>): ... this.
21761 (vec_extract<mode>): Renamed to ...
21762 (vec_extract<mode><Vel>): ... this.
21763 * config/aarch64/iterators.md (Vel): New mode attribute.
21764 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21765 Add element mode after mode in gen_vec_extract* calls.
21766 * config/s390/vector.md (non_vec_l): New mode attribute.
21767 (vec_extract<mode>): Renamed to ...
21768 (vec_extract<mode><non_vec_l>): ... this.
21769 (vec_init<mode>): Renamed to ...
21770 (vec_init<mode><non_vec_l>): ... this.
21771 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21772 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21774 * config/arm/iterators.md (V_elem_l): New mode attribute.
21775 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21776 (vec_extract<mode><V_elem_l>): ... this.
21777 (vec_extractv2di): Renamed to ...
21778 (vec_extractv2didi): ... this.
21779 (vec_init<mode>): Renamed to ...
21780 (vec_init<mode><V_elem_l>): ... this.
21781 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21782 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21783 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21784 Add element mode after gen_vec_extract* calls.
21785 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21786 (vec_init<mode><unitmode>): ... this.
21787 (vec_extract<mode>): Renamed to ...
21788 (vec_extract<mode><unitmode>): ... this.
21789 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21790 (vec_init<mode><unitmode>): ... this.
21791 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21792 (vec_initv2sfsf): ... this.
21793 (vec_extractv2sf): Renamed to ...
21794 (vec_extractv2sfsf): ... this.
21795 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21796 Add element mode after gen_vec_extract* calls.
21797 * config/mips/mips.md (unitmode): New mode iterator.
21798 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21799 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21800 * config/spu/spu.md (inner_l): New mode attribute.
21801 (vec_init<mode>): Renamed to ...
21802 (vec_init<mode><inner_l>): ... this.
21803 (vec_extract<mode>): Renamed to ...
21804 (vec_extract<mode><inner_l>): ... this.
21805 * config/sparc/sparc.md (veltmode): New mode iterator.
21806 (vec_init<VMALL:mode>): Renamed to ...
21807 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21808 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21809 (vec_initv2sisi): ... this.
21810 (vec_initv2sf): Renamed to ...
21811 (vec_initv2sfsf): ... this.
21812 (vec_extractv2sf): Renamed to ...
21813 (vec_extractv2sfsf): ... this.
21814 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21815 (vec_init<mode>): Renamed to ...
21816 (vec_init<mode><VEC_base_l>): ... this.
21817 (vec_extract<mode>): Renamed to ...
21818 (vec_extract<mode><VEC_base_l>): ... this.
21819 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21820 (vec_initv2sfsf): ... this.
21821 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21822 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21823 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21824 gen_vec_init* calls.
21826 2017-08-01 Richard Biener <rguenther@suse.de>
21828 PR tree-optimization/81297
21829 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21830 TREE_OVERFLOW from INTEGER_CSTs.
21832 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21834 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21836 2017-07-31 Carl Love <cel@us.ibm.com>
21838 * config/rs6000/rs6000-c: Add support for built-in functions
21839 vector signed char vec_xl_be (signed long long, signed char *);
21840 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21841 vector signed int vec_xl_be (signed long long, signed int *);
21842 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21843 vector signed long long vec_xl_be (signed long long, signed long long *);
21844 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21845 vector signed short vec_xl_be (signed long long, signed short *);
21846 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21847 vector double vec_xl_be (signed long long, double *);
21848 vector float vec_xl_be (signed long long, float *);
21849 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21850 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21851 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21853 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21854 (altivec_expand_builtin): Add switch statement to call
21855 altivec_expand_xl_be for each builtin.
21856 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21857 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21858 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21859 __builtin_vsx_le_be_v16qi.
21860 * doc/extend.texi: Update the built-in documentation file for the
21861 new built-in functions.
21863 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21866 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
21868 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21870 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21872 * config.gcc: Add z14.
21873 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
21874 CPU model numbers for z13s and z14.
21875 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
21877 * config/s390/s390-opts.h (enum processor_type): Rename
21878 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21879 * config/s390/s390.c (processor_table): Add field for CPU name to
21880 be passed to Binutils.
21881 (s390_asm_output_machine_for_arch): Use the new field in
21882 processor_table for Binutils.
21883 (s390_expand_builtin): Replace arch12 with z14.
21884 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21885 (s390_get_sched_attrmask): Likewise.
21886 (s390_get_unit_mask): Likewise.
21887 * config/s390/s390.opt: Add z14 to processor_type enum.
21889 2017-07-31 Martin Jambor <mjambor@suse.cz>
21892 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
21893 regardless of optimization level.
21895 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
21896 Martin Liska <mliska@suse.cz>
21898 * predict.def: Remove old comment and adjust probability.
21899 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
21900 PREDICT statements.
21902 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21905 * config/i386/i386.c (ix86_function_naked): New function.
21906 (ix86_can_use_return_insn_p): Return false for naked functions.
21907 (ix86_expand_prologue): Skip prologue for naked functions.
21908 (ix86_expand_epilogue): Skip epilogue for naked functions
21909 and emit trap instruction.
21910 (ix86_warn_func_return): New function.
21911 (ix86_attribute_table): Add "naked" attribute specification.
21912 (TARGET_WARN_FUNC_RETURN): Define.
21913 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
21915 2017-07-31 Martin Liska <mliska@suse.cz>
21917 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
21918 (dump_gimple_bb_header): Always dump BB info.
21919 (pp_cfg_jump): Do not append info about BB when dumping a jump.
21921 2017-07-31 Martin Liska <mliska@suse.cz>
21924 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
21925 also with current_function_decl non-null equality.
21927 2017-07-31 Jakub Jelinek <jakub@redhat.com>
21930 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
21931 change type to the element type, instead add eltype variable and
21932 use it where we are interested in the element type.
21934 PR tree-optimization/81603
21935 * ipa-polymorphic-call.c
21936 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
21937 offset arithmetic in offset_int, bail out if the resulting bit offset
21938 doesn't fit into shwi.
21940 2017-07-31 Martin Liska <mliska@suse.cz>
21942 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
21943 (gimplify_save_expr): Fix comment.
21945 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
21948 * config/i386/i386.c (ix86_function_arg): Update arguments for
21950 (ix86_compute_frame_layout): Set the initial stack offset to
21951 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
21952 INCOMING_FRAME_SP_OFFSET.
21953 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
21954 stack before exception handler returns.
21955 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
21956 the 'ERROR_CODE' for exception handler.
21958 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
21960 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
21961 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
21962 (ASM_OUTPUT_REG_POP): Ditto.
21963 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
21964 instead of asm_fprintf to output pure string.
21966 2017-07-29 Jakub Jelinek <jakub@redhat.com>
21968 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
21969 to imported_module_or_decl hook.
21970 (debug_nothing_tree_tree_tree_bool): Remove.
21971 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
21972 * debug.c (do_nothing_debug_hooks): Use
21973 debug_nothing_tree_tree_tree_bool_bool instead of
21974 debug_nothing_tree_tree_tree_bool.
21975 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21976 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
21977 * sdbout.c (sdb_debug_hooks): Likewise.
21978 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
21979 (gen_namespace_die): Add DW_AT_export_symbols attribute if
21981 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
21982 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
21983 attribute, don't add anything.
21985 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21987 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
21988 (fold_build2_stat_loc): Likewise.
21989 (fold_build3_stat_loc): Likewise.
21990 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
21991 (fold_build1_loc): Remove macro.
21992 (fold_build2_loc): Likewise.
21993 (fold_build3_loc): Likewise.
21995 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21997 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
21998 (gimple_build_debug_bind_source_stat): Likewise.
21999 * gimple.h (gimple_build_debug_bind): Remove macro.
22000 (gimple_build_debug_bind_source): Likewise.
22002 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22004 * bitmap.c (bitmap_alloc): Adjust.
22005 (bitmap_gc_alloc): Likewise.
22006 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22008 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22010 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22011 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22012 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22013 (bitmap_gc_alloc_stat): Likewise.
22014 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22016 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22018 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22019 * rtl.h (shallow_copy_rtx): Remove macro.
22021 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22023 * emit-rtl.c (gen_raw_REG): Adjust.
22024 * gengenrtl.c (gendef): Likewise.
22025 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22026 * rtl.h (rtx_alloc): Remove macro.
22028 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22030 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22031 (build_tree_list_stat): Likewise.
22032 * tree.h (build_tree_list): Remove macro.
22033 (build_tree_list_vec): Likewise.
22035 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22037 * tree.c (make_vector_stat): Remove _stat from name.
22038 (build_vector_stat): Likewise.
22039 * tree.h (make_vector_stat): Remove macro.
22040 (build_vector_stat): Likewise.
22042 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22044 * tree.h (build_var_debug_value): Remove prototype.
22046 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22048 * tree.c (tree_cons_stat): Remove _stat from name.
22049 * tree.h (tree_cons): Remove macro.
22051 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22053 * tree.c (build_vl_exp_stat): Remove _stat from name.
22054 * tree.h (build_vl_exp): Remove macro.
22056 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22058 * tree.c (build_decl_stat): Remove _stat from name.
22059 * tree.h (build_decl): Remove macro.
22061 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22063 * gimple.c (gimple_build_with_ops_stat): Adjust.
22064 (gimple_alloc_stat): Remove _stat from name.
22065 * gimple.h (gimple_alloc): Remove macro.
22067 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22069 * tree.c (make_tree_vec_stat): Remove _stat from name.
22070 (grow_tree_vec_stat): Likewise.
22071 * tree.h (make_tree_vec_stat): Adjust prototype.
22072 (grow_tree_vec_stat): Likewise.
22073 (make_tree_vec): Remove macro.
22074 (grow_tree_vec): Likewise.
22076 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22078 * fold-const.c (fold_build1_stat_loc): Adjust.
22079 (fold_build2_stat_loc): Likewise.
22080 (fold_build3_stat_loc): Likewise.
22081 * tree.c (build0_stat): Remove _stat from name.
22082 (build1_stat): Likewise.
22083 (build2_stat): Likewise.
22084 (build3_stat): Likewise.
22085 (build4_stat): Likewise.
22086 (build5_stat): Likewise.
22087 * tree.h (build1_loc): Remove macro, and rename _stat function
22089 (build2_loc): Likewise.
22090 (build3_loc): Likewise.
22091 (build4_loc): Likewise.
22092 (build5_loc): Likewise.
22094 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22096 * tree.c (make_int_cst_stat): Remove _stat from name.
22097 * tree.h (make_int_cst_stat): Adjust prototype.
22098 (make_int_cst): Remove macro.
22100 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22102 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22103 * tree.h (make_tree_binfo_stat): Adjust prototype.
22104 (make_tree_binfo): Remove.
22106 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22108 * tree.c (copy_node_stat): Rename to copy_node.
22109 (build_distinct_type_copy): Adjust.
22110 * tree.h (copy_node_stat): Adjust prototype.
22111 (copy_node): Remove macro.
22113 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22115 * tree.c (make_node_stat): rename to make_node.
22116 (build_tree_list_stat): Adjust.
22117 (build0_stat): Likewise.
22118 (build2_stat): Likewise.
22119 (build3_stat): Likewise.
22120 (build4_stat): Likewise.
22121 (build5_stat): Likewise.
22122 (build_decl_stat): Likewise.
22123 * tree.h (make_node_stat): Adjust prototype.
22124 (make_node): remove macro.
22126 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
22128 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
22129 (PPC_FEATURE2_SCV): Likewise.
22130 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
22132 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22134 * config/aarch64/aarch64.c
22135 (aarch64_internal_mov_immediate): Add new special pattern.
22136 * config/aarch64/aarch64.md (*movdi_aarch64):
22137 Add reg/32bit const mov case.
22139 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22140 Richard Sandiford <richard.sandiford@linaro.org>
22142 * config/aarch64/aarch64.md (mov<mode>): Generalize.
22143 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
22144 Add integer and movi cases.
22145 (movi-split-hf-df-sf split, fp16): New.
22146 (enabled): Added TARGET_FP_F16INST.
22147 * config/aarch64/iterators.md (GPF_HF): New.
22148 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
22150 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22152 * config/aarch64/aarch64.c
22153 (aarch64_simd_container_mode): Add prototype.
22154 (aarch64_expand_mov_immediate): Add HI support.
22155 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
22156 (aarch64_can_const_movi_rtx_p): New.
22157 (aarch64_preferred_reload_class):
22158 Remove restrictions of using FP registers for certain SIMD operations.
22159 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
22160 (aarch64_valid_floating_const): Add integer move validation.
22161 (aarch64_simd_imm_scalar_p): Remove.
22162 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
22163 (aarch64_legitimate_constant_p): Expand list of supported cases.
22164 * config/aarch64/aarch64-protos.h
22165 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
22166 (aarch64_reinterpret_float_as_int): New.
22167 (aarch64_simd_imm_scalar_p): Remove.
22168 * config/aarch64/constraints.md (Uvi): New.
22169 (Dd): Split into Ds and new Dd.
22170 * config/aarch64/aarch64.md (*movsi_aarch64):
22172 (*movdi_aarch64): Add SIMD mov case.
22174 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22176 * tree-predcom.c: (struct chain): Handle store-store chain in which
22177 stores for elimination only store loop invariant values.
22178 (execute_pred_commoning_chain): Ditto.
22179 (prepare_initializers_chain_store_elim): Ditto.
22180 (prepare_finalizers): Ditto.
22181 (is_inv_store_elimination_chain): New function.
22182 (initialize_root_vars_store_elim_1): New function.
22184 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22186 * tree-predcom.c: Revise general description of the pass.
22187 (enum chain_type): New enum type for store elimination.
22188 (struct chain): New field supporting store elimination.
22189 (struct component): Ditto.
22190 (dump_chain): Dump store-stores chain.
22191 (release_chain): Release resources.
22192 (split_data_refs_to_components): Compute and create component
22193 contains only stores for elimination.
22194 (get_chain_last_ref_at): New function.
22195 (make_invariant_chain): Initialization.
22196 (make_rooted_chain): Specify chain type in parameter and record it.
22197 (add_looparound_copies): Skip for store-stores chain.
22198 (determine_roots_comp): Compute type of chain and pass it to
22200 (initialize_root_vars_store_elim_2): New function.
22201 (finalize_eliminated_stores): New function.
22202 (remove_stmt): Handle store for elimination.
22203 (execute_pred_commoning_chain): Execute predictive commoning on
22204 store-store chains.
22205 (determine_unroll_factor): Skip unroll for store-stores chain.
22206 (prepare_initializers_chain_store_elim): New function.
22207 (prepare_initializers_chain): Hanlde store-store chain.
22208 (prepare_finalizers_chain, prepare_finalizers): New function.
22209 (tree_predictive_commoning_loop): Return integer value indicating
22210 if loop is unrolled or lcssa form is corrupted.
22211 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
22213 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22215 * tree-predcom.c (initialize_root): Delete.
22216 (execute_pred_commoning_chain): Initialize root vars and replace
22217 reference of non-combined chain directly, rather than call above
22220 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22222 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
22223 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
22225 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22227 * tree-predcom.c (struct chain): New field init_seq.
22228 (release_chain): Release init_seq.
22229 (prepare_initializers_chain): Record intialization stmts in above
22231 (insert_init_seqs): New function.
22232 (tree_predictive_commoning_loop): Call insert_init_seqs.
22234 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22236 * tree-predcom.c (determine_roots_comp): Skip trivial components.
22238 2017-07-28 Richard Biener <rguenther@suse.de>
22240 * match.pd: Remove superfluous :c.
22241 * genmatch.c (simplify::id): Add member.
22242 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
22244 (current_id): New global.
22245 (dt_node::parent): Move from ...
22246 (dt_operand::parent): ... here. Add for_id member.
22247 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
22248 (decision_tree::find_node): Relax order requirement when
22249 merging DT_TRUE nodes to ones inbetween the current simplify
22250 and the one we try to merge with. Add diagnostic whenever
22251 we need to enforce pattern order by not merging.
22252 (decision_tree::insert): Set current_id.
22253 (decision_tree::print_node): Dump parent node and for_id.
22254 (parser::last_id): Add member.
22255 (parser::push_simplify): Assign unique id.
22256 (parser::parser): Initialize last_id.
22258 2017-07-28 Martin Liska <mliska@suse.cz>
22261 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
22262 gimple_build_debug_bind.
22264 2017-07-28 Richard Biener <rguenther@suse.de>
22266 PR tree-optimization/81502
22267 * match.pd: Add pattern combining BIT_INSERT_EXPR with
22269 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
22271 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
22272 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
22273 for BIT_FIELD_REF args.
22274 * fold-const.c (make_bit_field_ref): Likewise.
22275 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
22277 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22280 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
22281 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
22282 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
22283 Or it into SANITIZER_UNDEFINED.
22284 * ubsan.c: Include gimple-fold.h and varasm.h.
22285 (ubsan_expand_ptr_ifn): New function.
22286 (instrument_pointer_overflow): New function.
22287 (maybe_instrument_pointer_overflow): New function.
22288 (instrument_object_size): Formatting fix.
22289 (pass_ubsan::execute): Call instrument_pointer_overflow
22290 and maybe_instrument_pointer_overflow.
22291 * internal-fn.c (expand_UBSAN_PTR): New function.
22292 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
22293 * sanitizer.def (__ubsan_handle_pointer_overflow,
22294 __ubsan_handle_pointer_overflow_abort): New builtins.
22295 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
22296 * internal-fn.def (UBSAN_PTR): New internal function.
22297 * opts.c (sanitizer_opts): Add pointer-overflow.
22298 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
22299 * fold-const.c (build_range_check): Compute pointer range check in
22300 integral type if pointer arithmetics would be needed. Formatting
22303 2017-07-28 Martin Liska <mliska@suse.cz>
22306 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
22307 parameters that are of a variable-length.
22309 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
22311 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
22313 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
22314 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
22315 (CRT_CALL_STATIC_FUNCTION): Likewise.
22316 (ASM_DEFAULT_SPEC): New define.
22317 (ASM_SPEC32): Likewise.
22318 (ASM_SPEC64): Likewise.
22319 (ASM_SPEC_COMMON): Likewise.
22320 (ASM_SPEC): Likewise.
22321 (INVALID_64BIT): Likewise.
22322 (LINK_OS_DEFAULT_SPEC): Likewise.
22323 (LINK_OS_SPEC32): Likewise.
22324 (LINK_OS_SPEC64): Likewise.
22325 (POWERPC_LINUX): Likewise.
22326 (PTRDIFF_TYPE): Likewise.
22327 (RESTORE_FP_PREFIX): Likewise.
22328 (RESTORE_FP_SUFFIX): Likewise.
22329 (SAVE_FP_PREFIX): Likewise.
22330 (SAVE_FP_SUFFIX): Likewise.
22331 (SIZE_TYPE): Likewise.
22332 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
22333 (TARGET_64BIT): Likewise.
22334 (TARGET_64BIT): Likewise.
22335 (TARGET_AIX): Likewise.
22336 (WCHAR_TYPE_SIZE): Likewise.
22337 (WCHAR_TYPE): Undefine.
22338 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
22339 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
22340 (CPP_OS_RTEMS_SPEC): Delete.
22341 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
22342 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
22344 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
22346 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22348 PR tree-optimization/81578
22349 * tree-parloops.c (build_new_reduction): Bail out if
22350 reduction_code isn't one of the standard OpenMP reductions.
22351 Move the details printing after that decision.
22353 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22355 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
22356 related to reload_in_progress.
22357 (splat_input_operand): Likewise.
22358 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
22360 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
22362 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
22363 (TARGET_INSTANTIATE_DECLS): Likewise.
22364 (legitimate_indexed_address_p): Delete reload_in_progress code.
22365 (rs6000_debug_legitimate_address_p): Likewise.
22366 (rs6000_eliminate_indexed_memrefs): Likewise.
22367 (rs6000_emit_le_vsx_store): Likewise.
22368 (rs6000_emit_move_si_sf_subreg): Likewise.
22369 (rs6000_emit_move): Likewise.
22370 (register_to_reg_type): Likewise.
22371 (rs6000_pre_atomic_barrier): Likewise.
22372 (rs6000_machopic_legitimize_pic_address): Likewise.
22373 (rs6000_allocate_stack_temp): Likewise.
22374 (rs6000_address_for_fpconvert): Likewise.
22375 (rs6000_address_for_altivec): Likewise.
22376 (rs6000_secondary_memory_needed_rtx): Delete function.
22377 (rs6000_check_sdmode): Likewise.
22378 (rs6000_alloc_sdmode_stack_slot): Likewise.
22379 (rs6000_instantiate_decls): Likewise.
22380 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
22381 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
22382 Delete reload_in_progress.
22383 (*vec_reload_and_plus_<mptrsize>): Likewise.
22384 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
22385 (vsx_div_v2di): Likewise.
22386 (vsx_udiv_v2di): Likewise.
22388 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22390 * config/rs6000/rs6000.opt (mlra): Replace with stub.
22391 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
22392 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
22393 (rs6000_debug_reg_global): Delete print of LRA status.
22394 (rs6000_option_override_internal): Delete dead LRA related code.
22395 (rs6000_lra_p): Delete function.
22396 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
22398 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
22400 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
22401 * config/riscv/rtems.h: New file.
22403 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22404 Sudakshina Das <sudi.das@arm.com>
22406 * config/aarch64/aarch64.md
22407 (define_split for and<mode>3nr_compare): Move
22408 non aarch64_logical_operand to a register.
22409 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
22410 register immediate operand to a register.
22411 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
22413 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22415 PR middle-end/81564
22416 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
22418 2017-07-27 Richard Biener <rguenther@suse.de>
22420 PR tree-optimization/81573
22421 PR tree-optimization/81494
22422 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
22423 multi defuse cycle case.
22425 2017-07-27 Richard Biener <rguenther@suse.de>
22427 PR tree-optimization/81571
22428 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
22431 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
22433 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
22434 earlier and only if MASK_FPU is set. Adjust formatting.
22436 2017-07-27 Martin Liska <mliska@suse.cz>
22438 * opt-functions.awk: Add validation of value of Init.
22439 * optc-gen.awk: Pass new argument.
22441 2017-07-27 Martin Liska <mliska@suse.cz>
22443 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
22444 Fix wrong condition.
22446 2017-07-27 Martin Liska <mliska@suse.cz>
22448 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
22449 BBs and edges seen by autoFDO.
22451 2017-07-27 Richard Biener <rguenther@suse.de>
22453 PR tree-optimization/81502
22454 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
22455 with incompatible but same sized type.
22456 (execute_update_addresses_taken): Likewise.
22458 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
22460 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
22461 flag_tree_loop_vectorize rather than flag_tree_vectorize.
22463 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22466 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
22467 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
22468 Change s_operand to memory_operand.
22470 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
22472 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
22473 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
22474 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
22475 Emit instructions rather than returning an expression. Handle TFmode
22476 and KFmode by casting to TImode.
22477 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
22478 (rs6000_emit_le_vsx_store): Likewise.
22479 * config/rs6000/vsx.md (VSX_TI): New iterator.
22480 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
22481 (*vsx_le_undo_permute_<mode>): Likewise.
22482 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
22483 emit the split sequence.
22484 (*vsx_le_perm_store_<mode>): Likewise.
22486 2017-07-27 Jakub Jelinek <jakub@redhat.com>
22488 PR tree-optimization/81555
22489 PR tree-optimization/81556
22490 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
22491 if true, force CHANGED for the recursive invocation.
22492 (reassociate_bb): Remember original length of ops array, pass
22493 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
22495 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
22496 attributes for noipa attribute. For naked attribute use
22497 lookup_attribute first before lookup_attribute_spec.
22498 * final.c (rest_of_handle_final): Disable IPA RA for functions with
22500 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
22501 for functions with noipa attribute.
22502 (cgraph_externally_visible_p): Return true for functions with noipa
22504 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
22505 for functions with noipa attribute.
22506 * doc/extend.texi: Document noipa function attribute.
22507 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
22508 also for functions with noipa attribute.
22509 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
22511 2017-07-26 Andrew Pinski <apinski@cavium.com>
22513 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
22514 vec_unalign_load_cost and vec_unalign_store_cost.
22516 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
22518 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
22519 -mvsx-small-integer option.
22520 (ISA_3_0_MASKS_IEEE): Likewise.
22521 (OTHER_VSX_VECTOR_MASKS): Likewise.
22522 (POWERPC_MASKS): Likewise.
22523 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
22524 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
22525 code, only testing for DImode being allowed in non-VSX floating
22527 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
22528 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
22530 (rs6000_option_override_internal): Delete -mvsx-small-integer.
22531 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
22532 TARGET_P8_VECTOR test.
22533 (rs6000_secondary_reload_simple_move): Likewise.
22534 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
22535 since TARGET_P9_VECTOR was already tested.
22536 (rs6000_opt_masks): Remove -mvsx-small-integer.
22537 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
22538 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22540 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
22541 test for TARGET_VEXTRACTUB was used, and that uses
22543 (p9 extract splitter): Likewise.
22544 (vsx_extract_<mode>_di_p9): Likewise.
22545 (vsx_extract_<mode>_store_p9): Likewise.
22546 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
22547 for TARGET_P9_VECTOR was used. Delete code that is now dead with
22548 the elimination of TARGET_VSX_SMALL_INTEGER.
22549 (vsx_extract_<mode>_p8): Likewise.
22550 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
22551 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
22552 (vsx_set_<mode>_p9): Likewise.
22553 (vsx_set_v4sf_p9): Likewise.
22554 (vsx_set_v4sf_p9_zero): Likewise.
22555 (vsx_insert_extract_v4sf_p9): Likewise.
22556 (vsx_insert_extract_v4sf_p9_2): Likewise.
22557 * config/rs6000/rs6000.md (sign extend splitter): Change
22558 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
22559 (floatsi<mode>2_lfiwax_mem): Likewise.
22560 (floatunssi<mode>2_lfiwzx_mem): Likewise.
22561 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
22562 since a test for TARGET_P9_VECTOR was used.
22563 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22564 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
22565 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22566 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22567 TARGET_P8_VECTOR test.
22568 (fix_trunc<mode>si2_stfiwx): Likewise.
22569 (fix_trunc<mode>si2_internal): Likewise.
22570 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
22571 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22573 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22574 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22575 TARGET_P8_VECTOR test.
22576 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22577 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
22578 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22580 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22581 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
22582 since a test for TARGET_P9_VECTOR was used.
22583 (splitter for loading small constants): Likewise.
22585 2017-07-26 Andrew Pinski <apinski@cavium.com>
22587 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
22590 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
22593 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
22594 (fp_valid_at): Likewise.
22596 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22598 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
22599 (qdf24xx_addrcost_table): Likewise.
22600 (cortexa57_tunings): Update to use generic_branch_cost.
22601 (cortexa72_tunings): Likewise.
22602 (cortexa73_tunings): Likewise.
22603 (qdf24xx_tunings): Likewise.
22605 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22607 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
22608 (thunderx2t99_branch_cost): Likewise.
22609 (cortexa35_tunings): Update to use generic_branch_cost.
22610 (cortexa53_tunings): Likewise.
22611 (cortexa57_tunings): Likewise.
22612 (cortexa72_tunings): Likewise.
22613 (cortexa73_tunings): Likewise.
22614 (thunderx2t99_tunings): Likewise.
22616 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22618 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22619 (sparc_option_override): Honour MASK_FSMULD.
22620 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22621 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22622 * config/sparc/sparc.opt (mfsmuld): New option.
22623 * doc/invoke.texi (mfsmuld): Document option.
22625 2017-07-26 Marek Polacek <polacek@redhat.com>
22627 PR middle-end/70992
22628 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22630 2017-07-26 Richard Biener <rguenther@suse.de>
22632 * gimple-match-head.c (do_valueize): Return OP if valueize
22634 (get_def): New helper to get at the def stmt of a SSA name
22635 if valueize allows.
22636 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22637 do_valueize to get at the def stmt.
22638 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22640 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22642 PR middle-end/46932
22643 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22645 2017-07-26 Martin Liska <mliska@suse.cz>
22648 * function.c (expand_function_start): Make expansion of
22649 nonlocal_goto_save_area after parm_birth_insn.
22651 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22653 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22654 from all CPU target flags enable members.
22656 2017-07-26 Richard Biener <rguenther@suse.de>
22658 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22659 (decision_tree::gen): Make 'type' const.
22660 (write_predicate): Likewise.
22662 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22664 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22665 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22666 (rs6000_option_override_internal): Likewise.
22667 (rs6000_expand_vector_set): Likewise.
22668 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22669 (TARGET_UPPER_REGS_SF): Likewise.
22670 (TARGET_UPPER_REGS_DI): Likewise.
22671 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22672 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22673 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22674 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22675 (Splitters for DI constants in Altivec registers): Likewise.
22676 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22677 (vsx_set_v4sf_p9): Likewise.
22678 (vsx_set_v4sf_p9_zero): Likewise.
22679 (vsx_insert_extract_v4sf_p9): Likewise.
22680 (vsx_insert_extract_v4sf_p9_2): Likewise.
22682 2017-07-25 Carl Love <cel@us.ibm.com>
22684 * doc/extend.texi: Update the built-in documentation file for the
22685 existing built-in functions
22686 vector signed char vec_cnttz (vector signed char);
22687 vector unsigned char vec_cnttz (vector unsigned char);
22688 vector signed short vec_cnttz (vector signed short);
22689 vector unsigned short vec_cnttz (vector unsigned short);
22690 vector signed int vec_cnttz (vector signed int);
22691 vector unsigned int vec_cnttz (vector unsigned int);
22692 vector signed long long vec_cnttz (vector signed long long);
22693 vector unsigned long long vec_cnttz (vector unsigned long long);
22695 2017-07-25 Andrew Pinski <apinski@cavium.com>
22697 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22698 accesses where the use is for the first operand of a BIT_INSERT.
22700 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22703 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22704 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22706 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22708 * config/i386/gstabs.h: Delete.
22709 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22711 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22713 * config/i386/i386.c (ix86_decompose_address): Do not check for
22714 register RTX when looking at index_reg or base_reg.
22715 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22717 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22719 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22720 to update EH info here.
22722 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22724 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22726 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22728 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22730 2017-07-25 Torsten Duwe <duwe@suse.de>
22732 * common.opt: Introduce -fpatchable-function-entry
22733 command line option, and its variables function_entry_patch_area_size
22734 and function_entry_patch_area_start.
22735 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22736 including a two-value parser.
22737 * target.def (print_patchable_function_entry): New target hook.
22738 * targhooks.h (default_print_patchable_function_entry): New function.
22739 * targhooks.c (default_print_patchable_function_entry): Likewise.
22740 * toplev.c (process_options): Switch off IPA-RA if
22741 patchable function entries are being generated.
22742 * varasm.c (assemble_start_function): Look at the
22743 patchable-function-entry command line switch and current
22744 function attributes and maybe generate NOP instructions by
22745 calling the print_patchable_function_entry hook.
22746 * doc/extend.texi: Document patchable_function_entry attribute.
22747 * doc/invoke.texi: Document -fpatchable_function_entry
22748 command line option.
22749 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22751 * doc/tm.texi: Re-generate.
22753 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22756 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22757 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22759 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22761 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22763 2017-07-25 Richard Biener <rguenther@suse.de>
22765 PR tree-optimization/81455
22766 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22767 not walk in cycles when looking for guards.
22769 2017-07-25 Richard Biener <rguenther@suse.de>
22771 PR tree-optimization/81529
22772 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22773 when optimizing backedge uses.
22775 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22777 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22779 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22780 to dl_section_ref. On AIX, append an expression to subtract
22781 the size of the section length to dl_section_ref.
22783 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22785 * configure.ac: If any of the config.* scripts fail, exit 1.
22786 * configure: Regenerate.
22788 2017-07-25 Richard Biener <rguenther@suse.de>
22790 PR middle-end/81546
22791 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22792 of immediate uses, be more verbose on errors.
22794 2017-07-25 Richard Biener <rguenther@suse.de>
22796 PR tree-optimization/81510
22797 * tree-vect-loop.c (vect_is_simple_reduction): When the
22798 reduction stmt is not inside the loop bail out.
22800 2017-07-25 Richard Biener <rguenther@suse.de>
22802 PR tree-optimization/81303
22803 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22804 profitability check against LOOP_VINFO_NITERSM1.
22806 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22808 * domwalk.c (cmp_bb_postorder): Simplify.
22809 (sort_bbs_postorder): New function. Use it...
22810 (dom_walker::walk): ...here to optimize common cases.
22812 2017-07-25 Martin Liska <mliska@suse.cz>
22815 * ipa-visibility.c (function_and_variable_visibility): Make the
22816 redirection just on target that supports aliasing.
22817 Fix GNU coding style.
22819 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22822 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22824 * config/arm/rtems.h: Likewise.
22825 * config/bfin/rtems.h: Likewise.
22826 * config/i386/rtemself.h: Likewise.
22827 * config/lm32/rtems.h: Likewise.
22828 * config/m32c/rtems.h: Likewise.
22829 * config/m68k/rtemself.h: Likewise.
22830 * config/microblaze/rtems.h: Likewise.
22831 * config/mips/rtems.h: Likewise.
22832 * config/moxie/rtems.h: Likewise.
22833 * config/nios2/rtems.h: Likewise.
22834 * config/powerpcspe/rtems.h: Likewise.
22835 * config/rs6000/rtems.h: Likewise.
22836 * config/rtems.h: Likewise.
22837 * config/sh/rtems.h: Likewise.
22838 * config/sh/rtemself.h: Likewise.
22839 * config/sparc/rtemself.h: Likewise.
22841 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22844 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22845 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22846 * tree-ssa-structalias.c (alias_get_name): Same.
22848 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22851 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22852 instructions if no du chain is found.
22854 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22856 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22858 2017-07-25 Richard Biener <rguenther@suse.de>
22860 PR middle-end/81505
22861 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
22864 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22866 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
22867 upper-regs options.
22868 (ISA_2_7_MASKS_SERVER): Likewise.
22869 (ISA_3_0_MASKS_IEEE): Likewise.
22870 (OTHER_P8_VECTOR_MASKS): Likewise.
22871 (OTHER_VSX_VECTOR_MASKS): Likewise.
22872 (POWERPC_MASKS): Likewise.
22873 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
22874 duplicate list of options.
22875 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
22876 explicit -mupper-regs options.
22877 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
22878 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
22879 alias for -mupper-regs-df.
22880 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
22881 (rs6000_init_hard_regno_mode_ok): Likewise.
22882 (rs6000_option_override_internal): Likewise.
22883 (rs6000_opt_masks): Likewise.
22884 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
22885 options in terms of whether -mvsx or -mpower8-vector was used.
22886 (TARGET_UPPER_REGS_DI): Likewise.
22887 (TARGET_UPPER_REGS_SF): Likewise.
22888 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
22889 -mupper-regs-* options.
22891 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22893 * passes.c (emergency_dump_function): Print some empty lines and a
22894 header before the RTL dump.
22896 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22898 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
22900 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
22903 * config/aarch64/aarch64.c (aarch64_classify_symbol):
22904 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
22906 2017-07-24 Carl Love <cel@us.ibm.com>
22908 * config/rs6000/rs6000-c.c: Add support for built-in functions
22909 vector float vec_extract_fp32_from_shorth (vector unsigned short);
22910 vector float vec_extract_fp32_from_shortl (vector unsigned short);
22911 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
22912 vec_extract_fp_from_shortl): Add defines for the two builtins.
22913 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
22914 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
22916 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
22917 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
22918 * doc/extend.texi: Update the built-in documentation file for the
22919 new built-in function.
22921 2017-07-24 Jakub Jelinek <jakub@redhat.com>
22924 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
22926 * doc/generic.texi: Likewise.
22927 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22928 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22930 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
22932 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
22933 (aarch64_mls_elt_merge<mode>): Likewise.
22935 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
22937 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22938 having __cxa_atexit.
22940 2017-07-23 Michael Collison <michael.collison@arm.com>
22942 * config/arm/arm.c (arm_option_override): Deprecate
22943 use of -mstructure-size-boundary.
22944 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
22945 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
22947 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22950 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22951 Reduce cost estimate for direct moves.
22953 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
22956 * config/i386/i386.c (ix86_option_override_internal): Disable
22957 BMI, BMI2 and TBM instructions for -m16.
22959 2017-07-21 Carl Love <cel@us.ibm.com>
22961 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22962 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22963 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22964 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22965 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22966 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22967 VMULOSW): New enum "unspec" values.
22968 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22969 altivec_vmulosw): New patterns.
22970 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22971 VMULOSW): Add definitions.
22973 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
22975 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
22976 (qdf24xx): Likewise.
22977 * config/aarch64/aarch64-options-extensions.def (rdma); New.
22978 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
22979 (AARCH64_FL_V8_1): Renumber.
22980 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
22981 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
22982 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
22983 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
22984 rdma to feature modifiers list.
22986 2017-07-21 Yury Gribov <tetra2005@gmail.com>
22988 PR middle-end/56727
22989 * ipa-visibility (function_and_variable_visibility): Convert
22990 recursive PLT call to direct call if appropriate.
22992 2017-07-21 Andrew Pinski <apinski@cavium.com>
22994 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
22995 operand 1 to see if the types precision matches.
22996 * fold-const.c (operand_equal_p): Likewise.
22998 2017-07-21 Richard Biener <rguenther@suse.de>
23000 PR tree-optimization/81303
23001 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23002 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23003 (vect_peeling_hash_get_lowest_cost): Adjust.
23004 (vect_enhance_data_refs_alignment): Likewise. Use
23005 vect_get_peeling_costs_all_drs to compute the penalty for no
23006 peeling to match up costs.
23008 2017-07-21 Richard Biener <rguenther@suse.de>
23010 PR tree-optimization/81500
23011 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23012 we didn't identify a reduction path.
23014 2017-07-21 Tom de Vries <tom@codesourcery.com>
23015 Cesar Philippidis <cesar@codesourcery.com>
23017 PR gcov-profile/81442
23018 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23021 2017-07-21 Tom de Vries <tom@codesourcery.com>
23024 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23026 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23027 nvptx_override_options_after_change.
23029 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23031 * dwarf2out.c (output_file_names): Avoid double testing for
23032 dwarf_version >= 5.
23034 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23036 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
23038 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23040 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23042 (try_crossjump_to_edge): Do not punt on partitioned functions.
23044 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23046 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23047 Put all BBs reachable only via paths crossing cold region to cold
23049 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23051 2016-07-21 Richard Biener <rguenther@suse.de>
23053 PR tree-optimization/81303
23054 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23055 into account prologue and epilogue iterations when raising
23056 min_profitable_iters to sth at least covering one vector iteration.
23058 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23060 * config/arm/arm.c (arm_test_cpu_arch_dat):
23063 2017-07-20 Nathan Sidwell <nathan@acm.org>
23065 Remove TYPE_METHODS.
23066 * tree.h (TYPE_METHODS): Delete.
23067 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23068 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23069 (dbxout_type_methods): Scan TYPE_FIELDS.
23070 (dbxout_type): Don't check TYPE_METHODS here.
23071 * function.c (use_register_for_decl): Always ignore register for
23072 class types when not optimizing.
23073 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23074 * tree.c (free_lang_data_in_type): Stitch out member functions and
23075 templates from TYPE_FIELDS.
23076 (build_distinct_type_copy, verify_type_variant,
23077 verify_type): Member fns are on TYPE_FIELDS.
23078 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23079 * tree-pretty-print.c (dump_generic_node): Likewise.
23081 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23084 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23085 V2TImode and V4TImode.
23086 (ix86_expand_vector_extract): Likewise.
23087 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23088 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23089 (ssescalarmode): Handle V4TImode and V2TImode.
23090 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23091 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23092 (VEXTRACTI128_MODE): New mode iterator.
23093 (splitter for *vec_extractv?ti first element): New.
23094 (VEC_INIT_MODE): New mode iterator.
23095 (vec_init<mode>): Consolidate 3 expanders into one using
23096 VEC_INIT_MODE mode iterator.
23098 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23100 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23101 non_spilled_static_chain_regno_p.
23103 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23105 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
23107 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
23109 * bb-reorder.c (connect_traces): Allow copying of blocks within
23112 2017-07-20 Richard Biener <rguenther@suse.de>
23114 * gimple.h (gimple_phi_result): Add gphi * overload.
23115 (gimple_phi_result_ptr): Likewise.
23116 (gimple_phi_arg): Likewise. Adjust index assert to only
23117 allow actual argument accesses rather than all slots available
23119 (gimple_phi_arg_def): Add gphi * overload.
23120 * tree-phinodes.c (make_phi_node): Initialize only actual
23122 (resize_phi_node): Clear memory not covered by old node,
23123 do not initialize excess argument slots.
23124 (reserve_phi_args_for_new_edge): Initialize new argument slot
23127 2017-07-20 Bin Cheng <bin.cheng@arm.com>
23129 PR tree-optimization/81388
23131 2016-07-21 Bin Cheng <bin.cheng@arm.com>
23133 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
23134 by removing computation of may_be_zero.
23136 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23137 Tom de Vries <tom@codesourcery.com>
23139 PR middle-end/81030
23140 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
23141 when gimple level profile disagrees with what RTL expander did.
23143 2017-07-20 Richard Biener <rguenther@suse.de>
23145 PR tree-optimization/61171
23146 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
23147 (vect_analyze_stmt): Add slp instance parameter.
23148 (vectorizable_reduction): Likewise.
23149 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
23150 (vect_is_simple_reduction): Deal with chains not detected
23151 as SLP reduction chain, specifically not properly associated
23152 chains containing a mix of plus/minus.
23153 (get_reduction_op): Remove.
23154 (get_initial_defs_for_reduction): Simplify, pass in whether
23155 this is a reduction chain, pass in the SLP node for the PHIs.
23156 (vect_create_epilog_for_reduction): Get the SLP instance as
23158 (vectorizable_reduction): Get the SLP instance as arg.
23159 During analysis remember the SLP node with the PHIs in the
23160 instance. Simplify getting at the vectorized reduction PHIs.
23161 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
23162 through SLP instance.
23163 (vect_slp_analyze_operations): Likewise.
23164 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
23165 (vect_transform_stmt): Likewise.
23167 2017-07-20 Tom de Vries <tom@codesourcery.com>
23169 PR tree-optimization/81489
23170 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
23171 read of phi arg location to before loop that modifies phi.
23173 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23175 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
23178 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23180 PR middle-end/81331
23181 * except.c (execute): Fix ordering issue.
23183 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23185 PR rtl-optimization/81423
23186 * combine.c (make_compound_operation_int): Don't try to optimize
23187 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
23189 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23191 PR rtl-optimization/81423
23192 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
23193 with a constant that is -1 in the truncated to mode.
23195 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23197 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
23198 (determine_unlikely_bbs): ... here.
23199 * predict.h (propagate_unlikely_bbs_forward): Declare.
23200 * cfgexpand.c (pass_expand::execute): Use it.
23201 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
23203 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
23204 propagate_unlikely_bbs_forward.
23206 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23208 PR middle-end/81331
23209 * except.c (maybe_add_nop_after_section_switch): New function.
23212 2017-07-19 Tom de Vries <tom@codesourcery.com>
23214 * gimple.h (gimple_phi_set_arg): Make assert more strict.
23216 2017-07-19 Tom de Vries <tom@codesourcery.com>
23218 * gimple.h (gimple_phi_arg): Make assert more strict.
23220 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
23222 * config.gcc (powerpc*-*-*): Add mmintrin.h.
23223 * config/rs6000/mmintrin.h: New file.
23224 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
23226 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23228 PR tree-optimization/81346
23229 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
23231 2017-07-19 Tom de Vries <tom@codesourcery.com>
23233 * config/nvptx/nvptx.md (VECIM): Add V2DI.
23235 2017-07-19 Tom de Vries <tom@codesourcery.com>
23237 * config/nvptx/nvptx-modes.def: Add V2DImode.
23238 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
23239 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
23240 (nvptx_output_mov_insn): Handle lack of mov.b128.
23241 (nvptx_print_operand): Handle 'H' and 'L' codes.
23242 (nvptx_vector_mode_supported): Allow V2DImode.
23243 (nvptx_preferred_simd_mode): New function.
23244 (nvptx_data_alignment): New function.
23245 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
23246 nvptx_preferred_simd_mode.
23247 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
23249 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
23251 2017-07-19 Tom de Vries <tom@codesourcery.com>
23253 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
23254 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
23255 (nvptx_vector_mode_supported): New function. Allow V2SImode.
23256 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
23257 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
23258 (mov<VECIM>_insn): New define_insn.
23259 (define_expand "mov<VECIM>): New define_expand.
23261 2017-07-19 Tom de Vries <tom@codesourcery.com>
23263 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
23265 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23267 PR tree-optimization/81346
23268 * fold-const.h (fold_div_compare, range_check_type): Declare.
23269 * fold-const.c (range_check_type): New function.
23270 (build_range_check): Use range_check_type.
23271 (fold_div_compare): No longer static, rewritten into
23272 a match.pd helper function.
23273 (fold_comparison): Don't call fold_div_compare here.
23274 * match.pd (X / C1 op C2): New optimization using fold_div_compare
23275 as helper function.
23277 2017-07-19 Nathan Sidwell <nathan@acm.org>
23279 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
23280 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
23281 * tree.c (find_decls_types_r, verify_type): Use
23282 TYPE_{MIN,MAX}_VALUE_RAW.
23283 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
23284 (hash_tree): Likewise.
23285 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
23287 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
23290 2017-07-18 Tom de Vries <tom@codesourcery.com>
23292 PR middle-end/81464
23293 * omp-expand.c (expand_omp_for_static_chunk): Handle
23294 equal-argument loop exit phi.
23296 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
23299 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
23300 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
23301 operand 2 predicate.
23302 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
23303 operand 2 predicate.
23304 (ror,rol -> rorx splitters): Use const_int_operand as
23305 operand 2 predicate.
23307 2017-06-18 Richard Biener <rguenther@suse.de>
23309 PR tree-optimization/81410
23310 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
23311 the gap in the ! slp_perm SLP case after each group.
23313 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23315 PR middle-end/81463
23316 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
23319 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23321 PR middle-end/81462
23322 * predict.c (set_even_probabilities): Cleanup; do not affect
23323 probabilities that are already known.
23324 (combine_predictions_for_bb): Call even when count is set.
23326 2017-07-18 Nathan Sidwell <nathan@acm.org>
23328 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
23331 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23334 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
23335 optimization for loop niter analysis.
23337 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
23340 * config/avr/avr.c (avr_optimize_casesi): Don't use
23341 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
23343 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
23345 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
23346 body_cost_vec from _vect_peel_extended_info.
23347 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
23348 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
23351 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23353 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
23355 2017-07-18 Richard Biener <rguenther@suse.de>
23357 PR tree-optimization/80620
23358 PR tree-optimization/81403
23359 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
23360 info when re-using a VN table entry.
23362 2017-07-18 Richard Biener <rguenther@suse.de>
23364 PR tree-optimization/81418
23365 * tree-vect-loop.c (vectorizable_reduction): Properly compute
23366 vectype_in. Verify that with lane-reducing reduction operations
23367 we have a single def-use cycle.
23369 2017-07-17 Carl Love <cel@us.ibm.com>
23371 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
23373 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23374 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23375 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23376 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23377 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23378 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23379 VMULOSW): New enum "unspec" values.
23380 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23381 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23382 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23383 altivec_vmulosw): New patterns.
23384 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23385 VMULOSW): Add definitions.
23387 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
23389 * config/alpha/alpha.c: Include predict.h.
23391 2017-07-17 Yury Gribov <tetra2005@gmail.com>
23393 * tree-vrp.c (compare_assert_loc): Fix comparison function
23394 to return predictable results.
23396 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23398 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
23400 (subdi3): Likewise.
23401 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
23402 * doc/invoke.texi (mexpand-adddi): Update text.
23404 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23406 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
23407 that also clobbers the CC register. The old expand code is moved
23409 (*arc_clzsi2): ... here.
23410 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
23411 the CC register. The old expand code is moved to ...
23412 (arc_ctzsi2): ... here.
23414 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23416 * config/arc/arc.opt (mindexed-loads): Use initial value
23417 TARGET_INDEXED_LOADS_DEFAULT.
23418 (mauto-modify-reg): Use initial value
23419 TARGET_AUTO_MODIFY_REG_DEFAULT.
23420 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23421 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23422 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23423 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23425 2017-07-17 Martin Liska <mliska@suse.cz>
23428 * opts.c (finish_options): Do not allow -fgnu-tm
23429 w/ -fsanitize={kernel-,}address. Say sorry.
23431 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23434 * tree-loop-distribution.c (classify_partition): Only assert on
23435 numer of iterations.
23436 (merge_dep_scc_partitions): Delete prameter. Update function call.
23437 (distribute_loop): Remove code handling loop with unknown niters.
23438 (pass_loop_distribution::execute): Skip loop with unknown niters.
23440 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23443 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
23444 function sort_partitions_by_post_order.
23446 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23448 PR tree-optimization/81374
23449 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
23450 the max index of basic blocks, rather than number of basic blocks.
23452 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23454 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
23456 (arc_legitimate_pic_operand_p): Likewise.
23457 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
23459 (arc_needs_pcl_p): Likewise.
23460 (arc_legitimate_pc_offset_p): Likewise.
23461 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
23462 function is also used in constrains.md.
23463 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
23464 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
23465 PLUS. Only return true/false in known cases, otherwise assert.
23466 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
23467 is already called in arc_legitimate_constant_p.
23468 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
23470 (LEGITIMATE_PIC_OPERAND_P): Use
23471 arc_raw_symbolic_reference_mentioned_p function.
23472 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
23477 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23478 Andrew Burgess <andrew.burgess@embecosm.com>
23480 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
23481 (arc_return_address_register): New function.
23482 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
23483 (arc_handle_fndecl_attribute): Add naked attribute.
23484 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
23485 (TARGET_WARN_FUNC_RETURN): Likewise.
23486 (arc_allocate_stack_slots_for_args): New function.
23487 (arc_warn_func_return): Likewise.
23488 (machine_function): Change type fn_type.
23489 (arc_compute_function_type): Consider new naked function type,
23490 change function return type.
23491 (arc_must_save_register): Adapt to handle new
23492 arc_compute_function_type's return type.
23493 (arc_expand_prologue): Likewise.
23494 (arc_expand_epilogue): Likewise.
23495 (arc_return_address_regs): Delete.
23496 (arc_return_address_register): New function.
23497 (arc_epilogue_uses): Use above function.
23498 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
23499 (arc_function_type): Change encoding, add naked type.
23500 (ARC_INTERRUPT_P): Change to handle the new encoding.
23501 (ARC_FAST_INTERRUPT_P): Likewise.
23502 (ARC_NORMAL_P): Define.
23503 (ARC_NAKED_P): Likewise.
23504 (arc_compute_function_type): Delete prototype.
23505 * config/arc/arc.md (in_ret_delay_slot): Use
23506 arc_return_address_register function.
23507 (simple_return): Likewise.
23508 (p_return_i): Likewise.
23510 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23512 PR tree-optimization/81428
23513 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
23514 can't be built for those types.
23516 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23518 Remove stuff dead since r239246.
23520 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
23521 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
23522 (avr_inform_devices): Remove dead stuff.
23524 2017-07-17 Tamar Christina <tamar.christina@arm.com>
23526 * config/arm/arm_neon.h: Fix softp typo.
23528 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23530 PR tree-optimization/81365
23531 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
23532 aggregate moves onto bb predecessor edges, make sure there are no
23533 loads that could alias the lhs in between the start of bb and the
23536 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23539 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
23540 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
23541 [LSHIFTRT, outer_code = TRUNCATE]: Same.
23543 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23545 PR tree-optimization/81396
23546 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
23547 (init_symbolic_number): Initialize it to 1.
23548 (perform_symbolic_merge): Add n_ops from both operands into the new
23550 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
23551 without base_addr as useless if they need more than one operation.
23552 (bswap_replace): Handle !bswap case for NULL base_addr.
23554 2017-07-17 Tom de Vries <tom@codesourcery.com>
23557 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
23560 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
23562 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
23563 conditional builtin define __FIX_LEON3FT_B2BST.
23565 2017-07-17 Daniel Cederman <cederman@gaisler.com>
23567 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
23568 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
23571 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23573 PR rtl-optimization/81424
23574 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
23575 to remove potential trapping from operands if -fnon-call-exceptions.
23577 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23579 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
23580 profile_proability for scalling.
23581 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
23583 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23585 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
23587 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23589 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
23590 fixpoint arithmetics.
23592 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23594 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
23595 fixpoint arithmetics.
23597 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23599 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
23600 fixpoint arithmetics.
23602 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23604 * profile-count.h (profile_probability::from_reg_br_prob_note,
23605 profile_probability::to_reg_br_prob_note): New functions.
23606 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
23607 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
23608 * predict.c (probability_reliable_p): Update.
23609 (edge_probability_reliable_p): Update.
23610 (br_prob_note_reliable_p): Update.
23611 (invert_br_probabilities): Update.
23612 (add_reg_br_prob_note): New function.
23613 (combine_predictions_for_insn): Update.
23614 * asan.c (asan_clear_shadow): Update.
23615 * cfgbuild.c (compute_outgoing_frequencies): Update.
23616 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23617 (update_br_prob_note): Update.
23618 (rtl_verify_edges): Update.
23619 (purge_dead_edges): Update.
23620 (fixup_reorder_chain): Update.
23621 * emit-rtl.c (try_split): Update.
23622 * ifcvt.c (cond_exec_process_insns): Update.
23623 (cond_exec_process_if_block): Update.
23624 (dead_or_predicable): Update.
23625 * internal-fn.c (expand_addsub_overflow): Update.
23626 (expand_neg_overflow): Update.
23627 (expand_mul_overflow): Update.
23628 * loop-doloop.c (doloop_modify): Update.
23629 * loop-unroll.c (compare_and_jump_seq): Update.
23630 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23631 * predict.h: Update.
23632 * reorg.c (mostly_true_jump): Update.
23634 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23635 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23636 * config/arc/arc.c: (emit_unlikely_jump): Update.
23637 * config/arm/arm.c: (emit_unlikely_jump): Update.
23638 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23639 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23640 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23641 (ix86_print_operand): Update.
23642 (ix86_split_fp_branch): Update.
23643 (predict_jump): Update.
23644 * config/ia64/ia64.c (ia64_print_operand): Update.
23645 * config/mmix/mmix.c (mmix_print_operand): Update.
23646 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23647 (rs6000_expand_split_stack_prologue): Update.
23648 * config/rs6000/rs6000.c: Update.
23649 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23650 (s390_expand_vec_movstr): Update.
23651 (s390_expand_cs_tdsi): Update.
23652 (s390_expand_split_stack_prologue): Update.
23653 * config/sh/sh.c (sh_print_operand): Update.
23654 (expand_cbranchsi4): Update.
23655 (expand_cbranchdi4): Update.
23656 * config/sparc/sparc.c (output_v9branch): Update.
23657 * config/spu/spu.c (get_branch_target): Update.
23658 (ea_load_store_inline): Update.
23659 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23660 * config/tilepro/tilepro.c: Update.
23662 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23664 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23665 (gimplify_save_expr): Likewise.
23667 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23669 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23671 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23673 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23675 * timevar.def (TV_IPA_FNSUMMARY): Define.
23677 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23679 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23680 to back store errata sensitive sequence from being generated.
23681 (sqrtdf2_fix): Likewise.
23683 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23685 * tree-ssa-threadupdate.c (compute_path_counts,
23686 update_joiner_offpath_counts): Use profile_probability.
23688 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23691 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23693 * config/arm/arm-c.c (arm_cpu_builtins): Define
23694 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23696 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23698 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23699 array entries to represent __ieee128 versions of the
23700 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23701 scalar_extract_sig, and scalar_insert_exp built-in functions.
23702 (altivec_resolve_overloaded_builtin): Add special case handling
23703 for the __builtin_scalar_insert_exp function, as represented by
23704 the P9V_BUILTIN_VEC_VSIEDP constant.
23705 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23706 exponent support for __ieee128 argument.
23707 (VSESQP): Add scalar extract signature support for __ieee128
23709 (VSTDCNQP): Add scalar test negative support for __ieee128
23711 (VSIEQP): Add scalar insert exponent support for __int128 argument
23712 with __ieee128 result.
23713 (VSIEQPF): Add scalar insert exponent support for __ieee128
23714 argument with __ieee128 result.
23715 (VSTDCQP): Add scalar test data class support for __ieee128
23717 (VSTDCNQP): Add overload support for scalar test negative with
23718 __ieee128 argument.
23719 (VSTDCQP): Add overload support for scalar test data class
23720 __ieee128 argument.
23721 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23722 UNSPEC_VSX_SXSIGDP.
23723 (UNSPEC_VSX_SIEXPQP): New constant.
23724 (xsxexpqp): New insn for VSX scalar extract exponent quad
23726 (xsxsigqp): New insn for VSX scalar extract significand quad
23728 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23729 precision with floating point argument.
23730 (xststdcqp): New expand for VSX scalar test data class quad
23732 (xststdcnegqp): New expand for VSX scalar test negative quad
23734 (xststdcqp): New insn to match expansions for VSX scalar test data
23735 class quad precision and VSX scalar test negative quad precision.
23736 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23737 special case operand checking to enforce that second operand of
23738 VSX scalar test data class with quad precision argument is a 7-bit
23740 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23741 prototypes and descriptions of __ieee128 versions of
23742 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23743 scalar_test_data_class, and scalar_test_neg built-in functions.
23745 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23747 PR tree-optimization/81162
23748 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23749 replace a negate with an add.
23751 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23753 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23755 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23757 * config/arm/arm-c.c (arm_cpu_builtins): Define
23758 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23760 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23762 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23763 (armv8-r): Set ARM Cortex-R52 as default CPU.
23764 * config/arm/arm-tables.opt: Regenerate.
23765 * config/arm/arm-tune.md: Regenerate.
23766 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23768 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23769 extension for -mcpu=cortex-r52.
23771 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23773 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23774 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23775 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23776 (fp-armv8): Define it as FP_ARMv8 only.
23777 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23778 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23780 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23781 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23782 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23783 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23784 than TARGET_FPU_ARMV8.
23785 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23786 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23787 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23789 * config/arm/neon.md (neon_vrint): Likewise.
23790 (neon_vcvt): Likewise.
23791 (neon_<fmaxmin_op><mode>): Likewise.
23792 (<fmaxmin><mode>3): Likewise.
23793 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23794 * config/arm/predicates.md (arm_cond_move_operator): Check against
23795 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23797 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23799 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23800 to top of function.
23802 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23804 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23805 loop in comment with memset.
23807 2017-07-14 Martin Liska <mliska@suse.cz>
23809 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23810 * dwarf2out.c (is_java): Remove the function.
23811 (output_pubname): Remove usage of the function.
23812 (lower_bound_default): Remove usage of DW_LANG_Java.
23813 (gen_compile_unit_die): Likewise.
23814 * gcc.c: Remove compiler defaults for .java and .zip files.
23815 * gimple-expr.c (remove_suffix): Change as there's no longer
23816 extension than 4-letter one.
23817 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23818 (gimplify_save_expr): Likewise.
23819 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23820 as it's possible even for other languages than Java.
23821 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23822 * lto-opts.c (lto_write_options): Remove reference to Java.
23823 * opts.c (strip_off_ending): Update file extension handling.
23824 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23825 * tree-eh.c (lower_resx): Likewise.
23826 * tree.c (free_lang_data_in_type): Remove dead code.
23827 (find_decls_types_r): Likewise.
23828 (build_common_builtin_nodes): Remove Java from a comment.
23829 (verify_type): Remove dead code.
23830 * varasm.c (assemble_external): Remove Java from a comment.
23832 2017-07-14 Martin Liska <mliska@suse.cz>
23834 * opts.c (finish_options): Add quotes.
23835 (common_handle_option): Likewise.
23837 2017-07-14 Martin Liska <mliska@suse.cz>
23839 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23840 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23841 Remove N_SO_PASCAL.
23842 * dwarf2out.c (lower_bound_default): Do not handle
23844 (gen_compile_unit_die): Likewise.
23845 * gcc.c: Remove default extension binding for GNU Pascal.
23846 * stmt.c: Remove Pascal language from a comment.
23847 * xcoffout.c: Likewise.
23849 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23852 * diagnostic-show-locus.c (fixit_cmp): New function.
23853 (layout::layout): Sort m_fixit_hints.
23854 (column_range::column_range): Assert that the values are valid.
23855 (struct char_span): New struct.
23856 (correction::overwrite): New method.
23857 (struct source_line): New struct.
23858 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23859 calls in terms of classes source_line and char_span, and
23860 correction::overwrite.
23861 (selftest::test_overlapped_fixit_printing_2): New function.
23862 (selftest::diagnostic_show_locus_c_tests): Call it.
23864 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
23866 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
23867 early if there is no lhs.
23869 2017-07-13 Martin Liska <mliska@suse.cz>
23871 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
23872 (gen_reference_type_die): Likewise.
23873 * stor-layout.c: Remove Pascal-related comment.
23875 2017-07-13 Martin Liska <mliska@suse.cz>
23877 * opts.c (finish_options): Add quotes to error messages.
23878 (parse_sanitizer_options): Likewise.
23880 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
23882 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
23884 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
23886 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
23888 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23890 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
23892 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
23894 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23897 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
23898 provides the hardware capability bits, define the macro
23899 __BUILTIN_CPU_SUPPORTS__.
23900 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
23901 if GLIBC does not provide the hardware capability bits. Add a
23902 gcc_unreachable call if the built-in cpu function is neither
23903 __builtin_cpu_is nor __builtin_cpu_supports.
23904 (rs6000_get_function_versions_dispatcher): Change the warning
23905 that an old GLIBC is used which does not export the capability
23906 bits to be an error.
23907 * doc/extend.texi (target_clones attribute): Document the
23908 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
23909 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
23910 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
23911 the macros defined by GCC if the newer GLIBC is available.
23913 2017-07-12 Jeff Law <law@redhat.com>
23915 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
23916 remaining includes slightly.
23917 * config/riscv/riscv-builtins.c: Include profile-count.h.
23919 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23922 * config/avr/avr.c (avr_set_current_function): In diagnostic
23923 messages: Quote keywords and (parts of) identifiers.
23924 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
23927 2017-07-12 Carl Love <cel@us.ibm.com>
23929 * config/rs6000/rs6000-c.c: Add support for built-in functions
23930 vector bool char vec_revb (vector bool char);
23931 vector bool short vec_revb (vector short char);
23932 vector bool int vec_revb (vector bool int);
23933 vector bool long long vec_revb (vector bool long long);
23934 * doc/extend.texi: Update the built-in documentation file for the
23935 new built-in functions.
23937 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23939 * config/s390/s390.md: Remove movcc splitter.
23941 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23943 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
23944 load/store on condition.
23946 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23949 * config/avr/avr.c (avr_encode_section_info)
23950 [progmem && !TREE_READONLY]: Error if progmem object needs
23953 2017-07-11 Michael Collison <michael.collison@arm.com>
23955 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
23958 2017-07-11 Carl Love <cel@us.ibm.com>
23960 * config/rs6000/rs6000-c.c: Add support for builtins
23961 vector unsigned int vec_parity_lsbb (vector signed int);
23962 vector unsigned int vec_parity_lsbb (vector unsigned int);
23963 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
23964 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
23965 vector unsigned long long vec_parity_lsbb (vector signed long long);
23966 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
23967 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
23968 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
23969 * doc/extend.texi: Update the built-in documentation file for the
23970 new built-in functions.
23972 2017-07-11 David Malcolm <dmalcolm@redhat.com>
23974 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
23975 (layout::m_primary_loc): New field.
23976 (layout::layout): Initialize new field. Move location filtering
23977 logic from here to...
23978 (layout::maybe_add_location_range): ...this new method. Add
23979 support for filtering to just the lines already specified by other
23981 (layout::will_show_line_p): New method.
23982 (gcc_rich_location::add_location_if_nearby): New method.
23983 (selftest::test_add_location_if_nearby): New test function.
23984 (selftest::diagnostic_show_locus_c_tests): Call it.
23985 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
23988 2017-07-11 Tom de Vries <tom@codesourcery.com>
23990 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
23991 (bb_first_real_insn): New function.
23992 (nvptx_single): Add extra initialization of broadcasted condition
23995 2017-07-11 Nathan Sidwell <nathan@acm.org>
23997 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
23999 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24001 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24002 Binutils doc as TEXI will mess them up.
24003 * doc/invoke.texi (AVR Options): Same here.
24005 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24007 * config/sparc/sparc.opt (mfix-ut700): New option.
24008 (mfix-gr712rc): Likewise.
24009 (sparc_fix_b2bst): New variable.
24010 * doc/invoke.texi (SPARC options): Document them.
24011 (ARM options): Fix warnings.
24012 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24013 instructions to prevent sequences that can trigger the store-store
24014 errata for certain LEON3FT processors.
24015 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24016 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24017 * config/sparc/sparc.md (fix_b2bst): New attribute.
24018 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24020 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24023 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24025 (*rsqrtsf2_sse): Ditto.
24027 (div<mode>3): Macroize insn from divdf3 and divsf3
24028 using MODEF mode iterator.
24030 2017-07-10 Martin Sebor <msebor@redhat.com>
24032 PR tree-optimization/80397
24033 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24034 instead of testing for equality to INTEGER_TYPE.
24036 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
24038 * config.gcc: Remove uclibc from arc target spec.
24040 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24042 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24044 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24047 * lto-wrapper.c (remove_option): New function.
24048 (merge_and_complain): Merge PIC/PIE options more realistically.
24050 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24052 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24056 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24057 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24058 * config.in: Regenerate.
24059 * configure: Regenerate.
24060 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24061 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24062 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24063 (TARGET_GASISR_PROLOGUES): ...target mask.
24064 * common/config/avr/avr-common.c
24065 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24066 Set -mgas-isr-prologues.
24067 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24068 INSERT_PASS_BEFORE for it.
24069 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24070 * config/avr/avr.c (avr_option_override)
24071 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24072 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24073 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24074 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24075 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24077 (make_avr_pass_pre_proep): New function.
24078 (emit_push_sfr) <treg>: Add argument to function and use it
24079 instead of TMP_REG.
24080 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24081 and set machine->gasisr.yes.
24082 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24083 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24084 __gcc_isr.n_pushed to .L__stack_usage.
24085 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24086 (avr_asm_final_postscan_insn): ...this new static function.
24087 * config/avr/avr.h (machine_function)
24088 <is_no_gccisr, use_L__stack_usage>: New fields.
24089 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24090 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24091 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24092 (gasisr, *gasisr): New expander and insn.
24093 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24094 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24095 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24097 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24099 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24102 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24104 Move jump-tables out of .text again.
24107 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
24108 (ASM_OUTPUT_ADDR_VEC): New function.
24109 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
24110 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
24111 INSN_ADDRESSes as asm comment.
24112 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
24113 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
24114 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
24115 * config/avr/avr.md (*tablejump): Adjust comment.
24116 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
24117 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
24119 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
24120 (avr_output_addr_vec): New proto.
24121 (avr_log_t) <insn_addresses>: New field.
24123 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
24126 * config/i386/i386.c (ix86_function_arg_advance): Set
24127 outgoing_args_on_stack to true if there are outgoing arguments
24129 (ix86_function_arg): Likewise.
24130 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
24131 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
24132 * config/i386/i386.h (machine_function): Add
24133 outgoing_args_on_stack.
24135 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
24137 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
24138 supporting pthreds.
24139 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
24141 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
24143 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
24144 (REAL_H): Remove $(MACHMODE_H).
24145 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
24147 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
24148 $(MACHMODE_H) and double-int.h.
24149 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
24151 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
24154 2017-07-07 Andrew Pinski <apinski@cavium.com>
24156 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
24157 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
24159 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
24161 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
24162 Add warning if GCC was not configured to link against a GLIBC that
24163 exports the hardware capability bits.
24164 (make_resolver_func): Make resolver function private and not a
24165 COMDAT function. Create the name with clone_function_name instead
24166 of make_unique_name.
24169 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
24170 correct operand in doing the split.
24172 2017-07-07 Carl Love <cel@us.ibm.com>
24174 * config/rs6000/rs6000-c: Add support for built-in function
24175 vector unsigned short vec_pack_to_short_fp32 (vector float,
24177 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
24178 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
24179 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
24180 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
24181 (convert_4f32_8i16): Add define_expand.
24182 * doc/extend.texi: Update the built-in documentation file for the
24183 new built-in function.
24185 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24187 * config/sparc/m8.md: New file.
24188 * config/sparc/sparc.md: Include m8.md.
24190 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24192 * config/sparc/sparc.opt: New option -mvis4b.
24193 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
24194 (sparc_option_override): Handle VIS4B.
24195 (enum sparc_builtins): Define
24196 SPARC_BUILTIN_DICTUNPACK{8,16,32},
24197 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
24198 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
24199 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
24200 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
24201 (check_constant_argument): New function.
24202 (sparc_vis_init_builtins): Define builtins
24203 __builtin_vis_dictunpack{8,16,32},
24204 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
24205 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
24206 __builtin_vis_fpcmpde{8,16,32}shl and
24207 __builtin_vis_fpcmpur{8,16,32}shl.
24208 (sparc_expand_builtin): Check that the constant operands to
24209 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
24210 constant and in range.
24211 * config/sparc/sparc-c.c (sparc_target_macros): Handle
24213 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
24214 (SPARC_IMM5_P): Likewise.
24215 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
24216 (enabled): Handle vis4b.
24217 (UNSPEC_DICTUNPACK): New unspec.
24218 (UNSPEC_FPCMPSHL): Likewise.
24219 (UNSPEC_FPUCMPSHL): Likewise.
24220 (UNSPEC_FPCMPDESHL): Likewise.
24221 (UNSPEC_FPCMPURSHL): Likewise.
24222 (cpu_feature): New CPU feature `vis4b'.
24223 (dictunpack{8,16,32}): New insns.
24224 (FPCSMODE): New mode iterator.
24225 (fpcscond): New code iterator.
24226 (fpcsucond): Likewise.
24227 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
24228 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
24229 (fpcmpde{8,16,32}{si,di}shl): Likewise.
24230 (fpcmpur{8,16,32}{si,di}shl): Likewise.
24231 * config/sparc/constraints.md: Define constraints `q' for unsigned
24232 2-bit integer constants and `t' for unsigned 5-bit integer
24234 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
24236 (imm5_operand_dictunpack16): Likewise.
24237 (imm5_operand_dictunpack32): Likewise.
24238 (imm2_operand): Likewise.
24239 * doc/invoke.texi (SPARC Options): Document -mvis4b.
24240 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
24241 ditunpack* and fpcmp*shl builtins.
24243 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24245 * config.gcc: Handle m8 in --with-{cpu,tune} options.
24246 * config.in: Add HAVE_AS_SPARC6 define.
24247 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
24249 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
24251 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
24252 (CPP_CPU_SPEC): Handle m8.
24253 (ASM_CPU_SPEC): Likewise.
24254 * config/sparc/sparc-opts.h (enum processor_type): Add
24256 * config/sparc/sparc.c (m8_costs): New struct.
24257 (sparc_option_override): Handle TARGET_CPU_m8.
24258 (sparc32_initialize_trampoline): Likewise.
24259 (sparc64_initialize_trampoline): Likewise.
24260 (sparc_issue_rate): Likewise.
24261 (sparc_register_move_cost): Likewise.
24262 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
24263 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
24264 (ASM_CPU64_DEFAULT_SPEC): Likewise.
24265 (CPP_CPU_SPEC): Handle M8.
24266 (ASM_CPU_SPEC): Likewise.
24267 (AS_M8_FLAG): Define.
24268 * config/sparc/sparc.md: Add m8 to the cpu attribute.
24269 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
24270 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
24272 * configure: Regenerate.
24273 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
24276 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24278 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
24280 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
24281 ("*movdi_insn_sp32"): Do not set v3pipe.
24282 ("*movsi_insn"): Likewise.
24283 ("*movdi_insn_sp64"): Likewise.
24284 ("*movsf_insn"): Likewise.
24285 ("*movdf_insn_sp32"): Likewise.
24286 ("*movdf_insn_sp64"): Likewise.
24287 ("*zero_extendsidi2_insn_sp64"): Likewise.
24288 ("*sign_extendsidi2_insn"): Likewise.
24289 ("*mov<VM32:mode>_insn"): Likewise.
24290 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24291 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24292 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24293 ("<vlop:code><VL:mode>3"): Likewise.
24294 ("*not_<vlop:code><VL:mode>3"): Likewise.
24295 ("*nand<VL:mode>_vis"): Likewise.
24296 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
24297 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
24298 ("one_cmpl<VL:mode>2"): Likewise.
24299 ("faligndata<VM64:mode>_vis"): Likewise.
24300 ("alignaddrsi_vis"): Likewise.
24301 ("alignaddrdi_vis"): Likweise.
24302 ("alignaddrlsi_vis"): Likewise.
24303 ("alignaddrldi_vis"): Likewise.
24304 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24305 ("bmaskdi_vis"): Likewise.
24306 ("bmasksi_vis"): Likewise.
24307 ("bshuffle<VM64:mode>_vis"): Likewise.
24308 ("cmask8<P:mode>_vis"): Likewise.
24309 ("cmask16<P:mode>_vis"): Likewise.
24310 ("cmask32<P:mode>_vis"): Likewise.
24311 ("pdistn<P:mode>_vis"): Likewise.
24312 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24314 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24316 * config/sparc/sparc.md ("subtype"): New insn attribute.
24317 ("*wrgsr_sp64"): Set insn subtype.
24318 ("*rdgsr_sp64"): Likewise.
24319 ("alignaddrsi_vis"): Likewise.
24320 ("alignaddrdi_vis"): Likewise.
24321 ("alignaddrlsi_vis"): Likewise.
24322 ("alignaddrldi_vis"): Likewise.
24323 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24324 ("fexpand_vis"): Likewise.
24325 ("fpmerge_vis"): Likewise.
24326 ("faligndata<VM64:mode>_vis"): Likewise.
24327 ("bshuffle<VM64:mode>_vis"): Likewise.
24328 ("cmask8<P:mode>_vis"): Likewise.
24329 ("cmask16<P:mode>_vis"): Likewise.
24330 ("cmask32<P:mode>_vis"): Likewise.
24331 ("fchksm16_vis"): Likewise.
24332 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
24333 ("fmean16_vis"): Likewise.
24334 ("fp<plusminus_insn>64_vis"): Likewise.
24335 ("<plusminus_insn>v8qi3"): Likewise.
24336 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24337 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
24338 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
24339 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
24340 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
24341 ("*movqi_insn"): Likewise.
24342 ("*movhi_insn"): Likewise.
24343 ("*movsi_insn"): Likewise.
24344 ("movsi_pic_gotdata_op"): Likewise.
24345 ("*movdi_insn_sp32"): Likewise.
24346 ("*movdi_insn_sp64"): Likewise.
24347 ("movdi_pic_gotdata_op"): Likewise.
24348 ("*movsf_insn"): Likewise.
24349 ("*movdf_insn_sp32"): Likewise.
24350 ("*movdf_insn_sp64"): Likewise.
24351 ("*zero_extendhisi2_insn"): Likewise.
24352 ("*zero_extendqihi2_insn"): Likewise.
24353 ("*zero_extendqisi2_insn"): Likewise.
24354 ("*zero_extendqidi2_insn"): Likewise.
24355 ("*zero_extendhidi2_insn"): Likewise.
24356 ("*zero_extendsidi2_insn_sp64"): Likewise.
24357 ("ldfsr"): Likewise.
24358 ("prefetch_64"): Likewise.
24359 ("prefetch_32"): Likewise.
24360 ("tie_ld32"): Likewise.
24361 ("tie_ld64"): Likewise.
24362 ("*tldo_ldub_sp32"): Likewise.
24363 ("*tldo_ldub1_sp32"): Likewise.
24364 ("*tldo_ldub2_sp32"): Likewise.
24365 ("*tldo_ldub_sp64"): Likewise.
24366 ("*tldo_ldub1_sp64"): Likewise.
24367 ("*tldo_ldub2_sp64"): Likewise.
24368 ("*tldo_ldub3_sp64"): Likewise.
24369 ("*tldo_lduh_sp32"): Likewise.
24370 ("*tldo_lduh1_sp32"): Likewise.
24371 ("*tldo_lduh_sp64"): Likewise.
24372 ("*tldo_lduh1_sp64"): Likewise.
24373 ("*tldo_lduh2_sp64"): Likewise.
24374 ("*tldo_lduw_sp32"): Likewise.
24375 ("*tldo_lduw_sp64"): Likewise.
24376 ("*tldo_lduw1_sp64"): Likewise.
24377 ("*tldo_ldx_sp64"): Likewise.
24378 ("*mov<VM32:mode>_insn"): Likewise.
24379 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24380 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24382 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24384 * config/sparc/sparc.md ("type"): New insn type viscmp.
24385 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
24387 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
24388 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
24389 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24390 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
24392 ("n7_vis_logical_11cycle"): Likewise.
24393 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
24394 * config/sparc/niagara2.md ("niag3_vis": Likewise.
24395 * config/sparc/niagara.md ("niag_vis"): Likewise.
24396 * config/sparc/ultra3.md ("us3_fga"): Likewise.
24397 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
24399 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24401 * config/sparc/sparc.md: New instruction type `bmask'.
24402 (bmaskdi_vis): Use the `bmask' type.
24403 (bmasksi_vis): Likewise.
24404 * config/sparc/ultra3.md (us3_array): Likewise.
24405 * config/sparc/niagara7.md (n7_array): Likewise.
24406 * config/sparc/niagara4.md (n4_array): Likewise.
24407 * config/sparc/niagara2.md (niag2_vis): Likewise.
24408 (niag3_vis): Likewise.
24409 * config/sparc/niagara.md (niag_vis): Likewise.
24411 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24413 * ipa-comdats.c: Remove optimize check from gate.
24414 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
24415 for functions not optimized.
24416 (ipa_fn_summary_read): Skip optimize check.
24417 (ipa_fn_summary_write): Likewise.
24418 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
24420 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
24422 (can_inline_edge_p): Check flag_pcc_struct_return for match.
24423 (check_callers): Give up on caller which is not optimized.
24424 (inline_small_functions): Likewise.
24425 (ipa_inline): Do not give up when not optimizing.
24426 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
24427 away unoptimizes cdtors.
24428 (whole_program_function_and_variable_visibility): Do
24429 ipa_discover_readonly_nonaddressable_vars in LTO mode.
24430 * ipa.c (process_references): Do not check optimize.
24431 (symbol_table::remove_unreachable_nodes): Update optimize check.
24432 (set_writeonly_bit): Update optimize check.
24433 (pass_ipa_cdtor_merge::gate): Do not check optimize.
24434 (pass_ipa_single_use::gate): Remove.
24436 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24438 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
24439 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
24440 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
24441 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
24442 permute_load, permute_store, adjust_extract, adjust_splat,
24443 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
24444 replace_swap_with_copy, dump_swap_insn_table,
24445 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
24446 recombine_lvx_pattern, recombine_stvx_pattern,
24447 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
24448 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
24449 to file rs6000-p8swap.c.
24450 * config/rs6000/rs6000-p8swap.c: New file.
24451 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
24452 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
24453 and rs6000*-*-* targets.
24455 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24457 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
24459 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24461 * lto-wrapper.c (merge_and_complain): Do not merge
24462 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
24463 fsigned_zeros, ftrapping_math, fwrapv.
24464 (append_compiler_options): Do not track these options.
24465 (append_linker_options): Likewie
24467 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24469 * cgraphunit.c (cgraph_node::finalize_function): When
24470 !flag_toplevel_reorde set no_reorder flag.
24471 (varpool_node::finalize_decl): Likewise.
24472 (symbol_table::compile): Drop no toplevel reorder path.
24474 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24476 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
24477 edges; zero probability is not better than uninitialized.
24479 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24481 * asan.h (asan_sanitize_allocas_p): Declare.
24482 * asan.c (asan_sanitize_allocas_p): New function.
24483 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
24484 (handle_builtin_alloca): Likewise.
24485 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
24486 if !asan_sanitize_allocas_p.
24487 * params.def (asan-instrument-allocas): Add new option.
24488 * params.h (ASAN_PROTECT_ALLOCAS): Define.
24489 * opts.c (common_handle_option): Disable allocas sanitization for
24492 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24494 * asan.c: Include gimple-fold.h.
24495 (get_last_alloca_addr): New function.
24496 (handle_builtin_stackrestore): Likewise.
24497 (handle_builtin_alloca): Likewise.
24498 (asan_emit_allocas_unpoison): Likewise.
24499 (get_mem_refs_of_builtin_call): Add new parameter, remove const
24500 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
24501 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
24502 (instrument_builtin_call): Pass gimple iterator to
24503 get_mem_refs_of_builtin_call.
24504 (last_alloca_addr): New global.
24505 * asan.h (asan_emit_allocas_unpoison): Declare.
24506 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
24507 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
24508 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
24509 if function calls alloca.
24510 * gimple-fold.c (replace_call_with_value): Remove static keyword.
24511 * gimple-fold.h (replace_call_with_value): Declare.
24512 * internal-fn.c: Include asan.h.
24513 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
24514 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
24516 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24518 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
24519 (C_SELFTEST_FLAGS): New.
24520 (CPP_SELFTEST_FLAGS): New.
24521 (SELFTEST_DEPS): New, from deps of s-selftest.
24522 (C_SELFTEST_DEPS): New, from deps of s-selftest.
24523 (CPP_SELFTEST_DEPS): New.
24524 (selftest): Add dependency on s-selftest-c++.
24525 (s-selftest): Rename to...
24526 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
24527 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
24528 than SELFTEST_FLAGS.
24529 (selftest-gdb): Rename to...
24530 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
24532 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
24533 (selftest-valgrind): Rename to...
24534 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
24536 (selftest-valgrind): Reintroduce as an alias for
24537 selftest-c-valgrind.
24538 (s-selftest-c++): New.
24539 (selftest-c++-gdb): New.
24540 (selftest-c++-valgrind): New.
24542 2017-07-06 Olivier Hainque <hainque@adacore.com>
24544 * gcc.c (process_command): When deciding if undefined variables
24545 should be ignored when processing specs, accept "gcc -v" as well.
24547 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24549 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
24550 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
24552 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24554 * config/arm/arm-cpus.in (armv8-r): Add new entry.
24555 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
24556 * config/arm/arm-tables.opt: Regenerate.
24557 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
24559 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
24561 2017-07-06 Carl Love <cel@us.ibm.com>
24563 * ChangeLog: Clean up from mid air collision
24565 2017-07-06 Carl Love <cel@us.ibm.com>
24567 * config/rs6000/rs6000-c.c: Add support for built-in functions
24568 vector signed int vec_subc (vector signed int, vector signed int);
24569 vector signed __int128 vec_subc (vector signed __int128,
24570 vector signed __int128);
24571 vector unsigned __int128 vec_subc (vector unsigned __int128,
24572 vector unsigned __int128);
24573 vector signed int vec_sube (vector signed int, vector signed int,
24574 vector signed int);
24575 vector unsigned int vec_sube (vector unsigned int,
24576 vector unsigned int,
24577 vector unsigned int);
24578 vector signed __int128 vec_sube (vector signed __int128,
24579 vector signed __int128,
24580 vector signed__int128);
24581 vector unsigned __int128 vec_sube (vector unsigned __int128,
24582 vector unsigned __int128,
24583 vector unsigned __int128);
24584 vector signed int vec_subec (vector signed int, vector signed int,
24585 vector signed int);
24586 vector unsigned int vec_subec (vector unsigned int,
24587 vector unsigned int,
24588 vector unsigned int);
24589 vector signed __int128 vec_subec (vector signed __int128,
24590 vector signed __int128,
24591 vector signed__int128);
24592 vector unsigned __int128 vec_subec (vector unsigned __int128,
24593 vector unsigned __int128,
24594 vector unsigned __int128);
24595 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
24596 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
24597 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
24598 BU_ALTIVEC_OVERLOAD_X definitions.
24599 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
24600 * doc/extend.texi: Update the built-in documentation file for the new
24601 built-in functions.
24603 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24606 * diagnostic-show-locus.c (layout::layout): Use start and finish
24607 spelling location for the start and finish of each range.
24608 * genmatch.c (linemap_client_expand_location_to_spelling_point):
24609 Add unused aspect param.
24610 * input.c (expand_location_1): Add "aspect" param, and use it
24611 to access the correct part of the location.
24612 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
24614 (expand_location_to_spelling_point): Likewise.
24615 (linemap_client_expand_location_to_spelling_point): Add "aspect"
24616 param, and pass it to expand_location_1.
24618 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24620 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24621 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24622 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24623 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24624 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24625 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24626 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24627 _mm_maskz_getmant_ss): New intrinsics.
24628 (__builtin_ia32_getexpss128_mask): Changed to ...
24629 __builtin_ia32_getexpss128_round ... this.
24630 (__builtin_ia32_getexpsd128_mask): Changed to ...
24631 __builtin_ia32_getexpsd128_round ... this.
24632 * config/i386/i386-builtin-types.def
24633 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24634 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24635 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24636 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24637 __builtin_ia32_getmantss_mask_round): New builtins.
24638 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24639 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24640 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24641 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24642 * config/i386/sse.md
24643 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24644 avx512f_sgetexp<mode><mask_scalar_name>
24645 <round_saeonly_scalar_name> ... this.
24646 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24647 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24648 vgetexp<ssescalarmodesuffix>
24649 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24650 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24651 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24652 avx512f_vgetmant<mode><mask_scalar_name>
24653 <round_saeonly_scalar_name> ... this.
24654 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24655 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24656 vgetmant<ssescalarmodesuffix>
24657 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24658 %0<mask_scalar_operand4>, %1, %2
24659 <round_saeonly_scalar_mask_op4>, %3} ... this.
24660 * config/i386/subst.md (mask_scalar_operand4,
24661 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24662 round_saeonly_scalar_nimm_predicate): New subst attributes.
24664 2017-07-06 Julia Koval <julia.koval@intel.com>
24666 * config/i386/i386.c (ix86_erase_embedded_rounding):
24667 Remove code for old rounding pattern.
24669 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24671 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24673 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24675 * doc/sourcebuild.texi (Test Directives, Variants of
24676 dg-require-support): Add documentation for dg-require-stack-check.
24678 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24680 * config/i386/subst.md (mask_scalar, round_scalar,
24681 round_saeonly_scalar): New meta-templates.
24682 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24683 round_scalar_mask_operand3, round_scalar_mask_op3,
24684 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24685 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24686 round_saeonly_scalar_constraint,
24687 round_saeonly_scalar_prefix): New subst attribute.
24688 * config/i386/sse.md
24689 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24690 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24691 <round_scalar_name> ... this.
24692 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24693 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24694 <round_scalar_name> ... this.
24695 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24696 <sse>_vm<code><mode>3<mask_scalar_name>
24697 <round_saeonly_scalar_name> ... this.
24698 (v<plusminus_mnemonic><ssescalarmodesuffix>
24699 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24700 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24701 v<plusminus_mnemonic><ssescalarmodesuffix>
24702 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24703 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24704 (v<multdiv_mnemonic><ssescalarmodesuffix>
24705 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24706 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24707 v<multdiv_mnemonic><ssescalarmodesuffix>
24708 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24709 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24710 (v<maxmin_float><ssescalarmodesuffix>
24711 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24712 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24713 v<maxmin_float><ssescalarmodesuffix>
24714 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24715 %0<mask_scalar_operand3>, %1, %<iptr>2
24716 <round_saeonly_scalar_mask_op3>} ... this.
24718 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24720 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24721 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24723 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24724 Alan Hayward <alan.hayward@arm.com>
24725 David Sherwood <david.sherwood@arm.com>
24727 * combine.c (simplify_if_then_else): Remove "enum" before
24729 * compare-elim.c (can_eliminate_compare): Likewise.
24730 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24732 (aarch64_lookup_simd_builtin_type): Likewise.
24733 (aarch64_simd_builtin_type): Likewise.
24734 (aarch64_init_simd_builtin_types): Likewise.
24735 (aarch64_simd_expand_args): Likewise.
24736 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24738 (aarch64_reverse_mask): Likewise.
24739 (aarch64_simd_emit_reg_reg_move): Likewise.
24740 (aarch64_gen_adjusted_ldpstp): Likewise.
24741 (aarch64_ccmp_mode_to_code): Likewise.
24742 (aarch64_operands_ok_for_ldpstp): Likewise.
24743 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24744 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24746 (aarch64_min_divisions_for_recip_mul): Likewise.
24747 (aarch64_reassociation_width): Likewise.
24748 (aarch64_get_condition_code_1): Likewise.
24749 (aarch64_simd_emit_reg_reg_move): Likewise.
24750 (aarch64_simd_attr_length_rglist): Likewise.
24751 (aarch64_reverse_mask): Likewise.
24752 (aarch64_operands_ok_for_ldpstp): Likewise.
24753 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24754 (aarch64_gen_adjusted_ldpstp): Likewise.
24755 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24757 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24758 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24759 (arm_lookup_simd_builtin_type): Likewise.
24760 (arm_simd_builtin_type): Likewise.
24761 (arm_init_simd_builtin_types): Likewise.
24762 (arm_expand_builtin_args): Likewise.
24763 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24764 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24765 (ft32_setup_incoming_varargs): Likewise.
24766 (ft32_function_arg): Likewise.
24767 (ft32_function_arg_advance): Likewise.
24768 (ft32_pass_by_reference): Likewise.
24769 (ft32_arg_partial_bytes): Likewise.
24770 (ft32_valid_pointer_mode): Likewise.
24771 (ft32_addr_space_pointer_mode): Likewise.
24772 (ft32_addr_space_legitimate_address_p): Likewise.
24773 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24775 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24776 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24777 (iamcu_alignment): Likewise.
24778 (canonicalize_vector_int_perm): Likewise.
24779 (ix86_noce_conversion_profitable_p): Likewise.
24780 (ix86_mpx_bound_mode): Likewise.
24781 (ix86_operands_ok_for_move_multiple): Likewise.
24782 * config/microblaze/microblaze-protos.h
24783 (microblaze_expand_conditional_branch_reg): Likewise.
24784 * config/microblaze/microblaze.c
24785 (microblaze_expand_conditional_branch_reg): Likewise.
24786 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24788 (rs6000_reassociation_width): Likewise.
24789 (rs6000_invalid_binary_op): Likewise.
24790 (fusion_p9_p): Likewise.
24791 (emit_fusion_p9_load): Likewise.
24792 (emit_fusion_p9_store): Likewise.
24793 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24795 (riscv_hard_regno_mode_ok_p): Likewise.
24796 (riscv_address_insns): Likewise.
24797 (riscv_split_symbol): Likewise.
24798 (riscv_legitimize_move): Likewise.
24799 (riscv_function_value): Likewise.
24800 (riscv_hard_regno_nregs): Likewise.
24801 (riscv_expand_builtin): Likewise.
24802 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24803 (riscv_build_integer): Likewise.
24804 (riscv_split_integer): Likewise.
24805 (riscv_legitimate_constant_p): Likewise.
24806 (riscv_cannot_force_const_mem): Likewise.
24807 (riscv_regno_mode_ok_for_base_p): Likewise.
24808 (riscv_valid_base_register_p): Likewise.
24809 (riscv_valid_offset_p): Likewise.
24810 (riscv_valid_lo_sum_p): Likewise.
24811 (riscv_classify_address): Likewise.
24812 (riscv_legitimate_address_p): Likewise.
24813 (riscv_address_insns): Likewise.
24814 (riscv_load_store_insns): Likewise.
24815 (riscv_force_binary): Likewise.
24816 (riscv_split_symbol): Likewise.
24817 (riscv_force_address): Likewise.
24818 (riscv_legitimize_address): Likewise.
24819 (riscv_move_integer): Likewise.
24820 (riscv_legitimize_const_move): Likewise.
24821 (riscv_legitimize_move): Likewise.
24822 (riscv_address_cost): Likewise.
24823 (riscv_subword): Likewise.
24824 (riscv_output_move): Likewise.
24825 (riscv_canonicalize_int_order_test): Likewise.
24826 (riscv_emit_int_order_test): Likewise.
24827 (riscv_function_arg_boundary): Likewise.
24828 (riscv_pass_mode_in_fpr_p): Likewise.
24829 (riscv_pass_fpr_single): Likewise.
24830 (riscv_pass_fpr_pair): Likewise.
24831 (riscv_get_arg_info): Likewise.
24832 (riscv_function_arg): Likewise.
24833 (riscv_function_arg_advance): Likewise.
24834 (riscv_arg_partial_bytes): Likewise.
24835 (riscv_function_value): Likewise.
24836 (riscv_pass_by_reference): Likewise.
24837 (riscv_setup_incoming_varargs): Likewise.
24838 (riscv_print_operand): Likewise.
24839 (riscv_elf_select_rtx_section): Likewise.
24840 (riscv_save_restore_reg): Likewise.
24841 (riscv_for_each_saved_reg): Likewise.
24842 (riscv_register_move_cost): Likewise.
24843 (riscv_hard_regno_mode_ok_p): Likewise.
24844 (riscv_hard_regno_nregs): Likewise.
24845 (riscv_class_max_nregs): Likewise.
24846 (riscv_memory_move_cost): Likewise.
24847 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24848 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24849 (rl78_addr_space_address_mode): Likewise.
24850 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24852 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24853 (rs6000_reassociation_width): Likewise.
24854 (rs6000_invalid_binary_op): Likewise.
24855 (fusion_p9_p): Likewise.
24856 (emit_fusion_p9_load): Likewise.
24857 (emit_fusion_p9_store): Likewise.
24858 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24859 (ok_for_simple_move_operands): Likewise.
24860 (ok_for_simple_move_strict_operands): Likewise.
24861 (ok_for_simple_arith_logic_operands): Likewise.
24862 (visium_legitimize_reload_address): Likewise.
24863 (visium_select_cc_mode): Likewise.
24864 (output_cbranch): Likewise.
24865 (visium_split_double_move): Likewise.
24866 (visium_expand_copysign): Likewise.
24867 (visium_expand_int_cstore): Likewise.
24868 (visium_expand_fp_cstore): Likewise.
24869 * config/visium/visium.c (visium_pass_by_reference): Likewise.
24870 (visium_function_arg): Likewise.
24871 (visium_function_arg_advance): Likewise.
24872 (visium_libcall_value): Likewise.
24873 (visium_setup_incoming_varargs): Likewise.
24874 (visium_legitimate_constant_p): Likewise.
24875 (visium_legitimate_address_p): Likewise.
24876 (visium_legitimize_address): Likewise.
24877 (visium_secondary_reload): Likewise.
24878 (visium_register_move_cost): Likewise.
24879 (visium_memory_move_cost): Likewise.
24880 (prepare_move_operands): Likewise.
24881 (ok_for_simple_move_operands): Likewise.
24882 (ok_for_simple_move_strict_operands): Likewise.
24883 (ok_for_simple_arith_logic_operands): Likewise.
24884 (visium_function_value_1): Likewise.
24885 (rtx_ok_for_offset_p): Likewise.
24886 (visium_legitimize_reload_address): Likewise.
24887 (visium_split_double_move): Likewise.
24888 (visium_expand_copysign): Likewise.
24889 (visium_expand_int_cstore): Likewise.
24890 (visium_expand_fp_cstore): Likewise.
24891 (visium_split_cstore): Likewise.
24892 (visium_select_cc_mode): Likewise.
24893 (visium_split_cbranch): Likewise.
24894 (output_cbranch): Likewise.
24895 (visium_print_operand_address): Likewise.
24896 * expmed.c (flip_storage_order): Likewise.
24897 * expmed.h (emit_cstore): Likewise.
24898 (flip_storage_order): Likewise.
24899 * genrecog.c (validate_pattern): Likewise.
24900 * hsa-gen.c (gen_hsa_addr): Likewise.
24901 * internal-fn.c (expand_arith_overflow): Likewise.
24902 * ira-color.c (allocno_copy_cost_saving): Likewise.
24903 * lra-assigns.c (find_hard_regno_for_1): Likewise.
24904 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
24905 (process_invariant_for_inheritance): Likewise.
24906 * lra-eliminations.c (move_plus_up): Likewise.
24907 * omp-low.c (lower_oacc_reductions): Likewise.
24908 * simplify-rtx.c (simplify_subreg): Likewise.
24909 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
24910 (TARGET_CHKP_BOUND_MODE): Likewise..
24911 * targhooks.c (default_chkp_bound_mode): Likewise.
24912 (default_setup_incoming_vararg_bounds): Likewise.
24913 * targhooks.h (default_chkp_bound_mode): Likewise.
24914 (default_setup_incoming_vararg_bounds): Likewise.
24915 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
24916 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
24917 (have_whole_vector_shift): Likewise.
24918 * tree-vect-stmts.c (vectorizable_load): Likewise.
24919 * doc/tm.texi: Regenerate.
24921 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24923 Graceful degrade if Binutils PR21472 is not available.
24926 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
24927 .rodata in flash test fails.
24928 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
24929 * confgure: Regenerate.
24930 * config.in: Regenerate.
24931 * config/avr/avr.c (avr_asm_named_section)
24932 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
24933 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
24934 (avr_asm_init_sections): Same.
24936 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24938 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
24939 (fma<VH:mode>4_intrinsic): Likewise.
24940 (*fmsub<VCVTF:mode>4): Likewise.
24941 (*fmsub<VH:mode>4_intrinsic): Likewise.
24943 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24946 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
24947 Don't depend on "optimize > 0".
24948 (out_movhi_r_mr, out_movqi_mr_r): Same.
24949 (out_movhi_mr_r, out_movqi_r_mr): Same.
24950 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
24951 io_address_operand on "optimize > 0".
24953 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24955 * tree-loop-distribution.c: Add general explanantion on the pass.
24956 (generate_loops_for_partition): Mark distributed loop.
24957 (pg_add_dependence_edges): New parameter. Handle alias data
24958 dependence specially and record it in the parameter if asked.
24959 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
24960 (init_partition_graph_vertices, add_partition_graph_edge): New.
24961 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
24962 (free_partition_graph_vdata, build_partition_graph): New.
24963 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
24964 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
24965 (data_ref_segment_size, latch_dominated_by_data_ref): New.
24966 (compute_alias_check_pairs, version_loop_by_alias_check): New.
24967 (version_for_distribution_p, finalize_partitions): New.
24968 (distribute_loop): Handle alias data dependence specially. Factor
24969 out loop fusion code as functions and call these functions.
24971 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24973 * tree-loop-distribution.c (classify_partition): New parameter and
24974 better handle reduction statement.
24975 (rdg_build_partitions): Revise comment.
24976 (distribute_loop): Compute statements in all partitions and pass it
24977 to classify_partition.
24979 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24981 * tree-loop-distribution.c (enum partition_type): New.
24982 (struct partition): New field type.
24983 (partition_merge_into): Add parameter. Update partition type.
24984 (data_dep_in_cycle_p, update_type_for_merge): New functions.
24985 (build_rdg_partition_for_vertex): Compute partition type.
24986 (rdg_build_partitions): Dump partition type.
24987 (distribute_loop): Update calls to partition_merge_into.
24989 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24991 * tree-loop-distribution.c (struct ddr_hasher): New.
24992 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
24994 (classify_partition): Call get_data_dependence.
24995 (pg_add_dependence_edges): Ditto.
24996 (distribute_loop): Release data dependence hash table.
24998 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25000 * tree-loop-distribution.c (ref_base_address): Delete.
25001 (similar_memory_accesses): Rename ...
25002 (share_memory_accesses): ... to this. Check if partitions access
25003 the same memory reference.
25004 (distribute_loop): Call share_memory_accesses.
25006 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25008 * tree-loop-distribution.c (struct partition): New field recording
25009 its data reference.
25010 (partition_alloc, partition_free): Init and release data refs.
25011 (partition_merge_into): Merge data refs.
25012 (build_rdg_partition_for_vertex): Collect data refs for partition.
25013 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25015 (distribute_loop): Remve data refs from vertice data of partition
25018 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25020 * tree-loop-distribution.c (params.h): Include header file.
25021 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25022 (datarefs_vec): New global var.
25023 (create_rdg_vertices): Use datarefs_vec directly.
25024 (free_rdg): Don't free data references.
25025 (build_rdg): Update use. Don't free data references.
25026 (distribute_loop): Compute global variable for data references.
25027 Bail out if there are too many data references.
25029 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25031 * tree-loop-distribution.c (loop_nest): New global var.
25032 (build_rdg): Use loop directly, rather than loop nest.
25033 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25035 (distribute_loop): Compute global variable loop nest. Update use.
25037 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25039 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
25040 (partition_merge_into): New parameter. Dump reason for fusion.
25041 (distribute_loop): Update use of partition_merge_into.
25043 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25045 * tree-loop-distribution.c (bb_top_order_index): New.
25046 (bb_top_order_index_size, bb_top_order_cmp): New.
25047 (stmts_from_loop): Use topological order.
25048 (pass_loop_distribution::execute): Compute and release topological
25049 order for basic blocks.
25051 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25053 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25056 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25058 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25059 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25060 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25061 * internal-fn.def (LOOP_DIST_ALIAS): New.
25062 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25063 (fold_loop_internal_call): ... this.
25064 (vect_loop_dist_alias_call): New function.
25065 (set_uid_loop_bbs): Call fold_loop_internal_call.
25066 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25069 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25072 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25073 Require dead FLAGS_REG at the beginning of a peephole.
25075 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25078 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25079 arguments in the call to __builtin_ia32_sbb_u32.
25080 (_subborrow_u64): Swap _X and _Y arguments in the call to
25081 __builtin_ia32_sbb_u64.
25083 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25086 * tree-vrp.c (compare_assert_loc): Turn into a function template
25087 with stable template parameter. Only test if a->e is NULL,
25088 !a->e == !b->e has been verified already. Use e == NULL or
25089 e != NULL instead of e or ! e tests. If stable is true, don't use
25090 iterative_hash_expr, on the other side allow a or b or both NULL
25091 and sort the NULLs last.
25092 (process_assert_insertions): Sort using compare_assert_loc<false>
25093 instead of compare_assert_loc, later sort using
25094 compare_assert_loc<true> before calling process_assert_insertions_for
25095 in a loop. Use break instead of continue once seen NULL pointer.
25097 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25099 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25100 Cortex-R7 and Cortex-R8 processors.
25102 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25104 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25105 uninitialized while src is not.
25107 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
25109 * common/config/arm/arm-common.c: Adjust include path for
25111 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
25112 (arm-cpu.h): Create in build directory. Adjust dependency rules.
25113 (arm-cpu-data.h): Likewise.
25114 (arm-cpu-cdata.h): Likewise.
25115 * config/arm/arm-cpu.h: Delete.
25116 * config/arm/arm-cpu-cdata.h: Delete.
25117 * config/arm/arm-cpu-data.h: Delete.
25119 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
25121 * config/arm/arm-cpus.in (cortex-a55): New.
25122 (cortex-a75): Likewise.
25123 (cortex-a75.cortex-a55): Likewise.
25124 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
25126 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
25127 * config/arm/arm-cpu-cdata.h: Regenerate.
25128 * config/arm/arm-cpu-data.h: Regenerate.
25129 * config/arm/arm-cpu.h: Regenerate.
25130 * config/arm/arm-tables.opt: Regenerate.
25131 * config/arm/arm-tune.md: Regenerate.
25133 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25135 * haifa-sched.c (sched_create_recovery_edges): Update profile.
25137 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25139 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
25142 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
25144 PR tree-optimization/81292
25145 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
25146 full_string_p, also call adjust_related_strinfos if the adjustment
25147 is simple, otherwise invalidate related strinfos.
25149 2017-07-04 Martin Liska <mliska@suse.cz>
25152 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
25153 newly created variable as DECL_IGNORED_P.
25155 2017-07-04 Martin Liska <mliska@suse.cz>
25158 * ipa-inline.c (inline_small_functions):
25159 Use xstrdup_for_dump.
25161 2017-07-04 Tom de Vries <tom@codesourcery.com>
25163 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
25165 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25168 * config/darwin.c (darwin_function_switched_text_sections):
25171 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
25173 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
25175 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25177 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
25179 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25181 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
25182 min_profitable_iters, and th as inclusive lower bounds.
25183 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
25184 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
25185 for min_profitable_iters and min_profitable_estimate.
25186 (vect_transform_loop): Treat th as an inclusive lower bound.
25187 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25189 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25192 * config/darwin.c (darwin_function_switched_text_sections):
25193 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
25194 in two pieces, and suppress the use of buf.
25196 2017-07-03 Nathan Sidwell <nathan@acm.org>
25198 * hash-table.h (hash_table_mod1): Fix indentation.
25200 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25202 PR middle-end/81290
25203 * predict.c (force_edge_cold): Be more careful about propagation
25205 * profile-count.h (profile_probability::guessed,
25206 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
25208 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
25210 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
25212 * doc/invoke.texi (rcpc architecture extension): Document it.
25214 2017-07-03 Richard Biener <rguenther@suse.de>
25216 PR tree-optimization/60510
25217 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
25218 the scalar reduction PHI and use it.
25219 (vectorizable_reduction): Properly guard the single_defuse_cycle
25220 path for non-SLP reduction chains where we cannot use it.
25221 Rework reduc_def/index and vector type deduction. Rework
25222 vector operand gathering during reduction op code-gen.
25223 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
25224 chains dissolve the chain and leave it to non-SLP reduction
25227 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25229 * tree-data-ref.h (dr_alignment): Declare.
25230 * tree-data-ref.c (dr_alignment): New function.
25231 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
25232 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
25234 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
25236 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25238 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
25239 and base_misalignment fields.
25240 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
25241 * tree-data-ref.c: Include builtins.h.
25242 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
25243 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
25244 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
25245 * tree-vect-data-refs.c: Include tree-cfg.h.
25246 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
25247 fields instead of calculating an alignment here.
25248 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
25249 innermost_loop_behavior fields.
25251 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25253 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
25255 (DR_STEP_ALIGNMENT): New macro.
25256 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
25257 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
25258 (create_data_ref): Print it.
25259 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
25260 to tell whether the step preserves vector (mis)alignment.
25261 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25262 Move the check for an integer step and generalise to all INTEGER_CST.
25263 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
25264 Print the outer step alignment.
25266 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25268 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
25269 with offset_alignment.
25270 (DR_ALIGNED_TO): Delete.
25271 (DR_OFFSET_ALIGNMENT): New macro.
25272 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
25273 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
25274 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
25275 (create_data_ref): Likewise.
25276 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25277 (vect_analyze_data_refs): Likewise.
25278 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
25279 creating dummy innermost behavior.
25281 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25283 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
25284 with a "innermost_loop_behavior *" and refeence tree.
25285 * tree-data-ref.c (dr_analyze_innermost): Likewise.
25286 (create_data_ref): Update call accordingly.
25287 * tree-predcom.c (find_looparound_phi): Likewise.
25289 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25291 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
25292 fields with dr_wrt_vec_loop.
25293 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
25294 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
25295 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
25296 (vect_dr_behavior): New function.
25297 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25298 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
25299 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
25300 track whether the step preserves the misalignment.
25301 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25302 Use vect_dr_behavior.
25303 (vect_setup_realignment): Update call accordingly.
25304 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
25305 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
25306 call to vect_create_addr_base_for_vector_ref.
25307 (vect_create_cond_for_align_checks): Likewise.
25308 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
25309 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
25310 (vect_recog_mask_conversion_pattern): Likewise.
25311 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
25312 (new_stmt_vec_info): Remove redundant zeroing.
25314 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25316 * common/config/arm/arm-common.c (arm_be8_option): New function.
25317 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
25318 (ISA_ARMv6): Add isa_bit_be8.
25319 * config/arm/arm.h (arm_be8_option): Add prototype.
25320 (BE8_SPEC_FUNCTION): New define.
25321 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
25322 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
25323 (mlittle-endian): Similarly.
25324 (mbe8, mbe32): New options.
25325 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
25326 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
25328 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25330 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
25332 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25334 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
25335 (cleanup_tree_cfg_bb): Use it.
25336 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
25338 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
25340 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25343 * loop-doloop.c (add_test): Update profile.
25345 2017-07-03 Martin Liska <mliska@suse.cz>
25348 * sanopt.c (rewrite_usage_of_param): New function.
25349 (sanitize_rewrite_addressable_params): Likewise.
25350 (pass_sanopt::execute): Call rewrite_usage_of_param.
25352 2017-07-03 Richard Biener <rguenther@suse.de>
25354 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
25355 back to using VIEW_CONVERT_EXPR.
25357 2017-07-03 Martin Liska <mliska@suse.cz>
25360 * doc/extend.texi: Document when a resolver function is
25361 generated for target_clones.
25363 2017-07-03 Martin Liska <mliska@suse.cz>
25365 * asan.c (asan_emit_stack_protection): Unpoison just red zones
25366 and shadow memory of auto variables which are subject of
25367 use-after-scope sanitization.
25368 (asan_expand_mark_ifn): Add do set only when is_poison.
25370 2016-07-03 Richard Biener <rguenther@suse.de>
25372 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
25374 (vect_force_simple_reduction): Record reduction def -> phi mapping.
25375 (vectorizable_reduction): Perform reduction PHI creation when
25376 visiting a reduction PHI and adjust and simplify code generation
25377 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
25378 (vect_transform_loop): Visit reduction PHIs.
25379 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
25380 defs into the SLP tree.
25381 (vect_build_slp_tree): Reduction defs terminate the recursion.
25382 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
25384 (vect_get_vec_defs_for_stmt_copy): Export.
25385 (vect_get_vec_defs): Likewise.
25386 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
25388 (vect_get_vec_defs_for_stmt_copy): Declare.
25389 (vect_get_vec_defs): Likewise.
25391 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25393 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
25394 parameter with a "loop" parameter and use it instead of the
25395 loop containing DR_STMT. Don't check simple_iv when doing
25396 BB analysis. Describe the two analysis modes in the comment.
25398 2017-07-03 Tom de Vries <tom@codesourcery.com>
25400 PR tree-optimization/69468
25401 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
25402 (find_same_succ_bb): Handle ignore_edge_flags.
25404 2017-07-03 Tom de Vries <tom@codesourcery.com>
25406 PR tree-optimization/81192
25407 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
25409 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
25411 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
25413 2017-07-03 Tom de Vries <tom@codesourcery.com>
25415 PR tree-optimization/81192
25416 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
25417 BB_SAME_SUCC (bb) == NULL.
25419 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25421 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
25424 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25426 * dumpfile.c: Include profile-count.h
25427 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
25429 (insert_cond_bb): Update profile.
25430 * tree-cfg.h (insert_cond_bb): Update prototype.
25431 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
25432 * tree-dump.c: Do not include tree-cfg.
25434 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25436 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
25438 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25440 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
25443 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25445 * tree-complex.c (expand_complex_div_wide): update profile.
25447 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25448 Alan Hayward <alan.hayward@arm.com>
25449 David Sherwood <david.sherwood@arm.com>
25451 * Makefile.in (MACHMODE_H): Remove insn-modes.h
25452 (CORETYPES_H): New define.
25453 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
25454 (insn-modes-inline.h, s-modes-inline-h): New rules.
25455 (generated_files): Add insn-modes-inline.h.
25456 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
25457 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
25458 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
25459 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
25460 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
25461 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
25462 (build/gencodes.o, build/genconditions.o): Likewise.
25463 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
25464 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
25465 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
25466 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
25467 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
25468 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
25469 * coretypes.h: Include everything up to real.h for generators.
25470 Include insn-modes.h first. Include wide-int-print.h after
25471 wide-int.h. Include insn-modes-inline.h and then machmode.h.
25472 * machmode.h: Don't include insn-modes.h here.
25473 * function-tests.c: Remove includes of signop.h, machmode.h,
25474 double-int.h and wide-int.h.
25476 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
25478 * optc-save-gen.awk: Likewise.
25479 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
25480 * godump.c: Remove include of wide-int-print.h.
25481 * pretty-print.h: Likewise.
25482 * wide-int-print.cc: Likewise.
25483 * wide-int.cc: Likewise.
25484 * hash-map-tests.c: Remove include of signop.h.
25485 * hash-set-tests.c: Likewise.
25486 * rtl-tests.c: Likewise.
25487 * mkconfig.sh: Remove include of machmode.h.
25488 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
25490 (emit_insn_modes_inline_h): ...this new function. Emit the code
25491 into an insn-modes-inline.h header file, adding appropriate
25492 include guards and end comments.
25493 (emit_insn_modes_c_header): Remove include of machmode.h.
25494 (emit_min_insn_modes_c_header): Include coretypes.h rather than
25496 (main): Handle -i flag and call emit_insn_modes_inline_h when
25499 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25501 * tree-ssa-strlen.c (strinfo): Rename the length field to
25502 nonzero_chars. Add a full_string_p field.
25503 (compare_nonzero_chars, zero_length_string_p): New functions.
25504 (get_addr_stridx): Add an offset_out parameter.
25505 Use compare_nonzero_chars.
25506 (get_stridx): Update accordingly. Use compare_nonzero_chars.
25507 (new_strinfo): Update after above changes to strinfo.
25508 (set_endptr_and_length): Set full_string_p.
25509 (get_string_length): Update after above changes to strinfo.
25510 (unshare_strinfo): Update call to new_strinfo.
25511 (maybe_invalidate): Likewise.
25512 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
25513 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
25514 as a uhwi instead of an shwi. Update after above changes to
25515 strinfo and new_strinfo.
25516 (zero_length_string): Assert that chainsi contains full strings.
25517 Use zero_length_string_p. Update call to new_strinfo.
25518 (adjust_related_strinfos): Update after above changes to strinfo.
25519 Copy full_string_p from origsi.
25520 (adjust_last_stmt): Use zero_length_string_p.
25521 (handle_builtin_strlen): Update after above changes to strinfo and
25522 new_strinfo. Install the lhs as the string length if the previous
25523 entry didn't describe a full string.
25524 (handle_builtin_strchr): Update after above changes to strinfo
25526 (handle_builtin_strcpy): Likewise.
25527 (handle_builtin_strcat): Likewise.
25528 (handle_builtin_malloc): Likewise.
25529 (handle_pointer_plus): Likewise.
25530 (handle_builtin_memcpy): Likewise. Track nonzero characters
25531 that aren't necessarily followed by a nul terminator.
25532 (handle_char_store): Likewise.
25534 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25536 PR tree-optimization/80769
25537 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
25538 for malloc and calloc. Document the new invariant that all related
25539 strinfos have delayed lengths or none do.
25540 (verify_related_strinfos): Move earlier in file.
25541 (set_endptr_and_length): New function, split out from...
25542 (get_string_length): ...here. Also set the lengths of related
25544 (zero_length_string): Assert that chainsi has known (rather than
25546 (adjust_related_strinfos): Likewise.
25548 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25550 PR tree-optimization/81136
25551 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
25552 assert that two references with the same misalignment have the same
25553 compile-time misalignment if those compile-time misalignments
25556 2017-07-01 Andi Kleen <ak@linux.intel.com>
25558 * print-tree.c (print_node): Print all attributes.
25560 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25562 * cfg.c (scale_bbs_frequencies): New function.
25563 * cfg.h (scale_bbs_frequencies): Declare it.
25564 * cfgloopanal.c (single_likely_exit): Cleanup.
25565 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
25567 (scale_loop_profile): Likewise.
25568 (loop_version): Likewise.
25569 (create_empty_loop_on_edge): Update.
25570 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
25571 scale_loop_frequencies, scale_loop_profile, loopify,
25572 loop_version): Update prototypes.
25573 * modulo-sched.c (sms_schedule): Update.
25574 * predict.c (unlikely_executed_edge_p): Also check probability.
25575 (probably_never_executed_edge_p): Fix typo.
25576 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25577 * tree-parloops.c (gen_parallel_loop): Update.
25578 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
25579 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25580 * tree-ssa-loop-split.c (split_loop): Update.
25581 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25582 * tree-vect-loop-manip.c (vect_do_peeling): Update.
25583 (vect_loop_versioning): Update.
25584 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25586 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25588 * trans-mem.c (split_bb_make_tm_edge): Update profile.
25590 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25592 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
25593 to keep profile consistent.
25595 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25597 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
25598 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
25599 * profile-count.h (max_safe_multiplier): Make unsigned.
25600 (profile_count::guessed_zero): New.
25602 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25604 * bb-reorder.c (fix_up_crossing_landing_pad,
25605 fix_crossing_conditional_branches): Use make_single_succ_edge
25606 to keep profile consistent.
25608 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25610 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
25613 2017-07-01 Jakub Jelinek <jakub@redhat.com>
25616 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
25617 the right scopes, make sure cond_jump isn't preserved between multiple
25618 iterations. Search for fallthru edge whenever there are 3+ edges and
25619 use find_fallthru_edge for it.
25621 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25623 Patch by Alexander Monakov <amonakov@ispras.ru>
25624 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25625 probabilities consistently.
25627 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25629 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25630 emit_cmp_and_jump_insns.
25632 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25635 * tree-inline.c (expand_call_inline): Combine profile statuses.
25637 2017-06-30 Andrew Pinski <apinski@cavium.com>
25639 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25640 fold_stmt returned true.
25642 2017-06-30 Nathan Sidwell <nathan@acm.org>
25644 * ggc.h (empty_string): Delete.
25645 * cfgexpand.c (expand_asm_stmt): Use plain "".
25646 * optabs.c (expand_asm_memory_barrier): Likewise.
25647 * stringpool.c (empty_string): Delete.
25648 (digit_vector, digit_string): Delete.
25649 (ggc_alloc_string): Use plain "", don't optimize single digit
25650 strings. Use ggc_alloc_atomic.
25652 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25654 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25655 comparison set and one other set, use the cost of the non-comparison
25658 2017-06-30 Nathan Sidwell <nathan@acm.org>
25660 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25663 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25665 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25666 loops. Remove now unneeded calls to gimple_switch_set_label() that
25667 just set removed labels to NULL_TREE.
25669 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25671 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25672 (set_range_info): ...here. Only call set_range_info_raw if domain
25674 (set_nonzero_bits): Call set_range_info_raw.
25675 * tree-ssanames.h (set_range_info_raw): New.
25677 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25680 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25681 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25682 of nonimmediate_operand and <store_mask_constraint> instead of m
25683 for the input operand. For V8FI iterator, always split if input
25684 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25685 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25686 <store_mask_predicate> instead of register_operand and
25687 <store_mask_constraint> instead of v for the input operand. Make
25688 sure both operands aren't MEMs for if not <mask_applied>.
25690 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25692 * lto-wrapper.c (copy_file) Close both file descriptors before
25695 2017-06-30 Martin Liska <mliska@suse.cz>
25698 * multiple_target.c (create_dispatcher_calls): Make ifunc
25699 also for function that don't have calls or are not referenced.
25701 2017-06-30 Richard Biener <rguenther@suse.de>
25703 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25704 analyze the first scalar stmt. Move vector type computation
25705 for the BB case here from ...
25706 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25707 live operation processing in the SLP case properly.
25709 2017-06-30 Richard Biener <rguenther@suse.de>
25711 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25713 2017-06-30 Martin Liska <mliska@suse.cz>
25716 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25717 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25719 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25721 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25724 2017-06-30 Richard Biener <rguenther@suse.de>
25726 PR tree-optimization/81249
25727 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25728 condition reduction result to original scalar type.
25730 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25732 * profile-count.h (enum profile_quality): Fix typos and whitespace
25735 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25737 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25738 type for branch probabilities.
25740 2017-06-29 Julian Brown <julian@codesourcery.com>
25741 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25743 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25744 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25745 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25746 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25748 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25750 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25751 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25752 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25753 CC usage from generic code to here.
25754 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25755 CC usage into the target macros.
25757 2017-06-29 Maya Rashish <coypu@sdf.org>
25759 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25762 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25764 * arm/arm-builtins.c: Include profile-count.h
25765 * except.c (sjlj_emit_function_enter): Use
25766 profile_probability::unlikely.
25768 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25770 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25771 and tocrel_offset be pointer args rather than implicitly using
25773 (legitimate_constant_pool_address_p, rs6000_emit_move,
25774 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25775 tocrel_offset and use in toc_relative_expr_p call.
25776 (print_operand, print_operand_address): Use static tocrel_base_oac
25777 and tocrel_offset_oac.
25778 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25781 2017-06-29 Maya Rashish <coypu@sdf.org>
25783 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25785 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25787 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25788 objects, take into account only the alignment of 'op0' and 'mode1' if
25791 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25793 * ccmp.c (ccmp_tree_comparison_p): New function.
25794 (ccmp_candidate_p): Update to use above function.
25795 (get_compare_parts): New function.
25796 (expand_ccmp_next): Update to use new functions.
25797 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25799 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25800 take mode as argument.
25801 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25802 * expr.c (expand_expr_real_1): Pass mode as argument.
25804 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25806 * combine.c (combine_instructions): Print insns to dump_file, together
25809 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25811 * asan.c (asan_emit_stack_protection): Update.
25812 (create_cond_insert_point): Update.
25813 * auto-profile.c (afdo_propagate_circuit): Update.
25814 * basic-block.h (struct edge_def): Turn probability to
25815 profile_probability.
25816 (EDGE_FREQUENCY): Update.
25817 * bb-reorder.c (find_traces_1_round): Update.
25818 (better_edge_p): Update.
25819 (sanitize_hot_paths): Update.
25820 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25821 (make_single_succ_edge): Update.
25822 (check_bb_profile): Update.
25823 (dump_edge_info): Update.
25824 (update_bb_profile_for_threading): Update.
25825 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25826 probabilitycount to 0.
25827 * cfgbuild.c (compute_outgoing_frequencies): Update.
25828 * cfgcleanup.c (try_forward_edges): Update.
25829 (outgoing_edges_match): Update.
25830 (try_crossjump_to_edge): Update.
25831 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25832 (expand_gimple_tailcall): Update.
25833 (construct_init_block): Use make_single_succ_edge.
25834 (construct_exit_block): Use make_single_succ_edge.
25835 * cfghooks.c (verify_flow_info): Update.
25836 (redirect_edge_succ_nodup): Update.
25837 (split_edge): Update.
25838 (account_profile_record): Update.
25839 * cfgloopanal.c (single_likely_exit): Update.
25840 * cfgloopmanip.c (scale_loop_profile): Update.
25841 (set_zero_probability): Remove.
25842 (duplicate_loop_to_header_edge): Update.
25843 * cfgloopmanip.h (loop_version): Update prototype.
25844 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25845 (force_nonfallthru_and_redirect): Update.
25846 (update_br_prob_note): Update.
25847 (rtl_verify_edges): Update.
25848 (purge_dead_edges): Update.
25849 (rtl_lv_add_condition_to_bb): Update.
25850 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25851 * cgraphunit.c (init_lowered_empty_function): Update.
25852 (cgraph_node::expand_thunk): Update.
25853 * cilk-common.c: Include profile-count.h
25854 * dojump.c (inv): Remove.
25855 (jumpifnot): Update.
25856 (jumpifnot_1): Update.
25857 (do_jump_1): Update.
25859 (do_jump_by_parts_greater_rtx): Update.
25860 (do_compare_rtx_and_jump): Update.
25861 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25862 do_jump_1. do_compare_rtx_and_jump): Update prototype.
25863 * dwarf2cfi.c: Include profile-count.h
25864 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
25865 (sjlj_emit_dispatch_table): Likewise.
25866 * explow.c: Include profile-count.h
25867 * expmed.c (emit_store_flag_force): Update.
25868 (do_cmp_and_jump): Update.
25869 * expr.c (compare_by_pieces_d::generate): Update.
25870 (compare_by_pieces_d::finish_mode): Update.
25871 (emit_block_move_via_loop): Update.
25872 (store_expr_with_bounds): Update.
25873 (store_constructor): Update.
25874 (expand_expr_real_2): Update.
25875 (expand_expr_real_1): Update.
25876 * expr.h (try_casesi, try_tablejump): Update prototypes.
25877 * gimple-pretty-print.c (dump_probability): Update.
25878 (dump_profile): New.
25879 (dump_gimple_label): Update.
25880 (dump_gimple_bb_header): Update.
25881 * graph.c (draw_cfg_node_succ_edges): Update.
25882 * hsa-gen.c (convert_switch_statements): Update.
25883 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
25884 (find_if_case_1): Update.
25885 (find_if_case_2): Update.
25886 * internal-fn.c (expand_arith_overflow_result_store): Update.
25887 (expand_addsub_overflow): Update.
25888 (expand_neg_overflow): Update.
25889 (expand_mul_overflow): Update.
25890 (expand_vector_ubsan_overflow): Update.
25891 * ipa-cp.c (good_cloning_opportunity_p): Update.
25892 * ipa-split.c (split_function): Use make_single_succ_edge.
25893 * ipa-utils.c (ipa_merge_profiles): Update.
25894 * loop-doloop.c (add_test): Update.
25895 (doloop_modify): Update.
25896 * loop-unroll.c (compare_and_jump_seq): Update.
25897 (unroll_loop_runtime_iterations): Update.
25898 * lra-constraints.c (lra_inheritance): Update.
25899 * lto-streamer-in.c (input_cfg): Update.
25900 * lto-streamer-out.c (output_cfg): Update.
25901 * mcf.c (adjust_cfg_counts): Update.
25902 * modulo-sched.c (sms_schedule): Update.
25903 * omp-expand.c (expand_omp_for_init_counts): Update.
25904 (extract_omp_for_update_vars): Update.
25905 (expand_omp_ordered_sink): Update.
25906 (expand_omp_for_ordered_loops): Update.
25907 (expand_omp_for_generic): Update.
25908 (expand_omp_for_static_nochunk): Update.
25909 (expand_omp_for_static_chunk): Update.
25910 (expand_cilk_for): Update.
25911 (expand_omp_simd): Update.
25912 (expand_omp_taskloop_for_outer): Update.
25913 (expand_omp_taskloop_for_inner): Update.
25914 * omp-simd-clone.c (simd_clone_adjust): Update.
25915 * optabs.c (expand_doubleword_shift): Update.
25916 (expand_abs): Update.
25917 (emit_cmp_and_jump_insn_1): Update.
25918 (expand_compare_and_swap_loop): Update.
25919 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
25920 * predict.c (predictable_edge_p): Update.
25921 (edge_probability_reliable_p): Update.
25922 (set_even_probabilities): Update.
25923 (combine_predictions_for_insn): Update.
25924 (combine_predictions_for_bb): Update.
25925 (propagate_freq): Update.
25926 (estimate_bb_frequencies): Update.
25927 (force_edge_cold): Update.
25928 * profile-count.c (profile_count::dump): Add missing space into dump.
25929 (profile_count::debug): Add newline.
25930 (profile_count::differs_from_p): Explicitly convert to unsigned.
25931 (profile_count::stream_in): Update.
25932 (profile_probability::dump): New member function.
25933 (profile_probability::debug): New member function.
25934 (profile_probability::differs_from_p): New member function.
25935 (profile_probability::differs_lot_from_p): New member function.
25936 (profile_probability::stream_in): New member function.
25937 (profile_probability::stream_out): New member function.
25938 * profile-count.h (profile_count_quality): Rename to ...
25939 (profile_quality): ... this one.
25940 (profile_probability): New.
25941 (profile_count): Update.
25942 * profile.c (compute_branch_probabilities): Update.
25943 * recog.c (peep2_attempt): Update.
25944 * sched-ebb.c (schedule_ebbs): Update.
25945 * sched-rgn.c (find_single_block_region): Update.
25946 (compute_dom_prob_ps): Update.
25947 (schedule_region): Update.
25948 * sel-sched-ir.c (compute_succs_info): Update.
25949 * stmt.c (struct case_node): Update.
25950 (do_jump_if_equal): Update.
25951 (get_outgoing_edge_probs): Update.
25952 (conditional_probability): Update.
25953 (emit_case_dispatch_table): Update.
25954 (expand_case): Update.
25955 (expand_sjlj_dispatch_table): Update.
25956 (emit_case_nodes): Update.
25957 * targhooks.c: Update.
25958 * tracer.c (better_p): Update.
25959 (find_best_successor): Update.
25960 * trans-mem.c (expand_transaction): Update.
25961 * tree-call-cdce.c: Update.
25962 * tree-cfg.c (gimple_split_edge): Upate.
25963 (move_sese_region_to_fn): Upate.
25964 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
25965 * tree-eh.c (lower_resx): Upate.
25966 (cleanup_empty_eh_move_lp): Upate.
25967 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25968 * tree-inline.c (copy_edges_for_bb): Update.
25969 (copy_cfg_body): Update.
25970 * tree-parloops.c (gen_parallel_loop): Update.
25971 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
25972 (gimple_gen_time_profiler): Update.
25973 * tree-ssa-dce.c (remove_dead_stmt): Update.
25974 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25975 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
25976 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
25977 (unloop_loops): Update.
25978 (try_peel_loop): Update.
25979 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25980 * tree-ssa-loop-split.c (connect_loops): Update.
25981 (split_loop): Update.
25982 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25983 (hoist_guard): Update.
25984 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
25985 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
25986 (value_replacement): Update.
25987 * tree-ssa-reassoc.c (branch_fixup): Update.
25988 * tree-ssa-tail-merge.c (replace_block_by): Update.
25989 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
25990 (create_edge_and_update_destination_phis): Update.
25991 (compute_path_counts): Update.
25992 (recompute_probabilities): Update.
25993 (update_joiner_offpath_counts): Update.
25994 (freqs_to_counts_path): Update.
25995 (duplicate_thread_path): Update.
25996 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
25997 (struct switch_conv_info): Update.
25998 (gen_inbound_check): Update.
25999 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26000 (vect_do_peeling): Update.
26001 (vect_loop_versioning): Update.
26002 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26003 (optimize_mask_stores): Update.
26004 * ubsan.c (ubsan_expand_null_ifn): Update.
26005 * value-prof.c (gimple_divmod_fixed_value): Update.
26006 (gimple_divmod_fixed_value_transform): Update.
26007 (gimple_mod_pow2): Update.
26008 (gimple_mod_pow2_value_transform): Update.
26009 (gimple_mod_subtract): Update.
26010 (gimple_mod_subtract_transform): Update.
26011 (gimple_ic): Update.
26012 (gimple_stringop_fixed_value): Update.
26013 (gimple_stringops_transform): Update.
26014 * value-prof.h: Update.
26016 2017-06-29 Carl Love <cel@us.ibm.com>
26018 * config/rs6000/rs6000-c.c: Add support for built-in functions
26019 vector signed int vec_signed (vector float);
26020 vector signed long long vec_signed (vector double);
26021 vector signed int vec_signed2 (vector double, vector double);
26022 vector signed int vec_signede (vector double);
26023 vector signed int vec_signedo (vector double);
26024 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26025 instruction generator.
26026 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26027 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26028 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26030 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26031 vunsignede_v2df): Add define_expands.
26032 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26033 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26034 VEC_UNSIGNEDO): Add definitions.
26035 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26036 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26037 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
26038 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
26039 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
26040 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26041 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26042 vec_unsignede, vec_unsignedo): Add builtin defines.
26043 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26045 * doc/extend.texi: Update the built-in documentation file for the
26046 new built-in functions.
26048 2017-06-29 Richard Biener <rguenther@suse.de>
26050 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26051 reduction chains to LOOP_VINFO_REDUCTIONS.
26052 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26053 SLP reductions after processing reduction chains.
26055 2017-06-29 Nathan Sidwell <nathan@acm.org>
26057 * builtins.c (fold_builtin_FUNCTION): Use
26058 lang_hooks.decl_printable_name.
26060 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26062 PR middle-end/81194
26063 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26064 with only one label.
26065 * stmt.c (expand_case): Assert NCASES is greater than one.
26067 2017-06-29 Richard Biener <rguenther@suse.de>
26069 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26071 (group_case_labels): Likewise.
26072 (find_taken_edge): Push sanity checking on val to workers...
26073 (find_taken_edge_cond_expr): ... here
26074 (find_taken_edge_switch_expr): ... and here, handle cases
26075 with just a default label.
26076 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26077 (group_case_labels): Likewise.
26078 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26079 group_case_labels does anything cleanup the CFG again.
26081 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26083 PR tree-optimization/81196
26084 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26085 exit condition comparing two IVs.
26087 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26089 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26090 profile to the dummy entry at the end of the list of architectures.
26091 * config/arm/arm-cpu-cdata.h: Regenerated.
26093 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26094 Michael Collison <michael.collison@arm.com>
26097 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26099 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26100 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26101 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26102 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26103 of shift when the shift amount is masked with constant equal to
26104 the size of the mode.
26105 * config/aarch64/predicates.md (subreg_lowpart_operator): New
26108 2017-06-29 Martin Liska <mliska@suse.cz>
26110 * config/i386/i386.opt: Change range from [1,5] to [0,5].
26112 2017-06-29 Yury Gribov <tetra2005@gmail.com>
26115 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
26118 (edge_growth_cache_entry::edge_growth_cache_entry): New
26120 (reset_edge_growth_cache): Update to use constructor.
26122 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26124 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
26125 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26126 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26128 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
26130 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
26131 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
26133 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
26135 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
26136 (*-linux-uclibc*): Add t-uclibc tmake_file.
26137 * config/t-musl: New.
26138 * config/t-uclibc: New.
26140 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
26142 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
26144 (gen_comm_data): Emit architectural setting of arch_prof.
26145 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
26147 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
26148 (armv8-m.base, armv8-m.main): Likewise.
26149 * arm-protos.h (arm_build_target): Add profile field.
26150 (arch_option): Likewise.
26151 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
26153 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
26154 arm_active_target.profile.
26156 2017-06-28 Richard Biener <rguenther@suse.de>
26158 PR middle-end/81227
26159 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
26160 TYPE_OVERFLOW_WRAPS.
26161 * match.pd (negate_expr_p): Likewise.
26162 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
26163 fold_build2, not fold_binary.
26165 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26167 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
26168 Convert memory address to Pmode.
26169 (aarch64_print_operand): Assert MEM operands are always Pmode.
26171 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26174 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
26175 Remove redundant if.
26176 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
26177 * config/arm/aarch-common-protos.h
26178 (aarch_forward_to_shift_is_not_shifted_re): Remove.
26179 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
26181 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
26184 * multiple_target.c (create_dispatcher_calls): Set the default
26185 clone to be static, not public.
26187 2017-06-28 Richard Biener <rguenther@suse.de>
26189 * tree-vect-loop.c (vectorizable_reduction): Move special
26190 cond reduction IV var creation ...
26191 (vect_create_epilog_for_reduction): ... here. Remove induction_index
26192 parameter. Use STMT_VINFO_VECTYPE.
26193 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
26196 2017-06-28 Martin Liska <mliska@suse.cz>
26199 * ipa-visibility.c (non_local_p): Handle visibility.
26201 2017-06-28 Martin Liska <mliska@suse.cz>
26204 * common.opt: Add IntegerRange to various options.
26205 * opt-functions.awk (integer_range_info): New function.
26206 * optc-gen.awk: Add integer_range_info to cl_options struct.
26207 * opts-common.c (decode_cmdline_option): Handle
26208 CL_ERR_INT_RANGE_ARG.
26209 (cmdline_handle_error): Likewise.
26210 * opts.c (print_filtered_help): Show valid interval in
26211 when --help is provided.
26212 * opts.h (struct cl_option): Add range_min and range_max fields.
26213 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
26215 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
26217 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
26218 (x * C EQ/NE y * C): New transformation.
26220 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
26222 * genmultilib (combination_space): Accept '+' in option names.
26224 2017-06-28 Martin Liska <mliska@suse.cz>
26227 * asan.c (instrument_derefs): Bail out inner references
26228 that are hard register variables.
26230 2017-06-28 Jakub Jelinek <jakub@redhat.com>
26233 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
26234 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
26236 2017-06-28 Richard Biener <rguenther@suse.de>
26238 * tree-vectorizer.h (vect_get_vec_defs): Remove.
26239 (vect_get_slp_defs): Adjust.
26240 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
26242 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
26244 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
26245 get_initial_defs_for_reduction instead of vect_get_vec_defs.
26246 (vectorizable_reduction): Adjust.
26247 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
26249 (vect_get_slp_defs): Likewise.
26250 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
26251 (vectorizable_bswap): Adjust.
26252 (vectorizable_call): Likewise.
26253 (vectorizable_conversion): Likewise.
26254 (vectorizable_assignment): Likewise.
26255 (vectorizable_shift): Likewise.
26256 (vectorizable_operation): Likewise.
26257 (vectorizable_store): Likewise.
26258 (vectorizable_condition): Likewise.
26259 (vectorizable_comparison): Likewise.
26261 2017-06-28 Michael Collison <michael.collison@arm.com>
26264 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
26266 (arm_gen_movmemqi): Removed unused variable 'i'.
26267 Convert 'for' loop into 'while' loop.
26268 (arm_expand_prologue): Remove last unnecessary set of insn.
26269 (thumb_pop): Remove unused variable 'pushed_words'.
26270 (thumb_exit): Remove last unnecessary set of regs_to_pop.
26272 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26274 * config/s390/predicates.md: Use s390_rel_address_ok_p.
26275 * config/s390/s390-protos.h: Add prototype of
26276 s390_rel_address_ok_p.
26277 * config/s390/s390.c (s390_got_symbol): New function.
26278 (s390_rel_address_ok_p): New function.
26279 (legitimize_pic_address): Use s390_rel_address_ok_p.
26280 (s390_load_got): Use s390_got_symbol.
26281 (s390_option_override): Issue error if
26282 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
26283 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
26285 * config/s390/s390.opt: New option mpic-data-is-text-relative.
26287 2017-06-27 Andrew Pinski <apinski@cavium.com>
26289 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
26290 (X * copysign (1.0, X)): New pattern.
26291 (X * copysign (1.0, -X)): New pattern.
26292 (copysign (-1.0, CST)): New pattern.
26294 2017-06-27 Joseph Myers <joseph@codesourcery.com>
26296 * genmultilib (combination_space): Remove variable.
26297 Validate reuse rules against regular expression for any sequence
26298 of multilib options in any order.
26300 2017-06-27 Michael Collison <michael.collison@arm.com>
26302 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
26303 call aarch64_split_simd_combine.
26304 * (aarch64_combine_internal<mode>): Delete pattern.
26305 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
26306 Allow register and subreg operands.
26308 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26310 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
26311 specific need, just fallback on defaults.
26312 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
26314 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26315 Olivier Hainque <hainque@adacore.com>
26317 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
26319 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
26320 targets. Pick a default if no particular attempt applied.
26321 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
26324 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26326 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
26327 (x86_64-wrs-vxworks7): Likewise.
26329 2017-06-27 Marek Polacek <polacek@redhat.com>
26332 * ubsan.c (instrument_null): Check get_base_address's result for null.
26334 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26336 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
26338 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26340 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
26341 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
26342 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
26343 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
26344 New function types.
26345 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
26346 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
26347 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
26348 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
26349 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
26350 BUILT_IN_FEUPDATEENV): New builtins.
26351 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
26352 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
26353 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
26354 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
26356 (builtin_structptr_types): Adjust size.
26357 * tree.c (builtin_structptr_types): Add four entries.
26359 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26360 Olivier Hainque <hainque@adacore.com>
26362 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
26363 (TLS_SYM): New local macro, forcing reference to __tls__ on
26364 link command lines for VxWorks 7 RTPs, triggering initialization
26366 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
26367 OS features TLS support, true for RTPs on VxWorks 7.
26368 * config/vxworks.c (vxworks_override_options): Setup emutls
26371 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26373 * predict.c (test_prediction_value_range): Use -1U instead of -1
26374 to avoid narrowing conversion warning.
26375 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
26376 to avoid narrowing conversion warning.
26377 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
26379 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
26381 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26383 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
26384 64bit configurations.
26385 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
26386 (SIZE_TYPE): Likewise.
26387 * config/vxworks.c (vxworks_emutls_var_fields): Use
26388 long_unsigned_type_node instead of unsigned_type_node as the offset
26389 field type, which is "pointer" mode in emutls.c.
26391 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26394 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
26396 PR middle-end/81207
26397 * gimple-fold.c (replace_call_with_call_and_fold): Handle
26398 gimple_vuse copying separately from gimple_vdef copying.
26400 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26402 * value-prof.c (free_hist): Remove call to memset and the enclosing if
26405 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26406 Olivier Hainque <hainque@adacore.com>
26408 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
26409 for all vxworks7 targets.
26410 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
26411 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
26412 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
26413 variations for VX6/VX7 and 32/64bits later on in ...
26414 (VXWORKS_LIB_SPEC): Leverage new macros.
26415 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
26416 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
26418 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26420 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
26421 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
26423 2017-06-26 Carl Love <cel@us.ibm.com>
26425 * config/rs6000/rs6000-c.c: Add support for built-in functions
26426 vector bool char vec_reve (vector bool char);
26427 vector signed char vec_reve (vector signed char);
26428 vector unsigned char vec_reve (vector unsigned char);
26429 vector bool int vec_reve (vector bool int);
26430 vector signed int vec_reve (vector signed int);
26431 vector unsigned int vec_reve (vector unsigned int);
26432 vector bool long long vec_reve (vector bool long long);
26433 vector signed long long vec_reve (vector signed long long);
26434 vector unsigned long long vec_reve (vector unsigned long long);
26435 vector bool short vec_reve (vector bool short);
26436 vector signed short vec_reve (vector signed short);
26437 vector double vec_reve (vector double);
26438 vector float vec_reve (vector float);
26439 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
26440 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
26441 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
26442 (altivec_vreve): New pattern.
26443 * config/rs6000/altivec.h (vec_reve): New define.
26444 * doc/extend.texi (vec_rev): Update the built-in documentation file
26445 for the new built-in functions.
26447 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26449 PR tree-optimization/71815
26450 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
26452 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
26454 (slsr_process_phi): Likewise.
26455 (replace_uncond_cands_and_profitable_phis): Don't replace a
26456 multiply candidate with a stride of 1 (copy or cast).
26457 (phi_incr_cost): Call uses_consumed_by_stmt rather than
26459 (lowest_cost_path): Likewise.
26460 (total_savings): Likewise.
26462 2017-06-26 Richard Biener <rguenther@suse.de>
26465 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
26466 Use def_builtin_pure for all gather builtins.
26468 2017-06-26 Richard Biener <rguenther@suse.de>
26470 PR tree-optimization/81203
26471 * tree-tailcall.c (find_tail_calls): Do not move stmts into
26472 non-dominating BBs.
26474 2017-06-26 Marek Polacek <polacek@redhat.com>
26477 * doc/invoke.texi: Document -Wmultistatement-macros.
26479 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
26481 * doc/sourcebuild.texi (ARM-specific attributes): Document new
26482 arm_neon_ok_no_float_abi effective target.
26484 2017-06-26 Richard Biener <rguenther@suse.de>
26486 PR tree-optimization/80928
26487 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
26488 (copy_bbs): Set BB_DUPLICATED flag early.
26489 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
26491 (execute_on_shrinking_pred): Likewise.
26492 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
26493 BB_DUPLICATED blocks.
26494 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
26495 iterate over all PHIs considering removal of *gsi.
26497 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
26499 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
26502 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26504 * config/rs6000/rs6000-string.c: (expand_block_clear,
26505 do_load_for_compare, select_block_compare_mode,
26506 compute_current_alignment, expand_block_compare,
26507 expand_strncmp_align_check, expand_strn_compare,
26508 expand_block_move, rs6000_output_load_multiple)
26509 Move functions related to string/block move/compare
26510 to a separate file.
26511 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
26512 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
26513 for this function which is now used in two files.
26514 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
26515 * config.gcc: Add rs6000-string.o to extra_objs for
26516 targets powerpc*-*-* and rs6000*-*-*.
26518 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
26521 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
26522 32-bit, since indexed is not valid for DImode.
26523 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
26524 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
26525 (define_peephole2 for Altivec d-form load): Add 32-bit support.
26526 (define_peephole2 for Altivec d-form store): Likewise.
26529 * multiple_target.c (create_dispatcher_calls): Only create the
26530 dispatcher call if the function is the default clone of a
26531 versioned function.
26533 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
26535 PR middle-end/80902
26536 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
26537 a call, force the call to not be a tail call.
26539 2017-06-23 Jeff Law <law@redhat.com>
26541 * doc/contrib.texi: Add entry for Steven Pemberton's work on
26544 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
26546 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
26547 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
26548 handling for early expansion of vector shifts (sl,sr,sra,rl).
26549 (builtin_function_type): Add vector shift right instructions
26550 to the unsigned argument list.
26552 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
26554 rtl-optimizatoin/79286
26555 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
26556 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
26557 trap. PIC register plus a const unspec without offset can never trap.
26559 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
26561 * tree.h (builtin_structptr_type): New type.
26562 (builtin_structptr_types): Declare new array.
26563 * tree.c (builtin_structptr_types): New array.
26564 (free_lang_data, build_common_tree_nodes): Use it.
26566 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
26569 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
26572 2017-06-22 Matt Turner <mattst88@gmail.com>
26574 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
26575 Lake models to skylake case. Assume skylake for unknown
26576 models with clflushopt.
26578 2017-06-22 Jeff Law <law@redhat.com>
26580 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
26581 frame sizes that do not satisfy aarch64_uimm12_shift.
26583 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
26585 * profile-count.h (apply_probability,
26586 apply_scale, probability_in): Fix checks for zero.
26588 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26590 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
26591 * doc/cppdiropts.texi (-I @var{dir}): Document it.
26593 2016-06-22 Richard Biener <rguenther@suse.de>
26595 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26596 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
26597 REDUC_MAX_EXPR support.
26598 (vectorizable_reduction): Likewise.
26599 (vect_create_epilog_for_reduction): Likewise.
26601 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
26603 * match.pd (A / (1 << B) -> A >> B): New.
26604 * generic-match-head.c: Include optabs-tree.h.
26605 * gimple-match-head.c: Likewise.
26606 * optabs-tree.h (target_supports_op_p): New.
26607 * optabs-tree.c (target_supports_op_p): New.
26609 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26611 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
26612 $gcc_cv_ld --help output.
26613 (gcc_cv_ld_demangle): Likewise.
26614 (gcc_cv_ld_eh_frame_hdr): Likewise.
26615 (gcc_cv_ld_pie): Likewise.
26616 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
26617 (gcc_cv_ld_buildid): Likewise.
26618 (gcc_cv_ld_sysroot): Likewise.
26619 (ld_bndplt_support): Likewise.
26620 (ld_pushpopstate_support): Likewise.
26621 * configure: Regenerate.
26622 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26624 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26627 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26628 operands indexes to avoid gap between operands and match_dups.
26630 2017-06-21 Andrew Pinski <apinski@cavium.com>
26632 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26633 Increment Arith_shift and Arith_shift_reg by 1.
26634 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26636 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26637 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26638 (aarch64_strip_extend): Add new argument and test for it.
26639 (aarch64_cheap_mult_shift_p): New function.
26640 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26641 add a cost if it is true.
26642 Update calls to aarch64_strip_extend.
26643 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26645 2017-06-21 Andrew Pinski <apinski@cavium.com>
26647 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26649 (thunderxt88): Likewise.
26650 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26651 (thunderx_prefetch_tune): New variable.
26652 (thunderx2t99_prefetch_tune): Update for the correct values.
26653 (thunderxt88_tunings): New variable.
26654 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26655 generic_prefetch_tune.
26656 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26658 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26660 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26661 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26662 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26663 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26664 (aarch64_atomic_cas<mode>, GPI): Likewise.
26666 2017-06-21 Martin Liska <mliska@suse.cz>
26668 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26669 statements on cold and hot labels.
26670 * predict.c (tree_estimate_probability_bb): Remove the
26671 prediction from this place.
26673 2017-06-21 Martin Liska <mliska@suse.cz>
26675 PR tree-optimization/79489
26676 * gimplify.c (maybe_add_early_return_predict_stmt): New
26678 (gimplify_return_expr): Call the function.
26679 * predict.c (tree_estimate_probability_bb): Remove handling
26681 * predict.def: Update comment about early return predictor.
26682 * gimple-predict.h (is_gimple_predict): New function.
26683 * predict.def: Change default value of early return to 66.
26684 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26686 * passes.def: Put pass_strip_predict_hints to the beginning of
26689 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26691 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26692 FUNCTION_DECL declarations.
26693 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26695 (dwaf2out_decl): Likewise.
26696 * godump.c (go_early_global_decl): Skip call to the real debug hook
26697 for FUNCTION_DECL declarations.
26698 * passes.c (rest_of_decl_compilation): Skip call to the
26699 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26700 -fdump-go-spec is passed.
26702 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26704 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26705 Reorder for compactness.
26706 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26707 (def_builtin_pure, def_builtin_pure2): New functions.
26708 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26710 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26712 * match.pd (nop_convert): New predicate.
26713 ((A +- CST1) +- CST2): Allow some NOP conversions.
26715 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26718 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26719 with ctors/dtors if GOVD_SHARED is set.
26721 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26723 * config/aarch64/aarch64.md (movti_aarch64):
26724 Emit mov rather than orr.
26725 (movtf_aarch64): Likewise.
26726 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26727 Emit mov rather than orr.
26729 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26731 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26732 Swap alternatives, make integer dup more expensive.
26734 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26736 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26737 Return true for non-tls symbols.
26739 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26741 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26742 (cortex-a75): Likewise.
26743 (cortex-a75.cortex-a55): Likewise.
26744 * config/aarch64/aarch64-tune.md: Regenerate.
26745 * doc/invoke.texi (-mtune): Document new values for -mtune.
26747 2017-06-21 Tom de Vries <tom@codesourcery.com>
26749 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26750 stack_size feature.
26751 (Effective-Target Keywords, Other attributes): Suggest using
26752 dg-add-options stack_size feature to get stack limit in stack_size
26753 effective target documentation.
26755 2017-06-21 Julian Brown <julian@codesourcery.com>
26756 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26758 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26759 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26760 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26762 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26763 attribute type list for neon_multiply.
26764 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26765 attribute type list for neon_multiply.
26766 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26767 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26768 attribute type list for neon_multiply.
26769 * config/arm/types.md (crypto_pmull): Add.
26770 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26771 attribute type list.
26773 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26775 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26778 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26780 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26781 to make sure not to dereference a NULL cost_classes_ptr pointer.
26783 2017-06-20 Carl Love <cel@us.ibm.com>
26785 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26786 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26787 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26788 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26789 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26790 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26791 VMULOSW): New enum "unspec" values.
26792 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26793 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26794 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26795 altivec_vmulosw): New patterns.
26796 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26797 VMULOSW): Add definitions.
26799 2017-06-20 Julia Koval <julia.koval@intel.com>
26801 * config/i386/i386.c: Fix rounding expand for new pattern.
26802 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26804 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26806 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26807 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26809 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26811 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26814 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26816 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26817 architecture, then by implementer ID.
26818 * config/aarch64/aarch64-tune.md: Regenerate.
26820 2017-06-20 Richard Biener <rguenther@suse.de>
26822 PR middle-end/81097
26823 * fold-const.c (split_tree): Fold to type before negating.
26825 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26827 * diagnostic-show-locus.c
26828 (selftest::test_fixit_deletion_affecting_newline): New function.
26829 (selftest::diagnostic_show_locus_c_tests): Call it.
26831 2017-06-20 Andreas Schwab <schwab@suse.de>
26834 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26837 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26839 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26840 __ARM_FEATURE_COPROC according to support.
26842 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26844 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26845 Rewritten to avoid overflow for > 32-bit pointers.
26848 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26849 by removing enum keyword.
26850 (ubsan_type_descriptor): Likewise. Formatting fix.
26853 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26854 splitter): Require TARGET_SSE2 in the condition.
26856 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26859 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26860 for doing vector set of SFmode on ISA 3.0.
26861 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26862 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
26864 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
26865 SFmode value into a V4SF variable that was extracted from another
26866 V4SF variable without converting the element to double precision
26867 and back to single precision vector format.
26868 (vsx_insert_extract_v4sf_p9_2): Likewise.
26870 2017-06-19 Jakub Jelinek <jakub@redhat.com>
26872 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
26873 in UWHI to avoid undefined overflow.
26876 * ubsan.h (enum ubsan_encode_value_phase): New.
26877 (ubsan_encode_value): Change second argument to
26878 enum ubsan_encode_value_phase with default value of
26879 UBSAN_ENCODE_VALUE_GENERIC.
26880 * ubsan.c (ubsan_encode_value): Change second argument to
26881 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
26882 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
26883 create_tmp_var_raw instead of create_tmp_var and use a
26885 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
26886 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
26887 ubsan_encode_value callers.
26890 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
26891 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
26892 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
26894 2017-06-19 Richard Biener <rguenther@suse.de>
26896 PR middle-end/81118
26897 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
26898 estimates if we changed anything.
26900 2017-06-19 Richard Biener <rguenther@suse.de>
26902 PR tree-optimization/80887
26903 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
26904 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
26905 simplified lookups, then reset mprts_hook.
26906 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
26908 (try_to_simplify): Likewise.
26910 2017-06-19 Martin Liska <mliska@suse.cz>
26913 * gimplify.c (gimplify_switch_expr):
26914 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
26916 2017-06-19 Martin Liska <mliska@suse.cz>
26918 * doc/install.texi: Document that PGO runs in 4 stages.
26920 2017-06-19 Martin Liska <mliska@suse.cz>
26923 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
26924 to dispatcher function name.
26925 * multiple_target.c (replace_function_decl): New function.
26926 (create_dispatcher_calls): Redirect both edges and references.
26928 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
26930 * profile-count.c (profile_count::dump): Dump quality.
26931 (profile_count::differs_from_p): Update for unsigned val.
26932 * profile-count.h (profile_count_quality): New enum.
26933 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
26935 2017-06-19 Richard Biener <rguenther@suse.de>
26937 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
26938 struct function as arg.
26939 (estimate_numbers_of_iterations): Export overload with loop arg.
26940 (free_numbers_of_iterations_estimates_loop): Use an overload of
26941 free_numbers_of_iterations_estimates instead.
26942 * tree-cfg.c (remove_bb): Adjust.
26943 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
26944 * tree-parloops.c (gen_parallel_loop): Likewise.
26945 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
26947 (tree_unroll_loops_completely): Likewise.
26948 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
26949 Use an overload instead and export.
26950 (estimated_loop_iterations): Adjust.
26951 (max_loop_iterations): Likewise.
26952 (likely_max_loop_iterations): Likewise.
26953 (estimate_numbers_of_iterations): Take struct function as arg
26955 (loop_exits_before_overflow): Adjust.
26956 (free_numbers_of_iterations_estimates_loop): Use an overload.
26957 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
26958 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
26960 2017-06-19 Richard Biener <rguenther@suse.de>
26963 * ipa-prop.c (find_constructor_constant_at_offset): Handle
26964 RANGE_EXPR conservatively.
26966 2017-06-16 Carl Love <cel@us.ibm.com>
26968 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26969 definitions for vec_float, vec_float2, vec_floato,
26970 vec_floate built-ins.
26971 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
26972 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
26974 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
26975 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
26976 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
26977 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
26978 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
26979 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
26980 vec_floato): Add builtin defines.
26981 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
26982 Update the built-in documentation file for the new built-in
26985 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26987 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
26988 (mthumb): Mark as the negative of -marm.
26990 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26992 * doc/invoke.texi (ARM Options, -mcpu): Document supported
26994 (ARM Options, -mtune): Document that this accepts the same
26995 extension options as -mcpu.
26996 (ARM Options, -mfpu): Document addition of -mfpu=auto.
26998 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27000 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27001 permitted extensions.
27003 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27005 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27006 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27007 (armv8-m.main): Add option +nodsp.
27008 * config/arm/arm-cpu-cdata.h: Regenerated.
27010 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27012 * config/arm/t-fuchsia: New file.
27013 * config.gcc (arm*-*-fuchsia*): Use it.
27015 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27017 * config/arm/t-symbian: Rewrite for new option infrastructure.
27019 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27021 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27022 (MULTILIB_REQUIRED): Likewise.
27024 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27026 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27027 (MULTILIB_RESUE): Likewise.
27028 (MULTILIB_MATCHES): Likewise.
27029 (MULTLIB_REQUIRED): Likewise.
27031 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27033 * config/arm/t-rtems: Rewrite for new option framework.
27035 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27037 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
27038 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
27039 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
27040 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27041 * config/arm/t-multilib: ... here.
27042 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27043 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27044 armv7-a and armv8*-a when A-profile libraries have not been built.
27045 * config/arm/t-rmprofile: Rewrite.
27047 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27049 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27050 with a backslash. Remove the backslash after substituting unescaped
27052 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27054 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27056 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27057 the driver to pass through the default setting of -mfloat-abi.
27058 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27060 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27061 (all_feat_combs): New rule.
27062 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27064 * config/arm/t-aprofile: Rewrite.
27066 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27068 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27069 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27070 fpu is not specified by the user/command-line.
27071 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27072 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27073 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27074 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27075 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27076 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27078 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27080 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27081 * config/arm/t-arm-elf: Rewritten.
27083 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27085 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27086 have some floating-point instructions.
27087 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27088 (TARGET_MAYBE_HARD_FLOAT): New macro.
27089 * config/arm/arm-builtins.c (arm_init_builtins): Use
27090 TARGET_MAYBE_HARD_FLOAT.
27091 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27093 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27095 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27096 (configargs.h): Include it.
27097 (arm_print_hint_for_fpu_option): New function.
27098 (arm_parse_fpu_option): New function.
27099 (candidate_extension): New class.
27100 (arm_canon_for_multilib): New function.
27101 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27102 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27103 (ARCH_CANONICAL_SPECS): New macro.
27104 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
27106 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27108 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
27109 are set after handling multilib fragments. Set target_cpu_default2
27112 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27114 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
27116 (arm*-*-*): Set target_cpu_default2 to a quoted string.
27117 * config/arm/parsecpu.awk (check_cpu): Validate any extension
27119 (check_arch): Likewise.
27120 * config/arm/arm.c (arm_configure_build_target): Handle
27121 TARGET_CPU_DEFAULT being a string constant. Scan any feature
27122 options in the default.
27124 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27126 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
27127 when an option is an alias of another.
27128 * config/arm/parsecpu.awk (optalias): New parser token.
27129 (gen_comm_data): Mark non-alias options as such. Emit entries
27130 for extension aliases.
27131 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
27132 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
27133 (armv6kz, armv6zk, armv6t2): Likewise.
27134 (armv7): Make vfpv3-d16 an alias.
27135 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
27137 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
27138 Sort in canonical order.
27139 (armv8-a): Sort in canonical order.
27140 (armv8.1-a, armv8.2-a): Likewise.
27141 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
27143 (cortex-a9): Sort in canonical order.
27144 * config/arm/arm.c (selftests.h): Include it.
27145 (arm_test_cpu_arch_data): New function.
27146 (arm_run_self_tests): New function.
27147 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
27148 (targetm): Move declaration to the end of the file.
27149 * arm-cpu-cdata.h: Regenerated.
27151 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27153 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
27154 call to target_mode_check describing the type of option passed.
27155 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
27156 (arm_target_thumb_only): Use arm_parse_arch_option_name or
27157 arm_parse_cpu_option_name to match parameters against list of
27159 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
27160 arm_arch_core_flags data structure.
27161 * config/arm/arm-cpu_cdata.h: Regenerated.
27163 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27165 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
27167 (arm_print_hint_for_cpu_option): Likewise.
27168 (arm_print_hint_for_arch_option): Likewise.
27169 (arm_parse_cpu_option_name): Likewise.
27170 (arm_parse_arch_option_name): Likewise.
27171 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
27172 of entries in the all_fpus list.
27173 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
27174 (arm_parse_cpu_option_name): Declare.
27175 (arm_parse_arch_option_name): Declare.
27176 (arm_parse_option_features): Declare.
27177 (arm_intialize_isa): Declare.
27178 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
27180 (gen_comm_data): ... here. Make definitions non-static.
27181 * config/arm/arm-cpu-data.h: Regenerated.
27182 * config/arm/arm-cpu-cdata.h: Regenerated.
27184 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27186 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
27187 (cpu_arch_extension): New structure.
27188 (cpu_arch_option, arch_option, cpu_option): New structures.
27189 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
27190 architecture types.
27191 (gen_data): Generate new format data tables.
27192 * config/arm/arm.c (cpu_tune): New structure.
27193 (cpu_option, processors): Delete.
27194 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
27195 (arm_print_hint_for_cpu_option): ... this and ...
27196 (arm_print_hint_for_arch_option): ... this.
27197 (arm_parse_arch_cpu_name): Delete. Replace with ...
27198 (arm_parse_cpu_option_name): ... this and ...
27199 (arm_parse_arch_option_name): ... this.
27200 (arm_unrecognized_feature): Change type of target parameter to
27202 (arm_parse_arch_cpu_features): Delete. Replace with ...
27203 (arm_parse_option_features): ... this.
27204 (arm_configure_build_target): Rework to use new configuration data
27206 (arm_print_tune_info): Rework for new configuration data tables.
27207 * config/arm/arm-cpu-data.h: Regenerated.
27208 * config/arm/arm-cpu.h: Regenerated.
27210 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27212 * Makefile.in (OBJS): Move sbitmap.o from here ...
27213 (OBJS-libcommon): ... to here.
27215 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27217 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
27218 (ISA_ALL_CRYPTO): New macro.
27219 (ISA_ALL_SIMD): New macro
27220 (ISA_ALL_FP): New macro.
27221 * config/arm/arm.c (fpu_bitlist): Update initializer.
27222 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
27224 (arm9e): Add fpu. Add option for nofp
27225 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
27226 (arm926ej-s, arm1026ej-s): Likewise.
27227 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
27228 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
27229 neon-fp16, neon-vfpv4, nofp and nosimd.
27230 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
27231 (cortex-a8): Add fpu. Add option for nofp.
27232 (cortex-a9): Add fpu. Add options for nosimd and nofp.
27233 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
27234 (cortex-r4f): Add fpu.
27235 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
27236 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
27238 (cortex-r8): Likewise.
27239 (cortex-m4): Add fpu. Add option for nofp.
27240 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
27241 (cortex-a17.cortex-a7): Likewise.
27242 (cortex-a32): Add fpu. Add options for crypto and nofp.
27243 (cortex-a35, cortex-a53): Likewise.
27244 (cortex-a57): Add fpu. Add option for crypto.
27245 (cortex-a72, cortex-a73): Likewise.
27246 (exynos-m1): Likewise.
27247 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
27248 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
27249 (cortex-m33): Add fpu. Add option for nofp.
27250 * config/arm/arm-cpu-cdata.h: Regenerated
27251 * config/arm/arm-cpu-data.h: Regenerated.
27253 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27255 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
27256 (armv5te, armv5tej): Likewise.
27257 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
27258 (armv7): Add options fp and vfpv3-d16.
27259 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
27260 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
27262 (armv7ve): Likewise.
27263 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
27264 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
27265 (armv8-a): Add nocrypto option.
27266 (armv8.1-a, armv8.2-a): Likewise.
27267 (armv8-m.main): add options fp, fp.dp and nofp.
27269 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27271 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
27273 (armv8-a+crc): Delete.
27274 (armv8.1-a): Add options simd, crypto and nofp.
27275 (armv8.2-a): Add options fp16, simd, crypto and nofp.
27276 (armv8.2-a+fp16): Delete.
27277 (armv8-m.main): Add option dsp.
27278 (armv8-m.main+dsp): Delete.
27279 (cortex-a8): Add fpu. Add nofp option.
27280 (cortex-a9): Add fpu. Add nofp and nosimd options.
27281 * config/arm/parsecpu.awk (gen_data): Generate option tables and
27282 link to main cpu and architecture data structures.
27283 (gen_comm_data): Only put isa attributes from the main architecture
27285 (option): New statement for architecture and CPU entries.
27286 * arm.c (struct cpu_option): New structure.
27287 (struct processors): Add entry for options.
27288 (arm_unrecognized_feature): New function.
27289 (arm_parse_arch_cpu_name): Ignore any characters after the first
27291 (arm_parse_arch_cpu_feature): New function.
27292 (arm_configure_build_target): Separate out any CPU and architecture
27293 features and parse separately. Don't error out if -mfpu=auto is
27294 used with only an architecture string.
27295 (arm_print_asm_arch_directives): New function.
27296 (arm_file_start): Call it.
27297 * config/arm/arm-cpu-cdata.h: Regenerated.
27298 * config/arm/arm-cpu-data.h: Likewise.
27299 * config/arm/arm-tables.opt: Likewise.
27301 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27303 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
27304 assembler when it is not -mfpu=auto.
27306 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27308 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
27309 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
27310 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
27311 (ASM_CPU_SPEC): Rewrite.
27312 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
27313 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
27314 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
27315 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
27316 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
27317 copied string is NUL-terminated. Also strip any characters prefixed
27319 (arm_rewrite_selected_arch): New function.
27320 (arm_rewrite_march): New function.
27322 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27324 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
27325 (x_arm_cpu_string, x_arm_tune_string): Likewise.
27326 (march, mcpu, mtune): Convert to string-based options.
27327 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
27328 (arm_parse_arch_cpu_name): New function.
27329 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
27330 identify selected architecture or CPU.
27331 (arm_option_save): New function.
27332 (TARGET_OPTION_SAVE): Redefine.
27333 (arm_option_restore): Restore string options.
27334 (arm_option_print): Print string options.
27336 2017-06-16 Martin Sebor <msebor@redhat.com>
27338 PR tree-optimization/80933
27339 PR tree-optimization/80934
27340 * builtins.c (fold_builtin_3): Do not handle bcmp here.
27341 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
27342 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
27343 (gimple_fold_builtin): Call them.
27345 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27347 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
27348 as unlikely; update profile.
27350 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27352 * predict.c (force_edge_cold): Handle declaring edges impossible
27355 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27357 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
27359 (try_unroll_loop_completely): Fix reporting.
27361 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27363 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
27365 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
27368 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
27369 if given a non-constant argument for an intrinsic which requires a
27372 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27374 * profile.c (compare_freqs): New function.
27375 (branch_prob): Sort edge list.
27376 (find_spanning_tree): Assume that the list is priority sorted.
27378 2017-06-16 Richard Biener <rguenther@suse.de>
27380 PR tree-optimization/81090
27381 * passes.def (pass_record_bounds): Remove.
27382 * tree-pass.h (make_pass_record_bounds): Likewise.
27383 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
27384 make_pass_record_bounds): Likewise.
27385 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
27386 not free niter estimates at the beginning but at the end.
27387 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
27389 2017-06-16 Richard Biener <rguenther@suse.de>
27391 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
27392 initializer to workaround ICE in host GCC 4.8.
27394 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27396 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
27398 (clone_inlined_nodes): Update.
27400 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27402 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
27403 prefetch settings, and enable prefetching by default at -O3.
27405 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27407 * config/aarch64/aarch64.c (aarch64_override_options_internal):
27408 Set flag_prefetch_loop_arrays according to tuning data.
27410 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27412 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
27413 New tune structure.
27414 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
27415 [Unrelated to main purpose of the patch] Place the pointer field last
27416 to enable type checking errors when tune structure are wrongly merged.
27417 * config/aarch64/aarch64.c (generic_prefetch_tune,)
27418 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
27419 (thunderx2t99_prefetch_tune): New tune constants.
27420 (tune_params *_tunings): Update all tunings (no functional change).
27421 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
27422 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
27423 from tunings structures.
27425 2017-06-16 Jakub Jelinek <jakub@redhat.com>
27428 * ubsan.c (instrument_null): Add T argument, use it instead
27429 of computing it based on IS_LHS.
27430 (instrument_object_size): Likewise.
27431 (pass_ubsan::execute): Adjust instrument_null and
27432 instrument_object_size callers to pass gimple_get_lhs or
27433 gimple_assign_rhs1 result to it. Use instrument_null instead of
27434 calling get_base_address and instrument_mem_ref. Handle
27435 aggregate call arguments for object-size sanitization.
27437 2017-06-16 Yury Gribov <tetra2005@gmail.com>
27439 PR tree-optimization/81089
27440 * tree-vrp.c (is_masked_range_test): Validate operands of
27443 2017-06-15 Martin Sebor <msebor@redhat.com>
27446 * dumpfile.c (dump_register): Avoid calling memset to initialize
27447 a class with a default ctor.
27448 * gcc.c (struct compiler): Remove const qualification.
27449 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
27450 * hash-table.h: Ditto.
27451 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
27453 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
27454 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
27456 * params.h (struct param_info): Make struct members non-const.
27457 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
27458 with default initialization.
27459 * vec.h (vec_copy_construct, vec_default_construct): New helper
27461 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
27462 with vec_copy_construct.
27463 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
27464 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
27465 * doc/invoke.texi (-Wclass-memaccess): Document.
27467 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27469 * emit-rtl.h (is_leaf): Update comment about local
27470 register allocator.
27472 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27475 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
27476 for a variable to have a section before checking if the section has a
27478 Set section to.persistent if persistent attribute is set.
27479 Warn if .persistent attribute is used on an automatic variable.
27481 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
27483 PR rtl-optimization/80474
27484 * reorg.c (update_block): Do not ignore instructions in a delay slot.
27486 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
27488 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
27491 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
27493 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
27494 (casesi): Emit bounds checking as RTL.
27495 (casesi_internal_mips16_<mode>): Remove bounds checking.
27497 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
27499 * config/xtensa/xtensa.c (xtensa_option_override): Append
27500 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
27501 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
27502 xtensa_doloop_hooks): Define unconditionally.
27503 (xtensa_reorg_loops): Only call reorg_loops in the presence of
27505 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
27506 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
27507 for it in xtensa_option_override.
27508 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
27509 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
27511 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
27513 * doc/cppopts.texi: Document '-' special value to -MF.
27515 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
27517 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
27518 (cortex_a53_fconst): Likewise.
27519 (cortex_a53_fpmul): Likewise.
27520 (cortex_a53_f_load_64): Likewise.
27521 (cortex_a53_f_load_many): Likewise.
27522 (cortex_a53_advsimd_alu): Likewise.
27523 (cortex_a53_advsimd_alu_q): Likewise.
27524 (cortex_a53_advsimd_mul): Likewise.
27525 (cortex_a53_advsimd_mul_q): Likewise.
27526 (fpmac bypass): Add new bypass for fpmac-fpmac case.
27527 Add missing fmul, r2f_cvt and fconst cases.
27529 2017-06-14 Richard Biener <rguenther@suse.de>
27531 PR middle-end/81088
27532 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
27534 (fold_binary_loc): When associating do not treat pre-existing
27535 TREE_OVERFLOW on literal constants as a reason to allow
27536 TREE_OVERFLOW on associated literal constants.
27538 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27540 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
27541 (MASK_FEATURES): New macro.
27542 * config/sparc/sparc.c (sparc_option_override): Remove the special
27543 handling of -mfpu and generalize it to all MASK_FEATURES switches.
27545 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27547 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
27548 a division of 0 if non-call exceptions are enabled.
27550 2017-06-14 Andrew Pinski <apinski@cavium.com>
27551 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27554 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
27555 Improve vector initialization code gen for only variable case.
27557 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27559 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
27561 2017-06-14 Richard Biener <rguenther@suse.de>
27563 PR tree-optimization/81083
27564 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
27567 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27569 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
27570 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
27571 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
27572 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
27573 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
27574 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
27576 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27578 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
27579 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
27581 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27583 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
27585 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27587 * config/rs6000/t-rtems: Don't handle SPE.
27589 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27591 * config/rs6000/t-linux: Don't handle SPE.
27593 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27595 * config/rs6000/eabispe.h: Delete file.
27597 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27599 * config/rs6000/t-spe: Delete file.
27601 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27603 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
27604 (rs6000_legitimate_offset_address_p): Return false for anything in
27605 V2SImode or V2SFmode.
27607 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27609 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
27610 except V2SF and V2SI. Rearrange the vector modes, and add comments.
27611 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
27613 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
27614 (rs6000_legitimate_offset_address_p): Ditto.
27615 (rs6000_emit_move): Ditto.
27616 (rs6000_init_builtins): Remove V4HI_type_node.
27618 2017-06-13 Martin Liska <mliska@suse.cz>
27621 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27622 (gate_asan): Likewise.
27623 * asan.h (asan_no_sanitize_address_p): Remove the function.
27624 (sanitize_flags_p): New function.
27625 * builtins.def: Fix coding style.
27626 * common.opt: Use renamed enum value.
27627 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27628 * doc/extend.texi: Document no_sanitize attribute.
27629 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27630 to SANITIZE_UNDEFINED_NONDEFAULT.
27631 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27632 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27633 Use sanitize_flags_p.
27634 * gimplify.c (gimplify_function_tree): Likewise.
27635 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27636 * opts.c (parse_no_sanitize_attribute): New function.
27637 (common_handle_option): Use renamed enum value.
27638 * opts.h (parse_no_sanitize_attribute): Declare.
27639 * tree.c (sanitize_flags_p): New function.
27640 * tree.h: Declared here.
27641 * tsan.c: Use sanitize_flags_p.
27642 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27643 (instrument_mem_ref): Likewise.
27644 (instrument_bool_enum_load): Likewise.
27645 (do_ubsan_in_current_function): Remove the function.
27646 (pass_ubsan::execute): Use sanitize_flags_p.
27647 * ubsan.h: Remove do_ubsan_in_current_function
27648 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27649 (dump_function_to_file): Use it here.
27651 2017-06-13 Martin Jambor <mjambor@suse.cz>
27653 PR tree-optimization/80803
27654 PR tree-optimization/81063
27655 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27656 (propagate_subaccesses_across_link): Enqueue subtree whenever
27657 necessary instead of relying on the caller.
27659 2017-06-13 Martin Jambor <mjambor@suse.cz>
27661 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27662 that have a first_link.
27663 (sort_and_splice_var_accesses): Do not check first_link before
27665 (subtree_mark_written_and_enqueue): Likewise.
27666 (propagate_all_subaccesses): Likewise and do not stop at first
27667 parent with a first_link.
27669 2017-06-13 Martin Jambor <mjambor@suse.cz>
27671 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27674 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27676 * match.pd: New pattern.
27678 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27680 * tree-vrp.c (is_masked_range_test): New function.
27681 (register_edge_assert_for): Determine ranges for
27684 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27686 PR tree-optimization/67328
27687 * fold-const.c (maskable_range_p): New function.
27688 (build_range_check): Generate bittests if possible.
27690 2017-06-13 Martin Liska <mliska@suse.cz>
27692 * gimple-pretty-print.c (dump_probability): Add new argument.
27693 (dump_edge_probability): Dump both probability and count.
27694 (dump_gimple_label): Likewise.
27695 (dump_gimple_bb_header): Likewise.
27697 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27700 * config/avr/avr-devices.c: Fix indentation.
27701 * config/avr/gen-avr-mmcu-specs.c: Dito.
27703 2017-06-13 Richard Biener <rguenther@suse.de>
27705 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27706 instead get vector type from stmt_info.
27707 (vectorizable_reduction): Adjust. Remove dead code.
27709 2017-06-13 Richard Biener <rguenther@suse.de>
27711 PR middle-end/81065
27712 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27713 case of C * (x * C2 + C3).
27714 (fold_addr_of_array_ref_difference): Properly fold index difference.
27716 2017-06-12 David S. Miller <davem@davemloft.net>
27719 * config/sparc/sparc.md (return expander): Emit frame blockage if
27720 function uses alloca.
27722 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27724 * combine.c (make_field_assignment): Check len rather than the mode
27725 precision when calling force_to_mode.
27727 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27729 Support multilibs and devices that see flash in RAM address range.
27732 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27733 (avr_mcu_t) <flash_pm_offset>: New field.
27734 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27735 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27736 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27737 (AVR_TINY_PM_OFFSET): Remove macro.
27738 * config/avr/avr.opt (-mshort-calls): New option.
27739 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27740 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27741 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27742 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27743 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27744 instead of avr_arch->have_jmp_call.
27745 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27746 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27747 avr_arch->flash_pm_offset to define.
27748 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27749 new field flash_pm_offset. Add entry for avrxmega3.
27750 (avr_texinfo): Add entry for avrxmega3.
27751 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27752 attiny212, attiny214,
27753 attiny412, attiny414, attiny416, attiny417,
27754 attiny814, attiny816, attiny817,
27755 attiny1614, attiny1616, attiny1617,
27756 attiny3214, attiny3216, attiny3217.
27757 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27758 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27759 (avr_print_operand_address) [AVR_TINY]: Same.
27760 (avr_asm_init_sections) <readonly_data_section>: Only patch
27761 callback if avr_arch->flash_pm_offset = 0.
27762 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27763 for rodata if avr_arch->flash_pm_offset != 0.
27764 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27765 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27766 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27767 (m_options): Append opt_rcall.
27768 (m_dirnames): Append dir_rcall.
27769 * config/avr/t-multilib: Regenerate.
27771 * configure.ac [target=avr]: Check whether avrxmega3 default
27772 linker description file works as needed.
27773 * configure: Regenerate.
27774 * doc/avr-mmcu.texi: Regenerate.
27775 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27776 <__AVR_ARCH__>: Document avrxmega3 and 103.
27777 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27778 <__AVR_SHORT_CALLS__>: Document it.
27779 <__AVR_PM_BASE_ADDRESS__>: Document it.
27780 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27781 (AVR Variable Attributes) <progmem>: Document this is
27782 not needed for avrxmega3.
27783 (AVR Named Address Spaces) <__flash>: Dito.
27785 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27787 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27789 2017-06-12 Doug Rupp <rupp@adacore.com>
27791 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27792 Append vxworks-stdint.h to the tm_file list.
27793 * config/vxworks-stdint.h: New file.
27795 2017-06-12 Martin Liska <mliska@suse.cz>
27797 PR tree-optimization/81041
27798 * tree-profile.c (gimple_gen_ic_func_profiler):
27799 Create an extra BB in profile-generate
27800 (gimple_gen_time_profiler): Likewise.
27802 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27804 PR tree-optimization/81003
27805 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27806 (update_range_test): Use it instead of force_gimple_operand_gsi.
27808 2017-06-12 Richard Biener <rguenther@suse.de>
27810 PR tree-optimization/81053
27811 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27812 with backedge value not defined in loop. Simplify def stmt
27815 2017-06-11 Tom de Vries <tom@codesourcery.com>
27818 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27820 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27821 nvptx_cannot_force_const_mem.
27823 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27825 * opts.c (finish_options): Move test for flag_split_stack after
27826 it has been initialized.
27828 2017-06-11 Jason Merrill <jason@redhat.com>
27830 * tree.h (id_equal): New.
27831 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27832 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27833 instead of strcmp of IDENTIFIER_POINTER.
27835 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27837 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27838 (mark_all_inlined_calls_cdtor): Fix formating.
27839 (inline_transform): Rescale profile before inlining.
27841 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27843 * cgraph.h (cgraph_edge::clone): Update prototype.
27844 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27845 (cgraph_node::create_clone): Update.
27846 (cgraph_node::create_version_clone): Update.
27847 * tree-inline.c (copy_bb): Update.
27848 (expand_call_inline): Update.
27850 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27852 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27853 factored out from ...
27854 (rs6000_emit_prologue): ... here.
27856 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27858 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27859 factored out from ...
27860 (rs6000_emit_prologue): ... here.
27862 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27864 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
27866 (handle_missing_profiles): Fix computation of tp_first_run.
27867 (counts_to_freqs): Do not touch freqs when count is 0.
27869 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27871 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
27874 2017-06-10 Tom de Vries <tom@codesourcery.com>
27876 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27877 attributes): Document signal effective target.
27879 2017-06-10 Tom de Vries <tom@codesourcery.com>
27881 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27882 Document effective target stack_size.
27884 2017-06-09 David Malcolm <dmalcolm@redhat.com>
27886 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
27887 to the edit_context if they can be auto-applied.
27889 2017-06-9 Ian Lance Taylor <iant@golang.org>
27891 * opts.c (finish_options): If -fsplit-stack, disable implicit
27892 -forder-blocks-and-partition.
27893 * doc/invoke.texi (Optimize Options): Document that when using
27894 -fsplit-stack -forder-blocks-and-partition is not implicitly
27897 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27899 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
27900 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
27901 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
27902 * builtins.def (abort, trap, unreachable): Declare cold.
27903 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
27904 * tree-core.h (ECF_COLD): New.
27905 * tree.c (set_call_expr_flags): Handle ECF_COLD.
27906 (build_common_builtin_nodes): Mark unreachable and abort as cold.
27908 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27910 * predict.c (unlikely_executed_stmt_p): Cleanup.
27912 2017-06-09 Richard Biener <rguenther@suse.de>
27914 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
27915 model if the ref is always written to.
27917 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27919 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
27921 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27923 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
27926 2017-06-09 Tom de Vries <tom@codesourcery.com>
27929 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
27930 "target cannot support label values" when encountering LABEL_REF.
27932 2017-06-09 Martin Liska <mliska@suse.cz>
27934 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
27935 (gimple_gen_ic_func_profiler): Emit direct comparison
27936 of __gcov_indirect_call_callee with NULL.
27937 (gimple_gen_time_profiler): Change probability from
27938 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
27940 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27942 * profile.c (edge_gcov_counts): Turn to pointer.
27943 (compute_branch_probabilities, compute_branch_probabilities): Update.
27944 (branch_prob): Do not clear edge_gcov_count.
27945 * profile.h (edge_gcov_counts): Turn to pointer.
27946 (edge_gcov_count): Update.
27948 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27950 * gimple.h (gimple_check_failed): Mark cold.
27952 2017-06-09 Richard Biener <rguenther@suse.de>
27954 PR tree-optimization/66623
27955 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
27956 refactor check_reduction into two parts, properly computing
27957 whether we have to check reduction validity for outer loop
27960 2017-06-09 Richard Biener <rguenther@suse.de>
27962 PR tree-optimization/79483
27963 * graphite-scop-detection.c (order): New global.
27964 (get_order): Compute bb to order mapping that satisfies code
27965 generation constraints.
27966 (cmp_pbbs): New helper.
27967 (build_scops): Start domwalk at entry block, sort generated
27970 2017-06-09 Richard Biener <rguenther@suse.de>
27972 PR middle-end/81007
27973 * ipa-polymorphic-call.c
27974 (ipa_polymorphic_call_context::restrict_to_inner_class):
27975 Skip FIELD_DECLs with error_mark_node type.
27976 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
27979 2017-06-09 Martin Liska <mliska@suse.cz>
27981 * predict.c (struct branch_predictor): New struct.
27982 (test_prediction_value_range): New test.
27983 (predict_c_tests): New function.
27984 * selftest-run-tests.c (selftest::run_tests): Run the function.
27985 * selftest.h: Declare new tests.
27987 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
27990 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
27991 gen_add3_insn did not fail.
27992 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
27993 r0, construct that number in a temporary reg and add that reg to r0.
27994 If asked to put the result in r0 as well, fail.
27996 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
27998 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27999 for early expansion of vec_eqv.
28001 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28003 PR middle-end/81005
28004 * ubsan.c (instrument_null): Avoid pointless code temporary.
28005 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28007 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28011 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28014 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28017 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28019 * predict.c (unlikely_executed_edge_p): Move ahead.
28020 (probably_never_executed_edge_p): Use it.
28022 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28024 PR middle-end/79988
28025 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28026 gimple_call_builtin_p call.
28028 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28030 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28031 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28032 rtl_check_failed_type2, rtl_check_failed_code1,
28033 rtl_check_failed_code2, rtl_check_failed_code_mode,
28034 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28035 rtvec_check_failed_bounds, rtl_check_failed_flag,
28036 _fatal_insn_not_found, _fatal_insn): Likewise.
28037 * tree.h (tree_contains_struct_check_failed,
28038 tree_check_failed, tree_not_check_failed,
28039 tree_class_check_failed, tree_range_check_failed,
28040 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28041 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28042 tree_operand_check_failed, omp_clause_check_failed,
28043 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28046 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28048 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28049 flag_branch_probabilities.
28050 * ipa-inline.c (edge_badness): Likewise.
28051 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28052 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28053 * predict.c (maybe_hot_frequency_p): Likewise.
28054 (probably_never_executed): Likewise.
28055 * sched-ebb.c (schedule_ebbs): Likewise.
28056 * sched-rgn.c (find_single_block_region): Likewise.
28057 * tracer.c (tail_duplicate): Likewise.
28059 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28061 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28062 longer requires x_flag_profile_use.
28064 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28066 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28067 instead of flag_reorder_blocks_and_partition.
28068 * dbxout.c (dbxout_function_end): Likewise.
28069 * dwarf2out.c (gen_subprogram_die): Likewise.
28070 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28071 * hw-doloop.c (reorg_loops): Likewise.
28072 * varasm.c (assemble_start_function,
28073 assemble_end_function): Likewise.
28074 (decide_function_section): Do not check for
28075 flag_reorder_blocks_and_partition.
28077 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28079 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28081 (chkp_get_hard_register_fake_addr_expr): Ditto.
28082 (chkp_build_addr_expr): Add check for hard reg case.
28083 (chkp_parse_array_and_component_ref): Ditto.
28084 (chkp_find_bounds_1): Ditto.
28085 (chkp_process_stmt): Don't generate bounds store for
28088 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28090 * predict.c (maybe_hot_bb_p): Do not check profile status.
28091 (maybe_hot_edge_p): Likewise.
28092 (probably_never_executed): Check for zero counts even if profile
28094 (unlikely_executed_edge_p): New function.
28095 (unlikely_executed_stmt_p): New function.
28096 (unlikely_executed_bb_p): New function.
28097 (set_even_probabilities): Use unlikely predicates.
28098 (combine_predictions_for_bb): Likewise.
28099 (predict_paths_for_bb): Likewise.
28100 (predict_paths_leading_to_edge): Likewise.
28101 (determine_unlikely_bbs): New function.
28102 (estimate_bb_frequencies): Use it.
28103 (compute_function_frequency): Use zero counts even if profile is
28105 * profile-count.h: Fix typo.
28107 2017-08-08 Julia Koval <julia.koval@intel.com>
28109 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
28110 _mm512_mask_cvtsepi16_storeu_epi8,
28111 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
28112 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
28113 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
28114 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
28115 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
28116 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
28117 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
28118 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
28119 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
28120 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
28121 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
28122 __builtin_ia32_pmovuswb256mem_mask,
28123 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
28124 __builtin_ia32_pmovwb512mem_mask): New builtins.
28126 2017-08-08 Julia Koval <julia.koval@intel.com>
28128 PR target/73350,80862
28129 * config/i386/subst.md (round): Fix round pattern.
28130 * config/i386/i386.c (ix86_erase_embedded_rounding):
28131 Fix erasing rounding for the fixed pattern.
28133 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28135 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
28137 2017-06-08 Martin Liska <mliska@suse.cz>
28139 PR gcov-profile/80911
28140 * gcov.c (block_info::block_info): New constructor.
28142 2017-06-07 Carl Love <cel@us.ibm.com>
28144 * config/rs6000/rs6000-c: The return type of the following
28145 built-in functions was implemented as int not long long. Fix sign
28146 of return value for the unsigned version of vec_mulo and vec_mule.
28147 vector unsigned long long vec_bperm (vector unsigned long long,
28148 vector unsigned char)
28149 vector signed long long vec_mule (vector signed int,
28151 vector unsigned long long vec_mule (vector unsigned int,
28152 vector unsigned int)
28153 vector signed long long vec_mulo (vector signed int,
28155 vector unsigned long long vec_mulo (vector unsigned int,
28156 vector unsigned int)
28157 * doc/extend.texi: Fix the documentation for the built-in
28160 2017-06-07 Carl Love <cel@us.ibm.com>
28163 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
28166 2017-06-07 Carl Love <cel@us.ibm.com>
28168 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
28169 support, Generate doublehv for signed int/float for BE case only.
28171 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
28173 * doc/invoke.texi (mcx16): Rewrite.
28175 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28177 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
28178 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
28179 *mov<mode>_softfloat, and an anonymous splitter): Use
28180 nonimmediate_operand instead of rs6000_nonimmediate_operand.
28182 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28184 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
28186 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28187 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
28188 (rs6000_debug_reg_global): Adjust.
28189 (rs6000_init_hard_regno_mode_ok): Adjust.
28190 (rs6000_dbx_register_number): Adjust.
28191 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
28192 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28193 Remove SPE_ACC and SPEFSCR.
28194 (REG_ALLOC_ORDER): Ditto.
28195 (FRAME_POINTER_REGNUM): Change to 111.
28196 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
28197 (REG_CLASS_NAMES): Ditto.
28198 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
28199 (REGISTER_NAMES): Ditto.
28200 (ADDITIONAL_REG_NAMES): Ditto.
28201 (rs6000_reg_names): Ditto.
28202 * config/rs6000/rs6000.md: Renumber some register number
28205 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28207 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
28209 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28210 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
28212 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
28213 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28214 Delete the SPE high registers.
28215 (REG_ALLOC_ORDER): Ditto.
28216 (enum reg_class): Remove SPE_HIGH_REGS.
28217 (REG_CLASS_NAMES): Ditto.
28218 (REG_CLASS_CONTENTS): Delete the SPE high registers.
28219 (REGISTER_NAMES): Ditto.
28220 (rs6000_reg_names): Ditto.
28221 * doc/tm.texi.in: Remove SPE as example.
28222 * doc/tm.texi: Regenerate.
28224 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28226 * config/rs6000/8540.md (ppc8540_brinc): Delete.
28227 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
28228 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
28229 * config/rs6000/rs6000.md (type): Remove "brinc".
28231 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28233 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
28234 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
28235 * config/rs6000/linuxspe.h: Delete file.
28236 * config/rs6000/rs6000.md: Don't include spe.md.
28237 * config/rs6000/spe.h: Delete file.
28238 * config/rs6000/spe.md: Delete file.
28239 * config/rs6000/t-rs6000: Remove spe.md.
28241 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28243 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
28244 (reg_or_none500mem_operand): Delete.
28245 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
28246 instead of reg_or_none500mem_operand.
28248 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28250 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28251 handling of SPE flags.
28252 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
28254 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28256 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
28258 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
28259 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
28260 paired_divv2sf3): Similar.
28261 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
28262 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
28263 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
28265 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
28266 Rename the paired_* instruction patterns.
28267 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
28269 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
28270 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
28271 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
28272 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
28273 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
28274 PAIRED_VECTOR_MODE.
28275 (struct machine_function): Delete field spe_insn_chain_scanned_p.
28276 (spe_func_has_64bit_regs_p): Delete.
28277 (spe_expand_predicate_builtin): Delete.
28278 (spe_expand_evsel_builtin): Delete.
28279 (TARGET_DWARF_REGISTER_SPAN): Do not define.
28280 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
28281 (invalid_e500_subreg): Delete.
28282 (rs6000_legitimize_address): Always force_reg op2 as well, for
28283 paired single memory accesses.
28284 (rs6000_member_type_forces_blk): Delete.
28285 (rs6000_spe_function_arg): Delete.
28286 (rs6000_expand_unop_builtin): Delete SPE handling.
28287 (rs6000_expand_binop_builtin): Ditto.
28288 (spe_expand_stv_builtin): Delete.
28289 (bdesc_2arg_spe): Delete.
28290 (spe_expand_builtin): Delete.
28291 (spe_expand_predicate_builtin): Delete.
28292 (spe_expand_evsel_builtin): Delete.
28293 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
28294 (spe_init_builtins): Delete.
28295 (spe_func_has_64bit_regs_p): Delete.
28296 (savres_routine_name): Delete "info" parameter. Adjust callers.
28297 (rs6000_emit_stack_reset): Ditto.
28298 (rs6000_dwarf_register_span): Delete.
28299 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
28300 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
28301 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
28303 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
28305 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
28306 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
28307 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
28308 mulv2sf3, divv2sf3): Delete expanders.
28310 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28312 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
28314 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28316 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
28317 * config/rs6000/rs6000.c: Ditto.
28319 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28321 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
28322 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
28323 comparison_operator.
28325 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28327 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
28328 * config/rs6000/rs6000.opt: Ditto.
28329 * config/rs6000/t-rtems: Ditto.
28331 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28333 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
28334 TARGET_E500_SINGLE by 0, simplify.
28335 * config/rs6000/rs6000.c: Ditto.
28336 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
28337 (spe_build_register_parallel): Delete.
28338 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
28339 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
28340 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
28341 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
28342 (E500_CONVERT): Delete.
28343 * config/rs6000/spe.md: Remove many patterns and all define_constants.
28345 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28347 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
28348 * config/rs6000/dfp.md: Ditto.
28349 (negdd2, *negdd2_fpr): Merge.
28350 (absdd2, *absdd2_fpr): Merge.
28351 (negtd2, *negtd2_fpr): Merge.
28352 (abstd2, *abstd2_fpr): Merge.
28353 * config/rs6000/e500.h: Delete file.
28354 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
28355 TARGET_FPRS by 1 and simplify.
28356 * config/rs6000/rs6000-c.c: Ditto.
28357 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
28358 TARGET_DF_SPE by 0.
28359 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
28361 * config/rs6000/rs6000.md: Ditto.
28362 (floatdidf2, *floatdidf2_fpr): Merge.
28363 (move_from_CR_gt_bit): Delete.
28364 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
28365 (E500_CR_IOR_COMPARE): Delete.
28366 (All patterns that require !TARGET_FPRS): Delete.
28367 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
28369 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28371 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
28373 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28375 * graphds.c (add_edge): Intitialize edge's attached data.
28376 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
28377 pointer parameter. Call pointed function on each edge during
28378 graph traversing. Skip traversing the edge when the function
28380 (graphds_dfs, graphds_scc): Ditto.
28381 (for_each_edge): New parameter. Pass the new parameter to callback
28383 * graphds.h (skip_edge_callback): New function pointer type.
28384 (graphds_dfs, graphds_scc): New function pointer parameter.
28385 (graphds_edge_callback, for_each_edge): New parameter.
28387 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28389 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
28390 out code checking if runtime alias check is possible to below ...
28391 Call the new function.
28392 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
28393 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
28395 2017-06-07 Marek Polacek <polacek@redhat.com>
28398 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
28399 TYPE_OVERFLOW_WRAPS check.
28401 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28403 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
28404 if versioning is required.
28405 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
28406 peeling with the check for versioning.
28408 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28410 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
28411 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
28412 Set true to new parameter if new ssa variable is defined.
28413 (vect_gen_vector_loop_niters): Refactor. Set range information
28414 for the new vector loop bound variable.
28415 (vect_do_peeling): Ditto.
28417 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28419 * tree-affine.c (ssa.h): Include header file.
28420 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
28421 has wrapping overflow behavior.
28423 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28425 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
28427 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28429 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
28430 (tree_to_aff_combination): ... here.
28432 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28434 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
28435 reg_pressure model function.
28436 (ivopts_global_cost_for_size): Delete.
28437 (determine_set_costs, iv_ca_recount_cost): Call new model function
28438 ivopts_estimate_reg_pressure.
28440 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28442 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
28443 expensive than udiv. Remove floating point cases from mod.
28445 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28447 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
28448 Increase idiv cost.
28450 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28452 * config/aarch64/aarch64.md
28453 (copysignsf3): Fix mask generation.
28455 2017-06-07 Jakub Jelinek <jakub@redhat.com>
28457 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
28459 (class dump_manager): Add register_dumps method.
28460 * dumpfile.c: Include langhooks.h.
28461 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
28462 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
28463 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
28464 (dump_manager::dump_register): Start with 512 entries instead of 32.
28465 (dump_manager::register_dumps): New method.
28466 * toplev.c (general_init): Instead of invoking register_dumps
28467 langhook, invoke register_dumps method on the dump manager.
28468 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
28471 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
28473 * doc/md.texi: Clarify the restrictions on a define_insn condition.
28474 Say that # requires an associated define_split to exist, and that
28475 the define_split must be suitable for use after register allocation.
28477 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28479 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
28480 (compute_outgoing_frequencies): Also initialize zero counts.
28481 (find_many_sub_basic_blocks): Do not produce uninitialized profile
28482 around loops; preserve more of profile when nothing changes.
28484 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
28486 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
28488 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
28489 * config/arm/arm-cpu-cdata.h: Regenerate.
28490 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
28491 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
28492 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
28493 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
28494 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
28496 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
28498 * config/arm/t-rmprofile: Likewise.
28499 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
28501 2017-06-06 David S. Miller <davem@davemloft.net>
28504 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
28505 blockage if function uses alloca.
28507 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28509 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
28510 New "uid" fields to hold pretty-print IDs of group and ref.
28511 Memory references are now identified as <group_id>:<ref_id>
28512 instead of using [random] addresses.
28513 (dump_mem_details): Simplify, no functional change.
28514 (dump_mem_ref): Simplify and make output more concise.
28515 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
28516 (find_or_create_group): Initialize group uid.
28517 (record_ref): Initialize ref uid. Improve debug output.
28518 (prune_group_by_reuse, should_issue_prefetch_p,)
28519 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
28520 (mark_nontemporal_store, determine_loop_nest_reuse):
28521 Improve debug output.
28523 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28525 * dbgcnt.def (prefetch): New debug counter.
28526 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
28527 (schedule_prefetches): Stop issueing prefetches if debug counter
28530 2017-06-06 Tom de Vries <tom@codesourcery.com>
28532 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
28533 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
28535 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28537 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
28538 Use aarch64_reg_or_zero predicate for operand 4.
28539 (aarch64_compare_and_swap<mode> define_insn_and_split):
28540 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
28541 (aarch64_store_exclusive<mode>): Likewise for operand 2.
28543 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
28545 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
28546 (arm_compute_save_core_reg_mask): This.
28547 (thumb1_compute_save_reg_mask): Rename into ...
28548 (thumb1_compute_save_core_reg_mask): This.
28549 (arm_compute_save_reg0_reg12_mask): Adapt comment.
28550 (arm_compute_frame_layout): Likewise.
28552 2017-06-06 Richard Biener <rguenther@suse.de>
28554 PR tree-optimization/80974
28555 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
28556 keep or clear leaders SSA info.
28558 2017-06-06 Tom de Vries <tom@codesourcery.com>
28560 * config/nvptx/nvptx.c (split_mode_p): New function.
28561 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
28563 2017-06-06 Tom de Vries <tom@codesourcery.com>
28565 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
28567 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28570 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
28573 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28575 * shrink-wrap.c (handle_simple_exit): Update profile.
28576 (try_shrink_wrapping): Upate profile.
28578 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28580 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
28581 (tree_guess_outgoing_edge_probabilities): New.
28582 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
28583 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
28585 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28587 * ipa-split.c (split_function): Initialize return bb profile.
28589 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28591 * profile.c (compute_branch_probabilities): Also initialize
28592 EXIT_BLOCK profile.
28594 2017-06-06 Richard Biener <rguenther@suse.de>
28596 PR tree-optimization/80928
28597 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
28598 (vect_analyze_loop_operations): Properly guard analysis for
28600 (vect_transform_loop): Likewise.
28601 (vect_analyze_loop_2): Also reset SLP type on PHIs.
28602 (vect_model_induction_cost): Do not cost for pure SLP.
28603 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
28604 of induction in inner loop vectorization.
28605 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
28606 (vect_get_and_check_slp_defs): Handle vect_induction_def.
28607 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
28609 (vect_analyze_slp_cost_1): Cost induction.
28610 (vect_detect_hybrid_slp_stmts): Handle PHIs.
28611 (vect_get_slp_vect_defs): Likewise.
28612 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
28613 (vect_transform_stmt): Handle SLP reductions.
28614 * tree-vectorizer.h (vectorizable_induction): Adjust.
28616 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28618 * config/rs6000/rs6000.c (make_resolver_func): Update
28619 init_lowered_empty_function call.
28621 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28623 * doc/invoke.texi: Document the -fprofile-abs-path option.
28624 * common.opt (fprofile-abs-path): New option.
28625 * gcov-io.h (gcov_write_filename): Declare.
28626 * gcov-io.c (gcov_write_filename): New function.
28627 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28628 * profile.c (output_location): Likewise.
28630 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28632 * shring-wrap.c: Revert accidental commit.
28634 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28636 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28638 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28640 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28642 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28643 profile in callgraph edge.
28644 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28645 (apply_scale): Likewise.
28646 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28647 Also scale profile when inlining function with zero profile.
28648 (initialize_cfun): Update exit block profile even when it is zero.
28649 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28650 when profile is read.
28652 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28654 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28655 (CLONE_*): New constants to define the processors we can generate
28656 code for with the target_clone attribute.
28657 (rs6000_clone_map): New array to identify which clone processors
28658 the current program is running on.
28659 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28660 target_clone attribute.
28661 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28662 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28663 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28664 (cpu_expand_builtin): Add support for target_clone attribute.
28665 (rs6000_valid_attribute_p): Allow "default" attribute.
28666 (get_decl_name): New debug function to simplify printing the
28667 current function name in debugging statements.
28668 (rs6000_clone_priority): New functions to support the target_clone
28669 attribute, and be able to generate code to switch between ISA 2.05
28670 through ISA 3.0 (power6 through power9).
28671 (rs6000_compare_version_priority): Likewise.
28672 (rs6000_get_function_versions_dispatcher): Likewise.
28673 (make_resolver_func): Likewise.
28674 (add_condition_to_bb): Likewise.
28675 (dispatch_function_versions): Likewise.
28676 (rs6000_generate_version_dispatcher_body): Likewise.
28677 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28678 (fusion_gpr_load_p): Fix a spacing issue.
28679 * doc/extend.texi (Common Function Attributes): Document that the
28680 PowerPC supports the target_clone attribute.
28682 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28684 * config/arm/arm.h: explain F symbol found in description of ARM
28685 register allocation in its legend.
28687 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28689 * config/mips/frame-header-opt.c: Include profile-count.h.
28690 * config/riscv/riscv.c: Include profile-count.h
28692 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28694 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28696 (sm_set_flag_if_changed): Add bbs field.
28697 (execute_sm_if_changed_flag_set): Pass BBS.
28698 (execute_sm): Update.
28700 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28702 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28705 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28707 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28708 (peephole2): New peephole2 to emit the above.
28709 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28711 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28713 * config/aarch64/aarch64.c (define_peephole2 above
28714 *sub_<shift>_<mode>): New peephole.
28716 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28718 * config/i386/i386.c (make_resolver_func): Update.
28719 * Makefile.in: Add profile-count.h and profile-count.o
28720 * auto-profile.c (afdo_indirect_call): Update to new API.
28721 (afdo_set_bb_count): Update.
28722 (afdo_propagate_edge): Update.
28723 (afdo_propagate_circuit): Update.
28724 (afdo_calculate_branch_prob): Update.
28725 (afdo_annotate_cfg): Update.
28726 * basic-block.h: Include profile-count.h
28727 (struct edge_def): Turn count to profile_count.
28728 (struct basic_block_def): Likewie.
28729 (REG_BR_PROB_BASE): Move to profile-count.h
28730 (RDIV): Move to profile-count.h
28731 * bb-reorder.c (max_entry_count): Turn to profile_count.
28732 (find_traces): Update.
28733 (rotate_loop):Update.
28734 (connect_traces):Update.
28735 (sanitize_hot_paths):Update.
28736 * bt-load.c (migrate_btr_defs): Update.
28737 * cfg.c (RDIV): Remove.
28738 (init_flow): Use alloc_block.
28739 (alloc_block): Uninitialize count.
28740 (unchecked_make_edge): Uninitialize count.
28741 (check_bb_profile): Update.
28742 (dump_edge_info): Update.
28743 (dump_bb_info): Update.
28744 (update_bb_profile_for_threading): Update.
28745 (scale_bbs_frequencies_int): Update.
28746 (scale_bbs_frequencies_gcov_type): Update.
28747 (scale_bbs_frequencies_profile_count): New.
28748 * cfg.h (update_bb_profile_for_threading): Update.
28749 (scale_bbs_frequencies_profile_count): Declare.
28750 * cfgbuild.c (compute_outgoing_frequencies): Update.
28751 (find_many_sub_basic_blocks): Update.
28752 * cfgcleanup.c (try_forward_edges): Update.
28753 (try_crossjump_to_edge): Update.
28754 * cfgexpand.c (expand_gimple_tailcall): Update.
28755 (construct_exit_block): Update.
28756 * cfghooks.c (verify_flow_info): Update.
28757 (dump_bb_for_graph): Update.
28758 (split_edge): Update.
28759 (make_forwarder_block): Update.
28760 (duplicate_block): Update.
28761 (account_profile_record): Update.
28762 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28763 (get_estimated_loop_iterations): Update.
28764 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28765 (single_likely_exit): Update.
28766 * cfgloopmanip.c (scale_loop_profile): Update.
28768 (set_zero_probability): Update.
28769 (lv_adjust_loop_entry_edge): Update.
28770 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28771 (purge_dead_edges): Update.
28772 (rtl_account_profile_record): Update.
28773 * cgraph.c (cgraph_node::create): Uninitialize count.
28774 (symbol_table::create_edge): Uninitialize count.
28775 (cgraph_update_edges_for_call_stmt_node): Update.
28776 (cgraph_edge::dump_edge_flags): Update.
28777 (cgraph_node::dump): Update.
28778 (cgraph_edge::maybe_hot_p): Update.
28779 * cgraph.h: Include profile-count.h
28780 (create_clone), create_edge, create_indirect_edge): Update.
28781 (cgraph_node): Turn count to profile_count.
28782 (cgraph_edge0: Likewise.
28783 (make_speculative, clone): Update.
28784 (create_edge): Update.
28785 (init_lowered_empty_function): Update.
28786 * cgraphclones.c (cgraph_edge::clone): Update.
28787 (duplicate_thunk_for_node): Update.
28788 (cgraph_node::create_clone): Update.
28789 * cgraphunit.c (cgraph_node::analyze): Update.
28790 (cgraph_node::expand_thunk): Update.
28791 * final.c (dump_basic_block_info): Update.
28792 * gimple-streamer-in.c (input_bb): Update.
28793 * gimple-streamer-out.c (output_bb): Update.
28794 * graphite.c (print_global_statistics): Update.
28795 (print_graphite_scop_statistics): Update.
28796 * hsa-brig.c: Include basic-block.h.
28797 * hsa-dump.c: Include basic-block.h.
28798 * hsa-gen.c (T sum_slice): Update.
28799 (convert_switch_statements):Update.
28800 * hsa-regalloc.c: Include basic-block.h.
28801 * ipa-chkp.c (chkp_produce_thunks): Update.
28802 * ipa-cp.c (struct caller_statistics): Update.
28803 (init_caller_stats): Update.
28804 (gather_caller_stats): Update.
28805 (ipcp_cloning_candidate_p): Update.
28806 (good_cloning_opportunity_p): Update.
28807 (get_info_about_necessary_edges): Update.
28808 (dump_profile_updates): Update.
28809 (update_profiling_info): Update.
28810 (update_specialized_profile): Update.
28811 (perhaps_add_new_callers): Update.
28812 (decide_about_value): Update.
28813 (ipa_cp_c_finalize): Update.
28814 * ipa-devirt.c (struct odr_type_warn_count): Update.
28815 (struct decl_warn_count): Update.
28816 (struct final_warning_record): Update.
28817 (possible_polymorphic_call_targets): Update.
28818 (ipa_devirt): Update.
28819 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28820 * ipa-icf.c (sem_function::merge): Update.
28821 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28822 * ipa-inline.c (compute_uninlined_call_time): Update.
28823 (compute_inlined_call_time): Update.
28824 (want_inline_small_function_p): Update.
28825 (want_inline_self_recursive_call_p): Update.
28826 (edge_badness): Update.
28827 (lookup_recursive_calls): Update.
28828 (recursive_inlining): Update.
28829 (inline_small_functions): Update.
28830 (dump_overall_stats): Update.
28831 (dump_inline_stats): Update.
28832 * ipa-profile.c (ipa_profile_generate_summary): Update.
28833 (ipa_propagate_frequency): Update.
28834 (ipa_profile): Update.
28835 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28836 * ipa-utils.c (ipa_merge_profiles): Update.
28837 * loop-doloop.c (doloop_modify): Update.
28838 * loop-unroll.c (report_unroll): Update.
28839 (unroll_loop_runtime_iterations): Update.
28840 * lto-cgraph.c (lto_output_edge): Update.
28841 (lto_output_node): Update.
28842 (input_node): Update.
28843 (input_edge): Update.
28844 (merge_profile_summaries): Update.
28845 * lto-streamer-in.c (input_cfg): Update.
28846 * lto-streamer-out.c (output_cfg): Update.
28847 * mcf.c (create_fixup_graph): Update.
28848 (adjust_cfg_counts): Update.
28849 (sum_edge_counts): Update.
28850 * modulo-sched.c (sms_schedule): Update.
28851 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28852 * predict.c (maybe_hot_count_p): Update.
28853 (probably_never_executed): Update.
28854 (dump_prediction): Update.
28855 (combine_predictions_for_bb): Update.
28856 (propagate_freq): Update.
28857 (handle_missing_profiles): Update.
28858 (counts_to_freqs): Update.
28859 (rebuild_frequencies): Update.
28860 (force_edge_cold): Update.
28861 * predict.h: Include profile-count.h
28862 (maybe_hot_count_p, counts_to_freqs): UPdate.
28863 * print-rtl-function.c: Do not include cfg.h
28864 * print-rtl.c: Include basic-block.h
28865 * profile-count.c: New file.
28866 * profile-count.h: New file.
28867 * profile.c (is_edge_inconsistent): Update.
28868 (correct_negative_edge_counts): Update.
28869 (is_inconsistent): Update.
28870 (set_bb_counts): Update.
28871 (read_profile_edge_counts): Update.
28872 (compute_frequency_overlap): Update.
28873 (compute_branch_probabilities): Update; Initialize and deinitialize
28875 (branch_prob): Update.
28876 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
28877 (edge_gcov_count): New.
28878 (bb_gcov_count): New.
28879 * shrink-wrap.c (try_shrink_wrapping): Update.
28880 * tracer.c (better_p): Update.
28881 * trans-mem.c (expand_transaction): Update.
28882 (ipa_tm_insert_irr_call): Update.
28883 (ipa_tm_insert_gettmclone_call): Update.
28884 * tree-call-cdce.c: Update.
28885 * tree-cfg.c (gimple_duplicate_sese_region): Update.
28886 (gimple_duplicate_sese_tail): Update.
28887 (gimple_account_profile_record): Update.
28888 (execute_fixup_cfg): Update.
28889 * tree-inline.c (copy_bb): Update.
28890 (copy_edges_for_bb): Update.
28891 (initialize_cfun): Update.
28892 (freqs_to_counts): Update.
28893 (copy_cfg_body): Update.
28894 (expand_call_inline): Update.
28895 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
28896 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
28897 (try_unroll_loop_completely): Update.
28898 (try_peel_loop): Update.
28899 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
28900 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
28901 * tree-ssa-loop-split.c (connect_loops): Update.
28902 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
28903 * tree-ssa-reassoc.c (branch_fixup): Update.
28904 * tree-ssa-tail-merge.c (replace_block_by): Update.
28905 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
28906 (compute_path_counts): Update.
28907 (update_profile): Update.
28908 (recompute_probabilities): Update.
28909 (update_joiner_offpath_counts): Update.
28910 (estimated_freqs_path): Update.
28911 (freqs_to_counts_path): Update.
28912 (clear_counts_path): Update.
28913 (ssa_fix_duplicate_block_edges): Update.
28914 (duplicate_thread_path): Update.
28915 * tree-switch-conversion.c (case_bit_test_cmp): Update.
28916 (struct switch_conv_info): Update.
28917 * tree-tailcall.c (decrease_profile): Update.
28918 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
28919 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
28920 * value-prof.c (check_counter): Update.
28921 (gimple_divmod_fixed_value): Update.
28922 (gimple_mod_pow2): Update.
28923 (gimple_mod_subtract): Update.
28924 (gimple_ic_transform): Update.
28925 (gimple_stringop_fixed_value): Update.
28926 * value-prof.h (gimple_ic): Update.
28928 2017-06-02 Carl Love <cel@us.ibm.com>
28930 * config/rs6000/rs6000-c: Add support for built-in functions
28931 vector double vec_doublee (vector signed int);
28932 vector double vec_doublee (vector unsigned int);
28933 vector double vec_doublee (vector float);
28934 vector double vec_doubleh (vector signed int);
28935 vector double vec_doubleh (vector unsigned int);
28936 vector double vec_doubleh (vector float);
28937 vector double vec_doublel (vector signed int);
28938 vector double vec_doublel (vector unsigned int);
28939 vector double vec_doublel (vector float);
28940 vector double vec_doubleo (vector signed int);
28941 vector double vec_doubleo (vector unsigned int);
28942 vector double vec_doubleo (vector float);.
28943 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
28944 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
28946 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
28947 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
28948 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
28950 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
28951 vec_doublel, vec_doubleh.
28952 * doc/extend.texi: Update the built-in documentation file for the
28953 new built-in functions.
28955 2017-06-02 David Malcolm <dmalcolm@redhat.com>
28958 * ipa-inline-analysis.c (free_growth_caches): Set
28959 edge_removal_hook_holder to NULL after removing it.
28961 2017-06-02 Sudakshina Das <sudi.das@arm.com>
28963 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
28964 comparision with zero.
28966 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28967 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28968 for early expansion of vec_min and vec_max builtins.
28969 (builtin_function_type): Add min/max unsigned variants to those
28970 identified as having unsigned arguments.
28972 2017-06-02 Olivier Hainque <hainque@adacore.com>
28974 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
28976 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28978 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
28979 Use VALL_F16 iterator rather than VALL.
28981 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28983 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
28984 Emit CBNZ inside loop when doing a strong exchange and comparing
28985 against zero. Generate the CC flags after the loop.
28987 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
28989 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
28990 (dl_section_ref): New.
28991 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
28992 On AIX, append an expression to subtract the size of the
28993 section length to dl_section_ref.
28995 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28997 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28998 for early expansion of vector absolute builtins.
29000 2017-06-02 Richard Biener <rguenther@suse.de>
29002 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29003 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29005 2017-06-02 Richard Biener <rguenther@suse.de>
29007 PR tree-optimization/80948
29008 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29009 stmt order as well.
29011 2017-06-02 Richard Biener <rguenther@suse.de>
29013 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29015 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29016 for inductions as relevant.
29018 2017-06-02 Richard Biener <rguenther@suse.de>
29020 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29021 (vectorizable_induction): ... this. Remove dead code.
29023 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29025 * builtins. (expand_builtin_alloca): Remove second parameter and
29026 infer its value from the first parameter instead.
29027 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29029 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29031 PR rtl-optimization/80903
29032 * loop-doloop.c (add_test): Unshare sequence.
29034 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29036 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
29038 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
29040 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29042 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29043 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29044 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29045 (xlogue_layout::get_stub_rtx): Make static.
29046 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29047 (xlogue_layout::compute_stub_managed_regs): Rename to...
29048 (xlogue_layout::count_stub_managed_regs): ...this.
29049 (xlogue_layout::is_stub_managed_reg): New function.
29050 (xlogue_layout::m_stub_names): Rename to...
29051 (xlogue_layout::s_stub_names): ...this, make static.
29052 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29053 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29054 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29055 xlogue_layout::s_stub_names): Instantiate statics.
29056 (stub_managed_regs): Remove.
29057 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29058 (disable_call_ms2sysv_xlogues): Rename to...
29059 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29060 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29062 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29063 change after reload_completed.
29064 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29066 (ix86_expand_prologue): Likewise.
29067 (ix86_expand_epilogue): Likewise.
29068 (ix86_expand_split_stack_prologue): Likewise.
29069 (ix86_compute_frame_layout): Remove frame parameter ...
29070 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29071 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29073 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29074 (ix86_frame): Move from here ...
29075 * config/i386/i386.h (ix86_frame): ... to here.
29076 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29077 complete ix86_frame data structure instead. Remove some_ld_name.
29079 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29081 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29082 symbols that hold a DECL_VALUE_EXPR.
29084 2017-06-01 Martin Jambor <mjambor@suse.cz>
29086 PR tree-optimization/80898
29087 * tree-sra.c (process_subtree_disqualification): Removed.
29088 (disqualify_candidate): Do not acll
29089 process_subtree_disqualification.
29090 (subtree_mark_written_and_enqueue): New function.
29091 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29092 RHS has been disqualified and re-queue LHS if necessary. Apart
29093 from that, ignore disqualified RHS.
29095 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29097 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29098 address fetch for z10 or later.
29100 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29102 * config/arc/arc.md (tst_movb): Add guard when splitting.
29104 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29106 * config/arc/arc.c (arc_can_eliminate): Test against
29107 arc_frame_pointer_needed.
29109 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29111 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
29112 to prevent store reordering.
29113 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
29114 (type): Add block type.
29115 (stack_tie): Define special instruction to be used in
29118 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29120 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
29121 constraint. It is not valid for the pattern.
29122 (noncommutative_binary_comparison): Likewise.
29124 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29126 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
29129 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29131 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
29132 be used by the reg-alloc.
29134 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29136 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
29137 reg-alloc when having mul64 or mul32x16 instructions.
29138 (mulsidi3): Likewise.
29139 (umulsidi3): Likewise.
29140 (mulsi32x16): New pattern.
29141 (mulsi64): Likewise.
29142 (mulsidi64): Likewise.
29143 (umulsidi64): Likewise.
29144 (MUL32x16_REG): Define.
29145 (mul64_600): Use MUL32x16_REG.
29146 (mac64_600): Likewise.
29147 (umul64_600): Likewise.
29148 (umac64_600): Likewise.
29150 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29152 * config/arc/arc.md (mulsi3_700): Make it commutative.
29154 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
29156 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
29158 (*sign_extendsidi2_insn): Likewise for movstosw.
29160 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29162 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
29163 the type of the input discriminant value. Convert the
29164 discriminant value of signedness vary.
29166 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
29168 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
29169 Add to -Wall section.
29171 2017-06-01 Richard Biener <rguenther@suse.de>
29173 PR middle-end/66313
29174 * fold-const.c (fold_plusminus_mult_expr): If the factored
29175 factor may be zero use a wrapping type for the inner operation.
29176 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
29177 and handle moved defs.
29178 (process_assignment): Properly guard the unary op case. Return a
29179 tri-state indicating that moving the stmt before the call may allow
29180 to continue. Pass through to_move.
29181 (find_tail_calls): Handle moving unrelated defs before
29184 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
29187 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
29188 splitter result in the canonical way.
29190 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
29192 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
29193 also for 32bit target. Update insn attributes.
29194 (zero-extendsidi2 splitter): Allow all registers for operand 1.
29196 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
29198 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
29199 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
29200 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
29201 (_mm_maskz_min_ss): New intrinsics.
29203 2017-05-31 Martin Liska <mliska@suse.cz>
29205 * tree-vect-loop.c (vect_create_epilog_for_reduction):
29206 Change comment style to one we normally use.
29207 (vectorizable_reduction): Likewise.
29208 (vectorizable_induction): Likewise.
29209 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
29210 (vectorizable_call): Likewise.
29211 (vectorizable_simd_clone_call): Likewise.
29212 (vectorizable_conversion): Likewise.
29213 (vectorizable_assignment): Likewise.
29214 (vectorizable_shift): Likewise.
29215 (vectorizable_operation): Likewise.
29216 (vectorizable_store): Likewise.
29217 (vectorizable_load): Likewise.
29218 * tree-vectorizer.h: Likewise.
29220 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
29222 * passes.c (emergency_dump_function): New.
29223 * tree-pass.h (emergency_dump_function): Declare.
29224 * plugin.c (plugins_internal_error_function): Remove.
29225 * plugin.h (plugins_internal_error_function): Remove declaration.
29226 * toplev.c (internal_error_function): New static function. Use it...
29227 (general_init): ...here.
29229 2017-05-31 Graham Markall <graham.markall@embecosm.com>
29231 * config/arc/arc.c (arc_print_operand): Handle constant operands.
29232 (arc_rtx_costs): Add costs for new patterns.
29233 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
29234 * config/arc/predicates.md: Add _1_2_3_operand predicate.
29236 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29238 * tree-ssa-strlen.c (get_next_strinfo): New function.
29239 (get_stridx_plus_constant): Use it.
29240 (zero_length_string): Likewise.
29241 (adjust_related_strinfos): Likewise.
29242 (adjust_last_stmt): Likewise.
29244 2017-05-31 Richard Biener <rguenther@suse.de>
29247 * config/i386/i386.c (ix86_expand_builtin): Remove assert
29248 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
29250 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29252 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
29253 loop_vinfo argument and use of dependence distance vectors.
29254 Check instead whether the two references differ only in their
29255 initial value and assume that they have the same alignment if the
29256 difference is a multiple of the vector alignment.
29257 (vect_analyze_data_refs_alignment): Update call accordingly.
29259 2017-05-31 Martin Liska <mliska@suse.cz>
29262 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
29264 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29266 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
29267 (create_intersect_range_checks): Move from ...
29268 * tree-data-ref.c (create_intersect_range_checks_index)
29269 (create_intersect_range_checks): ... to here.
29270 (create_runtime_alias_checks): New function factored from ...
29271 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
29272 here. Call above function.
29273 * tree-data-ref.h (create_runtime_alias_checks): New function.
29275 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29277 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
29278 segment length for dr_b and compute it in wide_int.
29280 2017-05-31 Richard Biener <rguenther@suse.de>
29282 PR tree-optimization/80906
29283 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
29284 and pass through iv_map.
29285 (copy_bb_and_scalar_dependences): Adjust.
29286 (translate_pending_phi_nodes): Likewise.
29287 (copy_loop_close_phi_args): Handle code-generating IVs instead
29290 2017-05-30 David Malcolm <dmalcolm@redhat.com>
29292 * diagnostic-color.c (color_dict): Add "type-diff".
29293 (parse_gcc_colors): Update comment.
29294 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
29295 -fdiagnostics-show-template-tree and -fno-elide-type.
29296 (GCC_COLORS): Add type-diff to example.
29298 (-fdiagnostics-show-template-tree): New.
29299 (-fno-elide-type): New.
29300 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
29301 the pp_format_decoder callback. Call any m_format_postprocessor's
29303 (pretty_printer::pretty_printer): Initialize
29304 m_format_postprocessor.
29305 (pretty_printer::~pretty_printer): Delete any
29306 m_format_postprocessor.
29307 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
29308 (class format_postprocessor): New class.
29309 (struct pretty_printer::format_decoder): Document the new parameters.
29310 (struct pretty_printer::m_format_postprocessor): New field.
29311 * tree-diagnostic.c (default_tree_printer): Update for new
29312 bool and const char ** params.
29313 * tree-diagnostic.h (default_tree_printer): Likewise.
29315 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
29317 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
29318 (lwa_operand): Delete rs6000_gen_cell_microcode test.
29319 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29320 rs6000_gen_cell_microcode code.
29321 (rs6000_final_prescan_insn): Delete.
29322 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
29323 "warn-cell-microcode" entries.
29324 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
29325 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
29326 throughout. Change cc_reg_not_micro_cr0_operand to
29327 cc_reg_not_cr0_operand throughout.
29328 (*extendhi<mode>2_noload): Delete.
29329 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
29330 (mwarn-cell-microcode): Delete.
29331 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
29332 -mgen-cell-microcode and -mwarn-cell-microcode.
29334 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
29337 * config/i386/constraints.md (Yd): New constraint.
29339 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
29340 and (?Yd, r) alternatives. Update insn attributes.
29341 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
29342 and (?*Yd, r) alternatives. Update insn attributes.
29343 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
29345 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
29347 * gimplify.c (gimplify_modify_expr): Don't create a
29348 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
29351 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
29353 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
29355 2017-05-30 Richard Biener <rguenther@suse.de>
29357 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
29358 and reduc_def fields.
29359 (STMT_VINFO_REDUC_TYPE): New define.
29360 (STMT_VINFO_REDUC_DEF): Likewise.
29361 (vect_force_simple_reduction): Adjust prototype.
29362 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
29363 (vect_is_simple_reduction): Remove check_reduction argument.
29364 (vect_force_simple_reduction): Adjust and set
29365 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29366 (vectorizable_reduction): Do not re-do reduction analysis
29367 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29368 * tree-parloops.c (gather_scalar_reductions): Adjust.
29370 2017-05-30 Richard Biener <rguenther@suse.de>
29372 PR middle-end/80901
29373 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
29376 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
29378 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
29379 Introduce unknown_misalignment parameter and remove vf.
29380 (vect_peeling_hash_get_lowest_cost):
29381 Pass unknown_misalignment parameter.
29382 (vect_enhance_data_refs_alignment):
29383 Fix unsupportable data ref treatment.
29385 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29387 * tree-vect-data-refs.c (vect_get_data_access_cost):
29388 Workaround for SLP handling.
29389 (vect_enhance_data_refs_alignment):
29390 Compute costs for doing no peeling at all, compare to the best
29391 peeling costs so far and avoid peeling if cheaper.
29393 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29395 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
29396 Return peeling info and set costs to zero for unlimited cost
29398 (vect_enhance_data_refs_alignment): Also inspect all datarefs
29399 with unknown misalignment. Compute and costs for unknown
29400 misalignment, compare them to the costs for known misalignment
29401 and choose the cheapest for peeling.
29403 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29405 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
29406 (vect_get_peeling_costs_all_drs): Create function.
29407 (vect_peeling_hash_get_lowest_cost):
29408 Use vect_get_peeling_costs_all_drs.
29409 (vect_peeling_supportable): Create function.
29410 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
29412 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29414 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
29415 DR_HAS_NEGATIVE_STEP.
29416 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
29417 (vect_enhance_data_refs_alignment): Use.
29418 (vect_duplicate_ssa_name_ptr_info): Use.
29419 * tree-vectorizer.h (dr_misalignment): Use.
29420 (known_alignment_for_access_p): Use.
29422 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
29425 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
29427 (has_section_name): New function.
29429 2017-05-30 Martin Liska <mliska@suse.cz>
29432 * auto-profile.c (get_function_decl_from_block): Fix
29435 2017-05-30 Richard Biener <rguenther@suse.de>
29437 PR middle-end/80876
29438 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
29440 2017-05-30 Martin Liska <mliska@suse.cz>
29442 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
29443 * dumpfile.h (struct dump_file_info): Remove ctors.
29445 2017-05-30 Martin Liska <mliska@suse.cz>
29447 * predict.def: Fix GNU coding style.
29449 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29451 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29452 Mark 'to' argument with ATTRIBUTE_UNUSED.
29454 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29456 * config/xtensa/xtensa.c (xtensa_emit_call): Use
29457 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
29458 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
29461 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
29463 * doc/install.texi (Options specification): Restore entry of
29464 --enable-sjlj-exceptions.
29466 2017-05-27 Michael Eager <eager@eagercon.com>
29469 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
29471 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
29473 * config/microblaze/microblaze.h
29474 (FIXED_REGISTERS): Update in macro.
29475 (CALL_USED_REGISTERS): Update in macro.
29477 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
29479 * doc/install.texi: Add links to macOS binary distributions.
29481 2017-05-27 Jakub Jelinek <jakub@redhat.com>
29485 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29487 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29489 2017-05-26 Martin Liska <mliska@suse.cz>
29491 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
29493 2017-05-26 Martin Liska <mliska@suse.cz>
29495 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
29496 always leading ';; '.
29497 (dump_bb_info): Likewise.
29498 (brief_dump_cfg): Likewise.
29499 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
29500 * dumpfile.c: Remove usage of TDF_VERBOSE.
29501 * dumpfile.h (enum dump_kind): Likewise.
29502 (dump_gimple_bb_header): Do not use TDF_COMMENT.
29503 * print-tree.c (debug_verbose): Remove.
29504 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
29505 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
29506 * tree-diagnostic.c (default_tree_printer): Replace
29507 TDF_DIAGNOSTIC with TDF_SLIM.
29509 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29511 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
29512 in parameter loop, rather than loop_vinfo.
29513 (create_intersect_range_checks): Ditto.
29514 (vect_create_cond_for_alias_checks): Update call to above functions.
29516 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29518 PR tree-optimization/80815
29519 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
29520 for merging runtime alias checks. Handle negative DR_STEPs.
29522 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29524 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
29526 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
29527 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
29528 out code pruning runtime alias checks.
29529 * tree-data-ref.c (prune_runtime_alias_test_list): New function
29530 factored out from above.
29531 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29533 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29535 (prune_runtime_alias_test_list): New decalaration.
29537 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29539 * tree-vect-data-refs.c (compare_tree): Rename and move ...
29540 * tree-data-ref.c (data_ref_compare_tree): ... to here.
29541 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
29542 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
29543 (operator==, comp_dr_with_seg_len_pair): Ditto.
29544 (vect_prune_runtime_alias_test_list): Ditto.
29546 2017-05-26 Martin Liska <mliska@suse.cz>
29549 * params.def: Bound partial-inlining-entry-probability param.
29551 2017-05-26 Marek Polacek <polacek@redhat.com>
29554 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
29557 2017-05-26 Richard Biener <rguenther@suse.de>
29559 PR tree-optimization/80842
29560 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
29563 2017-05-26 Richard Biener <rguenther@suse.de>
29565 PR tree-optimization/80844
29566 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
29568 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
29570 * doc/md.texi (Machine Constraints): Update x86 family
29571 machine constraints section to match 'config/i386/constraints.md'.
29573 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
29575 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
29577 2017-05-25 Nathan Sidwell <nathan@acm.org>
29579 * doc/invoke.texi (--enable-languages): Update documentation.
29581 2017-05-25 Martin Liska <mliska@suse.cz>
29583 * dumpfile.c: Add TDF_FOLDING.
29584 * dumpfile.h (enum dump_kind): Likewise.
29585 * genmatch.c (dt_simplify::gen_1): Use it.
29587 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29589 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
29591 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29593 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29594 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
29596 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29598 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
29599 * match.pd (X == C): Rewrite it here.
29600 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
29601 with_certain_nonzero_bits2): New predicates.
29602 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
29604 2017-05-24 Nathan Sidwell <nathan@acm.org>
29606 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
29609 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
29612 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29614 * config/powerpcspe: New port. Files are copied from the rs6000
29615 port, with "rs6000" in filenames replaced by "powerpcspe".
29617 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29619 PR rtl-optimization/80754
29620 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29622 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29624 * config/sparc/sparc.md (length): Return the correct value for -mflat
29625 sibcalls to match output_sibcall.
29627 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29631 * config/rs6000/rs6000.c (struct machine_function): Add new field
29633 (rs6000_get_separate_components): Init that field, use it.
29634 (rs6000_components_for_bb): Use the field.
29636 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29638 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29640 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29642 PR middle-end/80823
29643 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29645 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29648 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29649 address against address_operand predicate.
29650 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29652 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29654 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29655 parameters passed indirectly.
29657 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29659 * config/i386/i386.md (*movdi_internal): Remove SSE4
29660 alternative 18 (?r, *v). Update insn attributes.
29661 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29662 Update insn attributes.
29663 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29664 Update insn attributes.
29665 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29666 alternative 1 (r, v). Remove isa attribute.
29667 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29668 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29669 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29671 2017-05-23 Tom de Vries <tom@codesourcery.com>
29673 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29676 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29678 * cgraphunit.c (symbol_table::process_new_functions): Update.
29679 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29680 (inline_generate_summary): Rename to ...
29681 (ipa_fn_summary_generate): ... this one.
29682 (inline_read_summary): Rename to ...
29683 (ipa_fn_summary_read): ... this one.
29684 (inline_write_summary): Rename to ...
29685 (ipa_fn_summary_write): ... this one.
29686 (inline_free_summary): Rename to ...
29687 (ipa_free_fn_summary): ... this one.
29688 (pass_data_local_fn_summary, pass_local_fn_summary,
29689 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29690 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29691 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29692 make_pass_ipa_fn_summary): New.
29693 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29694 inline_write_summary, inline_free_summary): Remove.
29695 (ipa_free_fn_summary) : New.
29696 * ipa-inline.c (ipa_inline): Update.
29697 (pass_ipa_inline): Do not generate summaries.
29698 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29700 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29701 and add pass_ipa_fn_summary.
29702 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29704 (make_pass_inline_parameters): Remove.
29706 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29708 * omp-low.c (struct omp_context): Remove "default_kind" member.
29711 * omp-offload.c (execute_oacc_device_lower): Remove the
29712 parallelism dimensions function attributes for unparallelized
29713 OpenACC kernels constructs.
29715 2017-05-23 Martin Liska <mliska@suse.cz>
29717 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29719 (cgraph_edge::make_speculative): Likewise.
29720 (cgraph_edge::resolve_speculation): Likewise.
29721 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29722 (cgraph_node::dump): Likewise.
29723 * cgraph.h: Likewise.
29724 * cgraphunit.c (analyze_functions): Likewise.
29725 (symbol_table::compile): Likewise.
29726 * ipa-cp.c (print_all_lattices): Likewise.
29727 (determine_versionability): Likewise.
29728 (initialize_node_lattices): Likewise.
29729 (ipcp_verify_propagated_values): Likewise.
29730 (estimate_local_effects): Likewise.
29731 (update_profiling_info): Likewise.
29732 (create_specialized_node): Likewise.
29733 (perhaps_add_new_callers): Likewise.
29734 (decide_about_value): Likewise.
29735 (decide_whether_version_node): Likewise.
29736 (identify_dead_nodes): Likewise.
29737 (ipcp_store_bits_results): Likewise.
29738 * ipa-devirt.c (dump_targets): Likewise.
29739 (ipa_devirt): Likewise.
29740 * ipa-icf.c (sem_item::dump): Likewise.
29741 (sem_function::equals): Likewise.
29742 (sem_variable::equals): Likewise.
29743 (sem_item_optimizer::read_section): Likewise.
29744 (sem_item_optimizer::execute): Likewise.
29745 (congruence_class::dump): Likewise.
29746 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29747 (dump_inline_summary): Likewise.
29748 (estimate_node_size_and_time): Likewise.
29749 (inline_analyze_function): Likewise.
29750 * ipa-inline-transform.c (inline_call): Likewise.
29751 * ipa-inline.c (report_inline_failed_reason): Likewise.
29752 (want_early_inline_function_p): Likewise.
29753 (edge_badness): Likewise.
29754 (update_edge_key): Likewise.
29755 (inline_small_functions): Likewise.
29756 * ipa-profile.c (ipa_profile): Likewise.
29757 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29758 (ipa_make_edge_direct_to_target): Likewise.
29759 (remove_described_reference): Likewise.
29760 (ipa_impossible_devirt_target): Likewise.
29761 (propagate_controlled_uses): Likewise.
29762 (ipa_print_node_params): Likewise.
29763 (ipcp_transform_function): Likewise.
29764 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29765 (propagate_pure_const): Likewise.
29766 * ipa-reference.c (generate_summary): Likewise.
29767 (read_write_all_from_decl): Likewise.
29768 (propagate): Likewise.
29769 (ipa_reference_read_optimization_summary): Likewise.
29770 * ipa-utils.c (ipa_merge_profiles): Likewise.
29771 * ipa.c (walk_polymorphic_call_targets): Likewise.
29772 (symbol_table::remove_unreachable_nodes): Likewise.
29773 (ipa_single_use): Likewise.
29774 * passes.c (execute_todo): Likewise.
29775 * predict.c (drop_profile): Likewise.
29776 * symtab.c (symtab_node::get_dump_name): New function.
29777 (symtab_node::dump_name): Likewise.
29778 (symtab_node::dump_asm_name): Likewise.
29779 (symtab_node::dump_references): Likewise.
29780 (symtab_node::dump_referring): Likewise.
29781 (symtab_node::dump_base): Likewise.
29782 (symtab_node::debug_symtab): Likewise.
29783 * tree-sra.c (convert_callers_for_node): Likewise.
29784 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29785 * value-prof.c (init_node_map): Likewise.
29787 2017-05-23 Martin Liska <mliska@suse.cz>
29789 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29790 and symtab_node::debug_symtab to symbol_table::debug.
29791 * cgraphunit.c (analyze_functions): Use the renamed function.
29792 (symbol_table::compile): Likewise.
29793 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29794 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29795 * passes.c (execute_todo): Likewise.
29796 * symtab.c (symbol_table::dump): New function.
29797 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29799 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29801 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29802 that nonconst implies exec.
29804 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29806 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29807 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29808 (inline_edge_summary_vec): Turn into ...
29809 (ipa_call_summaries): ... this one.
29810 (redirect_to_unreachable, edge_set_predicate,
29811 evaluate_properties_for_edge, inline_summary_alloc,
29812 reset_ipa_call_summary, reset_inline_summary,
29813 inline_summary_t::duplicate): Update.
29814 (inline_edge_duplication_hook): Turn to ...
29815 (ipa_call_summary_t::duplicate): ... this one.
29816 (inline_edge_removal_hook): Turn to ...
29817 (ipa_call_summary_t::remove): ... this one.
29818 (dump_inline_edge_summary): Turn to ...
29819 (dump_ipa_call_summary): ... this one.
29820 (estimate_function_body_sizes): Update.
29821 (inline_update_callee_summaries): Update.
29822 (remap_edge_change_prob): Update.
29823 (remap_edge_summaries): Update.
29824 (inline_merge_summary): Update.
29825 (do_estimate_edge_time): Update.
29826 (inline_generate_summary): Update.
29827 (inline_read_section): Update.
29828 (inline_read_summary): Update.
29829 (inline_free_summary): Update.
29830 * ipa-inline.c (can_inline_edge_p): Update.
29831 (compute_inlined_call_time): Update.
29832 (want_inline_small_function_p): Update.
29833 (edge_badness): Update.
29834 (early_inliner): Update.
29835 * ipa-inline.h (inline_edge_summary): Turn to ...
29836 (ipa_call_summary): ... this one.
29837 (ipa_call_summary_t): New class.
29838 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29839 (ipa_call_summaries): New.
29840 (inline_edge_summary): Remove.
29841 (estimate_edge_growth): Update.
29842 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29843 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29844 * ipa-split.c (execute_split_functions): Update.
29845 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29847 2017-05-23 Tom de Vries <tom@codesourcery.com>
29849 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29850 attributes): Document rdrand effective target.
29852 2017-05-23 Tom de Vries <tom@codesourcery.com>
29854 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29855 attributes): Sort alphabetically.
29857 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29859 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29861 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29864 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
29865 V2DF/V2DI splat into two separate patterns, one that handles
29866 registers, and the other that only handles memory. Drop support
29867 for splatting from a GPR on ISA 2.07 and then splitting the
29868 splat into direct move and splat.
29869 (vsx_splat_<mode>_reg): Likewise.
29870 (vsx_splat_<mode>_mem): Likewise.
29872 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
29874 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
29876 2017-05-22 Jakub Jelinek <jakub@redhat.com>
29878 PR middle-end/80809
29879 * omp-low.c (finish_taskreg_remap): New function.
29880 (finish_taskreg_scan): If unit size of ctx->record_type
29881 is non-constant, unshare the size expression and replace
29882 decls in it with possible outer var refs.
29884 PR middle-end/80809
29885 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
29886 GOVD_SHARED rather than GOVD_PRIVATE with it.
29887 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
29888 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
29890 PR middle-end/80853
29891 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
29892 as last argument to build_outer_var_ref for pointer bases of array
29893 section reductions.
29895 2017-05-19 Martin Sebor <msebor@redhat.com>
29897 * print-tree.c (print_node): Print DECL_READ_P flag.
29899 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29901 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
29902 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
29903 * cgraph.c: Likewise.
29904 * cgraphunit.c: Likewise.
29905 * gengtype.c: Likewise.
29906 * ipa-cp.c: Likewise.
29907 * ipa-devirt.c: Likewise.
29908 * ipa-icf.c: Likewise.
29909 * ipa-predicate.c: Likewise.
29910 * ipa-profile.c: Likewise.
29911 * ipa-prop.c: Likewise.
29912 * ipa-split.c: Likewise.
29914 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
29915 edge_predicate_pool, dump_inline_hints,
29916 inline_summary::account_size_time, redirect_to_unreachable,
29917 edge_set_predicate, set_hint_predicate,
29918 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
29919 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
29920 inline_summary_t::remove, remap_hint_predicate_after_duplication,
29921 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
29922 ipa_call_summary_t::remove, initialize_growth_caches,
29923 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
29924 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
29925 mark_modified, unmodified_parm_1, unmodified_parm,
29926 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
29927 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
29928 compute_bb_predicates, will_be_nonconstant_expr_predicate,
29929 will_be_nonconstant_predicate, record_modified_bb_info,
29930 get_minimal_bb, record_modified, param_change_prob,
29931 phi_result_unknown_predicate, predicate_for_phi_result,
29932 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
29933 estimate_function_body_sizes, compute_inline_parameters,
29934 compute_inline_parameters_for_curren, pass_data_inline_parameters,
29935 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
29936 inline_update_callee_summaries, remap_edge_change_prob,
29937 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
29938 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
29939 inline_analyze_function, inline_summary_t::insert,
29940 inline_generate_summary, read_ipa_call_summary, inline_read_section,
29941 inline_read_summary, write_ipa_call_summary, inline_write_summary,
29942 inline_free_summary): Move to ipa-fnsummary.h
29943 (predicate_t): Remove.
29944 * ipa-fnsummary.c: New file.
29945 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
29946 (enum inline_hints_vals, inline_hints, agg_position_info,
29947 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
29948 inline_summaries, ipa_call_summary, ipa_call_summary_t,
29949 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
29950 dump_inline_summary, dump_inline_hints, inline_generate_summary,
29951 inline_read_summary, inline_write_summary, inline_free_summary,
29952 inline_analyze_function, initialize_inline_failed,
29953 inline_merge_summary, inline_update_overall_summary,
29954 compute_inline_parameters): Move to ipa-fnsummary.h
29955 * ipa-fnsummary.h: New file.
29956 * ipa-inline-transform.h: Include ipa-inline.h.
29957 * ipa-inline.c: LIkewise.
29959 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29961 * ipa-inline.c (edge_badness): Use inlined_time instead of
29962 inline_summaries->get.
29964 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29966 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
29968 2017-05-22 Nathan Sidwell <nathan@acm.org>
29970 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
29971 (fdump-lang): Document 'raw' option.
29972 * dumpfile.h (TDI_tu): Delete.
29973 * dumpfile.c (dump_files): Remove translation-unit.
29974 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
29976 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
29978 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
29979 command option from $(AWK) call.
29980 * config/avr/genmultilib.awk: Simplify and rewrite so that it
29981 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
29982 [FORMAT]: Remove handling of variable.
29983 * config/avr/t-multilib: Regenerate.
29985 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29987 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
29989 (dump_inline_summary): Do not print self_time.
29990 (estimate_function_body_sizes): Do not set self_time.
29991 (compute_inline_parameters): Likewise.
29992 (inline_read_section, inline_write_summary): Do not stream self_time.
29993 * ipa-inline.h (inline_summary): Drop self_time.
29995 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29997 * ipa-inline-analysis.c (account_size_time): Rename to ...
29998 (inline_summary::account_size_time): ... this one.
29999 (reset_ipa_call_summary): Turn to ...
30000 (ipa_call_summary::reset): ... this one.
30001 (reset_inline_summary): Turn to ...
30002 (inline_summary::reset): ... this one.
30003 (inline_summary_t::remove): Update.
30004 (inline_summary_t::duplicate): Update.
30005 (ipa_call_summary_t::remove): Update.
30006 (dump_inline_summary): Update.
30007 (estimate_function_body_sizes): Update.
30008 (compute_inline_parameters): Update.
30009 (estimate_node_size_and_time): Update.
30010 (inline_merge_summary): Update.
30011 (inline_update_overall_summary): Update.
30012 (inline_read_section): Update.
30013 (inline_write_summary): Update.
30014 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30015 add account_size_time and reset member functions.
30016 (ipa_call_summary): Add reset function.
30017 * ipa-predicate.h (predicate::operator &): Constify.
30019 2017-05-22 Richard Biener <rguenther@suse.de>
30021 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30023 2017-05-19 Jason Merrill <jason@redhat.com>
30025 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30027 2017-05-19 Marek Polacek <polacek@redhat.com>
30030 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30031 TYPE_OVERFLOW_WRAPS checks.
30033 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30035 * tree-core.h (enum omp_clause_default_kind): Add
30036 "OMP_CLAUSE_DEFAULT_PRESENT".
30037 * tree-pretty-print.c (dump_omp_clause): Handle it.
30038 * gimplify.c (enum gimplify_omp_var_data): Add
30039 "GOVD_MAP_FORCE_PRESENT".
30040 (gimplify_adjust_omp_clauses_1): Map it to
30041 "GOMP_MAP_FORCE_PRESENT".
30042 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30044 * gimplify.c (oacc_default_clause): Clarify.
30046 2017-05-19 Nathan Sidwell <nathan@acm.org>
30048 LANG_HOOK_REGISTER_DUMPS
30049 * toplev.c (general_init): Call register dump lang hook.
30050 * doc/invoke.texi: Document -fdump-lang option family.
30051 * dumpfile.c (dump_files): Remove class dump here.
30052 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30053 * dumpfile.h (tree_dump_index): Remove TDI_class.
30054 * langhooks-def.h (lhd_register_dumps): Declare.
30055 (LANG_HOOKS_REGISTER_DUMPS): Define.
30056 (LANG_HOOKS_INITIALIZER): Add it.
30057 * langhooks.c (lhd_register_dumps): Define.
30058 * langhooks.h (struct lang_hooks): Add register_dumps.
30060 2017-05-19 Nathan Sidwell <nathan@acm.org>
30062 * context.h (context::set_passes): New.
30063 * context.c (context::context): Do not create pass manager.
30064 * toplev.c (general_init): Create pass manager here.
30066 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30068 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30069 use this splitter if two add or or instructions would also work for
30070 the constant we want to generate.
30072 2017-05-19 Richard Biener <rguenther@suse.de>
30075 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30076 predicate evaluation.
30078 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30080 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30082 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30085 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30087 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30088 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30089 (inline_edge_summary_vec): Turn into ...
30090 (ipa_call_summaries): ... this one.
30091 (redirect_to_unreachable, edge_set_predicate,
30092 evaluate_properties_for_edge, inline_summary_alloc,
30093 reset_ipa_call_summary, reset_inline_summary,
30094 inline_summary_t::duplicate): Update.
30095 (inline_edge_duplication_hook): Turn to ...
30096 (ipa_call_summary_t::duplicate): ... this one.
30097 (inline_edge_removal_hook): Turn to ...
30098 (ipa_call_summary_t::remove): ... this one.
30099 (dump_inline_edge_summary): Turn to ...
30100 (dump_ipa_call_summary): ... this one.
30101 (estimate_function_body_sizes): Update.
30102 (inline_update_callee_summaries): Update.
30103 (remap_edge_change_prob): Update.
30104 (remap_edge_summaries): Update.
30105 (inline_merge_summary): Update.
30106 (do_estimate_edge_time): Update.
30107 (inline_generate_summary): Update.
30108 (inline_read_section): Update.
30109 (inline_read_summary): Update.
30110 (inline_free_summary): Update.
30111 * ipa-inline.c (can_inline_edge_p): Update.
30112 (compute_inlined_call_time): Update.
30113 (want_inline_small_function_p): Update.
30114 (edge_badness): Update.
30115 (early_inliner): Update.
30116 * ipa-inline.h (inline_edge_summary): Turn to ...
30117 (ipa_call_summary): ... this one.
30118 (ipa_call_summary_t): New class.
30119 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30120 (ipa_call_summaries): New.
30121 (inline_edge_summary): Remove.
30122 (estimate_edge_growth): Update.
30123 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30124 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30125 * ipa-split.c (execute_split_functions): Update.
30126 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30128 2017-05-19 Richard Biener <rguenther@suse.de>
30130 PR middle-end/80764
30131 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
30133 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
30135 * config/rs6000/rs6000.c (struct machine_function): Add field
30136 fpr_is_wrapped_separately.
30137 (rs6000_get_separate_components): Use 64 components. Handle the
30138 new FPR components.
30139 (rs6000_components_for_bb): Handle the FPR components.
30140 (rs6000_emit_prologue_components): Handle the FPR components.
30141 (rs6000_emit_epilogue_components): Handle the FPR components.
30142 (rs6000_set_handled_components): Handle the FPR components.
30143 (rs6000_emit_prologue): Don't output prologue code for those FPRs
30144 that are already separately shrink-wrapped.
30145 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
30146 that are already separately shrink-wrapped.
30148 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30151 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
30154 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
30155 (define_peephole2 for Altivec d-form load): Add peepholes to catch
30156 cases where the register allocator uses a move and an offsettable
30157 memory operation to/from a FPR register on ISA 2.06/2.07.
30158 (define_peephole2 for Altivec d-form store): Likewise.
30160 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
30163 * config/i386/mmx.md (*mov<mode>_internal): Enable
30164 alternatives 11, 12, 13 and 14 also for 32bit targets.
30165 Remove alternatives 15, 16, 17 and 18.
30166 * config/i386/sse.md (vec_concatv2di): Change
30167 alternative (!x, *y) to (x, ?!*Yn).
30169 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
30171 * dumpfile.h (enum dump_kind): Remove stray comma.
30173 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30175 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
30176 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
30177 predicate::num_conditions
30178 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
30179 (CHANGED): turn into predicate::changed.
30180 (agg_position_info): Move to ipa-predicate.h
30181 (add_condition, predicate::add_clause, predicate::operator &=,
30182 predicate::or_with, predicate::evaluate, predicate::probability,
30183 dump_condition, dump_clause, predicate::dump,
30184 predicate::remap_after_duplication, predicate::remap_after_inlining,
30185 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
30186 (evaluate_conditions_for_known_args): Update.
30187 (set_cond_stmt_execution_predicate): Update.
30188 * ipa-inline.h: Include ipa-predicate.h
30189 (condition, inline_param_summary, conditions, agg_position_info,
30190 predicate): Move to ipa-predicate.h
30191 * ipa-predicate.c: New file.
30192 * ipa-predicate.h: New file.
30194 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
30196 * final.c (leaf_function_p): Check we are not in a sequence.
30198 2017-05-18 Martin Liska <mliska@suse.cz>
30200 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
30201 * dumpfile.c (dump_register): Use new enum dump_kind.
30202 (get_dump_file_name): Likewise.
30203 (dump_enable_all): Likewise.
30204 (dump_switch_p_1): Likewise.
30205 (enable_rtl_dump_file): Remove usage of TDF_RTL.
30206 * dumpfile.h (enum dump_kind): New enum type.
30207 (struct dump_file_info): Create constructor and
30208 format fields and comments.
30209 * passes.c (pass_manager::register_one_dump_file):
30211 * statistics.c (statistics_early_init): Likewise.
30212 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
30213 TDF_TREE with TDF_SLIM.
30214 (gather_memory_references_ref): Likewise.
30216 2017-05-18 Martin Liska <mliska@suse.cz>
30218 * vec.h (struct vnull): Use it.
30220 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30222 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
30223 (true_predicate, false_predicate, true_predicate_p,
30224 false_predicate_p): Remove.
30225 (single_cond_predicate, not_inlined_predicate): Turn to member function
30227 (add_condition): Update.
30228 (add_clause): Turn to...
30229 (predicate::add_clause): ... this one; update; allow passing NULL
30231 (and_predicates): Turn to ...
30232 (predicate::operator &=): ... this one.
30233 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
30234 (or_predicates): Turn to ...
30235 (predicate::or_with): ... this one.
30236 (evaluate_predicate): Turn to ...
30237 (predicate::evaluate): ... this one.
30238 (predicate_probability): Turn to ...
30239 (predicate::probability): ... this one.
30240 (dump_condition): Update.
30241 (dump_predicate): Turn to ...
30242 (predicate::dump): ... this one.
30243 (account_size_time): Update.
30244 (edge_set_predicate): Update.
30245 (set_hint_predicate): UPdate.
30246 (evaluate_conditions_for_known_args): Update.
30247 (evaluate_properties_for_edge): Update.
30248 (remap_predicate_after_duplication): Turn to...
30249 (predicate::remap_after_duplication): ... this one.
30250 (remap_hint_predicate_after_duplication): Update.
30251 (inline_summary_t::duplicate): UPdate.
30252 (dump_inline_edge_summary): Update.
30253 (dump_inline_summary): Update.
30254 (set_cond_stmt_execution_predicate): Update.
30255 (set_switch_stmt_execution_predicate): Update.
30256 (compute_bb_predicates): Update.
30257 (will_be_nonconstant_expr_predicate): Update.
30258 (will_be_nonconstant_predicate): Update.
30259 (phi_result_unknown_predicate): Update.
30260 (predicate_for_phi_result): Update.
30261 (array_index_predicate): Update.
30262 (estimate_function_body_sizes): Update.
30263 (estimate_node_size_and_time): Update.
30264 (estimate_ipcp_clone_size_and_time): Update.
30265 (remap_predicate): Rename to ...
30266 (predicate::remap_after_inlining): ... this one.
30267 (remap_hint_predicate): Update.
30268 (inline_merge_summary): Update.
30269 (inline_update_overall_summary): Update.
30270 (estimate_size_after_inlining): Update.
30271 (read_predicate): Rename to ...
30272 (predicate::stream_in): ... this one.
30273 (read_inline_edge_summary): Update.
30274 (write_predicate): Rename to ...
30275 (predicate::stream_out): ... this one.
30276 (write_inline_edge_summary): Update.
30277 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
30278 (clause_t): Turn to uint32_t
30279 (predicate): Turn to class; implement constructor and operators
30281 (size_time_entry): Update.
30282 (inline_summary): Update.
30283 (inline_edge_summary): Update.
30285 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
30287 * fold-const.c (fold_binary_loc): Move transformation...
30288 * match.pd (C - X CMP X): ... here.
30290 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
30292 * config/sparc/sparc.c (sparc_option_override): Set function
30293 alignment for -mcpu=niagara7 to 64 to match the I$ line.
30294 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
30296 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
30298 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
30300 2017-05-18 Marek Polacek <polacek@redhat.com>
30303 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
30304 (pass_ubsan::execute): Call gimple_assign_single_p instead of
30305 gimple_assign_load_p.
30307 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
30309 PR middle-end/80692
30310 * real.c (do_compare): Give decimal_do_compare preference over
30311 comparing just the signs.
30313 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
30315 * doc/md.texi (Canonicalization of Instructions): Describe the
30316 canonical form of instructions that inherently set a condition
30319 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
30321 PR middle-end/80775
30322 * tree-cfg.c: Move deletion of unreachable case statements to after
30323 the merging of consecutive case labels.
30325 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
30327 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
30328 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
30329 restoring of callee-saved registers.
30331 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
30333 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
30334 * config/visium/visium.c (single_set_and_flags): Likewise.
30335 * config/visium/visium.md (Substitutions): Likewise.
30337 2017-05-17 Martin Liska <mliska@suse.cz>
30339 * cfg.c: Introduce dump_flags_t type and
30340 use it instead of int type.
30342 * cfghooks.c: Likewise.
30343 * cfghooks.h (struct cfg_hooks): Likewise.
30344 * cfgrtl.c: Likewise.
30345 * cfgrtl.h: Likewise.
30346 * cgraph.c (cgraph_node::get_body): Likewise.
30347 * coretypes.h: Likewise.
30348 * domwalk.c: Likewise.
30349 * domwalk.h: Likewise.
30350 * dumpfile.c (struct dump_option_value_info): Likewise.
30351 (dump_enable_all): Likewise.
30352 (dump_switch_p_1): Likewise.
30353 (opt_info_switch_p): Likewise.
30354 * dumpfile.h (enum tree_dump_index): Likewise.
30355 (struct dump_file_info): Likewise.
30356 * genemit.c: Likewise.
30357 * generic-match-head.c: Likewise.
30358 * gengtype.c (open_base_files): Likewise.
30359 * gimple-pretty-print.c: Likewise.
30360 * gimple-pretty-print.h: Likewise.
30361 * graph.c (print_graph_cfg): Likewise.
30362 * graphite-scop-detection.c (dot_all_sese): Likewise.
30363 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
30364 * loop-unroll.c (report_unroll): Likewise.
30365 * passes.c (pass_manager::register_one_dump_file): Likewise.
30366 * print-tree.c: Likewise.
30367 * statistics.c: Likewise.
30368 * tree-cfg.c: Likewise.
30369 * tree-cfg.h: Likewise.
30370 * tree-dfa.c: Likewise.
30371 * tree-dfa.h: Likewise.
30372 * tree-dump.c (dump_function): Likewise.
30373 * tree-dump.h (struct dump_info): Likewise.
30374 * tree-pretty-print.c: Likewise.
30375 * tree-pretty-print.h: Likewise.
30376 * tree-ssa-live.c: Likewise.
30377 * tree-ssa-live.h: Likewise.
30378 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
30379 * tree-vect-loop.c: Likewise.
30380 * tree-vect-slp.c: Likewise.
30382 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
30383 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30385 PR tree-optimization/80457
30386 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
30387 of all arguments to a statement as scalar_to_vec operations.
30388 (vectorizable_call): Adjust call to vect_model_simple_cost for
30390 (vectorizable_conversion): Likewise.
30391 (vectorizable_assignment): Likewise.
30392 (vectorizable_shift): Likewise.
30393 (vectorizable_operation): Likewise.
30394 (vectorizable_comparison): Likewise.
30395 (vect_is_simple_cond): Record the def types for operands.
30396 (vectorizable_condition): Likewise, call vect_model_simple_cost.
30397 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
30398 for statement argument count.
30400 2017-05-16 Carl Love <cel@us.ibm.com>
30402 * config/rs6000/rs6000-c: Add support for built-in functions
30403 vector unsigned long long vec_bperm (vector unsigned long long,
30404 vector unsigned char)
30405 vector signed long long vec_mule (vector signed int,
30407 vector unsigned long long vec_mule (vector unsigned int,
30408 vector unsigned int)
30409 vector signed long long vec_mulo (vector signed int,
30411 vector unsigned long long vec_mulo (vector unsigned int,
30412 vector unsigned int)
30413 vector signed char vec_sldw (vector signed char,
30414 vector signed char,
30416 vector unsigned char vec_sldw (vector unsigned char,
30417 vector unsigned char,
30419 vector signed short vec_sldw (vector signed short,
30420 vector signed short,
30422 vector unsigned short vec_sldw (vector unsigned short,
30423 vector unsigned short,
30425 vector signed int vec_sldw (vector signed int,
30428 vector unsigned int vec_sldw (vector unsigned int,
30429 vector unsigned int,
30431 vector signed long long vec_sldw (vector signed long long,
30432 vector signed long long,
30434 vector unsigned long long vec_sldw (vector unsigned long long,
30435 vector unsigned long long,
30437 * config/rs6000/rs6000-c: Add support for built-in functions
30438 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
30439 * config/rs6000/altivec.h: Add defintion for vec_sldw.
30440 * doc/extend.texi: Update the built-in documentation for the
30441 new built-in functions.
30443 2017-05-16 Marek Polacek <polacek@redhat.com>
30447 * tree.c (save_expr): Don't fold the expression.
30449 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
30451 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
30452 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
30453 and (?*y,m). Update insn attributes.
30455 2017-05-16 Martin Liska <mliska@suse.cz>
30457 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
30458 flags argument of print_gimple_stmt, print_gimple_expr,
30459 print_generic_stmt and print_generic_expr.
30460 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
30461 * coretypes.h: Likewise.
30462 * except.c (dump_eh_tree): Likewise.
30463 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
30464 * gimple-pretty-print.h: Likewise.
30465 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
30466 (backprop::push_to_worklist): Likewise.
30467 (backprop::pop_from_worklist): Likewise.
30468 (backprop::process_use): Likewise.
30469 (backprop::intersect_uses): Likewise.
30470 (note_replacement): Likewise.
30471 * gimple-ssa-store-merging.c
30472 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
30473 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
30474 (pass_store_merging::execute): Likewise.
30475 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
30476 (ssa_base_cand_dump_callback): Likewise.
30477 (dump_incr_vec): Likewise.
30478 (replace_refs): Likewise.
30479 (replace_mult_candidate): Likewise.
30480 (create_add_on_incoming_edge): Likewise.
30481 (create_phi_basis): Likewise.
30482 (insert_initializers): Likewise.
30483 (all_phi_incrs_profitable): Likewise.
30484 (introduce_cast_before_cand): Likewise.
30485 (replace_one_candidate): Likewise.
30486 * gimplify.c (gimplify_expr): Likewise.
30487 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
30488 (set_rename): Likewise.
30489 (rename_uses): Likewise.
30490 (copy_loop_phi_nodes): Likewise.
30491 (add_close_phis_to_merge_points): Likewise.
30492 (copy_loop_close_phi_args): Likewise.
30493 (copy_cond_phi_args): Likewise.
30494 (graphite_copy_stmts_from_block): Likewise.
30495 (translate_pending_phi_nodes): Likewise.
30496 * graphite-poly.c (print_pdr): Likewise.
30497 (dump_gbb_cases): Likewise.
30498 (dump_gbb_conditions): Likewise.
30499 (print_scop_params): Likewise.
30500 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
30501 (build_cross_bb_scalars_use): Likewise.
30502 (gather_bbs::before_dom_children): Likewise.
30503 * hsa-dump.c (dump_hsa_immed): Likewise.
30504 * ipa-cp.c (print_ipcp_constant_value): Likewise.
30505 (get_replacement_map): Likewise.
30506 * ipa-inline-analysis.c (dump_condition): Likewise.
30507 (estimate_function_body_sizes): Likewise.
30508 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
30509 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
30510 * ipa-prop.c (ipa_dump_param): Likewise.
30511 (ipa_print_node_jump_functions_for_edge): Likewise.
30512 (ipa_modify_call_arguments): Likewise.
30513 (ipa_modify_expr): Likewise.
30514 (ipa_dump_param_adjustments): Likewise.
30515 (ipa_dump_agg_replacement_values): Likewise.
30516 (ipcp_modif_dom_walker::before_dom_children): Likewise.
30517 * ipa-pure-const.c (check_stmt): Likewise.
30518 (pass_nothrow::execute): Likewise.
30519 * ipa-split.c (execute_split_functions): Likewise.
30520 * omp-offload.c (dump_oacc_loop_part): Likewise.
30521 (dump_oacc_loop): Likewise.
30522 * trans-mem.c (tm_log_emit): Likewise.
30523 (tm_memopt_accumulate_memops): Likewise.
30524 (dump_tm_memopt_set): Likewise.
30525 (dump_tm_memopt_transform): Likewise.
30526 * tree-cfg.c (gimple_verify_flow_info): Likewise.
30527 (print_loop): Likewise.
30528 * tree-chkp-opt.c (chkp_print_addr): Likewise.
30529 (chkp_gather_checks_info): Likewise.
30530 (chkp_get_check_result): Likewise.
30531 (chkp_remove_check_if_pass): Likewise.
30532 (chkp_use_outer_bounds_if_possible): Likewise.
30533 (chkp_reduce_bounds_lifetime): Likewise.
30534 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
30535 (chkp_mark_completed_bounds): Likewise.
30536 (chkp_register_incomplete_bounds): Likewise.
30537 (chkp_mark_invalid_bounds): Likewise.
30538 (chkp_maybe_copy_and_register_bounds): Likewise.
30539 (chkp_build_returned_bound): Likewise.
30540 (chkp_get_bound_for_parm): Likewise.
30541 (chkp_build_bndldx): Likewise.
30542 (chkp_get_bounds_by_definition): Likewise.
30543 (chkp_generate_extern_var_bounds): Likewise.
30544 (chkp_get_bounds_for_decl_addr): Likewise.
30545 * tree-chrec.c (chrec_apply): Likewise.
30546 * tree-data-ref.c (dump_data_reference): Likewise.
30547 (dump_subscript): Likewise.
30548 (dump_data_dependence_relation): Likewise.
30549 (analyze_overlapping_iterations): Likewise.
30550 * tree-inline.c (expand_call_inline): Likewise.
30551 (tree_function_versioning): Likewise.
30552 * tree-into-ssa.c (dump_defs_stack): Likewise.
30553 (dump_currdefs): Likewise.
30554 (dump_names_replaced_by): Likewise.
30555 (dump_update_ssa): Likewise.
30556 (update_ssa): Likewise.
30557 * tree-object-size.c (pass_object_sizes::execute): Likewise.
30558 * tree-parloops.c (build_new_reduction): Likewise.
30559 (try_create_reduction_list): Likewise.
30560 (ref_conflicts_with_region): Likewise.
30561 (oacc_entry_exit_ok_1): Likewise.
30562 (oacc_entry_exit_single_gang): Likewise.
30563 * tree-pretty-print.h: Likewise.
30564 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
30565 (get_scalar_evolution): Likewise.
30566 (add_to_evolution): Likewise.
30567 (get_loop_exit_condition): Likewise.
30568 (analyze_evolution_in_loop): Likewise.
30569 (analyze_initial_condition): Likewise.
30570 (analyze_scalar_evolution): Likewise.
30571 (instantiate_scev): Likewise.
30572 (number_of_latch_executions): Likewise.
30573 (gather_chrec_stats): Likewise.
30574 (final_value_replacement_loop): Likewise.
30575 (scev_const_prop): Likewise.
30576 * tree-sra.c (dump_access): Likewise.
30577 (disqualify_candidate): Likewise.
30578 (create_access): Likewise.
30579 (reject): Likewise.
30580 (maybe_add_sra_candidate): Likewise.
30581 (create_access_replacement): Likewise.
30582 (analyze_access_subtree): Likewise.
30583 (analyze_all_variable_accesses): Likewise.
30584 (sra_modify_assign): Likewise.
30585 (initialize_constant_pool_replacements): Likewise.
30586 (find_param_candidates): Likewise.
30587 (decide_one_param_reduction): Likewise.
30588 (replace_removed_params_ssa_names): Likewise.
30589 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
30590 * tree-ssa-copy.c (dump_copy_of): Likewise.
30591 (copy_prop_visit_cond_stmt): Likewise.
30592 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
30593 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
30594 (record_equivalences_from_stmt): Likewise.
30595 * tree-ssa-dse.c (compute_trims): Likewise.
30596 (delete_dead_call): Likewise.
30597 (delete_dead_assignment): Likewise.
30598 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
30599 (forward_propagate_into_cond): Likewise.
30600 (pass_forwprop::execute): Likewise.
30601 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
30602 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
30604 (move_computations_worker): Likewise.
30605 (execute_sm): Likewise.
30606 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
30607 (remove_exits_and_undefined_stmts): Likewise.
30608 (remove_redundant_iv_tests): Likewise.
30609 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
30610 (adjust_iv_update_pos): Likewise.
30611 * tree-ssa-math-opts.c (bswap_replace): Likewise.
30612 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
30613 (value_replacement): Likewise.
30614 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
30615 * tree-ssa-pre.c (print_pre_expr): Likewise.
30616 (get_representative_for): Likewise.
30617 (create_expression_by_pieces): Likewise.
30618 (insert_into_preds_of_block): Likewise.
30619 (eliminate_insert): Likewise.
30620 (eliminate_dom_walker::before_dom_children): Likewise.
30621 (eliminate): Likewise.
30622 (remove_dead_inserted_code): Likewise.
30623 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30624 * tree-ssa-reassoc.c (get_rank): Likewise.
30625 (eliminate_duplicate_pair): Likewise.
30626 (eliminate_plus_minus_pair): Likewise.
30627 (eliminate_not_pairs): Likewise.
30628 (undistribute_ops_list): Likewise.
30629 (eliminate_redundant_comparison): Likewise.
30630 (update_range_test): Likewise.
30631 (optimize_range_tests_var_bound): Likewise.
30632 (optimize_vec_cond_expr): Likewise.
30633 (rewrite_expr_tree): Likewise.
30634 (rewrite_expr_tree_parallel): Likewise.
30635 (linearize_expr): Likewise.
30636 (break_up_subtract): Likewise.
30637 (linearize_expr_tree): Likewise.
30638 (attempt_builtin_powi): Likewise.
30639 (attempt_builtin_copysign): Likewise.
30640 (transform_stmt_to_copy): Likewise.
30641 (transform_stmt_to_multiply): Likewise.
30642 (dump_ops_vector): Likewise.
30643 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30644 (print_scc): Likewise.
30645 (set_ssa_val_to): Likewise.
30646 (visit_reference_op_store): Likewise.
30647 (visit_use): Likewise.
30648 (sccvn_dom_walker::before_dom_children): Likewise.
30649 (run_scc_vn): Likewise.
30650 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30652 (expr_hash_elt::print): Likewise.
30653 (const_and_copies::pop_to_marker): Likewise.
30654 (const_and_copies::record_const_or_copy_raw): Likewise.
30655 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30656 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30657 (dump_predicates): Likewise.
30658 (find_uninit_use): Likewise.
30659 (warn_uninitialized_phi): Likewise.
30660 (pass_late_warn_uninitialized::execute): Likewise.
30661 * tree-ssa.c (verify_vssa): Likewise.
30662 (verify_ssa): Likewise.
30663 (maybe_optimize_var): Likewise.
30664 * tree-vrp.c (dump_value_range): Likewise.
30665 (dump_all_value_ranges): Likewise.
30666 (dump_asserts_for): Likewise.
30667 (register_edge_assert_for_2): Likewise.
30668 (vrp_visit_cond_stmt): Likewise.
30669 (vrp_visit_switch_stmt): Likewise.
30670 (vrp_visit_stmt): Likewise.
30671 (vrp_visit_phi_node): Likewise.
30672 (simplify_cond_using_ranges_1): Likewise.
30673 (fold_predicate_in): Likewise.
30674 (evrp_dom_walker::before_dom_children): Likewise.
30675 (evrp_dom_walker::push_value_range): Likewise.
30676 (evrp_dom_walker::pop_value_range): Likewise.
30677 (execute_early_vrp): Likewise.
30679 2017-05-16 Richard Biener <rguenther@suse.de>
30681 * dwarf2out.c (loc_list_from_tree_1): Do not create
30682 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30684 2017-05-16 Richard Biener <rguenther@suse.de>
30686 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30688 (note_variable_value_in_expr): If we resolved the decl ref
30689 do not push to the stack.
30691 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30693 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30694 operations in fast-math mode.
30695 (vaddq_f16): Likewise.
30696 (vmul_f16): Likewise.
30697 (vmulq_f16): Likewise.
30698 (vsub_f16): Likewise.
30699 (vsubq_f16): Likewise.
30700 * config/arm/neon.md (add<mode>3): New.
30702 (fma:<VH:mode>3): New. Also remove outdated comment.
30705 2017-05-16 Martin Liska <mliska@suse.cz>
30709 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30710 (odr_types_equivalent_p): Likewise.
30712 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30714 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30716 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30719 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30720 non-interunit SSE move alternatives with '?'.
30721 (zero-extendsidi peephole2): New peephole to skip intermediate
30722 general register in SSE zero-extend sequence.
30724 2017-05-15 Jeff Law <law@redhat.com>
30726 * reorg.c (relax_delay_slots): Create a new variable to hold
30727 the temporary target rather than clobbering TARGET_LABEL.
30729 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30730 missing argument to extract_bit_field call.
30731 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30733 2017-05-15 Martin Liska <mliska@suse.cz>
30736 * gcc.c (process_command): Do not allow empty argument of -o option.
30738 2017-05-15 Renlin Li <renlin.li@arm.com>
30740 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30741 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30742 * config/aarch64/constraints.md (Usf): Add long call check.
30743 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30744 (call_value): Likewise.
30745 (sibcall): Likewise.
30746 (sibcall_value): Likewise.
30748 (call_value_insn): New.
30749 (sibcall_insn): Update rtx pattern.
30750 (sibcall_value_insn): Likewise.
30751 (call_internal): Remove.
30752 (call_value_internal): Likewise.
30753 (sibcall_internal): Likewise.
30754 (sibcall_value_internal): Likewise.
30755 (call_reg): Likewise.
30756 (call_symbol): Likewise.
30757 (call_value_reg): Likewise.
30758 (call_value_symbol): Likewise.
30760 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30763 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30765 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30767 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30768 compatible with CCGOCmode and with CCZmode.
30770 2017-05-14 Martin Sebor <msebor@redhat.com>
30772 PR middle-end/77671
30773 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30774 (gimple_fold_builtin_snprintf): Same.
30775 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30776 (gimple_fold_builtin_snprintf): Same.
30777 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30778 of character types.
30779 (is_call_safe): New function.
30780 (try_substitute_return_value): Call it.
30781 (try_simplify_call): New function.
30782 (pass_sprintf_length::handle_gimple_call): Call it.
30784 2017-05-14 Martin Sebor <msebor@redhat.com>
30786 PR middle-end/80669
30787 * builtins.c (expand_builtin_stpncpy): Simplify.
30789 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30791 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30792 * config/i386/i386.h
30793 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30794 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30795 (struct machine_function): Add new members call_ms2sysv,
30796 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30797 (struct machine_frame_state): New fields sp_realigned and
30798 sp_realigned_offset.
30799 * config/i386/i386.c
30800 (enum xlogue_stub): New enum.
30801 (enum xlogue_stub_sets): New enum.
30802 (class xlogue_layout): New class.
30803 (struct ix86_frame): New fields stack_realign_allocate_offset,
30804 stack_realign_offset and outlined_save_offset. Modify comments to
30805 detail stack layout when using out-of-line stubs.
30806 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30807 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30808 -mcall-ms2sysv-xlogues.
30809 (stub_managed_regs): New static variable.
30810 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30811 registers managed by out-of-line stub.
30812 (disable_call_ms2sysv_xlogues): New function.
30813 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30814 m->call_ms2sysv when appropriate and compute frame layout for
30816 (sp_valid_at, fp_valid_at): New inline functions.
30817 (choose_basereg): New function.
30818 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30820 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30821 Use align parameter of choose_baseaddr to generated aligned SSE movs
30823 (pro_epilogue_adjust_stack): Modify to track
30824 machine_frame_state::sp_realigned.
30825 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30826 (ix86_nsaved_sseregs): Likewise.
30827 (ix86_emit_save_regs): Likewise.
30828 (ix86_emit_save_regs_using_mov): Likewise.
30829 (ix86_emit_save_sse_regs_using_mov): Likewise.
30830 (get_scratch_register_on_entry): Likewise.
30831 (gen_frame_set): New function.
30832 (gen_frame_load): Likewise.
30833 (gen_frame_store): Likewise.
30834 (emit_outlined_ms2sysv_save): Likewise.
30835 (emit_outlined_ms2sysv_restore): Likewise.
30836 (ix86_expand_prologue): Modify stack re-alignment code and call
30837 emit_outlined_ms2sysv_save when appropriate.
30838 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30839 parameter rtx_insn *insn, which allows the function to be used to only
30840 generate the notes.
30841 (ix86_expand_epilogue): Modify validity checks of frame and stack
30842 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30843 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30844 * config/i386/predicates.md
30845 (save_multiple): New predicate.
30846 (restore_multiple): Likewise.
30847 * config/i386/sse.md
30848 (save_multiple<mode>): New pattern.
30849 (save_multiple_realign<mode>): Likewise.
30850 (restore_multiple<mode>): Likewise.
30851 (restore_multiple_and_return<mode>): Likewise.
30852 (restore_multiple_leave_return<mode>): Likewise.
30853 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30855 2017-05-14 Julia Koval <julia.koval@intel.com>
30857 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30858 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30859 (__builtin_ia32_xsetbv): New builtins.
30860 * config/i386/i386.c (ix86_expand_special_args_builtin):
30862 (ix86_expand_builtin): Special expand for new intrinsics.
30863 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
30864 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
30865 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
30867 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30869 * cfganal.c (inverted_post_order_compute): Change argument type
30871 * cfganal.h (inverted_post_order_compute): Adjust prototype.
30872 * df-core.c (rest_of_handle_df_initialize): Adjust.
30873 (rest_of_handle_df_finish): Likewise.
30874 (df_analyze_1): Likewise.
30875 (df_analyze): Likewise.
30876 (loop_inverted_post_order_compute): Change argument to be a vec *.
30877 (df_analyze_loop): Adjust.
30878 (df_get_n_blocks): Likewise.
30879 (df_get_postorder): Likewise.
30880 * df.h (struct df_d): Change field to be a vec.
30881 * lcm.c (compute_laterin): Adjust.
30882 (compute_available): Likewise.
30883 * lra-lives.c (lra_create_live_ranges_1): Likewise.
30884 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
30885 * tree-ssa-pre.c (compute_antic): Likewise.
30887 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30889 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
30890 (depth_first_search::depth_first_search): Change structure init
30891 function to this constructor.
30892 (depth_first_search::add_bb): Rename function to this member.
30893 (depth_first_search::execute): Likewise.
30894 (flow_dfs_compute_reverse_finish): Adjust.
30896 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30898 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
30899 (longest_simple_path): Likewise.
30900 * shrink-wrap.c (spread_components): Likewise.
30901 (disqualify_problematic_components): Likewise.
30902 (emit_common_heads_for_components): Likewise.
30903 (emit_common_tails_for_components): Likewise.
30904 (insert_prologue_epilogue_for_components): Likewise.
30906 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30908 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
30911 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30913 * df-core.c (df_set_blocks): Start using auto_bitmap.
30914 (df_compact_blocks): Likewise.
30915 * df-problems.c (df_rd_confluence_n): Likewise.
30916 * df-scan.c (df_insn_rescan_all): Likewise.
30917 (df_process_deferred_rescans): Likewise.
30918 (df_update_entry_block_defs): Likewise.
30919 (df_update_exit_block_uses): Likewise.
30920 (df_entry_block_bitmap_verify): Likewise.
30921 (df_exit_block_bitmap_verify): Likewise.
30922 (df_scan_verify): Likewise.
30923 * lra-constraints.c (lra_constraints): Likewise.
30924 (undo_optional_reloads): Likewise.
30925 (lra_undo_inheritance): Likewise.
30926 * lra-remat.c (calculate_gen_cands): Likewise.
30927 (do_remat): Likewise.
30928 * lra-spills.c (assign_spill_hard_regs): Likewise.
30929 (spill_pseudos): Likewise.
30930 * tree-ssa-pre.c (bitmap_set_and): Likewise.
30931 (bitmap_set_subtract_values): Likewise.
30933 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30935 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
30936 management with auto_bitmap.
30937 (fix_inter_tick): Likewise.
30938 (fix_recovery_deps): Likewise.
30939 * ira.c (add_store_equivs): Likewise.
30940 (find_moveable_pseudos): Likewise.
30941 (split_live_ranges_for_shrink_wrap): Likewise.
30942 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
30943 (rtx_reuse_manager::seen_def_p): Likewise.
30944 (rtx_reuse_manager::set_seen_def): Likewise.
30945 * print-rtl.h (class rtx_reuse_manager): Likewise.
30947 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30949 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
30951 (migrate_btr_def): Likewise.
30952 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
30953 * df-core.c (loop_post_order_compute): Likewise.
30954 (loop_inverted_post_order_compute): Likewise.
30955 * hsa-common.h: Likewise.
30956 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
30957 * init-regs.c (initialize_uninitialized_regs): Likewise.
30958 * ipa-inline.c (resolve_noninline_speculation): Likewise.
30959 (inline_small_functions): Likewise.
30960 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
30961 * ira.c (combine_and_move_insns): Likewise.
30962 (build_insn_chain): Likewise.
30963 * loop-invariant.c (find_invariants): Likewise.
30964 * lower-subreg.c (propagate_pseudo_copies): Likewise.
30965 * predict.c (tree_predict_by_opcode): Likewise.
30966 (predict_paths_leading_to): Likewise.
30967 (predict_paths_leading_to_edge): Likewise.
30968 (estimate_loops_at_level): Likewise.
30969 (estimate_loops): Likewise.
30970 * shrink-wrap.c (try_shrink_wrapping): Likewise.
30971 (spread_components): Likewise.
30972 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
30973 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
30974 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
30975 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
30976 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
30977 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
30978 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
30979 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
30980 (mark_threaded_blocks): Likewise.
30981 (thread_through_all_blocks): Likewise.
30982 * tree-ssa.c (verify_ssa): Likewise.
30983 (execute_update_addresses_taken): Likewise.
30984 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
30986 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30988 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
30990 (post_order_compute): Likewise.
30991 (inverted_post_order_compute): Likewise.
30992 (pre_and_rev_post_order_compute_fn): Likewise.
30994 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30996 * genrecog.c (int_set::int_set): Explicitly construct our
30997 auto_vec base class.
30998 * vec.h (auto_vec::auto_vec): New constructor.
31000 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31002 * bitmap.h (class auto_bitmap): New constructor taking
31003 bitmap_obstack * argument.
31005 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31007 * bitmap.h (class auto_bitmap): Change type of m_bits to
31008 bitmap_head, and adjust ctor / dtor and member operators.
31010 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31012 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31013 when returned register mode doesn't match original mode.
31015 2017-05-12 Jeff Law <law@redhat.com>
31016 Jakub Jelinek <jakub@redhat.com>
31018 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31019 we look for cc setter after the compare-elim changes.
31020 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31021 within the vector to match what compare-elim now expects.
31022 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31023 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31025 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31026 after the compare-elim changes.
31027 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31028 the vector to match what compare-elim now expects.
31029 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31030 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31031 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31032 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31033 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31035 * config/visium/visium.c (single_set_and_flags): Fix where
31036 we look for cc setter after the compare-elim changes.
31037 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
31038 with the vector to match what compare-elim now expects.
31039 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
31040 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31041 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31042 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31043 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31044 (neg<mode>2_insn_set_overflow): Likewise.
31046 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31048 PR middle-end/79794
31049 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31050 maybe_expand_insn call, set ops[0].target. If still set after call,
31051 set alt_rtl. Add extra arg to recursive calls.
31052 (extract_bit_field): Add alt_rtl argument. Pass to
31054 * expmed.h (extract_bit_field): Fix prototype.
31055 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31056 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31057 to extract_bit_field_calls.
31058 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31059 Pass alt_rtl to extract_bit_field calls.
31060 * calls.c (store_unaligned_arguments_into_psuedos)
31061 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31062 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31064 * optabs.h (struct expand_operand): Add target bitfield.
31066 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31068 * compare-elim.c (try_eliminate_compare): Canonicalize
31069 operation with embedded compare to
31070 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31071 (set (reg) (operation)].
31073 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31075 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31078 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31079 cost of adding a carry flag for ADC instruction.
31080 [case MINUS]: Ignore the cost of subtracting a carry flag
31081 for SBB instruction.
31083 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31085 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31087 * config/rs6000/bmiintrin.h: New file.
31088 * config/rs6000/bmi2intrin.h: New file.
31089 * config/rs6000/x86intrin.h: New file.
31091 2017-05-12 Jeff Law <law@redhat.com>
31093 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31096 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31098 PR middle-end/80707
31099 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31101 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31103 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31104 early expansion of vector divide builtins.
31105 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
31106 builtins identified as having unsigned arguments.
31108 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31110 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
31111 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
31112 expansion of vector logical operations (and, andc, or, xor,
31115 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31117 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
31118 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
31120 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31122 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31123 early GIMPLE expansion of vector multiplies.
31125 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31127 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
31128 TARGET_HAVE_MOVT conditional.
31129 (movt splitter): Likewise.
31131 2017-05-12 Richard Biener <rguenther@suse.de>
31133 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
31134 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31135 Fold all stmts not inplace.
31137 2017-05-12 Richard Biener <rguenther@suse.de>
31139 PR tree-optimization/80713
31140 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
31141 inserted_exprs bit for not removed stmts.
31143 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
31145 PR middle-end/69921
31146 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
31147 parallelized" attribute for parallelized OpenACC kernels.
31148 * omp-offload.c (execute_oacc_device_lower): Use it.
31150 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
31151 Set "oacc kernels" attribute.
31152 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
31153 parameter. Adjust all users.
31154 (oacc_fn_attrib_kernels_p): Remove function.
31155 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
31156 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
31157 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
31158 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
31159 assert "oacc kernels" attribute is set.
31161 2017-05-11 Carl Love <cel@us.ibm.com>
31163 * config/rs6000/rs6000-c: Add support for built-in functions
31164 vector unsigned char vec_popcnt (vector signed char)
31165 vector unsigned char vec_popcnt (vector unsigned char)
31166 vector unsigned short vec_popcnt (vector signed short)
31167 vector unsigned short vec_popcnt (vector unsigned short)
31168 vector unsigned int vec_popcnt (vector signed int)
31169 vector unsigned int vec_popcnt (vector unsigned int)
31170 vector unsigned long long vec_popcnt (vector signed long long)
31171 vector unsigned long long vec_popcnt (vector unsigned long long)
31172 vector signed long long vec_slo (vector signed long long,
31173 vector signed char)
31174 vector signed long long vec_slo (vector signed long long,
31175 vector unsigned char)
31176 vector unsigned long long vec_slo (vector unsigned long long,
31177 vector signed char)
31178 vector unsigned long long vec_slo (vector unsigned long long,
31179 vector unsigned char)
31180 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
31181 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
31182 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
31183 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
31184 * doc/extend.texi: Update the built-in documentation file for the
31185 new built-in functions.
31187 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
31189 * attribs.h (sorted_attr_string): Move machine independent
31190 functions for target clone support from the i386 port to common
31191 code. Rename ix86_function_versions to common_function_versions.
31192 Rename make_name to make_unique_name.
31193 (common_function_versions): Likewise.
31194 (make_unique_name): Likewise.
31195 (make_dispatcher_decl): Likewise.
31196 (is_function_default_version): Likewise.
31197 * attribs.c (attr_strcmp): Likewise.
31198 (sorted_attr_string): Likewise.
31199 (common_function_versions): Likewise.
31200 (make_unique_name): Likewise.
31201 (make_dispatcher_decl): Likewise.
31202 (is_function_default_version): Likewise.
31203 * config/i386/i386.c (attr_strcmp): Likewise.
31204 (sorted_attr_string): Likewise.
31205 (ix86_function_versions): Likewise.
31206 (make_name): Likewise.
31207 (make_dispatcher_decl): Likewise.
31208 (is_function_default_version): Likewise.
31209 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
31211 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31214 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
31215 Account for direct move costs for vec_construct of integer
31218 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
31221 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
31222 (UNSPEC_STX_ATOMIC): Ditto.
31223 (loaddi_via_sse): New insn.
31224 (storedi_via_sse): Ditto.
31225 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
31226 Update corresponding peephole2 patterns.
31227 (atomic_storedi_fpu): Ditto.
31229 2017-05-11 Julia Koval <julia.koval@intel.com>
31231 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
31232 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
31234 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
31235 (__builtin_ia32_rsqrt14ss_mask): New builtins.
31236 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
31238 2017-05-11 Nathan Sidwell <nathan@acm.org>
31240 * graphite-poly.c: Include dumpfile.h.
31242 * dumpfle.h (dump_function): Declare here ...
31243 * tree-dump.h (dump_function): ... not here.
31244 * dumpfile.c: #include tree-cfg.h.
31245 (dump_function): Move here from ...
31246 * tree-dump.c (dump_function): ... here.
31247 * gimplify.c: #include splay-tree.h, not tree-dump.h.
31248 * graphite-poly.c: Don't include tree-dump.h.
31249 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
31250 * print-tree.c: Likewise.
31251 * stor-layout.c: Likewise.
31252 * tree-nested.c: Likewise.
31254 * dumpfile.c (dump_start): Use TDF_FLAGS.
31255 (dump_enable_all): Fix TDF_KIND check thinko.
31257 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
31259 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31260 array entries to represent two legal parameterizations of the
31261 overloaded __builtin_cmpb function, as represented by the
31262 P6_OV_BUILTIN_CMPB constant.
31263 (altivec_resolve_overloaded_builtin): Add special case handling
31264 for the __builtin_cmpb function, as represented by the
31265 P6_OV_BUILTIN_CMPB constant.
31266 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
31267 (BU_P6_64BIT_2): New macro.
31268 (BU_P6_OVERLOAD_2): New macro
31269 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
31270 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
31271 (CMPB): Add overload support to represent both 32-bit and 64-bit
31272 compare-bytes function.
31273 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
31274 support for TARGET_CMPB.
31275 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
31276 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31277 documentation of the __builtin_cmpb overloaded built-in function.
31279 2017-05-11 Richard Biener <rguenther@suse.de>
31281 PR tree-optimization/80705
31282 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
31283 bases are not vectorizable.
31285 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31287 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
31288 when counting register pressure.
31290 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31292 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
31293 (iv_ca_more_deps): Renamed to ...
31294 (iv_ca_compare_deps): ... this.
31295 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
31297 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31299 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
31301 (determine_group_iv_costs): ... here.
31302 (find_inv_vars_cb): Record inv var if it's not recorded before.
31304 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31306 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
31307 (get_shiftadd_cost): Ditto.
31309 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31311 * tree-ssa-address.c: Include header file.
31312 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
31314 (add_to_parts): Refactor.
31315 (addr_to_parts): New parameter. Update use of move_hint_to_base.
31316 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
31319 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31321 PR tree-optimization/53090
31322 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
31324 (extract_cond_operands): Detect condition with IV on both sides
31325 and return COMP_IV_EXPR_2.
31326 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
31327 (rewrite_use_compare): Simplify by removing call to function
31328 extract_cond_operands.
31330 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31332 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
31333 (extract_cond_operands): Detect condition comparing against non-
31334 invariant bound and return appropriate enum value.
31335 (find_interesting_uses_cond): Update use of extract_cond_operands.
31336 Handle its return value accordingly.
31337 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
31339 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31341 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
31342 nonlinear iv_use computation in loop invariant sensitive way.
31344 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31346 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
31347 (find_iv_candidates): Call relate_compare_use_with_all_cands.
31349 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31351 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
31352 (dump_cand): Support iv_cand.inv_exprs.
31353 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
31355 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
31358 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31360 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
31362 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
31363 as local function. Include necessary header files.
31364 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
31366 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31368 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
31370 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31372 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
31373 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
31374 RSHIFT_EXPR and BIT_NOT_EXPR.
31376 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31378 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
31379 (adjust_setup_cost): New parameter supporting round up adjustment.
31380 (struct address_cost_data): Delete.
31381 (force_expr_to_var_cost): Don't bound cost with spill_cost.
31382 (split_address_cost, ptr_difference_cost): Delete.
31383 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
31384 (struct ainc_cost_data): New struct.
31385 (get_address_cost_ainc): New function.
31386 (get_address_cost, get_computation_cost): Reimplement.
31387 (determine_group_iv_cost_address): Record inv_expr for all uses of
31389 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
31390 (iv_ca_has_deps): Reimplemented to ...
31391 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
31393 (iv_ca_extend): Call iv_ca_more_deps.
31395 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31397 * tree-ssa-address.c (struct mem_address): Move to header file.
31398 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
31399 * tree-ssa-address.h (struct mem_address): Move from C file.
31400 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
31402 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31404 * tree-affine.h (aff_combination_type): New interface.
31405 (aff_combination_zero_p): Remove static.
31406 (aff_combination_const_p): New interface.
31407 (aff_combination_singleton_var_p): New interfaces.
31409 2017-05-11 Richard Biener <rguenther@suse.de>
31411 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31412 Skip unreachable blocks and destinations.
31413 (eliminate): Move stmt removal and fixup ...
31414 (fini_eliminate): ... here. Skip inserted exprs.
31415 (pass_pre::execute): Move fini_pre after fini_eliminate.
31416 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
31417 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
31418 PRE to get rid of dead code that has invalid SSA form and
31419 split critical edges again.
31421 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31423 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31425 2017-05-11 Richard Biener <rguenther@suse.de>
31427 * passes.c (execute_function_todo): Verify loops if they are
31428 said to be up-to-date.
31429 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
31430 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
31432 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
31435 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
31436 handle calling assemble_external ourself.
31439 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
31440 modes with zero size. Enhance comment.
31442 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31444 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
31445 built-ins for vec_xl and vec_xst with short and char pointer
31448 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31450 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
31451 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
31452 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
31453 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
31454 (_mm_maskz_min_round_ss): New intrinsics.
31455 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
31456 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
31457 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
31458 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
31459 (__builtin_ia32_minss_mask_round): New builtins.
31460 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31461 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31462 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
31464 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
31465 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
31467 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
31470 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31472 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
31473 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
31474 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
31475 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
31476 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
31477 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
31478 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
31479 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31480 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31481 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
31482 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
31483 (__builtin_ia32_mulss_mask_round): New builtins.
31484 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31485 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31486 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
31488 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
31489 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31491 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31494 2017-05-10 Julia Koval <julia.koval@intel.com>
31496 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
31497 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
31498 (_mm256_setr_m128i): New intrinsics.
31500 2017-05-10 Julia Koval <julia.koval@intel.com>
31502 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
31503 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
31504 (_mm_maskz_rcp14_ss): New intrinsics.
31505 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
31506 (__builtin_ia32_rcp14ss_mask): New builtins.
31507 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
31509 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
31511 PR tree-optimization/51513
31512 * tree-cfg.c (gimple_seq_unreachable_p): New function.
31513 (assert_unreachable_fallthru_edge_p): Use it.
31514 (group_case_labels_stmt): Likewise.
31515 * tree-cfg.h: Prototype it.
31516 * stmt.c: Include cfghooks.h and tree-cfg.h.
31517 (emit_case_dispatch_table) <gap_label>: New local variable.
31518 Use it to fill dispatch table gaps.
31519 Test for default_label before updating probabilities.
31520 (expand_case) <default_label>: Remove unneeded initialization.
31521 Test for unreachable default case statement and remove its edge.
31522 Set default_label accordingly.
31523 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
31525 2017-05-10 Carl Love <cel@us.ibm.com>
31527 * config/rs6000/rs6000-c: Add support for built-in functions
31528 vector signed char vec_neg (vector signed char)
31529 vector signed short int vec_neg (vector short int)
31530 vector signed int vec_neg (vector signed int)
31531 vector signed long long vec_neg (vector signed long long)
31532 vector float vec_neg (vector float)
31533 vector double vec_neg (vector double)
31534 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
31536 * config/rs6000/altivec.h: Add define for vec_neg
31537 * doc/extend.texi: Update the built-in documentation for the
31538 new built-in functions.
31540 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31542 PR tree-optimization/77644
31543 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
31545 2017-05-10 Nathan Sidwell <nathan@acm.org>
31547 * dumpfile.h (TDI_lang_all): New.
31548 (TDF_KIND): New. Renumber others
31549 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
31551 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
31553 (get_dump_file_name): Adjust suffix generation.
31554 (dump_enable_all): Use TDF_KIND.
31555 * doc/invoke.texi (-fdump-lang-all): Document.
31557 * dumpfile.h: Tabify.
31559 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
31562 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
31563 Move member access before delete.
31565 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
31567 * tree-inline.c (expand_call_inline): Split block at stmt
31570 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
31573 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
31574 are now unused after splitting mov{sf,sd}_hardfloat.
31575 (f32_lr2): Likewise.
31576 (f32_lm): Likewise.
31577 (f32_lm2): Likewise.
31578 (f32_li): Likewise.
31579 (f32_li2): Likewise.
31580 (f32_lv): Likewise.
31581 (f32_sr): Likewise.
31582 (f32_sr2): Likewise.
31583 (f32_sm): Likewise.
31584 (f32_sm2): Likewise.
31585 (f32_si): Likewise.
31586 (f32_si2): Likewise.
31587 (f32_sv): Likewise.
31588 (f32_dm): Likewise.
31589 (f32_vsx): Likewise.
31590 (f32_av): Likewise.
31591 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
31592 For movsf, order stores so the VSX stores occur before the GPR
31593 store which encourages the register allocator to use a traditional
31594 FPR instead of a GPR. For movsd, order the stores so that the GPR
31595 store comes before the VSX stores to allow the power6 to work.
31596 This is due to the power6 not having a 32-bit integer store
31597 instruction from a FPR.
31598 (movsf_hardfloat): Likewise.
31599 (movsd_hardfloat): Likewise.
31601 2017-05-09 Martin Sebor <msebor@redhat.com>
31603 PR translation/80280
31604 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
31607 PR translation/80280
31608 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
31609 data member added in r247778.
31610 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
31612 2017-05-09 Nathan Sidwell <nathan@acm.org>
31614 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
31616 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31619 2017-05-09 Marek Polacek <polacek@redhat.com>
31621 * doc/invoke.texi: Fix typo.
31623 2017-05-09 Richard Biener <rguenther@suse.de>
31625 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31626 (vrp_val_is_min): Likewise.
31627 (set_value_range_to_value): Likewise.
31628 (set_value_range_to_nonnegative): Likewise.
31629 (gimple_assign_nonzero_p): Likewise.
31630 (gimple_stmt_nonzero_p): Likewise.
31631 (vrp_int_const_binop): Likewise. Remove unreachable case.
31632 (adjust_range_with_scev): Adjust comments.
31633 (compare_range_with_value): Likewise.
31634 (extract_range_from_phi_node): Likewise.
31635 (test_for_singularity): Likewise.
31637 2017-05-09 Richard Biener <rguenther@suse.de>
31639 * tree-vrp.c (get_single_symbol): Add assert that we don't
31640 get overflowed constants as invariant part.
31641 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31642 checks. Use wi::cmp instead of recursing for integer constants.
31643 (compare_values): Just ignore whether we assumed undefined
31644 overflow instead of failing the compare.
31645 (extract_range_for_var_from_comparison_expr): Add comment before the
31646 TREE_NO_WARNING sets.
31647 (test_for_singularity): Likewise.
31648 (extract_range_from_comparison): Do not disable optimization
31649 when we assumed undefined overflow.
31650 (extract_range_basic): Remove init of unused var.
31652 2017-05-09 Richard Biener <rguenther@suse.de>
31654 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31655 (extract_range_from_multiplicative_op_1): Adjust.
31656 (extract_range_from_binary_expr_1): Use int_const_binop.
31658 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31661 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31662 rs6000_store_data_bypass_p in seven define_bypass directives and
31663 in several comments.
31664 * config/rs6000/rs6000-protos.h: Add prototype for
31665 rs6000_store_data_bypass_p function.
31666 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31667 function implements slightly different (rs6000-specific) semantics
31668 than store_data_bypass_p, returning false rather than aborting
31669 with assertion error when arguments do not satisfy the
31670 requirements of store data bypass.
31671 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31672 rs6000_store_data_bypass_p.
31674 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31676 * config/xtensa/xtensa-protos.h
31677 (xtensa_initial_elimination_offset): New declaration.
31678 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31679 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31680 macro definition, add case for FRAME_POINTER_REGNUM when
31681 FRAME_GROWS_DOWNWARD.
31682 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31683 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31684 xtensa_initial_elimination_offset.
31686 2017-05-08 Nathan Sidwell <nathan@acm.org>
31688 * doc/invoke.texi: Alphabetize -fdump options.
31690 2017-05-08 Martin Sebor <msebor@redhat.com>
31692 PR translation/80280
31693 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31695 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31697 * target.def (compute_frame_layout): New optional target hook.
31698 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31699 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31700 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31702 * reload1.c (verify_initial_elim_offsets): Likewise.
31703 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31704 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31705 (arm_get_frame_offsets): Split up into this ...
31706 (arm_compute_frame_layout): ... and this function.
31708 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31710 * config/aarch64/constraints.md (Usa): New constraint.
31711 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31713 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31715 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31716 with_multilib_list after it has been checked.
31718 2017-05-08 Richard Biener <rguenther@suse.de>
31720 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31721 (bitmap_set_subtract_values): Likewise.
31723 2017-05-08 Richard Biener <rguenther@suse.de>
31725 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31726 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31728 (gimple_stmt_nonzero_warnv_p): Rename to ...
31729 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31731 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31732 (extract_range_basic): Adjust, do not disable propagation on
31733 strict overflow sensitive simplification.
31734 (vrp_visit_cond_stmt): Likewise.
31736 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31738 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31739 body size unconditionally.
31741 2017-05-07 Jeff Law <law@redhat.com>
31744 2017-05-06 Jeff Law <law@redhat.com>
31745 PR tree-optimization/78496
31746 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31749 PR tree-optimization/78496
31750 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31751 (simplify_stmt_using_ranges): Call it.
31752 (vrp_dom_walker::before_dom_children): Extract equivalences
31753 from an ASSERT_EXPR with an equality comparison against a
31756 2017-05-06 Jeff Law <law@redhat.com>
31758 PR tree-optimization/78496
31759 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31762 PR tree-optimization/78496
31763 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31764 (simplify_stmt_using_ranges): Call it.
31765 (vrp_dom_walker::before_dom_children): Extract equivalences
31766 from an ASSERT_EXPR with an equality comparison against a
31769 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31771 * lra-constraints.c (lra_copy_reg_equiv): New function.
31772 (split_reg): Use it to copy equivalence information from the
31773 original register to the spill register.
31775 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31777 PR rtl-optimization/75964
31778 * simplify-rtx.c (simplify_const_relational_operation): Remove
31779 invalid handling of comparisons of integer ABS.
31781 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31783 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31784 initialize to zero.
31785 (init_regs): Remove declaration.
31786 (function_arg_advance_32): Initialize error_p as boolean variable.
31788 2017-05-05 Nathan Sidwell <nathan@acm.org>
31790 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31791 lines. Use for (;;).
31793 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31795 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31796 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31797 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31798 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31799 VF=2 that require versioning.
31801 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31803 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31806 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31808 * diagnostic.h (diagnostic_override_option_index): Convert from
31809 macro to inline function.
31811 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31813 * diagnostic.c (last_module_changed_p): New function.
31814 (set_last_module): New function.
31815 (diagnostic_report_current_module): Convert macro usage to
31816 the above functions.
31817 * diagnostic.h (diagnostic_context::last_module): Strengthen
31818 from const line_map * to const line_map_ordinary *.
31819 (diagnostic_last_module_changed): Delete macro.
31820 (diagnostic_set_last_module): Delete macro.
31822 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31824 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31825 with diagnostic_report_diagnostic.
31826 (diagnostic_n_impl_richloc): Likewise.
31827 * diagnostic.h (report_diagnostic): Delete macro.
31828 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31829 with diagnostic_report_diagnostic.
31830 * substring-locations.c (format_warning_va): Likewise.
31832 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31834 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31835 save/restor of format_spec. Move option-printing code to...
31836 (print_option_information): ...this new function, and
31837 reimplement by simply printing to the pretty_printer,
31838 rather than appending to the format string.
31840 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31842 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31843 handling logic into...
31844 (update_effective_level_from_pragmas): ...this new function.
31846 2017-05-04 Andrew Waterman <andrew@sifive.com>
31848 * config/riscv/riscv.opt (mstrict-align): New option.
31849 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31850 (SLOW_UNALIGNED_ACCESS): Define.
31851 (riscv_slow_unaligned_access): Declare.
31852 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31854 (riscv_slow_unaligned_access): New variable.
31855 (rocket_tune_info): Set slow_unaligned_access to true.
31856 (optimize_size_tune_info): Set slow_unaligned_access to false.
31857 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31858 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31859 (riscv_option_override): Set riscv_slow_unaligned_access.
31860 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31862 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
31864 * config/riscv/riscv.md: Unify indentation.
31866 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
31871 * config/rs6000/rs6000.md (u code attribute): Add FIX and
31873 (extendsi<mode>2): Add support for doing sign extension via
31874 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
31875 don't have ISA 3.0 instructions.
31876 (extendsi<mode>2 splitter): Likewise.
31877 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
31878 generate the normal insns since SImode can now go in vector
31879 registers. Disallow the special UNSPECs needed for previous
31880 machines to hide SImode being used. Add new insns
31881 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
31882 (fix_trunc<mode>si2_stfiwx): Likewise.
31883 (fix_trunc<mode>si2_internal): Likewise.
31884 (fixuns_trunc<mode>si2): Likewise.
31885 (fixuns_trunc<mode>si2_stfiwx): Likewise.
31886 (fctiw<u>z_<mode>_smallint): Likewise.
31887 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
31888 of floating point to 32-bit integer from doing a direct move to
31889 the GPR registers to do a store.
31890 (fctiwz_<mode>): Break long line.
31892 2017-05-05 Bin Cheng <bin.cheng@arm.com>
31894 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
31895 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
31896 (addr_list, addr_offset_valid_p): New.
31897 (split_address_groups): Check offset validity with above function.
31898 (gt-tree-ssa-loop-ivopts.h): Include header file.
31900 2017-05-05 Nathan Sidwell <nathan@acm.org>
31902 * config.gcc (arm*-*-*): Add missing 'fi'.
31904 2017-05-05 Steve Ellcey <sellcey@cavium.com>
31906 * doc/invoke.texi (-fopt-info): Explicitly say order of options
31907 included in -fopt-info does not matter.
31908 * doc/optinfo.texi (-fopt-info): Fix description of default
31909 behavour. Explicitly say order of options included in -fopt-info
31912 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
31914 * config.gcc: Allow combinations of aprofile and rmprofile values for
31915 --with-multilib-list.
31916 * config/arm/t-multilib: New file.
31917 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
31918 variables. Remove setting of ISA and floating-point ABI in
31919 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
31920 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
31921 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
31922 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
31924 * config/arm/t-rmprofile: Likewise except for the matches changes.
31925 * doc/install.texi (--with-multilib-list): Document the combination of
31926 aprofile and rmprofile values and warn about pitfalls in doing that.
31928 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31930 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
31931 (movdi_aarch64): Likewise.
31933 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31935 PR tree-optimization/80632
31936 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
31938 (build_arrays): Initialize it for virtual phis.
31939 (fix_phi_nodes): Use it for virtual phis.
31941 PR tree-optimization/80558
31942 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
31943 [x, y] op z into [x op, y op z] for op & or | if conditions
31946 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
31947 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31950 * config/arm/arm.md (use_literal_pool): Remove.
31951 (64-bit immediate split): No longer takes cost into consideration
31952 if arm_disable_literal_pool is enabled.
31953 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
31954 used when arm_disable_literal_pool is enabled.
31955 (arm_max_const_double_inline_cost): Remove use of
31956 arm_disable_literal_pool.
31957 (push_minipool_fix): Add assert.
31958 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
31959 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
31960 (no_literal_pool_sf_immediate): New.
31962 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31964 PR tree-optimization/80613
31965 * tree-ssa-dce.c (propagate_necessity): Remove cases for
31966 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
31968 2017-05-05 Richard Biener <rguenther@suse.de>
31970 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
31972 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
31974 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
31975 of this flag from insn conditions due to removal from r247495.
31977 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31979 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
31981 (arm_early_store_addr_dep_ptr): Likewise.
31982 * config/arm/aarch-common-protos.h
31983 (arm_early_load_addr_dep_ptr): Add prototype.
31984 (arm_early_store_addr_dep_ptr): Likewise.
31985 * config/arm/cortex-a53.md: Add new bypasses.
31987 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31989 * tree.c (next_type_uid): Change type to unsigned.
31990 (type_hash_canon): Decrement back next_type_uid if
31991 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
31992 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
31995 2017-05-04 Martin Sebor <msebor@redhat.com>
31997 * builtins.c: Fix a trivial typo in a comment.
31999 PR middle-end/79234
32000 * builtins.c (check_sizes): Adjust to handle reading past the end.
32001 Avoid printing excessive upper bound of ranges. Use %E to print
32002 tree nodes instead of converting them to %wu.
32003 (expand_builtin_memchr): New function.
32004 (compute_dest_size): Rename...
32005 (compute_objsize): ...to this.
32006 (expand_builtin_memcpy): Adjust.
32007 (expand_builtin_mempcpy): Adjust.
32008 (expand_builtin_strcat): Adjust.
32009 (expand_builtin_strcpy): Adjust.
32010 (check_strncat_sizes): Adjust.
32011 (expand_builtin_strncat): Adjust.
32012 (expand_builtin_strncpy): Adjust and simplify.
32013 (expand_builtin_memset): Adjust.
32014 (expand_builtin_bzero): Adjust.
32015 (expand_builtin_memcmp): Adjust.
32016 (expand_builtin): Handle memcmp.
32017 (maybe_emit_chk_warning): Check strncat just once.
32019 2017-05-04 Martin Sebor <msebor@redhat.com>
32021 PR preprocessor/79214
32022 PR middle-end/79222
32023 PR middle-end/79223
32024 * builtins.c (check_sizes): Add inlining context and issue
32025 warnings even when -Wno-system-headers is set.
32026 (check_strncat_sizes): Same.
32027 (expand_builtin_strncat): Same.
32028 (expand_builtin_memmove): New function.
32029 (expand_builtin_stpncpy): Same.
32030 (expand_builtin): Handle memmove and stpncpy.
32032 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32034 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32035 which is not used any more.
32037 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32039 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32041 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32043 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32044 (cortexa53_tunings): Likewise.
32045 (cortexa57_tunings): Likewise.
32046 (cortexa72_tunings): Likewise.
32047 (cortexa73_tunings): Likewise.
32049 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32051 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32052 Set loop alignment to 8.
32054 2017-05-04 Martin Sebor <msebor@redhat.com>
32056 PR translation/80280
32057 * builtins.c (expand_builtin_object_size): Add missing quoting to
32058 %D and like directives.
32059 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32060 (hsa_type_for_tree_type): Same.
32061 (verify_function_arguments): Same.
32062 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32063 * varasm.c (get_section): Same.
32066 2017-05-04 Martin Sebor <msebor@redhat.com>
32068 PR translation/80280
32069 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32071 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32073 * config/aarch64/aarch64.c (generic_addrcost_table):
32074 Change HI/TI mode setting.
32076 2017-05-04 Martin Jambor <mjambor@suse.cz>
32078 PR tree-optimization/80622
32079 * tree-sra.c (comes_initialized_p): New function.
32080 (build_accesses_from_assign): Only set write lazily when
32081 comes_initialized_p is false.
32082 (analyze_access_subtree): Use comes_initialized_p.
32083 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32084 instead of testing for PARM_DECL.
32086 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32088 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32089 constraint on operand 0 to allow more general addressing modes.
32090 Adjust output template.
32091 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32093 * config/aarch64/aarch64-protos.h
32094 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32095 * config/aarch64/constraints.md (Dp): New address constraint.
32096 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32099 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32101 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32102 update use of estimate_ipcp_clone_size_and_time.
32103 (estimate_local_effects): Update use of
32104 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32105 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
32106 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
32107 Return nonspecialized time.
32109 2017-05-04 Richard Biener <rguenther@suse.de>
32111 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
32112 for the last VUSE which def dominates the PHI. Directly call
32114 (get_continuation_for_phi_1): Remove.
32116 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
32118 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
32119 to explain the use of truncating division. Cap the number of
32120 iterations to the maximum given by nb_iterations_upper_bound,
32123 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32125 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
32126 * configure: Regenerate.
32127 * config.in: Regenerate.
32128 * config/i386/driver-mingw32.c: new file.
32129 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
32130 * config.host: Link driver-mingw32.o on MinGW host.
32131 * doc/install.texi: Document new --enable-mingw-wildcard configure
32134 2017-05-04 Marek Polacek <polacek@redhat.com>
32136 PR tree-optimization/80612
32137 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
32139 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32140 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
32142 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
32143 (movt splitter): Likewise.
32144 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
32145 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
32146 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
32147 block for Thumb-1 with MOVT.
32148 (thumb2_legitimate_address_p): Move code block ...
32149 (can_avoid_literal_pool_for_label_p): ... into this new function.
32150 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
32152 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
32153 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
32154 "M-profile targets with the MOVT instruction".
32156 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32158 * config/arm/arm-builtins.c (arm_init_builtins): Rename
32159 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
32160 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
32162 2017-05-04 Martin Liska <mliska@suse.cz>
32164 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
32165 variable cond_code.
32167 2017-05-04 Richard Biener <rguenther@suse.de>
32169 * tree.c (array_at_struct_end_p): Handle arrays at struct
32170 end with flexarrays more conservatively. Refactor and treat
32171 arrays of arrays or aggregates more strict. Fix
32172 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
32173 * tree.c (array_at_struct_end_p): Adjust prototype.
32174 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
32175 * gimple-fold.c (get_range_strlen): Likewise.
32176 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
32178 2017-05-04 Richard Biener <rguenther@suse.de>
32180 PR tree-optimization/31130
32181 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
32183 (supports_overflow_infinity): Likewise.
32184 (is_negative_overflow_infinity): Likewise.
32185 (is_positive_overflow_infinity): Likewise.
32186 (is_overflow_infinity): Likewise.
32187 (stmt_overflow_infinity): Likewise.
32188 (overflow_infinity_range_p): Likewise.
32189 (usable_range_p): Remove as always returning true.
32190 (make_overflow_infinity): Remove.
32191 (negative_overflow_infinity): Likewise.
32192 (positive_overflow_infinity): Likewise.
32193 (avoid_overflow_infinity): Likewise.
32194 (set_value_range): Adjust accordingly.
32195 (set_value_range_to_nonnegative): Likewise, remove now unused
32196 overflow_infinity arg.
32197 (vrp_operand_equal_p): Adjust.
32198 (update_value_range): Likewise.
32199 (range_int_cst_singleton_p): Likewise.
32200 (operand_less_p): Likewise.
32201 (compare_values_warnv): Likewise.
32202 (extract_range_for_var_from_comparison_expr): Likewise.
32203 (vrp_int_const_binop): Likewise.
32204 (zero_nonzero_bits_from_vr): Likewise.
32205 (extract_range_from_multiplicative_op_1): Likewise.
32206 (extract_range_from_binary_expr_1): Likewise.
32207 (extract_range_from_unary_expr): Likewise.
32208 (extract_range_from_comparison): Likewise.
32209 (extract_range_basic): Likewise.
32210 (adjust_range_with_scev): Likewise.
32211 (compare_ranges): Likewise.
32212 (compare_range_with_value): Likewise.
32213 (dump_value_range): Likewise.
32214 (test_for_singularity): Likewise, remove strict_overflow_p parameter
32216 (simplify_cond_using_ranges): Adjust.
32218 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32220 * brig-builtins.def: Added a builtin for class_f64.
32221 * builtin-types.def: Added a builtin type needed by class_f64.
32223 2017-05-03 Jason Merrill <jason@redhat.com>
32225 * timevar.def: Add TV_CONSTEXPR.
32227 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32229 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
32231 2017-05-03 Martin Jambor <mjambor@suse.cz>
32233 * ipa-prop.c (ipa_update_after_lto_read): Removed.
32234 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
32235 * ipa-cp.c (ipcp_propagate_stage): Do not call
32236 ipa_update_after_lto_read.
32237 * ipa-inline.c (ipa_inline): Likewise.
32239 2017-05-03 Martin Jambor <mjambor@suse.cz>
32241 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
32242 tag. Added a default constructor and a destructor.
32243 (ipa_edge_args_sum_t): New class;
32244 (ipa_edge_args_sum): Declare.
32245 (ipa_edge_args_vector): Remove declaration.
32246 (IPA_EDGE_REF): Use ipa_edge_args_sum.
32247 (ipa_free_edge_args_substructures): Remove declaration.
32248 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
32249 (ipa_edge_args_info_available_for_edge_p): Likewise.
32250 * ipa-prop.c (ipa_edge_args_vector): Removed.
32251 (edge_removal_hook_holder): Likewise.
32252 (edge_duplication_hook_holder): Likewise.
32253 (ipa_edge_args_sum): New variable.
32254 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
32255 ipa_edge_args_vector.
32256 (ipa_free_edge_args_substructures): Likewise.
32257 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
32258 ipa_edge_args_vector.
32259 (ipa_edge_removal_hook): Turned into method
32260 ipa_edge_args_sum_t::remove.
32261 (ipa_edge_duplication_hook): Turned into method
32262 ipa_edge_args_sum_t::duplicate.
32263 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
32264 registering edge hooks.
32265 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
32266 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
32267 ipa_edge_args_sum instead of ipa_edge_args_vector.
32268 * ipa-profile.c (ipa_profile): Likewise.
32270 2017-05-03 Martin Jambor <mjambor@suse.cz>
32272 * symbol-summary.h (function_summary): New method exists.
32273 (function_summary::symtab_removal): Deallocate through release.
32274 (call_summary): New class.
32275 (gt_ggc_mx): New overload.
32276 (gt_pch_nx): Likewise.
32277 (gt_pch_nx): Likewise.
32279 2017-05-03 Jeff Law <law@redhat.com>
32281 PR tree-optimization/78496
32282 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
32283 from simplify_cond_using_ranges. Split off code to walk
32284 backwards through casts into ...
32285 (simplify_cond_using_ranges_2): New function.
32286 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
32287 (execute_vrp): After identifying jump threads, call
32288 simplify_cond_using_ranges_2.
32290 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
32293 * ipa-inline.h (inline_summary): Add ctor.
32294 (create_ggc): Do not use ggc_cleared_alloc.
32296 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
32297 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32299 * gcc.c (handle_braces): Support escaping in switch matching
32301 * doc/invoke.texi (Spec Files): Document it.
32302 Remove superfluous @code markup in items.
32304 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32306 * diagnostic-show-locus.c (struct column_range): New struct.
32307 (get_affected_columns): New function.
32308 (get_printed_columns): New function.
32309 (struct correction): New struct.
32310 (correction::ensure_capacity): New function.
32311 (correction::ensure_terminated): New function.
32312 (struct line_corrections): New struct.
32313 (line_corrections::~line_corrections): New dtor.
32314 (line_corrections::add_hint): New function.
32315 (layout::print_trailing_fixits): Reimplement in terms of the new
32317 (selftest::test_overlapped_fixit_printing): New function.
32318 (selftest::diagnostic_show_locus_c_tests): Call it.
32320 2017-05-03 Nathan Sidwell <nathan@acm.org>
32322 Canonicalize canonical type hashing
32323 * tree.h (type_hash_canon_hash): Declare.
32324 * tree.c (type_hash_list, attribute_hash_list): Move into
32325 type_hash_canon_hash.
32326 (build_type_attribute_qual_variant): Break out hash code calc into
32327 type_hash_canon_hash.
32328 (type_hash_canon_hash): New. Generic type hash computation.
32329 (build_range_type_1, build_array_type_1, build_function_type,
32330 build_method_type_directly, build_offset_type, build_complex_type,
32331 make_vector_type): Call it.
32333 2017-05-03 Richard Biener <rguenther@suse.de>
32335 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
32336 When all DRs have unknown misaligned do not always peel
32337 when there is a store but apply the same costing model as if
32338 there were only loads.
32340 2017-05-03 Richard Biener <rguenther@suse.de>
32343 PR tree-optimization/80492
32344 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32345 compare_base_decls returning dont-know properly.
32347 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
32349 * config/arm/iterators.md (CCSI): New mode iterator.
32350 (arch): New mode attribute.
32351 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
32352 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
32353 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
32354 code iterator for success result mode.
32355 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
32356 the corresponding new insn generators.
32358 2017-05-03 Bin Cheng <bin.cheng@arm.com>
32361 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32362 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32364 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
32366 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
32367 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
32368 (DDR_A): Wrap DDR argument in brackets.
32369 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
32370 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
32371 (DDR_REVERSED_P): Likewise.
32373 2017-05-03 Jakub Jelinek <jakub@redhat.com>
32375 PR tree-optimization/79472
32376 * tree-switch-conversion.c (struct switch_conv_info): Add
32377 contiguous_range and default_case_nonstandard fields.
32378 (collect_switch_conv_info): Compute contiguous_range and
32379 default_case_nonstandard fields, don't clear final_bb if
32380 contiguous_range and only the default case doesn't have the required
32382 (check_all_empty_except_final): Set default_case_nonstandard instead
32383 of failing if contiguous_range and the default case doesn't have empty
32385 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
32386 and only the default case doesn't have the required constants. Skip
32388 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
32389 if default_case_nonstandard.
32390 (build_constructors): Build constant 1 just once. Assert that default
32391 values aren't inserted in between cases if contiguous_range. Skip
32393 (build_arrays): Skip virtual phis.
32394 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
32395 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
32396 Handle virtual phis.
32397 (gen_inbound_check): Handle default_case_nonstandard case.
32398 (process_switch): Adjust check_final_bb caller. Call
32399 gather_default_values with the first non-default case instead of
32400 default case if default_case_nonstandard.
32402 2017-05-02 Nathan Sidwell <nathan@acm.org>
32404 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
32405 check. Fix formatting.
32407 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
32409 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
32410 errors when comparing specialized and unspecialized times.
32412 2017-05-02 David Malcolm <dmalcolm@redhat.com>
32414 * diagnostic-show-locus.c
32415 (layout::should_print_annotation_line_p): Make private.
32416 (layout::print_annotation_line): Make private.
32417 (layout::annotation_line_showed_range_p): Make private.
32418 (layout::show_ruler): Make private.
32419 (layout::print_source_line): Make private. Pass in line and
32420 line_width, rather than calling location_get_source_line. Drop
32422 (layout::print_leading_fixits): New method.
32423 (layout::print_any_fixits): Rename to...
32424 (layout::print_trailing_fixits): ...this, and make private.
32425 Don't print newline fixits.
32426 (diagnostic_show_locus): Move logic for printing one row into...
32427 (layout::print_line): ...this new function. Move the
32428 location_get_source_line call and error-handling from
32429 print_source_line to here. Call print_leading_fixits, and rename
32430 print_any_fixits to print_trailing_fixits.
32431 (selftest::test_fixit_insert_containing_newline): Update now that
32432 newlines are partially supported.
32433 (selftest::test_fixit_insert_containing_newline_2): New test.
32434 (selftest::test_fixit_replace_containing_newline): Update comments.
32435 (selftest::diagnostic_show_locus_c_tests): Call the new test.
32436 * edit-context.c (class added_line): New class.
32437 (class edited_line): Describe newline handling in comment.
32438 (edited_line::actually_edited_p): New method.
32439 (edited_line::print_content): Delete redundant decl.
32440 (edited_line::m_predecessors): New field.
32441 (edited_file::print_content): Call edited_line::print_content.
32442 (edited_file::print_diff): Update to support newlines.
32443 (edited_file::print_diff_hunk): Likewise.
32444 (edited_file::print_run_of_changed_lines): New function.
32445 (edited_file::print_diff_line): Convert to...
32446 (print_diff_line): ...this.
32447 (edited_file::get_effective_line_count): New function.
32448 (edited_line::edited_line): Initialize new field m_predecessors.
32449 (edited_line::~edited_line): Clean up m_predecessors.
32450 (edited_line::apply_fixit): Handle newlines.
32451 (edited_line::get_effective_line_count): New function.
32452 (edited_line::print_content): New function.
32453 (edited_line::print_diff_lines): New function.
32454 (selftest::test_applying_fixits_insert_containing_newline): New
32456 (selftest::test_applying_fixits_replace_containing_newline): New
32458 (selftest::insert_line): New function.
32459 (selftest::test_applying_fixits_multiple_lines): Add example of
32461 (selftest::edit_context_c_tests): Call the new tests.
32463 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32465 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
32466 parameter cand. Update dump information.
32467 (get_computation_cost): Update uses.
32469 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32471 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
32472 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
32473 (get_computation_at, rewrite_use_address): Update use of
32474 get_computation_aff.
32476 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32478 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
32479 (get_computation): Delete.
32480 (get_computation_cost): Implement like get_computation_cost_at.
32481 Use get_computation_at.
32482 (get_computation_cost_at): Delete.
32483 (rewrite_use_nonlinear_expr): Use get_computation_at.
32484 (rewrite_use_compare, remove_unused_ivs): Ditto.
32486 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32488 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
32490 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32492 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
32493 (ivopts_global_cost_for_size): Rename parameter and update uses.
32494 (iv_ca_recount_cost): Update uses.
32495 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
32496 candidates seperately in n_invs and n_cands.
32497 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
32499 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32501 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
32502 (find_inv_vars_cb): New.
32503 (find_depends): Renamed to ...
32504 (find_inv_vars): ... this.
32505 (add_candidate_1, force_var_cost): Call find_inv_vars.
32506 (split_address_cost, determine_group_iv_cost_cond): Ditto.
32508 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32510 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
32511 inv_vars. Add inv_exprs.
32512 (struct iv_cand): Rename depends_on to inv_vars.
32513 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
32514 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
32515 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
32516 (dump_cand): Dump inv_vars.
32517 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
32518 (record_invariant, find_depends, add_candidate_1): Ditto.
32519 (set_group_iv_cost, force_var_cost): Ditto.
32520 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
32521 (get_computation_cost_at, get_computation_cost): Ditto.
32522 (determine_group_iv_cost_generic): Ditto.
32523 (determine_group_iv_cost_address): Ditto.
32524 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
32525 (determine_group_iv_costs): Ditto.
32526 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
32527 (iv_ca_set_remove_invariants): Renamed to ...
32528 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
32529 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
32530 (iv_ca_set_add_invariants): Renamed to ...
32531 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
32532 (iv_ca_set_cp): Use iv_ca_set_add_invs.
32533 (iv_ca_has_deps): Support inv_vars and inv_exprs.
32534 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
32535 (create_new_ivs): Remove useless dump.
32537 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32539 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
32541 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
32542 (iv_ca_set_no_cp, create_new_iv): Ditto.
32544 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32546 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32548 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32550 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
32551 function tree_check2.
32553 2017-05-02 Martin Liska <mliska@suse.cz>
32555 * doc/gcov.texi: Add missing preposition.
32556 * gcov.c (function_info::function_info): Properly fill up
32557 all member variables.
32559 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32561 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
32563 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32565 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
32567 2017-05-02 Martin Liska <mliska@suse.cz>
32570 * lto-streamer-in.c (lto_read_tree_1): Remove
32571 LTO_STREAMER_DEBUG.
32572 * lto-streamer.c (struct tree_hash_entry): Likewise.
32573 (struct tree_entry_hasher): Likewise.
32574 (tree_entry_hasher::hash): Likewise.
32575 (tree_entry_hasher::equal): Likewise.
32576 (lto_streamer_init): Likewise.
32577 (lto_orig_address_map): Likewise.
32578 (lto_orig_address_get): Likewise.
32579 (lto_orig_address_remove): Likewise.
32580 * lto-streamer.h: Likewise.
32581 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
32582 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
32584 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
32586 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
32587 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
32588 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
32589 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
32590 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
32591 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
32592 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
32593 (mm_maskz_sub_ss): New intrinsics.
32594 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32595 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32596 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
32597 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
32598 (__builtin_ia32_subss_mask_round): New builtins.
32599 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32600 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32601 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
32603 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
32604 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32606 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32609 2017-05-02 Martin Jambor <mjambor@suse.cz>
32611 PR tree-optimization/78687
32612 * tree-sra.c (access): New field parent.
32613 (process_subtree_disqualification): New function.
32614 (disqualify_candidate): Call it.
32615 (build_accesses_from_assign): Reset write flag if creating an
32617 (build_access_subtree): Fill in parent field and also prpagate
32618 down grp_write flag.
32619 (create_artificial_child_access): New parameter set_grp_write, set
32620 grp_write to its value.
32621 (propagate_subaccesses_across_link): Also propagate grp_write flag
32623 (propagate_all_subaccesses): Push the closest parent back to work
32624 queue if add_access_to_work_queue returned true.
32626 2017-05-02 Richard Biener <rguenther@suse.de>
32628 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32629 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32630 -fstrict-overflow documentation.
32631 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32632 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32633 flag_strict_overflow.
32634 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32635 * lto-opts.c (lto_write_options): Do not stream it.
32636 * lto-wrapper.c (merge_and_complain): Do not handle it.
32637 * opts.c (default_options_table): Do not set -fstrict-overflow.
32638 (finish_options): Likewise do not clear it when sanitizing.
32639 * simplify-rtx.c (simplify_const_relational_operation): Do not
32640 test flag_strict_overflow.
32642 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32644 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32645 using enabled attribute.
32646 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32647 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32648 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32649 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32650 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32651 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32652 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32653 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32654 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32655 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32657 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32659 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32661 2017-05-02 Richard Biener <rguenther@suse.de>
32663 PR tree-optimization/80591
32665 2017-04-10 Richard Biener <rguenther@suse.de>
32667 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32670 2017-05-02 Richard Biener <rguenther@suse.de>
32672 PR tree-optimization/80549
32673 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32674 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32675 headers if they do not have a preheader.
32677 2017-05-02 Martin Liska <mliska@suse.cz>
32680 * common.opt: Fix typo.
32681 * doc/invoke.texi: Likewise.
32683 2017-05-01 Jan Beulich <jbeulich@suse.com>
32685 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32686 swapping, add (x,x,m,x,n) alternative.
32688 2017-05-01 Nathan Sidwell <nathan@acm.org>
32690 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32691 unnecessary unadjusted_alignment check.
32693 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32696 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32698 * gimplify.c (gimplify_cilk_detach): New function.
32699 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32700 * tree-core.h: Document EXPR_CILK_SPAWN.
32701 * tree.h (EXPR_CILK_SPAWN): Define.
32703 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32705 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32706 to use new fixit_hint representation, using the "replace" logic.
32707 (get_line_span_for_fixit_hint): Likewise.
32708 (layout::print_any_fixits): Likewise.
32709 (selftest::test_one_liner_many_fixits): Rename to...
32710 (selftest::test_one_liner_many_fixits_1): ...this, and update
32711 comment and expected output to reflect that the multiple fix-it
32712 hints are now consolidated into one insertion.
32713 (selftest::test_one_liner_many_fixits_2): New test.
32714 (selftest::test_diagnostic_show_locus_one_liner): Update for
32716 (selftest::test_fixit_consolidation): Update for fix-it API
32718 * diagnostic.c (print_parseable_fixits): Likewise.
32719 * edit-context.c (edited_line::m_line_events): Convert from
32720 auto_vec <line_event *> to auto_vec <line_event>.
32721 (class line_event): Convert from abstract base class to a concrete
32722 class, taking over the role of replace_event.
32723 (class insert_event): Delete.
32724 (class replace_event): Rename to class line_event. Convert to
32726 (edit_context::add_fixits): Reimplement.
32727 (edit_context::apply_insert): Delete.
32728 (edit_context::apply_replace): Rename to...
32729 (edit_context::apply_fixit): ...this. Convert to half-open range.
32730 (edited_file::apply_insert): Delete.
32731 (edited_file::apply_replace): Rename to...
32732 (edited_file::apply_fixit): ...this.
32733 (edited_line::~edited_line): Drop deletion of events.
32734 (edited_line::apply_insert): Delete.
32735 (edited_line::apply_replace): Rename to...
32736 (edited_line::apply_fixit): ...this. Convert to half-open range.
32737 Update for change to type of m_line_events.
32738 * edit-context.h (edit_context::apply_insert): Delete.
32739 (edit_context::apply_replace): Rename to...
32740 (edit_context::apply_fixit): ...this.
32742 2017-05-01 Martin Sebor <msebor@redhat.com>
32744 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32747 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32750 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32751 __get_cpuid_max returns 0.
32752 (__get_cpuid_count): Ditto.
32754 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32756 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32757 replacement expression is another instance of one of its arguments.
32759 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32762 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32763 check for stack push/pop autoinc.
32764 * config/i386/i386.c (ix86_agi_dependent): Return false
32765 if the only reason why modified_in_p returned true is that
32766 addr is SP based and set_insn is a push or pop.
32768 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32770 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32773 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32776 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32777 (account_size_time): Use two predicates - exec_pred and
32779 (evaluate_conditions_for_known_args): Compute both clause and
32781 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32782 (inline_summary_t::duplicate): Update.
32783 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32785 (compute_inline_parameters): Likewise.
32786 (estimate_edge_size_and_time): Update caluclation of time.
32787 (estimate_node_size_and_time): Compute both time and nonspecialized
32789 (estimate_ipcp_clone_size_and_time): Update.
32790 (inline_merge_summary): Update.
32791 (do_estimate_edge_time): Update.
32792 (do_estimate_edge_size): Update.
32793 (do_estimate_edge_hints): Update.
32794 (inline_read_section, inline_write_summary): Stream both new predicates.
32795 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32797 (compute_inlined_call_time): Cleanup.
32798 (big_speedup_p): Update.
32799 (edge_badness): Update.
32800 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32801 (size_time_entry): Replace predicate by exec_predicate and
32802 nonconst_predicate.
32803 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32804 (estimate_edge_time): Return also nonspec_time.
32805 (reset_edge_growth_cache): Update.
32807 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32809 PR rtl-optimization/80491
32810 * ifcvt.c (noce_process_if_block): When looking for x setter
32811 with missing else_bb, don't check only the insn right before
32812 cond_earliest, but look for the last insn that x is modified in
32813 within the same bb.
32815 PR rtl-optimization/80491
32816 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32818 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32820 PR tree-optimization/80487
32821 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32823 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32825 PR tree-optimization/79697
32826 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32827 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32828 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32830 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32831 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32833 2017-04-28 Martin Sebor <msebor@redhat.com>
32835 PR tree-optimization/80523
32836 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32837 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32839 (maybe_warn, format_directive, parse_directive): Use new functions.
32840 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32842 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32844 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32846 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32848 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32849 target_header_dir): Set correctly.
32850 * configure: Regenerated.
32851 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32852 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32853 instead of SYSTEM_HEADER_DIR.
32855 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32857 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32858 (estimate_local_effects): Likewise.
32859 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32860 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32861 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32862 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
32863 do_estimate_edge_time, estimate_edge_time): Likewise.
32864 * ipa-inline-analysis.c (estimate_node_size_and_time,
32865 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
32866 (estimate_time_after_inlining): Remove.
32868 2017-04-28 Martin Liska <mliska@suse.cz>
32870 * doc/gcov.texi: Enhance documentation of gcov.
32872 2017-04-28 Martin Liska <mliska@suse.cz>
32874 * doc/gcov.texi: Sort options in alphabetic order.
32875 * doc/gcov-dump.texi: Likewise.
32876 * doc/gcov-tool.texi: Likewise.
32877 * gcov.c (print_usage): Likewise.
32878 * gcov-dump.c (print_usage): Likewise.
32879 * gcov-tool.c (print_merge_usage_message): Likewise.
32880 (print_rewrite_usage_message): Likewise.
32881 (print_overlap_usage_message): Likewise.
32883 2017-04-28 Martin Liska <mliska@suse.cz>
32885 PR gcov-profile/53915
32886 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
32888 2017-04-28 Martin Liska <mliska@suse.cz>
32890 PR gcov-profile/79891
32891 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
32892 is marked by compiler as living on a line.
32893 (get_cycles_count): Remove usage of the union.
32894 (output_intermediate_file): Likewise.
32895 (find_source): Fix GNU coding style.
32896 (accumulate_line_counts): Remove old non-all block mode.
32897 (output_lines): Remove usage of the union.
32898 * profile.c (output_location): Include all BBs, even if
32899 belonging to a same line (and file) as a previous BB.
32901 2017-04-28 Martin Liska <mliska@suse.cz>
32903 * gcov.c (process_args): Handle new argument 'w'.
32904 (read_graph_file): Assign ID to BBs.
32905 (output_branch_count): Display BB # if verbose flag is set.
32906 (output_lines): Likewise for arcs.
32907 (print_usage): Add '--verbose' option help.
32908 * doc/gcov.texi: Document --verbose (-w) option.
32910 2017-04-28 Martin Liska <mliska@suse.cz>
32912 * gcov.c (struct block_location_info): New struct.
32913 (process_file): Fill up the new structure.
32914 (read_graph_file): Replace usage of encoding by the newly added
32916 (add_line_counts): Likewise.
32917 (accumulate_line_counts): Remove usage of the union.
32918 (function_info::function_info): New function.
32919 (function_info::~function_info): Likewise.
32920 (process_file): Call delete instead of release_function.
32921 (release_function): Release the function.
32922 (release_structures): Call delete instead of release_function.
32923 (solve_flow_graph): Replace usage of num_blocks.
32924 (find_exception_blocks): Likewise.
32925 (output_lines): Fix GNU coding style.
32927 2017-04-28 Martin Liska <mliska@suse.cz>
32930 * coverage.c (coverage_remove_note_file): New function.
32931 * coverage.h: Declare the function.
32932 * toplev.c (finalize): Clean if an error has been seen.
32934 2017-04-28 Martin Liska <mliska@suse.cz>
32936 PR gcov-profile/80031
32937 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
32938 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
32939 * gcov.c (read_graph_file): Read just number of blocks.
32940 * profile.c (branch_prob): Do not stream 0 flags per a basic
32943 2017-04-28 Martin Liska <mliska@suse.cz>
32945 * gcov-dump.c (tag_*): Add new argument to declarations.
32946 (dump_gcov_file): Likewise.
32947 (tag_blocks): Add and use new argument depth.
32948 (tag_arcs): Likewise.
32949 (tag_lines): Likewise.
32950 (tag_counters): Likewise.
32951 (tag_summary): Likewise.
32952 (dump_working_sets): Use depth to do a proper indentation.
32954 2017-04-28 Jakub Jelinek <jakub@redhat.com>
32957 * cgraph.h (symtab_node::debug_symtab): No longer inline.
32958 * symtab.c (symtab_node::debug_symtab): Move definition here.
32960 2017-04-28 Richard Biener <rguenther@suse.de>
32962 * lto-streamer.h (LTO_major_version): Bump to 7.
32964 2017-04-28 Richard Biener <rguenther@suse.de>
32966 * tree-vrp.c (assert_info): New struct.
32967 (add_assert_info): New helper.
32968 (register_edge_assert_for_2): Refactor to add asserts to a vector
32970 (register_edge_assert_for_1): Likewise.
32971 (register_edge_assert_for): Likewise.
32972 (finish_register_edge_assert_for): New helper actually registering
32973 asserts where live on edge.
32974 (find_conditional_asserts): Adjust.
32975 (find_switch_asserts): Likewise.
32976 (evrp_dom_walker::try_find_new_range): Generalize.
32977 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
32979 2017-04-27 Marek Polacek <polacek@redhat.com>
32982 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
32983 arg10 and arg11 to itype.
32985 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
32987 * doc/extend.texi (Object Size Checking): Improve grammar.
32989 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
32992 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
32993 that the logic for permitting reciprocal estimates matches that
32996 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32999 * tree.c (type_cache_hasher::equal): Only compare
33000 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33001 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33002 non-aggregate element types.
33003 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33004 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33006 2017-04-27 Richard Biener <rguenther@suse.de>
33008 PR middle-end/80533
33009 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33010 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33011 keeping a reference to a trailing array.
33013 2017-04-27 Richard Biener <rguenther@suse.de>
33015 PR middle-end/80539
33016 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33017 being in loop-closed SSA form conservatively.
33018 (chrec_fold_multiply_poly_poly): Likewise.
33020 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33022 PR middle-end/79665
33023 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33024 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33026 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33029 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33030 (aarch64_function_arg_alignment): Return unsigned int again, but still
33031 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33032 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33033 Don't emit -Wpsabi note.
33034 (aarch64_function_arg_boundary): Likewise.
33035 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33038 2017-04-26 Nathan Sidwell <nathan@acm.org>
33040 * tree.h (crc32_unsigned_n): Declare.
33041 (crc32_unsigned, crc32_unsigned): Make inline.
33042 * tree.c (crc32_unsigned_bits): Replace with ...
33043 (crc32_unsigned_n): ... this.
33044 (crc32_unsigned, crc32_byte): Remove.
33045 (crc32_string): Remove unnecessary braces.
33047 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33049 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33050 * ipa-inline-analysis.c (MAX_TIME): Remove.
33051 (account_size_time): Use sreal for time.
33052 (dump_inline_summary): Update.
33053 (estimate_function_body_sizes): Update.
33054 (estimate_edge_size_and_time): Update.
33055 (estimate_calls_size_and_time): Update.
33056 (estimate_node_size_and_time): Update.
33057 (inline_merge_summary): Update.
33058 (inline_update_overall_summary): Update.
33059 (estimate_time_after_inlining): Update.
33060 (inline_read_section): Update.
33061 (inline_write_summary): Update.
33062 * ipa-inline.c (compute_uninlined_call_time): Update.
33063 (compute_inlined_call_time): Update.
33064 (recursive_inlining): Update.
33065 (inline_small_functions): Update.
33066 (dump_overall_stats): Update.
33067 * ipa-inline.h: Include sreal.h.
33068 (size_time_entry): Turn time to sreal.
33069 (inline_summary): Turn self_time nad time to sreal.
33071 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33073 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33075 (sreal::stream_out, sreal::stream_in): New.
33076 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33078 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33080 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33083 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33086 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33088 (dimode_scalar_chain::compute_convert_gain): Ditto.
33089 (dimode_scalar_chain::make_vector_copies): Ditto.
33090 (dimode_scalar_chain::convert_reg): Ditto.
33091 (dimode_scalar_chain::convert_insn): Ditto.
33092 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33093 (VI248_AVX512BW_1): New mode iterator.
33094 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33095 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33098 2017-04-25 Martin Sebor <msebor@redhat.com>
33100 PR tree-optimization/80497
33101 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33102 constants are representable in HOST_WIDE_INT.
33103 (parse_directive): Ditto.
33105 2017-04-25 Martin Sebor <msebor@redhat.com>
33108 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
33109 (new_zero_array): Adjust signature.
33110 (dom_info::dom_init): Used unsigned rather that size_t.
33111 (dom_info::dom_info): Same.
33113 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33114 Jakub Jelinek <jakub@redhat.com>
33117 * config/arm/arm.c: Include gimple.h.
33118 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33119 returns negative, increment ncrn only if it returned positive.
33120 (arm_needs_doubleword_align): Return int instead of bool,
33121 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
33122 members, but if there is any such non-FIELD_DECL
33123 > PARM_BOUNDARY aligned decl, return -1 instead of false.
33124 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33125 returns negative, increment nregs only if it returned positive.
33126 (arm_setup_incoming_varargs): Likewise.
33127 (arm_function_arg_boundary): Emit -Wpsabi note if
33128 arm_needs_doubleword_align returns negative, return
33129 DOUBLEWORD_ALIGNMENT only if it returned positive.
33131 2017-04-25 Marek Polacek <polacek@redhat.com>
33134 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33135 first argument to type.
33137 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
33140 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
33141 type checks to test for compatibility instead of equality.
33143 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33144 Jakub Jelinek <jakub@redhat.com>
33147 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
33149 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
33150 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
33151 the alignment computation, but return their maximum in warn_alignment.
33152 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33153 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
33155 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
33156 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33159 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33161 * config/arc/simdext.md (dmpyh): Fix typo.
33163 2017-04-25 Richard Biener <rguenther@suse.de>
33165 PR tree-optimization/80492
33166 * alias.c (compare_base_decls): Handle registers with asm
33167 specification conservatively.
33168 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33169 compare_base_decls returning dont-know properly.
33171 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33173 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
33174 (legitimate_offset_address_p): New function.
33175 (arc_legitimate_address_p): Use above function.
33177 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33179 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
33181 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33183 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
33184 ACCH registers whenever they are available.
33186 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33188 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
33189 double regs fix when not used.
33191 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33193 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
33195 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
33196 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
33198 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33200 * config/arc/arc.c (arc_output_addsi): Check for h-register class
33201 when emitting short ADD instructions.
33203 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33205 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
33207 (cmpsi_cc_c_insn): Likewise.
33208 (cbranchsi4_scratch): Compute proper instruction length using
33209 compact_hreg_operand.
33210 * config/arc/predicates.md (compact_hreg_operand): New predicate.
33212 2017-04-25 Richard Biener <rguenther@suse.de>
33214 PR middle-end/80509
33215 * passes.c (pass_manager::pass_manager): Initialize
33216 m_name_to_pass_map.
33218 2017-04-25 Richard Biener <rguenther@suse.de>
33220 PR tree-optimization/79201
33221 * tree-ssa-sink.c (statement_sink_location): Handle calls.
33223 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33226 * config/s390/vector.md: Split MEM->GPR vector moves for
33227 non-s_operand addresses.
33229 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33232 * config/s390/predicates.md (reload_const_wide_int_operand): New
33234 * config/s390/s390.md ("movti"): Remove d/P alternative.
33235 ("movti_bigconst"): New pattern definition.
33237 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33240 * s390-protos.h (s390_expand_cs_hqi): Removed.
33241 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
33242 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
33243 modes as well as CCZ1mode and CCZmode.
33244 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
33245 signature of s390_emit_compare_and_swap.
33246 (s390_expand_cs_hqi): Likewise, make static.
33247 (s390_expand_cs_tdsi): Generate an explicit compare before trying
33248 compare-and-swap, in some cases.
33249 (s390_expand_cs): Wrapper function.
33250 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
33252 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
33253 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
33254 patterns for small and large integers. Forbid symref memory operands.
33255 Move expander to s390.c. Require cc register.
33256 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
33257 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
33258 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
33259 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
33260 symref memory operands. Remove CC mode and call s390_match_ccmode
33262 ("atomic_exchange<mode>"): Allow and implement all integer modes.
33264 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33266 * config/s390/s390.md (define_peephole2): New peephole to help
33267 combining the load-and-test pattern with volatile memory.
33269 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33271 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
33272 with CCZmode for TARGET_Z196.
33274 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33276 PR rtl-optimization/80501
33277 * combine.c (make_compound_operation_int): Set subreg_code to SET
33278 even for AND with mask of the sign bit of mode.
33280 PR rtl-optimization/80500
33281 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
33282 sum's initial value.
33284 2017-04-25 Julian Brown <julian@codesourcery.com>
33285 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33287 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
33289 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
33291 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
33293 2017-04-25 Julian Brown <julian@codesourcery.com>
33294 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33296 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
33297 (thunderx2t99_sha): New Reservation.
33299 2017-04-25 Julian Brown <julian@codesourcery.com>
33300 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33302 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
33303 type for 1-element load.
33305 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
33307 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
33309 2017-04-24 Martin Jambor <mjambor@suse.cz>
33311 PR tree-optimization/80293
33312 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
33313 char arrays not totally scalarizable if it is false.
33314 (analyze_all_variable_accesses): Pass correct value in the new
33315 parameter. Add a statistics counter.
33317 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
33319 PR middle-end/79931
33320 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
33322 2017-04-24 Richard Biener <rguenther@suse.de>
33324 PR tree-optimization/80494
33325 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
33326 out for complex types.
33328 2017-04-24 Richard Biener <rguenther@suse.de>
33330 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
33331 * tree-ssa-sccvn.c (print_scc): Print SCC size.
33332 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
33333 (DFS): Adjust and never fail.
33334 (sccvn_dom_walker::fail): Remove.
33335 (sccvn_dom_walker::before_dom_children): Adjust.
33336 (run_scc_vn): Likewise and never fail.
33337 * tree-ssa-pre.c (pass_pre::execute): Adjust.
33338 (pass_fre::execute): Likewise.
33340 2017-04-24 Richard Biener <rguenther@suse.de>
33342 PR tree-optimization/79725
33343 * tree-ssa-sink.c (statement_sink_location): Return whether
33344 failure reason was zero uses. Move that check later.
33345 (sink_code_in_bb): Deal with zero uses by removing the stmt
33348 2017-04-24 Richard Biener <rguenther@suse.de>
33351 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
33352 pointer-based references.
33354 2017-04-24 Richard Biener <rguenther@suse.de>
33357 * pass_manager.h (pass_manager::operator new): Remove.
33358 (pass_manager::operator delete): Likewise.
33359 * passes.c (pass_manager::operator new): Remove.
33360 (pass_manager::operator delete): Likewise.
33361 (pass_manager::pass_manager): Zero individual pass members.
33363 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
33366 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
33367 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
33368 Check "XEXP (src, 1)" operand here.
33369 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
33370 Check "XEXP (src, 1)" operand here.
33371 (dimode_scalar_chain::make_vector_copies): Detect count register
33372 of a shift instruction. Zero extend count register from QImode
33373 to DImode to satisfy vector shift pattern count operand predicate.
33374 Substitute vector shift count operand with a DImode copy.
33375 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
33378 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
33380 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
33381 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33382 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
33383 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33384 (UNSPEC_NOREX_MEM): Remove definition.
33386 2017-04-21 Richard Biener <rguenther@suse.de>
33388 PR tree-optimization/79547
33389 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
33390 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
33391 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
33392 without any constraints.
33394 2017-04-21 Richard Biener <rguenther@suse.de>
33396 PR tree-optimization/78847
33397 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
33399 2017-04-21 Richard Biener <rguenther@suse.de>
33401 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
33402 (build_distinct_type_copy): Likewise.
33403 (build_variant_type_copy): Likewise.
33404 * tree.c (build_qualified_type): Pass down mem-stat info.
33405 (build_distinct_type_copy): Likewise.
33406 (build_variant_type_copy): Likewise.
33408 2017-04-21 Richard Biener <rguenther@suse.de>
33410 PR tree-optimization/80237
33411 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
33413 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
33414 for a simplified result.
33416 2016-04-21 Richard Biener <rguenther@suse.de>
33418 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
33419 sth as strict as a simple_iv but a chrec without symbols and an
33420 operand defined in the loop we are peeling (and not some subloop).
33421 (propagate_constants_for_unrolling): Propagate all constants.
33423 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33426 * config/i386/i386.c (print_reg): Remove assert for disalowed
33427 regno values, call output_operand_lossage instead.
33429 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33432 * config/i386/constraints.md (Yc): New register constraint.
33433 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
33434 Use Yc constraint for alternative 2 of operand 0. Remove
33435 preferred_for_speed attribute.
33437 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33439 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
33440 lastprivate clauses in SIMT case.
33442 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
33444 * doc/invoke.texi (-Wextra-semi): Document new warning option.
33446 2017-04-20 Richard Biener <rguenther@suse.de>
33448 PR tree-optimization/57796
33449 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
33450 as N scalar stores.
33451 (vect_model_load_cost): Cost gathers as N scalar loads.
33453 2017-04-20 Richard Biener <rguenther@suse.de>
33455 * ggc-page.c (ggc_allocated_p): Rename to ...
33456 (safe_lookup_page_table_entry): ... this and return the lookup
33458 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
33460 2017-04-20 Richard Biener <rguenther@suse.de>
33462 PR tree-optimization/80453
33463 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
33464 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
33465 from the conditions.
33466 (vn_phi_eq): Pass them down.
33467 (vn_phi_lookup): Record them.
33468 (vn_phi_insert): Likewise.
33470 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
33472 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
33473 uninitialized variable warning to avoid buffer overrun.
33475 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33478 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
33479 is suppressed for '{ 0 }' in C.
33481 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33483 * BASE-VER: Set to 8.0.0.
33485 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
33487 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
33488 priority .init_array and .fini_array section with SECTION_NOTYPE
33491 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33493 PR middle-end/80423
33494 * tree.h (build_array_type): Add typeless_storage default argument.
33495 * tree.c (type_cache_hasher::equal): Also compare
33496 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
33497 (build_array_type): Add typeless_storage argument, set
33498 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
33500 (build_nonshared_array_type): Adjust build_array_type_1 caller.
33501 (build_array_type): Likewise. Add typeless_storage argument.
33503 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33504 Jakub Jelinek <jakub@redhat.com>
33506 PR tree-optimization/80426
33507 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
33508 operation on symbolic operands, also compute the overflow for the
33509 invariant part when the operation degenerates into a negation.
33511 2017-04-19 Jakub Jelinek <jakub@redhat.com>
33514 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
33515 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
33518 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
33520 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
33523 * config/avr/avr.c (tree.h): Include it.
33524 (cgraph.h): Include it.
33525 (avr_encode_section_info): Don't warn for uninitialized progmem
33526 variable if it's just an alias.
33528 2017-04-19 Richard Biener <rguenther@suse.de>
33531 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
33532 when needed by AutoPGO.
33534 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
33537 * doc/lto.texi: Remove an extra 'that'.
33539 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
33541 PR rtl-optimization/80429
33542 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
33543 are only used in debug insns.
33545 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33546 Vladimir Makarov <vmakarov@redhat.com>
33548 * config/sparc/predicates.md (input_operand): Add comment. Return
33549 true for any memory operand when LRA is in progress.
33550 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
33552 2017-04-18 Jeff Law <law@redhat.com>
33555 * mips.md ({return,simple_return}_internal): Do not overwrite
33558 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33560 PR tree-optimization/80443
33561 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
33562 instead of adding 1, subtract -1 and similarly instead of subtracting
33565 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
33567 PR rtl-optimization/80357
33568 * haifa-sched.c (tmp_bitmap): New variable.
33569 (model_recompute): Handle duplicate use records.
33570 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
33571 (free_global_sched_pressure_data): Free it.
33573 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
33576 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33577 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33578 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33579 instead of SYSTEM_HEADER_DIR.
33581 2017-04-18 Jeff Law <law@redhat.com>
33583 PR middle-end/80422
33584 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
33585 predecessors after walking up the insn chain.
33587 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33590 * dwarf2out.c (modified_type_die): Try harder not to emit internal
33591 sizetype type into debug info.
33593 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33596 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
33597 unneeded test for TARGET_UPPER_REGS_SF.
33598 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
33600 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33603 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
33604 instead of gsi_after_labels.
33606 2017-04-18 Jeff Law <law@redhat.com>
33608 * regcprop.c (maybe_mode_change): Avoid creating copies of the
33612 2017-04-13 Jeff Law <law@redhat.com>
33613 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33614 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33616 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33619 * config/avr/avr.c (intl.h): Include it.
33620 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33622 2017-04-18 Martin Liska <mliska@suse.cz>
33624 PR gcov-profile/78783
33625 * gcov-tool.c (gcov_output_files): Validate that destination
33626 file is either removed by the tool or by a user.
33628 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33629 Guy Benyei <guybe@mellanox.com>
33631 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33632 block, and do not negate it, the stored id is already negative.
33634 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33636 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33638 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33641 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33642 masks of options that should be turned off if the VSX vector
33643 options are turned off.
33644 (OTHER_P8_VECTOR_MASKS): Likewise.
33645 (OTHER_VSX_VECTOR_MASKS): Likewise.
33646 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33647 rs6000_disable_incompatible_switches to validate no type switches
33649 (rs6000_incompatible_switch): New function to disallow turning on
33650 other vector options if -mno-vsx, -mno-power8-vector, or
33651 -mno-power9-vector are specified.
33653 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33655 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33657 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33659 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33660 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33661 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33662 (ARG_POINTER_CFA_OFFSET): Likewise.
33664 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33666 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33667 conditions to take advantage of various optimizations.
33669 2017-04-13 Jeff Law <law@redhat.com>
33671 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33672 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33673 (zero_extendsidi2_dext): Likewise.
33675 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33678 * fold-const.c (fold_ternary_loc): Revert
33679 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33682 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33684 PR rtl-optimization/80343
33685 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33686 new scratch pseudo.
33688 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33691 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33692 to ubsan_encode_value.
33694 2017-04-13 Jeff Law <law@redhat.com>
33696 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33697 appearing in DEBUG_INSNs.
33699 2017-04-13 Martin Liska <mliska@suse.cz>
33701 PR gcov-profile/80413
33702 * gcov-io.c (gcov_write_string): Copy to buffer just when
33703 allocated size is greater than zero.
33705 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33708 * dwarf2out.c (decls_for_scope): Ignore declarations of
33709 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33711 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33714 * ipa-visibility.c (non_local_p): Fix typos.
33715 (localize_node): When localizing symbol in same comdat group,
33716 dissolve the group only when we know external symbols are going
33718 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33720 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33722 PR tree-optimization/79390
33723 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33724 order does not result in usable sequence, retry with reversed operand
33730 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33731 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33732 op0 instead of fold_convert_loc (loc, type, arg0).
33734 2017-04-12 Jeff Law <law@redhat.com>
33736 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33737 has a delay slot in the generated code.
33739 * config/cris/cris.md (cris_preferred_reload_class): Return
33740 GENNONACR_REGS rather than GENERAL_REGS.
33742 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33745 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33746 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33747 signedness of the result type.
33749 2017-04-12 Richard Biener <rguenther@suse.de>
33750 Jeff Law <law@redhat.com>
33752 PR tree-optimization/80359
33753 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33754 trim stores to TARGET_MEM_REFs.
33756 2017-04-12 Richard Biener <rguenther@suse.de>
33758 PR tree-optimization/79390
33759 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33760 threading case even more.
33762 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33765 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33766 for quad_address_p for TImode, instead of just not indexed_address.
33768 2017-04-12 Richard Biener <rguenther@suse.de>
33769 Bernd Edlinger <bernd.edlinger@hotmail.de>
33771 PR middle-end/79671
33772 * alias.c (component_uses_parent_alias_set_from): Handle
33773 TYPE_TYPELESS_STORAGE.
33774 (get_alias_set): Likewise.
33775 * tree-core.h (tree_type_common): Add typeless_storage flag.
33776 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33777 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33778 for types containing members with TYPE_TYPELESS_STORAGE.
33779 (place_field): Likewise.
33780 (layout_type): Likewise for ARRAY_TYPE.
33781 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33782 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33783 TYPE_TYPELESS_STORAGE.
33784 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33786 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33789 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33790 first argument to type.
33792 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33796 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33797 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33798 (rs6000_expand_binop_builtin): Likewise.
33799 (rs6000_expand_ternop_builtin): Likewise; also add missing
33800 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33801 vshasigma built-ins.
33802 * doc/extend.texi: Document that vec_xxpermdi's third argument
33803 must be a constant.
33805 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33807 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33808 Use shift_const cost parameter when calculating gain of STV shifts.
33810 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33812 PR rtl-optimization/70478
33813 * lra-constraints.c (process_alt_operands): Check memory for
33814 disfavoring memory insn operand.
33816 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33818 PR middle-end/80100
33819 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33820 left shift in unsigned HOST_WIDE_INT type.
33822 PR rtl-optimization/80385
33823 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33824 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33827 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33828 if they have any depend clauses.
33830 2017-04-11 Martin Liska <mliska@suse.cz>
33833 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33834 * ipa-split.c (split_function): Create a local comdat symbol
33835 if caller is in a comdat group.
33837 2017-04-11 Martin Liska <mliska@suse.cz>
33840 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33843 2017-04-11 Martin Sebor <msebor@redhat.com>
33845 PR middle-end/80364
33846 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33847 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33849 (directive::set_width, directive::set_precision, format_character):
33851 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33854 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33857 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33858 conflict, set target->arch_name instead of target->cpu_name.
33860 2017-04-11 Richard Biener <rguenther@suse.de>
33862 PR tree-optimization/80374
33863 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
33864 build_zero_cst, remove fold_convertible_p check again.
33866 2017-04-11 Martin Liska <mliska@suse.cz>
33869 * ubsan.c (instrument_object_size): Do not instrument register
33872 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33875 * config/i386/i386-builtin-types.def
33876 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
33877 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
33878 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
33879 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
33880 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
33881 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
33882 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
33883 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
33884 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
33885 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
33886 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
33887 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
33888 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
33889 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
33890 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
33891 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
33892 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
33893 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
33894 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
33895 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
33896 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
33897 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
33898 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
33899 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
33900 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
33901 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
33902 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
33903 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
33904 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
33905 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
33906 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
33907 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
33908 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
33909 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
33910 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
33911 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
33912 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
33913 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
33914 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
33915 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
33916 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
33917 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
33918 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
33919 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
33920 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
33921 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
33923 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
33924 flag to second_arg_count, handle 4 argument function type _COUNT
33925 aliases, handle second_arg_count on second argument rather than last.
33927 2017-04-10 Jeff Law <law@redhat.com>
33929 PR tree-optimization/80374
33930 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
33931 record anything if we can not convert integer_zero_node to the
33934 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33937 * config/rs6000/rs6000.c (rs6000_option_override_internal):
33938 Enhance special handling given to the TARGET_P9_MINMAX option in
33939 relation to certain other options.
33941 2017-04-10 Bin Cheng <bin.cheng@arm.com>
33943 PR tree-optimization/80153
33944 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
33945 remove POINTER_PLUS_EXPR's base part directly, rather than through
33948 2017-04-10 Richard Biener <rguenther@suse.de>
33949 Bin Cheng <bin.cheng@arm.com>
33951 PR tree-optimization/80153
33952 * tree-affine.c (aff_combination_to_tree): Get base pointer from
33953 the first element of pointer type aff_tree. Build result expr in
33955 (add_elt_to_tree): Convert to type unconditionally. Remove other
33956 fold_convert calls.
33957 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
33958 (rewrite_use_nonlinear_expr): Check invariant using iv information.
33960 2017-04-10 Richard Biener <rguenther@suse.de>
33962 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33965 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
33967 PR rtl-optimization/70478
33968 * lra-constraints.c (curr_small_class_check): New.
33969 (update_and_check_small_class_inputs): New.
33970 (process_alt_operands): Update curr_small_class_check. Disfavor
33971 alternative insn memory operands. Check available regs for small
33974 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33977 * config/mips/mips.opt (-mvirt): Update description.
33978 * doc/invoke.texi (-mvirt): Likewise.
33980 2017-04-10 Richard Biener <rguenther@suse.de>
33982 PR middle-end/80362
33983 * fold-const.c (fold_binary_loc): Look at unstripped ops when
33984 looking for NEGATE_EXPR in -A / -B to A / B folding.
33986 2017-04-10 Martin Liska <mliska@suse.cz>
33988 PR gcov-profile/80224
33989 * gcov.c (print_usage): Fix usage string.
33990 (get_gcov_intermediate_filename): Remove.
33991 (output_gcov_file): Use both for normal and intermediate format.
33992 (generate_results): Do not initialize special file for
33993 intermediate format.
33995 2017-04-10 Richard Biener <rguenther@suse.de>
33997 PR tree-optimization/80304
33998 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34001 2017-04-10 Nathan Sidwell <nathan@acm.org>
34004 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34005 (rs6000_init_builtins): Use it.
34007 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34009 * config/arm/arm.md (<mrc>): Add mode to SET source.
34010 (<mrrc>): Likewise.
34012 2017-04-10 Richard Biener <rguenther@suse.de>
34014 PR middle-end/80344
34015 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34017 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34020 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34021 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34022 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34023 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34024 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34025 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34026 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34027 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34028 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34029 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34030 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34031 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34032 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34033 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34034 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34035 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34036 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34037 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
34038 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
34039 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
34040 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34041 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34042 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34044 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34046 PR rtl-optimization/70478
34047 * lra-constraints.c: Reverse the last patch.
34049 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34051 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34052 Add comment for WCHAR_T.
34054 2017-04-08 Martin Liska <mliska@suse.cz>
34057 2017-04-07 Martin Liska <mliska@suse.cz>
34060 * ipa-split.c (split_function): Add function part to a same comdat
34063 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34066 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34068 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34070 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34071 (vec_pairing): ...to this.
34072 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34073 (rs6000_sched_init): Adjust for name change.
34074 (struct rs6000_sched_context): Likewise.
34075 (rs6000_init_sched_context): Likewise.
34076 (rs6000_set_sched_context): Likewise.
34078 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34084 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34086 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34087 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34088 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34090 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34092 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34094 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34096 PR rtl-optimization/70703
34097 * ira-color.c (update_conflict_hard_regno_costs): Use
34098 int64_t instead of HOST_WIDE_INT.
34100 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34102 PR rtl-optimization/70478
34103 * lra-constraints.c (process_alt_operands): Disfavor alternative
34104 insn memory operands.
34106 2017-04-07 Jeff Law <law@redhat.com>
34108 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
34109 CALL and NOTE_INSN_CALL_ARG_LOCATION.
34111 2017-04-07 Martin Liska <mliska@suse.cz>
34114 * config/aarch64/aarch64.c (aarch64_process_target_attr):
34115 Show error message instead of an ICE.
34117 2017-04-07 Martin Liska <mliska@suse.cz>
34120 * ipa-split.c (split_function): Add function part to a same comdat
34123 2017-04-07 Richard Biener <rguenther@suse.de>
34125 PR middle-end/80341
34126 * tree.c (get_unwidened): Also handle ! for_type case for
34128 * convert.c (do_narrow): Split out from ...
34129 (convert_to_integer_1): ... here. Do not pass final truncation
34130 type to get_unwidened for TRUNC_DIV_EXPR.
34132 2017-04-07 Richard Biener <rguenther@suse.de>
34134 * tree-affine.c (wide_int_ext_for_comb): Take type rather
34136 (aff_combination_const): Adjust.
34137 (aff_combination_scale): Likewise.
34138 (aff_combination_add_elt): Likewise.
34139 (aff_combination_add_cst): Likewise.
34140 (aff_combination_convert): Likewise.
34141 (add_elt_to_tree): Likewise. Remove unused argument.
34142 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
34144 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
34146 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
34148 * config/arm/arm.c (arm_default_short_enums): Use
34149 ARM_DEFAULT_SHORT_ENUMS.
34150 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
34152 2017-04-06 Jakub Jelinek <jakub@redhat.com>
34155 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
34156 members with redundant out-of-class redeclaration.
34158 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34161 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
34162 * config/i386/i386.md (*zero_extendsidi2):
34163 Add (?*x,*x) and (?*v,*v) alternatives.
34165 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34168 * config/i386/i386.c (ix86_expand_builtin)
34169 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
34170 mode from insn data. Convert operands to insn operand mode.
34171 Copy operands that don't satisfy insn predicate to a register.
34173 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
34175 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
34178 2017-04-06 Richard Biener <rguenther@suse.de>
34180 PR tree-optimization/80334
34181 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
34182 preserve alignment of accesses.
34184 2017-04-06 Richard Biener <rguenther@suse.de>
34186 PR tree-optimization/80262
34187 * tree-sra.c (build_ref_for_offset): Preserve address-space
34189 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
34190 Drop useless address-space information on MEM_REF offsets.
34192 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
34194 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
34196 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34198 PR rtl-optimization/70703
34199 * ira-color.c (update_conflict_hard_regno_costs): Use
34200 HOST_WIDE_INT instead of long.
34202 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
34205 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
34206 not defined for x86_64 target. Add -mmmx target option when __SSE2__
34208 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
34209 for x86_64 target. Handle -m3dnowa option.
34211 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34213 PR rtl-optimization/70703
34214 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
34215 (update_conflict_hard_regno_costs): Use long instead of unsigned
34216 arithmetic for cost calculation.
34218 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34219 Bernd Edlinger <bernd.edlinger@hotmail.de>
34222 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
34225 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
34228 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
34229 ptr_mode with Pmode throughout.
34230 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
34231 into probe_stack_range and use DImode.
34233 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
34236 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
34237 call_eh_return is true.
34239 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34241 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
34242 Initialize last_match_fntype_index.
34244 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34247 * tree-nvr.c: Include internal-fn.h.
34248 (pass_return_slot::execute): Ignore internal calls without
34251 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34252 Richard Biener <rguenther@suse.de>
34255 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
34256 captures used multiple times, except for the last use.
34257 * generic-match-head.c: Include gimplify.h.
34259 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34261 PR tree-optimization/79390
34262 * target.h (struct noce_if_info): Declare.
34263 * targhooks.h (default_noce_conversion_profitable_p): Declare.
34264 * target.def (noce_conversion_profitable_p): New target hook.
34265 * ifcvt.h (struct noce_if_info): New type, moved from ...
34266 * ifcvt.c (struct noce_if_info): ... here.
34267 (noce_conversion_profitable_p): Renamed to ...
34268 (default_noce_conversion_profitable_p): ... this. No longer
34270 (noce_try_store_flag_constants, noce_try_addcc,
34271 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
34272 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
34273 instead of noce_conversion_profitable_p.
34274 * config/i386/i386.c: Include ifcvt.h.
34275 (ix86_option_override_internal): Don't override
34276 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
34277 (ix86_noce_conversion_profitable_p): New function.
34278 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
34279 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
34280 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
34281 * doc/tm.texi: Regenerated.
34283 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34285 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
34288 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
34291 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
34292 instructions for small multiply cores.
34294 2017-04-04 Jeff Law <law@redhat.com>
34296 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
34298 (mips_expand_vec_perm_const): Initialize elements in orig_perm
34299 that are not set by the loop over the elements.
34301 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34304 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
34305 int mode, convert_modes it to mode as unsigned, otherwise use
34306 lowpart_subreg to mode rather than SImode.
34307 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
34308 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
34309 Use DImode instead of SImode for the shift count operand.
34310 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
34313 2017-04-04 Richard Biener <rguenther@suse.de>
34315 PR middle-end/80281
34316 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
34317 arithmetic done for the negate or the plus. Simplify.
34318 (A - (-B) -> A + B): Likewise.
34319 * fold-const.c (split_tree): Make sure to not negate pointers.
34321 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
34323 PR rtl-optimization/60818
34324 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
34325 a compare of comparisons with the thing compared if this results
34326 in a different machine mode.
34328 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
34330 * alias.c (base_alias_check): Fix typo in comment.
34331 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
34332 * cgraphunit.c (symbol_table::compile): Likewise.
34333 * collect2.c (maybe_run_lto_and_relink): Likewise.
34334 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
34335 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
34336 * config/avr/avr.c (avr_map_op_t): Likewise.
34337 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
34338 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
34339 * config/epiphany/epiphany.md (movcc): Likewise.
34340 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
34341 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
34343 * config/mips/mips.c (mips_save_restore_reg): Likewise.
34344 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
34345 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
34346 * config/sh/sh.c (sh_rtx_costs): Likewise.
34347 * fold-const.c (fold_truth_andor): Likewise.
34348 * genautomata.c (collapse_flag): Likewise.
34349 * gengtype.h (struct type::u::s): Likewise.
34350 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
34351 * input.c (FORMAT_AMOUNT): Likewise.
34352 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
34353 (known_aggs_to_agg_replacement_list): Likewise.
34354 * ipa-inline-analysis.c: Likewise.
34355 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
34356 * ipa-polymorphic-call.c
34357 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
34358 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
34359 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
34361 * modulo-sched.c (apply_reg_moves): Likewise.
34362 * omp-expand.c (build_omp_regions_1): Likewise.
34363 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
34364 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
34365 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
34366 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
34367 * value-prof.c: Likewise.
34368 * var-tracking.c (val_reset): Likewise.
34370 2017-04-03 Richard Biener <rguenther@suse.de>
34372 PR tree-optimization/80275
34373 * fold-const.c (split_address_to_core_and_offset): Handle
34376 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
34378 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
34379 descriptors is at least equal to that of functions.
34381 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34383 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
34385 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34388 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
34389 (mov<IMOD4:mode>): New expander.
34390 (*mov<IMOD4:mode>_internal): New insn and split pattern.
34392 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
34394 PR rtl-optimization/79405
34395 * fwprop.c (propagations_left): New variable.
34396 (forward_propagate_into): Decrement it.
34397 (fwprop_init): Initialize it.
34398 (fw_prop): If the variable has reached zero, stop propagating.
34399 (fwprop_addr): Ditto.
34401 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34404 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
34405 a FUNCTION_DECL, pass it as decl instead of origin to
34408 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
34410 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
34413 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
34416 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
34417 TARGET_VSX_SMALL_INTEGER.
34419 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34421 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
34422 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
34424 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34426 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
34427 extraction from odd-numbered MSA register.
34429 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34431 PR middle-end/80173
34432 * expmed.c (store_bit_field_1): Don't attempt to create
34433 a word subreg out of hard registers wider than word if they
34434 have HARD_REGNO_NREGS of 1 for their mode.
34436 PR middle-end/80163
34437 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
34438 conversions to integer types wider than word and pointer.
34441 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
34442 (rtx_equal_for_cselib_p): Pass 0 to it.
34443 * cselib.c (cselib_hasher::equal): Likewise.
34444 (rtx_equal_for_cselib_1): Add depth argument. If depth
34445 is 128, don't look up VALUE locs and punt. Increment
34446 depth in recursive calls when walking VALUE locs.
34448 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
34450 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
34451 (make_gcov_file_name): Use the canonical path name for generating
34453 (read_line): Fix handling of files with ascii null bytes.
34455 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
34457 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
34458 to initialise a vector register instead
34459 of using a const_int.
34461 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34463 PR translation/80189
34464 * gimplify.c (omp_default_clause): Use %qs instead of %s in
34465 diagnostic messages.
34467 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
34470 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
34471 (dfp_diex_<mode>): Update mode of operand 1.
34472 * doc/extend.texi (dxex, dxexq): Document change to return type.
34473 (diex, diexq): Document change to argument type.
34475 2017-03-30 Martin Jambor <mjambor@suse.cz>
34478 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
34479 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
34480 it reflects the signature changes performed at the callee side.
34481 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
34482 to cgraph_build_function_type_skip_args.
34483 (build_function_decl_skip_args): Adjust call to the above function.
34485 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34488 * config/i386/sse.md
34489 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
34490 register as dest whenever it is a MEM not rtx_equal_p to the
34491 corresponding dup operand, and when forcing into reg move the
34492 reg into the memory afterwards.
34493 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
34494 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
34495 for the force_reg mode.
34496 (avx512vl_vextractf128<mode>): Use register as dest either
34497 always when a MEM, or when it is a MEM not rtx_equal_p to the
34498 corresponding dup operand, or even not when it is a CONST_VECTOR
34499 depending on the mode and lo vs. hi.
34500 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
34502 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
34503 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
34504 Likewise. Require that operands[2] is even.
34505 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
34506 Remove extraneous parens. Require that operands[2] is a multiple
34508 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
34509 operands[0] is a MEM if <mask_applied>, the predicates/constraints
34510 disallow memory then.
34512 2017-03-30 Richard Biener <rguenther@suse.de>
34514 PR tree-optimization/77498
34515 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
34516 to non-constants over backedges.
34518 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
34520 PR rtl-optimization/80233
34521 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
34522 as last_combined_insn. Do not test for BARRIER_P separately.
34524 2017-03-29 Andreas Schwab <schwab@suse.de>
34527 * calls.c (prepare_call_address): Convert funexp to Pmode before
34528 copying to temp reg.
34530 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34532 PR tree-optimization/80158
34533 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
34534 Handle possible future case of more than one alternate
34536 (replace_rhs_if_not_dup): Likewise.
34537 (replace_one_candidate): Likewise.
34539 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
34541 PR rtl-optimization/80193
34542 * ira.c (ira): Do not check allocation for LRA.
34544 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
34546 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
34547 (nvptx_output_simt_exit): Declare.
34548 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
34549 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
34550 (init_softstack_frame): Move initialization of crtl->is_leaf to...
34551 (nvptx_declare_function_name): ...here. Emit declaration of local
34552 memory space buffer for omp_simt_enter insn.
34553 (nvptx_output_unisimt_switch): New.
34554 (nvptx_output_softstack_switch): New.
34555 (nvptx_output_simt_enter): New.
34556 (nvptx_output_simt_exit): New.
34557 * config/nvptx/nvptx.h (struct machine_function): New fields
34558 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
34559 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
34560 (UNSPECV_SIMT_EXIT): Ditto.
34561 (omp_simt_enter_insn): New insn.
34562 (omp_simt_enter): New expansion.
34563 (omp_simt_exit): New insn.
34564 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
34566 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
34567 (expand_GOMP_SIMT_ENTER_ALLOC): New.
34568 (expand_GOMP_SIMT_EXIT): New.
34569 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
34570 (GOMP_SIMT_ENTER_ALLOC): Ditto.
34571 (GOMP_SIMT_EXIT): Ditto.
34572 * target-insns.def (omp_simt_enter): New insn.
34573 (omp_simt_exit): Ditto.
34574 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
34576 (lower_rec_simd_input_clauses): Implement SIMT privatization.
34577 (lower_rec_input_clauses): Likewise.
34578 (lower_lastprivate_clauses): Handle SIMT privatization.
34580 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
34581 (ompdevlow_adjust_simt_enter): New.
34582 (find_simtpriv_var_op): New.
34583 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
34584 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
34586 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
34587 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
34588 (copy_decl_for_dup_finish): Ditto.
34590 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
34592 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
34595 * config/i386/i386.c (ix86_option_override_internal): Always
34596 allow -mpreferred-stack-boundary=3 for 64-bit targets.
34598 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34600 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
34602 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34604 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
34605 mark new edge's irreducible flag accordign to it.
34606 (vect_do_peeling): Check loop preheader edge's irreducible flag
34607 and pass it to function slpeel_add_loop_guard.
34609 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
34611 PR tree-optimization/80218
34612 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34613 Update block frequencies and counts.
34615 2017-03-28 Richard Biener <rguenther@suse.de>
34617 PR tree-optimization/78644
34618 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34619 of a simplification result we may not use it at all.
34621 2017-03-28 Richard Biener <rguenther@suse.de>
34624 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34625 without arguments, generate default definition of a SSA name.
34627 2017-03-28 Richard Biener <rguenther@suse.de>
34629 PR middle-end/80222
34630 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34631 TYPE_REF_CAN_ALIAS_ALL references.
34632 * fold-const.c (fold_indirect_ref_1): Likewise.
34634 2017-03-28 Martin Liska <mliska@suse.cz>
34637 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34638 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34640 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34641 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34643 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34644 (EXTRA_SPECS): Define.
34645 (SUBTARGET_EXTRA_SPECS): Likewise.
34646 (SUBTARGET_CPP_SPEC): Likewise.
34647 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34648 SUBTARGET_EXTRA_SPECS.
34649 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34651 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34653 * config/arc/simdext.md (vst64_insn): Update pattern.
34654 (vld32wh_insn): Likewise.
34655 (vld32wl_insn): Likewise.
34656 (vld64_insn): Likewise.
34657 (vld32_insn): Likewise.
34659 2017-03-28 Marek Polacek <polacek@redhat.com>
34662 * fold-const.c (fold_comparison): Use protected_set_expr_location
34663 instead of SET_EXPR_LOCATION.
34665 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34667 * tree.c (add_expr): Avoid name lookup warning.
34669 2017-03-27 Jeff Law <law@redhat.com>
34671 PR tree-optimization/80216
34672 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34673 function name. Limit recursion depth.
34674 (record_temporary_equivalences): Corresponding changes.
34676 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34678 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34681 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34684 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34686 * cfgcleanup.c (reg_note_cfa_p): New array.
34687 (insns_have_identical_cfa_notes): New function.
34688 (old_insns_match_p): Don't cross-jump in between /f
34689 and non-/f instructions. If both i1 and i2 are frame related,
34690 verify all CFA notes, their order and content.
34692 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34695 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34696 HImode and SImode with zero extend to DImode to one insn.
34697 (bswap<mode>2_extenddi): Likewise.
34698 (bswapsi2_extenddi): Likewise.
34699 (bswaphi2_extendsi): Likewise.
34700 (bswaphi2): Combine bswap HImode and SImode into one insn.
34701 Separate memory insns from swapping register.
34702 (bswapsi2): Likewise.
34703 (bswap<mode>2): Likewise.
34704 (bswaphi2_internal): Delete, no longer used.
34705 (bswapsi2_internal): Likewise.
34706 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34707 store, and gpr<-gpr swap insns.
34708 (bswap<mode>2_store): Likewise.
34709 (bswaphi2_reg): Register only splitter, combine with the splitter.
34710 (bswaphi2 splitter): Likewise.
34711 (bswapsi2_reg): Likewise.
34712 (bswapsi2 splitter): Likewise.
34713 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34714 the insns into load, store, and register/register insns.
34715 (bswapdi2_ldbrx): Likewise.
34716 (bswapdi2_load): Likewise.
34717 (bswapdi2_store): Likewise.
34718 (bswapdi2_reg): Likewise.
34720 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34722 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34723 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34725 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34728 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34730 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34731 special handling for target option conflicts between dform
34732 options (-mpower9-dform, -mpower9-dform-vector,
34733 -mpower9-dform-scalar) and -mno-direct-move.
34735 2017-03-27 Richard Biener <rguenther@suse.de>
34737 PR tree-optimization/80181
34738 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34740 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34742 * config/arc/predicates.md (move_double_src_operand): Replace the
34743 call to move_double_src_operand with a call to address_operand.
34745 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34747 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34748 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34749 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34751 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34753 * config/arc/predicates.md (long_immediate_loadstore_operand):
34754 Consider scaled addresses cases.
34756 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34758 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34759 restored when in interrupt.
34760 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34761 doesn't have delay slot.
34763 2017-03-27 Richard Biener <rguenther@suse.de>
34766 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34767 inlined thunk clones.
34769 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34772 * asan.c (instrument_derefs): Copy over last operand from
34773 original COMPONENT_REF to the new COMPONENT_REF with
34774 DECL_BIT_FIELD_REPRESENTATIVE.
34775 * ubsan.c (instrument_object_size): Likewise.
34777 2017-03-27 Richard Biener <rguenther@suse.de>
34779 PR tree-optimization/80170
34780 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34781 sure DR/SCEV didnt fold in constants we do not see when looking
34782 at the reference base alignment.
34784 2017-03-27 Richard Biener <rguenther@suse.de>
34786 PR middle-end/80171
34787 * gimple-fold.c (fold_ctor_reference): Properly guard against
34788 NULL return value from canonicalize_constructor_val.
34790 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34793 * config/i386/i386.c (ix86_expand_builtin)
34794 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34795 flags reg setting and flags reg using instructions.
34796 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34797 clobbering instructions to zero extend op2.
34799 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34801 * doc/install.texi (Configuration) <--with-aix-soname>:
34802 Update link to AIX ld.
34804 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34806 PR rtl-optimization/80160
34807 PR rtl-optimization/80159
34808 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34809 reg_alternate_class into account.
34811 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34814 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34815 to consider in curr_insn_transform.
34817 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34819 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34820 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34821 and emit_mode_inner.
34823 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34825 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34826 argument to the overloaded builtin variants. Use the new flag to
34827 deprecate certain builtin variants.
34828 * config/s390/s390-builtin-types.def: Add new builtin types.
34829 * config/s390/s390-builtins.h: Support new flags field for
34830 overloaded builtins.
34831 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34832 (s390_macro_to_expand): Enable vector float data type.
34833 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34834 builtins by incrementing the __VEC__ version number.
34835 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34837 (s390_resolve_overloaded_builtin): Emit error messages depending
34838 on the builtin flags.
34839 * config/s390/s390.c (s390_expand_builtin): Support additional
34840 flags argument. Change error message to match the messages
34841 emitted in s390-c.c.
34842 * config/s390/s390.md: New UNSPEC_* constants.
34843 (op_type): Add new instruction types.
34844 * config/s390/vecintrin.h: Add new builtins and test data class
34846 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34847 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34848 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34849 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34850 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34851 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34853 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34854 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34855 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34856 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34858 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34859 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34860 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34861 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
34863 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
34864 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
34865 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
34866 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
34867 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
34868 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
34869 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
34871 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
34872 ("vec_scatter_element<V_HW_4:mode>_DI")
34873 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
34874 ("vec_fpint<mode>", "vflls")
34875 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
34876 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
34877 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
34878 ("*vec_cmphe<mode>_cc"): ... these.
34880 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
34881 mode constant instead of magic value.
34883 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34885 * config/s390/s390.c (s390_expand_vec_compare): Support other
34886 vector floating point modes than just V2DF.
34887 (s390_expand_vcond): Likewise.
34888 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
34889 (s390_cannot_change_mode_class): Prevent mode changes between TF
34890 and V1TF in vector registers.
34891 * config/s390/s390.md (DF, SF): New mode attributes.
34892 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
34893 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
34894 SFmode support for VRs.
34895 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
34897 (VFT, VF_HW): New mode iterators.
34898 (vw, sdx): New mode attributes.
34899 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
34900 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
34901 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
34902 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
34903 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
34904 also the new vector floating point modes. Renaming to ...
34906 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
34907 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
34908 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
34909 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
34910 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
34911 ("vec_unordered<mode>"): ... these.
34913 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
34914 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
34915 ("*vec_extendv2df"): New insn definitions.
34917 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34919 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
34920 ("mulditi3_2", "*muldi3_sign"): New patterns.
34921 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
34922 rename the pattern definition.
34924 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34926 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
34928 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
34930 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34932 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
34933 instruction if possible.
34934 * config/s390/vector.md (vec_halfnumelts): New mode
34936 ("*vec_vllezlf<mode>"): New pattern.
34938 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34940 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
34941 ("popcountv4si2", "popcountv2di2"): Rename to ...
34942 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
34943 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
34945 ("popcount<mode>2_vxe"): New pattern.
34947 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34949 * common/config/s390/s390-common.c (processor_flags_table): Add
34951 * config.gcc: Add arch12.
34952 * config/s390/driver-native.c (s390_host_detect_local_cpu):
34953 Default to arch12 for unknown CPU model numbers.
34954 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
34955 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
34956 PROCESSOR_max sanity check.
34957 * config/s390/s390-opts.h (enum processor_type): Add
34959 * config/s390/s390.c (processor_table): Add arch12.
34960 (s390_expand_builtin): Add check for B_VXE flag.
34961 (s390_issue_rate): Add PROCESSOR_ARCH12.
34962 (s390_get_sched_attrmask): Likewise.
34963 (s390_get_unit_mask): Likewise.
34964 (s390_sched_score): Enable z13 scheduling for arch12.
34965 (s390_sched_reorder): Likewise.
34966 (s390_sched_variable_issue): Likewise.
34967 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
34969 (s390_tune_attr): Use z13 scheduling also for arch12.
34970 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
34971 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
34972 (TARGET_VXE_P): New macros.
34973 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
34974 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
34975 * config/s390/s390.opt: Add arch12 as processor_type.
34977 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34979 * config/s390/s390.md
34980 ("fixuns_truncdddi2", "fixuns_trunctddi2")
34981 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
34982 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
34984 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
34985 Rename expanders to ...
34987 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
34988 ("fixuns_truncdddi2_emu"): ... these.
34990 ("fixuns_trunc<mode>si2_emu"): New expander.
34992 ("*fixuns_truncdfdi2_z13"): Rename to ...
34993 ("*fixuns_truncdfdi2_vx"): ... this.
34995 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34997 * config/s390/2964.md: Remove the single element vector compare
34998 instructions which are no longer used.
34999 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35001 (s390_canonicalize_comparison): Remove handling of DFmode
35003 (s390_expand_vec_compare_scalar): Remove function.
35004 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35005 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35007 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35009 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35011 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35013 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35014 will anyway by matched by mov<mode>_64dfp.
35016 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35018 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35019 vlef/vstef. Add missing operand to vleif.
35021 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35023 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35024 pair for all vector types with 64 bit elements.
35025 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35026 * config/s390/vector.md (V_HW_64): ... here.
35027 (V_128_NOSINGLE): New mode iterator.
35028 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35029 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35030 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35031 ("*vec_load_pairv2di"): Change to ...
35032 ("*vec_load_pair<mode>"): ... this one.
35034 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35036 * config/s390/constraints.md: Add comments.
35037 (jKK): Reject element sizes > 8 bytes.
35038 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
35040 * config/s390/s390.md: Add the s_operand checks formerly in
35041 s390_split_ok_p to various splitters where they are still
35043 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35044 for 128 bit vectors. Plus two splitters.
35046 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35048 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35051 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35054 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35055 error if the boundary argument is not constant.
35057 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35059 PR rtl-optimization/80112
35060 * loop-doloop.c (doloop_condition_get): Don't check condition
35061 if cmp isn't SET with IF_THEN_ELSE src.
35063 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35065 PR tree-optimization/80158
35066 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35067 replacing a candidate statement, also replace it for the
35068 candidate's alternate interpretation.
35069 (replace_rhs_if_not_dup): Likewise.
35070 (replace_one_candidate): Likewise.
35072 2017-03-24 Richard Biener <rguenther@suse.de>
35074 PR tree-optimization/80167
35075 * graphite-isl-ast-to-gimple.c
35076 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35078 (translate_isl_ast_to_gimple::get_rename): Likewise.
35080 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35082 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35083 handling of certain combinations of target options, including the
35084 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35085 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35087 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35090 * config/arm/arm.md (*load_multiple): Add reload_completed to
35091 matching condition.
35093 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35094 Richard Biener <rguenth@suse.de>
35096 PR tree-optimization/79908
35097 PR tree-optimization/80136
35098 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35099 been cast away, gimplify_and_add suffices.
35101 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35103 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35105 2017-03-23 Richard Biener <rguenther@suse.de>
35107 PR tree-optimization/80032
35108 * gimplify.c (gimple_push_cleanup): Forced unconditional
35109 cleanups still have to go to the conditional_cleanups
35112 2017-03-22 Jakub Jelinek <jakub@redhat.com>
35114 PR tree-optimization/80072
35115 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
35117 (next_operand_entry_id): Change type to unsigned int.
35118 (sort_by_operand_rank): Make sure to return the right return value
35119 even if unsigned fields are bigger than INT_MAX.
35120 (struct oecount): Change cnt and id type to unsigned int.
35121 (oecount_hasher::equal): Formatting fix.
35122 (oecount_cmp): Make sure to return the right return value
35123 even if unsigned fields are bigger than INT_MAX.
35124 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
35127 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
35128 TREE_READONLY on result if writing it more than once.
35131 * doc/invoke.texi (-fsanitize=thread): Document that with
35132 -fnon-call-exceptions atomics are not able to throw
35136 * tsan.c: Include tree-eh.h.
35137 (instrument_builtin_call): Call maybe_clean_eh_stmt or
35138 maybe_clean_or_replace_eh_stmt where needed.
35139 (instrument_memory_accesses): Add cfg_changed argument.
35140 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
35141 if it returned true.
35142 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
35144 PR rtl-optimization/63191
35145 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
35146 wrapper function, moved the whole old content into ...
35147 (ix86_delegitimize_address_1): ... this. New inline function.
35148 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
35149 true as last argument instead of ix86_delegitimize_address.
35151 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35153 * config/aarch64/aarch64.c (generic_branch_cost): Copy
35154 cortexa57_branch_cost.
35156 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35158 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
35160 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35163 * doc/md.texi (Constraints): Document wA constraint.
35164 * config/rs6000/constraints.md (wA): New.
35165 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
35166 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
35167 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
35168 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
35170 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
35173 * gimplify.c (is_oacc_declared): New function.
35174 (oacc_default_clause): Use it to set default flags for acc declared
35175 variables inside parallel regions.
35176 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
35177 declared variables.
35178 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
35179 declare attribute to any decl as necessary.
35181 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
35184 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
35185 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
35186 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
35187 (arm_arch_lpae): This.
35188 * config/arm/arm.c (arm_arch7ve): Rename into ...
35189 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
35190 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
35193 2017-03-22 Martin Liska <mliska@suse.cz>
35196 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
35197 error message instead of an ICE.
35199 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35201 * doc/extend.texi (6.11 Additional Floating Types): Revise.
35203 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
35205 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35207 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35210 2017-03-21 Martin Sebor <msebor@redhat.com>
35212 * doc/extend.texi: Use "cannot" instead of "can't."
35213 * doc/hostconfig.texi: Same.
35214 * doc/install.texi: Same.
35215 * doc/invoke.texi: Same.
35216 * doc/loop.texi: Same.
35217 * doc/md.texi: Same.
35218 * doc/objc.texi: Same.
35219 * doc/rtl.texi: Same.
35220 * doc/tm.texi: Same.
35221 * doc/tm.texi.in: Same.
35222 * doc/trouble.texi: Same.
35224 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
35227 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
35228 (collect_checksum_attributes): Set it.
35229 (die_checksum_ordered): Use it.
35231 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35233 PR tree-optimization/79908
35234 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
35235 change: For a VA_ARG whose LHS has been cast away, use
35236 force_gimple_operand to construct the side effects.
35238 2017-03-21 David Malcolm <dmalcolm@redhat.com>
35240 PR translation/80001
35241 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
35242 more amenable to translation.
35243 (oacc_loop_auto_partitions): Likewise.
35245 2017-03-21 Marek Polacek <polacek@redhat.com>
35246 Martin Sebor <msebor@redhat.com>
35248 PR tree-optimization/80109
35249 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
35250 on INTEGRAL_TYPE_P.
35252 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35253 Segher Boessenkool <segher@kernel.crashing.org>
35256 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
35257 check reg_used_between_p between insn and one of succ or succ2
35258 depending on if succ is artificial insn not inserted into insn
35261 2017-03-21 Martin Liska <mliska@suse.cz>
35263 PR gcov-profile/80081
35264 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
35265 * doc/gcc.texi: Include gcov-dump stuff.
35266 * doc/gcov-dump.texi: New file.
35268 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
35270 PR rtl-optimization/79150
35271 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
35272 conditional jump, if the jump is the last insn of the loop.
35274 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35275 Richard Biener <rguenth@suse.de>
35277 PR tree-optimization/79908
35278 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35279 been cast away, use force_gimple_operand to construct the side
35282 2017-03-21 Martin Liska <mliska@suse.cz>
35284 PR libfortran/79956
35285 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
35288 2017-03-21 Brad Spengler <spender@grsecurity.net>
35291 * plugin.c (htab_hash_plugin): New function.
35292 (add_new_plugin): Use it and adjust.
35293 (parse_plugin_arg_opt): Adjust.
35294 (init_one_plugin): Likewise.
35296 2017-03-21 Richard Biener <rguenther@suse.de>
35298 PR tree-optimization/80032
35299 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
35300 if set force the cleanup to happen unconditionally.
35301 (gimplify_target_expr): Push inserted clobbers with force_uncond
35302 to avoid them being removed by control-dependent DCE.
35304 2017-03-21 Richard Biener <rguenther@suse.de>
35306 PR tree-optimization/80122
35307 * tree-inline.c (copy_bb): Do not expans va-arg packs or
35308 va_arg_pack_len when the inlined call stmt requires pack
35310 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
35312 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35315 * tsan.c (instrument_builtin_call): If the memory model argument
35316 is not a constant, assume it is valid.
35319 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
35322 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35324 PR rtl-optimization/79910
35325 * combine.c (can_combine_p): Do not allow combining an I0 or I1
35326 if its dest is used by an insn before I2 (other than the combined
35327 insns themselves, which are properly handled already).
35329 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35332 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35334 * combine.c (record_used_regs): New static function.
35335 (try_combine): Handle situations where there is an additional
35336 instruction between I2 and I3 which needs to have a LOG_LINK
35340 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35342 * combine.c (try_combine): Delete redundant i1 test. Call
35343 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35345 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35348 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
35349 alternatives 13/14.
35351 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35353 PR tree-optimization/80054
35354 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
35355 the optimization if a PHI or any of its arguments is not dominated
35356 by the candidate's basis. Use gphi* rather than gimple* as
35358 (replace_profitable_candidates): Clean up a gimple* variable that
35359 should be a gphi* variable.
35361 2017-03-20 Martin Sebor <msebor@redhat.com>
35364 * doc/extend.texi (attribute constructor): Document present limitation.
35366 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
35369 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
35370 __POWER9_VECTOR__ #ifdef control, change template definition to
35371 use Power9-specific built-in function.
35372 (vec_any_eq): Likewise.
35373 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
35374 to control outcomes from this test.
35375 (vector_ae_<mode>p): For VEC_F modes, likewise.
35377 2017-03-20 Ian Lance Taylor <iant@google.com>
35379 * config/i386/i386.c (ix86_function_regparm): Save an extra
35380 register for -fsplit-stack with DECL_STATIC_CHAIN.
35382 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35385 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
35386 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
35388 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35390 * config/riscv/riscv.c (riscv_print_operand): Use "fence
35392 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
35395 2017-03-20 Marek Polacek <polacek@redhat.com>
35398 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
35400 2017-03-20 Richard Biener <rguenther@suse.de>
35402 PR tree-optimization/80113
35403 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
35404 allocate extra SSA name for PHI def.
35405 (add_close_phis_to_outer_loops): Likewise.
35406 (add_close_phis_to_merge_points): Likewise.
35407 (copy_loop_close_phi_args): Likewise.
35408 (copy_cond_phi_nodes): Likewise.
35410 2017-03-20 Martin Liska <mliska@suse.cz>
35412 PR middle-end/79753
35413 * tree-chkp.c (chkp_build_returned_bound): Do not build
35414 returned bounds for a LHS that's not a BOUNDED_P type.
35416 2017-03-20 Martin Liska <mliska@suse.cz>
35420 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
35421 COMPLEX_CST and VECTOR_CST.
35423 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35426 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
35427 target operand. A new splitter adds the clobber statement in case
35428 the target operand is dead anyway.
35430 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
35432 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
35433 to age-old versions of binutils and glibc.
35435 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
35437 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
35439 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35441 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
35443 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35445 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
35446 requirement for binutils 2.13.
35448 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35450 * combine.c (try_combine): Delete redundant i1 test. Call
35451 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35453 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
35455 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
35456 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
35458 <riscv64-*-elf>: Re-arrange section
35459 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
35460 <riscv32-*-linux>: Likewise.
35461 <riscv64-*-elf>: Likewise
35462 <riscv64-*-linux>: Likewise.
35464 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
35467 * aarch64.opt(verbose-cost-dump): Fix typo.
35469 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
35472 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
35473 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
35475 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35477 * reload.c (find_reloads): When reloading a nonoffsettable address,
35478 use RELOAD_OTHER for it and its address reloads.
35480 PR rtl-optimization/79910
35481 * combine.c (record_used_regs): New static function.
35482 (try_combine): Handle situations where there is an additional
35483 instruction between I2 and I3 which needs to have a LOG_LINK
35486 2017-03-17 Jeff Law <law@redhat.com>
35488 PR tree-optimization/71437
35489 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
35490 conditional in the hash table first.
35491 (vrp_dom_walker::before_dom_children): Extract condition from
35492 ASSERT_EXPR. Record condition, its inverion and any implied
35493 conditions as well.
35495 2017-03-17 Marek Polacek <polacek@redhat.com>
35496 Markus Trippelsdorf <markus@trippelsdorf.de>
35498 PR tree-optimization/80079
35499 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
35502 2017-03-17 Richard Biener <rguenther@suse.de>
35504 PR middle-end/80075
35505 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
35506 Properly verify the LHS before the RHS possibly claims to be
35508 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
35511 2017-03-17 Martin Jambor <mjambor@suse.cz>
35513 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
35514 (List of -O2 options): Likewise.
35515 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
35518 2017-03-17 Tom de Vries <tom@codesourcery.com>
35520 * gcov-dump.c (print_usage): Print bug_report_url.
35522 2017-03-17 Richard Biener <rguenther@suse.de>
35524 PR middle-end/80050
35525 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
35526 (parser::peek): Likewise.
35528 2017-03-17 Richard Biener <rguenther@suse.de>
35530 PR tree-optimization/80048
35531 * sese.c (free_sese_info): Properly release rename_map and
35532 copied_bb_map elements.
35534 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
35536 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
35537 Add linked-list forward and backlinks. Insert on
35538 construction, remove on destruction.
35539 (class pass_store_merging): Add m_stores_head field.
35540 (pass_store_merging::terminate_and_process_all_chains):
35541 Iterate over m_stores_head list.
35542 (pass_store_merging::terminate_all_aliasing_chains):
35544 (pass_store_merging::execute): Check for debug stmts first.
35545 Push new chains onto the m_stores_head stack.
35547 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
35550 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
35551 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
35552 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
35554 2017-03-16 Jeff Law <law@redhat.com>
35556 PR tree-optimization/71437
35557 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
35558 member function. Implementation moved into after_dom_children
35559 member function and into the threader's thread_outgoing_edges
35561 (dom_opt_dom_walker::after_dom_children): Simplify by moving
35562 some code into new thread_outgoing_edges.
35563 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
35564 definition. Simplify marker handling (do it here). Assume we always
35565 have the available expression and the const/copies tables.
35566 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
35568 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
35569 * tree-vrp.c (equiv_stack): No longer file scoped.
35570 (vrp_dom_walker): New class.
35571 (vrp_dom_walker::before_dom_children): New member function.
35572 (vrp_dom_walker::after_dom_children): Likewise.
35573 (identify_jump_threads): Setup domwalker. Use it rather than
35574 walking edges in a random order by hand. Simplify setup/finalization.
35575 (finalize_jump_threads): Remove.
35576 (vrp_finalize): Do not call identify_jump_threads here.
35577 (execute_vrp): Do it here instead and call thread_through_all_blocks
35580 PR tree-optimization/71437
35581 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
35583 (simplify_stmt_for_jump_threading): Add basic_block argument. All
35585 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
35586 (dom_opt_dom_walker::thread_across_edge): Remove
35587 handle_dominating_asserts argument. All callers changed.
35588 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
35589 changes. Remove calls to lhs_of_dominating_assert. Other
35590 uses of handle_dominating_asserts turn into unconditional code
35591 (simplify_control_stmt_condition_1): Likewise.
35592 (simplify_control_stmt_condition): Likewise.
35593 (thread_through_normal_block, thread_across_edge): Likewise.
35594 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
35595 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
35596 object if it is not an SSA_NAME.
35597 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
35598 before calling into the VRP specific simplifiers.
35599 (identify_jump_threads): Remove handle_dominating_asserts
35602 2017-03-16 Jakub Jelinek <jakub@redhat.com>
35605 * tree-diagnostic.c (default_tree_printer): No longer static.
35606 * tree-diagnostic.h (default_tree_printer): New prototype.
35608 2017-03-16 Tamar Christina <tamar.christina@arm.com>
35610 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
35611 Change ins into fmov.
35613 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35615 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
35616 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
35617 Use h_con constraint for operand 1.
35618 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35619 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35621 2017-03-15 Jeff Law <law@redhat.com>
35623 PR tree-optimization/71437
35624 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35625 (record_temporary_equivalences): Use it.
35627 PR tree-optimization/71437
35628 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35629 tree-ssa-scopedtables.
35630 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35631 (record_conditions, record_cond, vuse_eq): Likewise.
35632 (record_edge_info): Adjust to API tweak of record_conditions.
35633 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35634 (record_temporary_equivalences, optimize_stmt): Likewise.
35635 (eliminate_redundant_computations): Likewise.
35636 (record_equivalences_from_stmt): Likewise.
35637 * tree-ssa-scopedtables.c: Include options.h and params.h.
35638 (vuse_eq): New function, moved from tree-ssa-dom.c
35639 (build_and_record_new_cond): Likewise.
35640 (record_conditions): Likewise. Accept vector of conditions rather
35641 than edge_equivalence structure for first argument.
35642 for the first argument.
35643 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35644 from tree-ssa-dom.c.
35645 (avail_exprs_stack::record_cond): Likewise.
35646 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35647 from tree-ssa-dom.c.
35648 (avail_exprs_stack): Add new member functions lookup_avail_expr
35650 (record_conditions): Declare.
35652 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35655 * lra-constraints.c (process_alt_operands): Increase reject for
35656 reloading an input/output operand.
35658 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35661 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35662 insns to convert from signed/unsigned char/short to IEEE 128-bit
35664 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35666 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35669 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35670 subreg of inner mode for values already in registers.
35672 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35674 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35675 iteration reg is used after the loop.
35677 2017-03-14 Martin Sebor <msebor@redhat.com>
35679 PR tree-optimization/79800
35680 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35681 precision in negative-positive range.
35682 (format_floating): Call non-const overload with adjusted precision.
35684 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35687 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35690 2017-03-14 Martin Sebor <msebor@redhat.com>
35692 PR middle-end/80020
35693 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35694 * builtins.def (aligned_alloc): Use it.
35697 * Makefile.in (GTFILES): Add calls.c.
35698 * calls.c: Include "gt-calls.h".
35700 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35702 PR rtl-optimization/79728
35703 * regs.h (struct target_regs): New field
35704 x_contains_allocatable_regs_of_mode.
35705 (contains_allocatable_regs_of_mode): New macro.
35706 * reginfo.c (init_reg_sets_1): Initialize it, and change
35707 contains_reg_of_mode so it includes global regs as well.
35708 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35709 rather than contains_regs_of_mode.
35711 2017-03-14 Martin Liska <mliska@suse.cz>
35713 * doc/invoke.texi: Document options that can't be combined with
35714 -fcheck-pointer-bounds.
35716 2017-03-14 Martin Liska <mliska@suse.cz>
35718 PR middle-end/79831
35719 * doc/invoke.texi (-Wchkp): Document the option.
35721 2017-03-14 Martin Liska <mliska@suse.cz>
35723 * Makefile.in: Install gcov-dump.
35725 2017-03-14 Martin Liska <mliska@suse.cz>
35727 * multiple_target.c (expand_target_clones): Bail out for
35728 an invalid attribute.
35730 2017-03-14 Richard Biener <rguenther@suse.de>
35732 * alias.c (struct alias_set_entry): Pack properly.
35733 * cfgloop.h (struct loop): Likewise.
35734 * cse.c (struct set): Likewise.
35735 * ipa-utils.c (struct searchc_env): Likewise.
35736 * loop-invariant.c (struct invariant): Likewise.
35737 * lra-remat.c (struct cand): Likewise.
35738 * recog.c (struct change_t): Likewise.
35739 * rtl.h (struct address_info): Likewise.
35740 * symbol-summary.h (function_summary): Likewise.
35741 * tree-loop-distribution.c (struct partition): Likewise.
35742 * tree-object-size.c (struct object_size_info): Likewise.
35743 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35744 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35745 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35746 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35747 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35748 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35749 (struct _stmt_vec_info): Likewise.
35751 2017-03-14 Martin Liska <mliska@suse.cz>
35754 * multiple_target.c (create_dispatcher_calls): Check that
35755 a target can create a function dispatcher.
35757 2017-03-14 Martin Liska <mliska@suse.cz>
35760 * multiple_target.c (expand_target_clones): Drop local.local
35761 flag for default implementation.
35763 2017-03-14 Richard Biener <rguenther@suse.de>
35765 PR tree-optimization/80030
35766 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35768 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35770 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35771 gcc_fallthrough() instead of __attribute__((fallthrough));
35773 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35775 * doc/gcc.texi: Remove "up" link to (DIR).
35776 * doc/gccint.texi: Ditto.
35778 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35780 * doc/install.texi (Specific) <avr>: Remove reference to
35783 2017-03-13 Jeff Law <law@redhat.com>
35785 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35786 attribute rather than comments.
35788 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35789 match_scratch operand is highest.
35791 2017-03-13 Martin Liska <mliska@suse.cz>
35793 PR middle-end/78339
35794 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35795 is a CHKP clone, use original declaration.
35797 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35799 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35800 (arc_conditional_register_usage): Use a different allocation order
35801 when optimizing for size.
35802 * common/config/arc/arc-common.c (arc_option_optimization_table):
35803 Section anchors default on when optimizing for size.
35805 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35807 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35809 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35811 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35812 * config/arc/arc.md (cpu_facility): Add cd variant.
35813 (*movqi_insn): Add code density variant.
35814 (*movhi_insn): Likewise.
35815 (*movqi_insn): Likewise.
35816 (*addsi3_mixed): Likewise.
35817 (subsi3_insn): Likewise.
35819 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35821 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35823 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35825 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35826 expressions with MINUS and UNARY ops.
35828 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35831 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35833 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35834 between vec_select and vector argument.
35835 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35836 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35837 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35838 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35839 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35840 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35842 2017-03-13 Richard Biener <rguenther@suse.de>
35845 * params.def (vect-max-peeling-for-alignment): Fix typo.
35847 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35849 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35850 issue that only occurred with binutils below 2.18.
35852 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35854 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35855 refer to binutils 2.11/2.12 minimum.
35857 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35859 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35860 ftp.kernel.org and simplify binutils requirement.
35862 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
35864 * doc/invoke.texi (Warning Options): Fix spelling of link-time
35866 (Optimize Options): Ditto. Also remove redundancy.
35868 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35870 PR translation/79848
35871 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
35873 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
35874 to G_ to avoid double translation.
35876 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35878 PR translation/79923
35879 * auto-profile.c (get_combined_location): Convert leading
35880 character of diagnostics to lower case and remove trailing period.
35881 (read_profile): Likewise for various diagnostics.
35882 * config/arm/arm.c (arm_option_override): Remove trailing period
35883 from various diagnostics.
35884 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
35885 (msp430_expand_delay_cycles): Likewise.
35887 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35890 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
35891 full command-line argument, rather than just "str".
35892 (aarch64_validate_march): Likewise.
35893 (aarch64_validate_mtune): Likewise.
35895 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
35897 PR rtl-optimization/78911
35898 * lra-assigns.c (must_not_spill_p): New function.
35899 (spill_for): Use it.
35901 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35903 PR tree-optimization/79981
35904 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
35905 ATOMIC_COMPARE_EXCHANGE ifn result.
35906 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
35907 IFN_ATOMIC_COMPARE_EXCHANGE.
35909 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35912 * opts.c (parse_sanitizer_options): Add missing question mark to
35913 "did you mean" message.
35915 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35917 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
35919 (VMULEUH_UNS): Likewise.
35920 (VMULOUB_UNS): Likewise.
35921 (VMULOUH_UNS): Likewise.
35922 * config/rs6000/rs6000.c (builtin_function_type): Remove
35923 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
35925 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35928 * read-rtl-function.c (function_reader::read_rtx_operand): Update
35929 x with result of extra_parsing_for_operand_code_0.
35930 (function_reader::extra_parsing_for_operand_code_0): Convert
35931 return type from void to rtx, returning x. When reading
35932 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
35933 larger size containing struct block_symbol.
35935 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
35937 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
35938 -mfloat128-hardware without -m64.
35940 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
35943 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
35944 entries to the case statement that marks unsigned arguments to
35945 overloaded functions.
35947 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
35949 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
35950 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
35952 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
35955 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
35956 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
35958 2017-03-10 Martin Liska <mliska@suse.cz>
35962 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
35963 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
35966 2017-03-10 Olivier Hainque <hainque@adacore.com>
35968 * tree-switch-conversion (array_value_type): Start by resetting
35969 candidate type to it's main variant.
35971 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35973 PR rtl-optimization/79909
35974 * combine.c (try_combine): Use simplify_replace_rtx on individual
35975 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
35976 of the whole CALL_INSN_FUNCTION_USAGE.
35978 PR tree-optimization/79972
35979 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
35980 get_range_info on SSA_NAMEs. Formatting fixes.
35982 2017-03-10 Richard Biener <rguenther@suse.de>
35983 Jakub Jelinek <jakub@redhat.com>
35985 PR tree-optimization/77975
35986 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
35987 edge to be constant.
35988 (get_val_for): For constant x return it. Formatting fix.
35989 (loop_niter_by_eval): Avoid pointless looping if the next iteration
35990 would use the same bases as the current one.
35992 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35994 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
35995 instead of vec_select for V1TImode.
35996 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
35998 (VSX_LE_128): Add V1TI to this mode iterator.
35999 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36000 (*vsx_le_perm_store_<mode>): Likewise.
36001 (pre-reload splitter for VSX stores): Likewise.
36002 (post-reload splitter for VSX stores): Likewise.
36003 (*vsx_xxpermdi2_le_<mode>): Likewise.
36004 (*vsx_lxvd2x2_le_<mode>): Likewise.
36005 (*vsx_stxvd2x2_le_<mode>): Likewise.
36007 2017-03-09 Michael Eager <eager@eagercon.com>
36009 Correct failures with --enable-checking=yes,rtl.
36011 * config/microblaze/microblaze.c (microblaze_expand_shift):
36012 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36013 test for const0_rtx.
36014 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36015 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36017 2017-03-09 Richard Biener <rguenther@suse.de>
36019 PR tree-optimization/79977
36020 * graphite-scop-detection.c (scop_detection::merge_sese):
36021 Handle the case of extra exits to blocks dominating the entry.
36023 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36025 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36028 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36030 PR rtl-optimization/79949
36031 * lra-constraints.c (process_alt_operands): Check memory when
36032 trying to predict a cycle. Print about the overall increase.
36034 2017-03-09 Richard Biener <rguenther@suse.de>
36036 PR middle-end/79971
36037 * gimple-expr.c (useless_type_conversion_p): Preserve
36038 TYPE_SATURATING for fixed-point types.
36040 2017-03-09 Richard Biener <rguenther@suse.de>
36043 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36044 alignment of BLKmode params.
36046 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36049 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36050 (VALL_NO_V2Q): Likewise.
36052 * config/aarch64/aarch64-simd.md
36053 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36055 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36056 VALL_NO_V2Q mode iterator.
36057 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36059 2017-03-09 Martin Liska <mliska@suse.cz>
36061 PR tree-optimization/79631
36062 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36063 tree_int_cst_sign_bit just for INTEGER constants.
36065 2017-03-09 Martin Liska <mliska@suse.cz>
36069 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36072 2017-03-09 Marek Polacek <polacek@redhat.com>
36075 * tree.c (inchash::add_expr): Handle TREE_VEC.
36077 2017-03-09 Martin Liska <mliska@suse.cz>
36080 (chkp_narrow_size_and_offset): New function.
36081 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36082 (void chkp_parse_bit_field_ref): New function.
36083 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36084 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36086 2017-03-09 Martin Liska <mliska@suse.cz>
36089 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36090 (chkp_find_bounds_1): Remove gcc_unreachable.
36092 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36095 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36096 BUILT_IN_SYNC*, determine the access type from the size suffix and
36097 always build a MEM_REF with that type. Handle forgotten
36098 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36101 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36102 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36103 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36104 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36105 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
36106 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
36107 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
36108 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
36109 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
36110 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
36111 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
36112 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
36113 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
36114 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
36115 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
36116 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
36117 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
36118 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
36119 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
36120 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
36121 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
36122 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
36123 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
36124 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
36125 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
36126 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
36127 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
36128 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
36129 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
36130 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
36131 definitions outside of __OPTIMIZE__ guarded section.
36134 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
36135 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
36136 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
36137 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
36140 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36142 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
36143 ("vfenez<mode>"): Add missing constraints.
36145 2017-03-08 Martin Sebor <msebor@redhat.com>
36148 * config/nds32/nds32.c (nds32_option_override):
36149 Fix misspelled diagnostic.
36151 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36154 * gimplify.c (gimplify_omp_for): Replace index var in outer
36155 taskloop statement with an artificial variable and add
36156 OMP_CLAUSE_PRIVATE clause for it.
36158 2017-03-08 Richard Biener <rguenther@suse.de>
36160 PR tree-optimization/79955
36161 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
36162 for accesses that are completely outside of the variable.
36164 2017-03-08 Andrew Haley <aph@redhat.com>
36166 PR tree-optimization/79943
36167 * tree-ssa-loop-split.c (compute_new_first_bound): When
36168 calculating the new upper bound, (END-BEG) should be added, not
36171 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36173 * config/avr/avr.md (setmemhi): Make sure match_dup
36174 operand number comes before match_scratch.
36176 2017-03-08 Richard Biener <rguenther@suse.de>
36178 PR tree-optimization/79920
36179 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
36180 with ncopies == 1 to ...
36181 (vect_transform_slp_perm_load): ... here. Properly compute
36182 all element loads by iterating VF times over the group. Do
36183 not handle ncopies (computed in a broken way) in
36184 vect_create_mask_and_perm.
36186 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36189 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
36190 is a uniform vector, use uniform_vector_p return value instead of
36191 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
36193 2017-03-07 Marek Polacek <polacek@redhat.com>
36195 PR middle-end/79809
36196 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
36197 (alloca_call_type): Likewise.
36199 2017-03-07 Martin Liska <mliska@suse.cz>
36201 * gcov.c (process_args): Put comment to correct location.
36203 2017-03-07 Martin Liska <mliska@suse.cz>
36205 PR middle-end/68270
36206 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
36207 Use array_at_struct_end_p instead of DECL_CHAIN (field).
36208 (chkp_narrow_bounds_for_field): Likewise.
36209 (chkp_parse_array_and_component_ref): Pass one more argument to
36212 2017-03-07 Richard Biener <rguenther@suse.de>
36214 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
36217 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
36219 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
36220 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
36222 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36225 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
36226 to end of description.
36227 (PARAM_MAX_STORES_TO_MERGE): Likewise.
36229 2017-03-07 Jakub Jelinek <jakub@redhat.com>
36231 PR rtl-optimization/79901
36232 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
36234 (*avx512f_<code><mode>3<mask_name>): ... this.
36235 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
36236 iterator instead of VI8_AVX2_AVX512BW.
36238 PR rtl-optimization/79901
36239 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
36240 min/max expander, expand it using expand_vec_cond_expr.
36243 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
36246 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36249 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
36250 to void * for PCH reasons.
36251 * dwarf2out.c (output_loc_operands, output_die): Cast
36252 v.val_vec.array to unsigned char *.
36254 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
36257 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
36260 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
36262 PR rtl-optimization/79571
36263 * lra-constraints.c (process_alt_operands): Calculate static
36264 reject and subtract it from overall when only addresses will be
36267 2017-03-06 Julia Koval <julia.koval@intel.com>
36270 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
36271 incoming stack boundary to 128 for 64-bit targets.
36273 2017-03-06 Richard Biener <rguenther@suse.de>
36275 PR tree-optimization/79894
36276 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
36277 to NULL after folding it.
36279 2017-03-06 Richard Biener <rguenther@suse.de>
36281 PR tree-optimization/79824
36282 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
36283 check disabling peeling for gaps.
36285 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
36287 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
36288 attributes): Document gettimeofday.
36290 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36292 * config/s390/s390.c (s390_option_override_internal): Set
36293 PARAM_MIN_VECT_LOOP_BOUND
36295 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36297 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
36298 * config/s390/s390.md: Likewise.
36300 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36303 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
36304 (<avx2_avx512>_perm<mode>): Rename to ...
36305 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
36307 (<avx512>_perm<mode>_mask): Rename to ...
36308 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
36310 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
36311 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
36312 instead of VI8F_256_512.
36313 (avx512f_perm<mode>): New define_expand.
36314 (avx512f_perm<mode>_mask): Likewise.
36315 (avx512f_perm<mode>_1<mask_name>): New define_insn.
36316 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
36318 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36320 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
36321 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
36323 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
36325 2017-03-06 Martin Liska <mliska@suse.cz>
36328 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
36329 when having a SSA NAME w/o VAR_DECL assigned to it.
36331 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36333 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
36334 msa_dpsub_<su>_d): Fix MODE for vec_select.
36336 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36338 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
36340 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
36342 2017-03-06 Richard Biener <rguenther@suse.de>
36344 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
36345 * plugin.c (register_plugin_info): Likewise.
36346 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
36348 2017-03-05 Jakub Jelinek <jakub@redhat.com>
36350 * config/i386/sse.md (sse_storehps, sse_storelps,
36351 avx_<castmode><avxsizesuffix>_<castmode>,
36352 avx512f_<castmode><avxsizesuffix>_<castmode>,
36353 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
36354 in condition that at least one operand is not a MEM.
36356 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36358 PR middle-end/79805
36359 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
36360 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
36362 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
36363 gimple_call_nothrow_p flag based on whether original builtin can throw.
36364 If it can, emit following stmts on the fallthrough edge.
36365 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
36366 don't create new bb if inserting just debug stmts on the edge, try to
36367 insert them on the fallthru bb or just reset debug stmts.
36369 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
36372 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
36373 restore recog_data (including the operand rtxes inside it) around
36374 the call to get_insn_template.
36376 2017-03-03 Martin Sebor <msebor@redhat.com>
36378 PR tree-optimization/79699
36379 * context.c (context::~context): Free MPFR caches to avoid
36380 a memory leak on program exit.
36382 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36384 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
36385 Use wide_int::ulow () instead of .elt (0).
36387 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36389 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
36390 (*pushxf): Limit oF constraint to 32bit targets and add oC
36391 constraint for 64bit targets.
36392 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
36393 (*pushdf): Change rmF constraint to rmC.
36395 2017-03-03 Martin Liska <mliska@suse.cz>
36397 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
36398 Remove unused variable.
36400 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36403 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
36404 is a memory operand, increase num_memory.
36405 (ix86_expand_args_builtin): Likewise.
36407 2017-03-03 Jan Hubicka <jh@suse.cz>
36410 * ipa-devirt.c (maybe_record_node): Properly handle
36411 __cxa_pure_virtual visibility.
36413 2017-03-03 Martin Liska <mliska@suse.cz>
36415 PR tree-optimization/79803
36416 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
36418 (pass_loop_prefetch::execute): Disabled optimization if an
36419 assumption about L1 cache size is not met.
36421 2017-03-03 Martin Liska <mliska@suse.cz>
36423 PR rtl-optimization/79574
36424 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
36425 (hash_scan_set): Likewise.
36426 (dump_hash_table): Likewise.
36427 (hoist_code): Likewise.
36429 2017-03-03 Richard Biener <rguenther@suse.de>
36431 * fixed-value.c (fixed_from_string): Restore use of elt (1)
36432 in place of uhigh ().
36433 (fixed_convert_from_real): Likewise.
36435 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36438 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
36440 2017-03-03 Richard Biener <rguenther@suse.de>
36442 PR middle-end/79818
36443 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
36444 TYPE_OVERFLOW_UNDEFINED check.
36446 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36448 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
36450 (vector_ae_<mode>_p): Likewise.
36451 (vector_nez_<mode>_p): Likewise.
36452 (vector_ne_v2di_p): Likewise.
36453 (vector_ae_v2di_p): Likewise.
36454 (vector_ne_<mode>_p): Likewise.
36455 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
36457 (vsx_tsqrt<mode>2_fe): Likewise.
36459 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
36462 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
36464 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36466 PR rtl-optimization/79780
36467 * cprop.c (one_cprop_pass): When second and further conditional trap
36468 in a single basic block is turned into an unconditional trap, turn it
36469 into a deleted note to avoid RTL verification failures.
36471 2017-03-02 Richard Biener <rguenther@suse.de>
36473 * fold-const.c (const_binop): Use ulow () instead of elt (0).
36475 2017-03-02 Richard Biener <rguenther@suse.de>
36477 PR tree-optimization/79345
36479 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
36480 param and abort the walk, returning -1 if it is hit.
36481 (walk_aliased_vdefs): Take a limit param and pass it on.
36482 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
36483 defaulting to 0 and return a signed int.
36484 * tree-ssa-uninit.c (struct check_defs_data): New struct.
36485 (check_defs): New helper.
36486 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
36487 about uninitialized memory.
36488 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
36489 bogus uninitialized warning.
36490 (fixed_convert_from_real): Likewise.
36492 2017-03-02 Bin Cheng <bin.cheng@arm.com>
36494 PR tree-optimization/66768
36495 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
36496 iv_use if base object can't be determined.
36498 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36500 PR tree-optimization/79345
36501 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
36502 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
36503 (get_pattern_stats): Initialize it.
36504 * genemit.c (gen_expand): Verify match_scratch numbers come after
36505 match_operand/match_dup numbers.
36506 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
36507 match_scratch numbers.
36508 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
36510 * config/s390/s390.md (trunctdsd2): Likewise.
36512 2017-03-02 Richard Biener <rguenther@suse.de>
36514 * wide-int.h (wide_int_storage::operator=): Implement in terms
36517 2017-03-02 Richard Biener <rguenther@suse.de>
36519 PR tree-optimization/79777
36520 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
36521 the to insert expression to sth existing.
36523 2017-03-01 Martin Sebor <msebor@redhat.com>
36525 PR middle-end/79692
36526 * gimple-ssa-sprintf.c
36527 (directive::known_width_and_precision): New function.
36528 (format_integer): Use it.
36529 (get_mpfr_format_length): Consider the full range of precision
36530 when computing %g output with the # flag. Set the likely byte
36531 count to 3 rather than 1 when precision is indeterminate.
36532 (format_floating): Correct the lower bound of precision.
36534 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36536 * doc/invoke.texi: Document default code model for 64-bit Linux.
36538 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36541 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
36542 udiv rather than div since input pattern is unsigned.
36544 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
36546 * config/i386/i386.c (print_reg): Warn for values of
36547 unsupported size in integer register.
36549 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
36552 * config/rs6000/predicates.md (current_file_function_operand): Do
36553 not allow self calls to be local if the function is replaceable.
36555 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
36558 * config/rs6000/altivec.h (vec_ctz and others): Change the
36559 preprocessor macro that controls conditional compilation from
36560 _ARCH_PWR9 to __POWER9_VECTOR__.
36561 (vec_all_ne): Change parameterization of __altivec_scalar_pred
36562 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
36563 control (instead of _ARCH_PWR9 control) so that template
36564 definition uses power9-specific function.
36565 (vec_any_eq): Likewise.
36566 (vec_all_ne): Change macro definition to use a power9-specific
36567 expansion under #ifdef __POWER9_VECTOR__ control (instead of
36568 _ARCH_PWR9 control).
36569 (vec_any_eq) Likewise.
36570 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
36571 expansion for CMPNEF to remove support for xvcmpnesp instruction.
36572 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
36573 support for xvcmpnedp instruction.
36574 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
36575 macro expansion so that Power9 implementation of vec_all_ne does
36576 not use the AltiVec predicate framework.
36577 (VCMPNEH_P): Likewise.
36578 (VCMPNEW_P): Likewise.
36579 (VCMPNED_P): Likewise.
36580 (VCMPNEFP_P): Likewise.
36581 (VCMPNEDP_P): Likewise.
36582 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
36583 implementation of vec_any_eq to not use AltiVec predicate
36585 (VCMPAEH_P): Likewise.
36586 (VCMPAEW_P): Likewise.
36587 (VCMPAED_P): Likewise.
36588 (VCMPAEFP_P): Likewise.
36589 (VCMPAEDP_P): Likewise.
36590 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
36591 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
36592 not use the AltiVec predicate framework.
36593 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
36594 of vec_any_eq to not use AltiVec predicate framework.
36595 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36596 support for predefined __POWER9_VECTOR__ macro to indicate that
36597 Power9 instruction selection is enabled.
36598 (altivec_overloaded_builtins): Remove extraneous
36599 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
36600 function argument types RS6000_BTI_bool_V16QI and
36601 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
36602 entry for overloaded function argument types RS6000_BTI_bool_V4SI
36603 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
36604 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
36605 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
36606 Power9 for implementations of vec_cmpne. Change the signature for
36607 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
36608 (representing vec_all_ne) to remove the previously described first
36609 argument of type RS6000_BTI_INTSI, as this was an artifact of
36610 reliance on the AltiVec predicate framework, which is no longer
36611 used in the implementation of these functions. Add
36612 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
36613 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
36614 since, unlike the AltiVec predicate framework implementation, we
36615 do not share function descriptors between vec_alle and vec_anyeq.
36616 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
36617 set of modes that receive special treatment even when
36618 TARGET_P9_VECTOR is true. The special treatment emits code that
36619 does not depend on Power9 instructions.
36620 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36621 define_expand to not rely on AltiVec predicate framework.
36622 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36624 (vector_ne_v2di_p): Change this define_expand to not rely on
36625 AltiVec predicate framework.
36626 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36628 (vector_ne_<mode>_p): Change this define_expand to not rely on
36629 AltiVec predicate framework.
36630 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36632 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36633 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36634 define_insn pattern.
36635 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36636 define_insn pattern because the xvcmpne<VSs>. instruction is not
36638 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36639 instruction is not supported.
36641 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36643 * config/nvptx/nvptx.c: Include intl.h.
36645 2017-03-01 Martin Jambor <mjambor@suse.cz>
36648 * ipa-prop.h (ipa_bits): Removed field known.
36649 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36650 to pointers. Adjusted their comments to warn about their sharing.
36651 (ipcp_transformation_summary): Change bits to a vector of pointers.
36652 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36653 (ipa_get_ipa_bits_for_value): Declare.
36654 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36655 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36656 (ipa_bits_hash_table): Likewise.
36657 (ipa_vr_ggc_hash_traits): Likewise.
36658 (ipa_vr_hash_table): Likewise.
36659 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36660 being pointers and vr_known being removed.
36661 (ipa_set_jf_unknown): Likewise.
36662 (ipa_get_ipa_bits_for_value): New function.
36663 (ipa_set_jfunc_bits): Likewise.
36664 (ipa_get_value_range): New overloaded functions.
36665 (ipa_set_jfunc_vr): Likewise.
36666 (ipa_compute_jump_functions_for_edge): Use the above functions to
36667 construct bits and vr parts of jump functions.
36668 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36669 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36671 (ipcp_grow_transformations_if_necessary): Also allocate
36672 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36674 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36675 them. Fix too long lines.
36676 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36677 vr_known being removed.
36678 (ipa_read_jump_function): Use new setter functions to construct bits
36679 and vr parts of jump functions or set them to NULL.
36680 (write_ipcp_transformation_info): Adjust for bits being pointers.
36681 (read_ipcp_transformation_info): Likewise.
36682 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36684 Include gt-ipa-prop.h.
36685 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36687 (ipcp_store_bits_results): Likewise.
36688 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36689 Do not write to existing jump functions but use a temporary instead.
36691 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36694 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36695 attempt to use its first operand as BIT_FIELD_REF base.
36697 2017-03-01 Richard Biener <rguenther@suse.de>
36699 PR middle-end/79721
36700 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36701 interpolating formula in wrapping arithmetic.
36702 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36704 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36706 PR tree-optimization/79734
36707 * tree-vect-generic.c (expand_vector_condition): Optimize
36708 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36709 Handle VEC_COND_EXPR where comparison has different inner width from
36710 type's inner width.
36712 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36714 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36715 markup, and similar issues. Remove @opindex entries for things
36716 that aren't options. Add missing -mmpy-option entries.
36718 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36720 PR tree-optimization/79737
36721 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36722 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36723 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36724 instead of byte_size. Formatting fix.
36725 (shift_bytes_in_array_right): Formatting fix.
36727 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36730 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36731 condition on optimize for the leaf function test.
36733 2017-02-28 Martin Liska <mliska@suse.cz>
36736 * read-rtl-function.c (function_reader::handle_unknown_directive):
36737 Bail out when one uses -flto.
36739 2017-02-28 Martin Liska <mliska@suse.cz>
36741 * common.opt: Replace space with tabular for options of <number>
36743 * config/i386/i386.opt: Show <number> value for
36744 -mlarge-data-threshold.
36745 * opts.c (print_filtered_help): Do not display number in hexadecimal
36748 2017-02-28 Martin Liska <mliska@suse.cz>
36750 * common.opt: Fix --help=option -Q for options which are of
36753 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36755 * config/i386/i386.c (print_reg): Error out for values
36756 of 8-bit size in invalid integer register.
36758 2017-02-28 Martin Sebor <msebor@redhat.com>
36760 PR tree-optimization/79691
36761 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36763 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36766 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36767 gcc_unreachable with output_operand_lossage.
36769 2017-02-28 Richard Biener <rguenther@suse.de>
36771 PR tree-optimization/79740
36772 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36774 (visit_nary_op): Insert the nary into the hashtable if we
36775 pattern-matched sth.
36776 * tree-ssa-pre.c (eliminate_insert): Robustify.
36778 2017-02-28 Richard Biener <rguenther@suse.de>
36780 PR middle-end/79731
36781 * fold-const.c (decode_field_reference): Reject out-of-bound
36784 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36786 * config/i386/i386.c: Include intl.h.
36787 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36788 instead of just cond ? "..." : "...".
36789 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36790 * coverage.c (read_counts_file): Likewise.
36791 * omp-offload.c: Include intl.h.
36792 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36793 of just cond ? "..." : "...".
36794 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36795 of just cond ? "..." : "...".
36797 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36800 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36802 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36803 'tune for' CPU name.
36804 * config/arm/arm-cpu-data.h: Regenerated.
36806 2017-02-28 Richard Biener <rguenther@suse.de>
36808 PR tree-optimization/79732
36809 * tree-inline.c (expand_call_inline): Do not shadow var.
36811 2017-02-28 Richard Biener <rguenther@suse.de>
36813 PR tree-optimization/79723
36814 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36815 address-space properly.
36817 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36819 * doc/optinfo.texi (Optimization groups): Fix option used for
36821 * doc/invoke.texi (-fopt-info): Document "omp".
36822 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36823 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36824 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36825 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36826 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36828 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36830 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36831 OPTGROUP_OMP with "omp".
36833 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36836 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36837 for arithmetic shift of unsigned V2DI.
36839 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36841 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36842 arc/linux.h headers.
36843 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36844 (LINK_SPEC): Likewise.
36845 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36846 (EXTRA_SPECS): Likewise.
36847 (STARTFILE_SPEC): Likewise.
36848 (ENDFILE_SPEC): Likewise.
36849 (LIB_SPEC): Likewise.
36850 (TARGET_SDATA_DEFAULT): Likewise.
36851 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36852 (MULTILIB_DEFAULTS): Likewise.
36853 (DWARF2_UNWIND_INFO): Likewise.
36854 * config/arc/big.h: New file.
36855 * config/arc/elf.h: Likewise.
36856 * config/arc/linux.h: Likewise.
36857 * config/arc/t-uClibc: Remove.
36859 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36861 PR tree-optimization/77536
36862 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
36863 (tree_transform_and_unroll_loop): Use above function to compute the
36864 estimated niter of unrolled loop and use it when scaling profile.
36865 Also use count info rather than frequency if it's non-zero.
36866 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
36867 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
36868 (vect_transform_loop): Call above function.
36870 2017-02-27 Richard Biener <rguenther@suse.de>
36872 PR tree-optimization/45397
36873 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
36874 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
36875 (visit_nary_op): Add pattern matching for CSEing sign-changed
36876 or truncated operations with wider ones.
36878 2017-02-27 Richard Biener <rguenther@suse.de>
36880 PR tree-optimization/79690
36881 * tree-vect-stmts.c (vectorizable_store): Use vector type
36882 built from the DR with address-space.
36884 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
36886 * doc/invoke.texi (Optimize Options): Refine the description
36887 of asan-use-after-return.
36889 2017-02-25 Alan Modra <amodra@gmail.com>
36891 PR rtl-optimization/79584
36892 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
36893 base, not ad->base_term, the reg within base. Remove assertion
36894 that ad->base == ad->base_term. Replace gen_int_mode using
36895 bogus mode with const0_rtx.
36897 2017-02-25 Jakub Jelinek <jakub@redhat.com>
36899 PR middle-end/79396
36900 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
36901 FMA_EXPR like tcc_binary or tcc_unary.
36903 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
36906 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
36908 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
36909 (output_loc_operands): Handle DW_OP_call_ref and
36910 DW_OP_GNU_variable_value.
36911 (struct variable_value_struct): New type.
36912 (struct variable_value_hasher): Likewise.
36913 (variable_value_hash): New variable.
36914 (string_types): Remove.
36915 (copy_loc_descr): New function.
36916 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
36917 (prepend_loc_descr_to_each): New function.
36918 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
36919 instead of add_loc_descr_to_each if the first argument is single
36920 location list and the second has multiple.
36921 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
36922 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
36923 when looking for variable value which doesn't have other location info.
36924 (loc_list_from_tree): Formatting fix.
36925 (gen_array_type_die): Simplify DW_AT_string_length handling.
36926 (adjust_string_types): Remove.
36927 (gen_subprogram_die): Don't call adjust_string_types nor test/set
36928 string_types. Call resolve_variable_values.
36929 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
36930 (resolve_addr_in_expr): Likewise. Add A argument.
36931 (copy_deref_exprloc): Remove deref argument. Adjust for the
36932 original expression being DW_OP_GNU_variable_value with optionally
36933 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
36934 optionally after it.
36935 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
36936 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
36937 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
36938 (variable_value_hasher::hash, variable_value_hasher::equal): New
36940 (resolve_variable_value_in_expr, resolve_variable_value,
36941 resolve_variable_values, note_variable_value_in_expr,
36942 note_variable_value): New functions.
36943 (dwarf2out_early_finish): Call note_variable_value on all toplevel
36946 2017-02-24 Jakub Jelinek <jakub@redhat.com>
36949 * opts.h (handle_generated_option): Add GENERATED_P argument.
36950 * opts-common.c (handle_option): Adjust function comment.
36951 (handle_generated_option): Add GENERATED_P argument, pass it to
36953 (control_warning_option): Pass false to handle_generated_option
36955 * opts.c (maybe_default_option): Pass true to handle_generated_option
36957 * optc-gen.awk: Likewise.
36959 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36961 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
36962 a REG, look at the REG it is a SUBREG of.
36963 (splitter for cmpeqsi_t): Ditto.
36965 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36967 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
36968 the special USEs with the pattern of the insn, not the insn itself.
36970 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
36973 * doc/invoke.texi: Document -mload-store-pairs.
36975 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36976 Sandra Loosemore <sandra@codesourcery.com>
36978 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
36979 argument isn't a CONST_INT.
36980 (nios2_alternate_compare_const): Assert op is a CONST_INT.
36981 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
36982 (nios2_validate_compare): Bypass alternate compare logic if *op2
36983 is not a CONST_INT.
36984 (ldstwm_operation_p): Return false if first_base is not a REG or
36985 if first_offset is not a CONST_INT.
36987 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36989 * config/cris/cris.md: Use correct operand in a define_peephole2.
36991 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36993 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
36995 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36997 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
36998 this_insn if it is an INSN or JUMP_INSN.
36999 (force_offsettable): Look at base, not at addr.
37000 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37001 on things that aren't necessarily CONST_INTs.
37003 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37005 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37006 -mfpmath=sse is the default also for x86-32 targets with SSE2
37007 instruction set when @option{-ffast-math} is enabled
37009 2017-02-24 Jeff Law <law@redhat.com>
37011 PR rtl-optimizatoin/79286
37012 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37015 2017-02-24 Richard Biener <rguenther@suse.de>
37017 PR tree-optimization/79389
37018 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37021 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37023 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37024 function comment to reflect reality.
37025 (loop_exits_before_overflow): Fix typo in function description.
37027 2017-02-24 Richard Biener <rguenther@suse.de>
37029 PR tree-optimization/79389
37030 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37031 properly that a threading opportunity exists. Detect conditional
37032 copy/constant propagation opportunities.
37034 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37036 * config/visium/visium.md (type): Add trap.
37037 (b): New mode attribute.
37038 (*btst): Rename into...
37039 (*btst<mode>): ...this and adjust.
37040 (*cbranchsi4_btst_insn): Rename into...
37041 (*cbranch<mode>4_btst_insn): ...this and adjust.
37042 (trap): New define_insn.
37044 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37046 PR tree-optimization/79389
37047 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37048 (noce_reversed_cond_code): New function.
37049 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37050 reversed_comparison_code. Formatting fix.
37051 (noce_try_store_flag): Test rev_cond != NULL in addition to
37052 reversed_comparison_code.
37053 (noce_try_store_flag_constants): Likewise.
37054 (noce_try_store_flag_mask): Likewise.
37055 (noce_try_addcc): Use rev_cond if non-NULL instead of
37056 reversed_comparison_code.
37057 (noce_try_cmove_arith): Likewise. Formatting fixes.
37058 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37059 (noce_find_if_block): Initialize rev_cond.
37060 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37061 instead of false as last argument never attempt to reverse it
37064 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37066 PR tree-optimization/79663
37067 * tree-predcom.c (combine_chains): Process refs in reverse order
37068 only for ZERO length chains, and add explaining comment.
37070 2017-02-23 Jeff Law <law@redhat.com>
37072 PR tree-optimization/79578
37073 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37074 in call to operand_equal_p.
37076 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37079 * config/i386/cpuid.h: Fix another undefined behavior.
37081 2017-02-23 Richard Biener <rguenther@suse.de>
37083 PR tree-optimization/79683
37084 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37085 vector types for data-refs.
37087 2017-02-23 Martin Liska <mliska@suse.cz>
37089 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37091 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37093 PR middle-end/79665
37094 * internal-fn.c (get_range_pos_neg): Moved to ...
37095 * tree.c (get_range_pos_neg): ... here. No longer static.
37096 * tree.h (get_range_pos_neg): New prototype.
37097 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37098 are known to be in between 0 and signed maximum inclusive, try to
37099 expand both unsigned and signed divmod and use the cheaper one from
37102 2017-02-22 Jeff Law <law@redhat.com>
37104 PR tree-optimization/79578
37105 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37106 to compare base operands.
37108 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
37111 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
37112 gpc_reg_operand instead of fpr_reg_operand.
37114 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
37116 * config/mips/mips.c (mips_return_in_memory): Force FP
37117 vector types to be returned in memory for o32 ABI.
37119 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37121 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
37122 instead of DW_TAG_member for static data member declarations and don't
37123 set no_linkage_name for static inline data members.
37124 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
37127 2017-02-22 Martin Liska <mliska@suse.cz>
37129 * doc/invoke.texi: Replace inequality signs with square brackets
37132 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
37135 * lra-constraints.c (simplify_operand_subreg): Handle
37136 WORD_REGISTER_OPERATIONS targets.
37138 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37141 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
37142 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
37143 elimination by swapping fld*.
37145 2017-02-22 Richard Biener <rguenther@suse.de>
37147 PR tree-optimization/79673
37148 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
37149 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
37150 irrelevant address-space qualifiers and avoiding a
37151 ADDR_SPACE_CONVERT_EXPR from fold_convert.
37153 2017-02-22 Richard Biener <rguenther@suse.de>
37155 PR tree-optimization/79666
37156 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
37157 to not symbolically negate if that may introduce undefined
37160 2017-02-22 Martin Liska <mliska@suse.cz>
37163 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
37164 * data-streamer-out.c (streamer_write_gcov_count_stream):
37166 * value-prof.c (stream_out_histogram_value): Make assert more
37167 precise based on type of counter.
37169 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
37172 * config/i386/i386.md (standard_x87sse_constant_load splitter):
37173 Use nonimmediate_operand instead of memory_operand for operand 1.
37174 (float-extend standard_x87sse_constant_load splitter): Ditto.
37176 2017-02-21 Jeff Law <law@redhat.com>
37178 PR tree-optimization/79621
37179 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
37180 blocks with edges to themselves.
37182 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37185 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
37186 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
37187 Use gimple_call_builtin_p.
37190 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
37191 on temporarily removed DEBUG_INSNs.
37193 PR tree-optimization/79649
37194 * tree-loop-distribution.c (classify_partition): Give up on
37195 non-generic address space loads/stores.
37197 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
37199 * doc/loop.texi (Loop manipulation): Remove nonexistent
37200 tree_ssa_loop_version from the documentation.
37201 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
37203 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37206 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
37207 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
37208 * config/rs6000/rs6000.c: Include except.h.
37209 (rs6000_expand_split_stack_prologue): Call
37210 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
37212 2017-02-21 Martin Jambor <mjambor@suse.cz>
37215 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
37216 have been analyzed.
37218 2017-02-21 Martin Jambor <mjambor@suse.cz>
37220 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
37221 for backward compatibility only.
37222 * doc/invoke.texi (Option Summary): Remove all references to
37223 -fipa-cp-alignment.
37225 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
37229 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37231 * lra-constraints.c (curr_insn_transform): Handle
37232 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37234 2017-02-21 Martin Liska <mliska@suse.cz>
37236 * config/i386/i386.opt: Replace -masm-dialect with -masm.
37238 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
37240 PR translation/79638
37241 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
37243 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
37246 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
37247 (arm_function_ok_for_sibcall): Return false for an indirect call by
37248 descriptor if all the argument registers are used.
37249 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
37250 alignment of the function.
37252 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37254 PR tree-optimization/61441
37255 * simplify-rtx.c (simplify_const_unary_operation): For
37256 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
37257 the sNaN unmodified.
37259 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
37261 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
37262 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
37263 instead of SYSTEM_HEADER_DIR.
37265 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
37266 Martin Liška <mliska@suse.cz>
37268 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
37269 Fix typos and grammar, use active voice, and clarify.
37271 2017-02-20 Marek Polacek <polacek@redhat.com>
37273 PR middle-end/79537
37274 * gimplify.c (gimplify_expr): Handle unused *&&L;.
37277 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
37279 2017-02-20 Jakub Jelinek <jakub@redhat.com>
37282 * config/i386/i386.c (ix86_expand_builtin): Handle
37283 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
37284 ix86_builtins_isa[fcode].isa as a requirement of those
37285 flags and any other flag in the bitmask.
37286 (ix86_init_mmx_sse_builtins): Use 0 instead of
37287 ~OPTION_MASK_ISA_64BIT as mask.
37288 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
37289 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
37290 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
37291 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
37293 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37296 * lra-constraints.c (split_reg): Check requested split mode
37297 is supported by the register.
37299 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37301 * lra-constraints.c (simplify_operand_subreg): Remove early
37304 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37307 * lra-constraints.c (curr_insn_transform): Tighten condition
37308 for converting SUBREG reloads from OP_OUT to OP_INOUT.
37310 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37313 * lra-constraints.c (curr_insn_transform): Handle
37314 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37316 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
37319 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
37321 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
37323 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
37326 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
37329 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37331 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
37334 2017-02-18 Jakub Jelinek <jakub@redhat.com>
37336 * final.c (last_columnnum, override_columnnum): New variables.
37337 (final_start_function): Set last_columnnum, pass it to begin_prologue
37338 hook and pass 0 to dwarf2out_begin_prologue.
37339 (final_scan_insn): Update override_columnnum. Pass last_columnnum
37340 to source_line debug hook.
37341 (notice_source_line): Compute last_columnnum and for debug_column_info
37342 return true on column changes.
37343 * debug.h (struct gcc_debug_hooks): Add column argument to
37344 source_line and begin_prologue hooks.
37345 (debug_nothing_int_charstar_int_bool): Remove prototype.
37346 (debug_nothing_int_int_charstar,
37347 debug_nothing_int_int_charstar_int_bool): New prototypes.
37348 (dwarf2out_begin_prologue): Add column argument.
37349 * debug.c (do_nothing_debug_hooks): Adjust source_line and
37350 begin_prologue hooks.
37351 (debug_nothing_int_charstar_int_bool): Remove.
37352 (debug_nothing_int_int_charstar,
37353 debug_nothing_int_int_charstar_int_bool): New functions.
37354 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
37355 through to dwarf2out_source_line.
37356 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
37357 (dwarf2out_source_line): Add column argument, emit it if requested.
37358 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
37360 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37361 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37362 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
37363 through to dwarf2out_begin_prologue.
37364 (vmsdbgout_source_line): Add column argument, pass it through to
37365 dwarf2out_source_line.
37366 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
37367 dbxout_source_line caller.
37368 (dbxout_source_line): Add column argument.
37370 * common.opt (gno-column-info, gcolumn-info): New options.
37371 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
37372 (check_die): Also test for multiple DW_AT_decl_column attributes.
37373 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
37374 DW_AT_decl_column if requested.
37375 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
37377 (gen_variable_die): Likewise.
37378 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
37379 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
37382 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
37383 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
37384 (ix86_handle_option): Handle OPT_m3dnowa.
37385 * doc/invoke.texi (-m3dnowa): Document.
37386 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
37387 -m3dnowa instead of -m3dnow -march=athlon.
37390 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
37391 instead of gcc_assert for K, r and R code checks. Formatting fixes.
37393 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37396 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
37397 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
37398 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
37399 generator for vsx_xxpermdi_<mode>_be.
37400 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
37401 force big-endian semantics.
37402 (vsx_xxpermdi_<mode>_be): New define_expand with same
37403 implementation as previous version of vsx_xxpermdi_<mode>.
37405 2017-02-17 Jakub Jelinek <jakub@redhat.com>
37407 PR tree-optimization/79327
37408 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
37409 variable, its initialization and use.
37411 2017-02-17 Julia Koval <julia.koval@intel.com>
37413 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
37414 (OPTION_MASK_ISA_PKU_UNSET): New.
37415 (ix86_handle_option): Handle -mrdpid.
37416 * config/i386/cpuid.h (bit_RDPID): New.
37417 * config/i386/driver-i386.c (host_detect_local_cpu):
37418 Detect RDPID feature.
37419 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
37420 * config/i386/i386-c.c (ix86_target_macros_internal):
37422 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
37423 (ix86_valid_target_attribute_inner_p): Add "rdpid".
37424 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
37425 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
37426 * config/i386/i386.md (define_insn "rdpid"): New.
37427 * config/i386/i386.opt Add -mrdpid.
37428 * config/i386/immintrin.h (_rdpid_u32): New.
37430 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
37432 PR rtl-optimization/79541
37433 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
37434 instead of transforming it into USE.
37436 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
37438 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
37439 If HONOR_SNANS (SFmode) force the input to a register.
37440 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
37441 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
37442 an frsp or similar insn.
37444 2017-02-17 Martin Liska <mliska@suse.cz>
37446 PR rtl-optimization/79577
37447 * params.def (selsched-max-sched-times): Increase minimum to 1.
37449 2017-02-17 Martin Liska <mliska@suse.cz>
37451 PR rtl-optimization/79574
37452 * gcse.c (want_to_gcse_p): Prevent integer overflow.
37454 2017-02-17 Martin Liska <mliska@suse.cz>
37456 PR tree-optimization/79529
37457 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
37458 ssa_defined_default_def_p to handle cases which are implicitly
37460 * tree-ssa.c (ssa_defined_default_def_p): New function.
37461 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
37462 which are implicitly defined.
37463 * tree-ssa.h (ssa_defined_default_def_p): Declare.
37465 2017-02-17 Richard Biener <rguenther@suse.de>
37467 PR middle-end/79576
37468 * params.def (max-ssa-name-query-depth): Limit to 10.
37470 2017-02-17 Richard Biener <rguenther@suse.de>
37472 PR tree-optimization/79552
37473 * tree-ssa-structalias.c (visit_loadstore): Properly verify
37476 2017-02-17 Richard Biener <rguenther@suse.de>
37479 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
37481 2017-02-17 Marek Polacek <polacek@redhat.com>
37483 PR middle-end/79536
37484 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
37485 (fold_negate_expr): New wrapper.
37487 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
37489 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
37490 Correct terminology and de-emphasize pre-standard behavior.
37492 2017-02-16 Alan Modra <amodra@gmail.com>
37494 PR rtl-optimization/79286
37495 * ira.c (def_dominates_uses): New function.
37496 (update_equiv_regs): Don't create an equivalence for insns that
37497 may trap where the register def does not dominate the use.
37499 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
37501 PR rtl-optimization/78127
37502 * lra.c (lra): Call lra_eliminate before finish the loop after
37505 2017-02-16 Richard Biener <rguenther@suse.de>
37507 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
37509 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
37510 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
37511 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
37512 (isl_val_int_from_wi): New function.
37513 (extract_affine_gmp): Rename to ...
37514 (extract_affine_wi): ... this, take a widest_int.
37515 (extract_affine_int): Just wrap extract_affine_wi.
37516 (add_param_constraints): Use isl_val_int_from_wi.
37517 (add_loop_constraints): Likewise, and extract_affine_wi.
37519 2017-02-15 Jeff Law <law@redhat.com>
37521 PR middle-end/79521
37522 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
37523 ira_init_register_move_cost_if_necessary.
37525 2017-02-15 Martin Sebor <msebor@redhat.com>
37527 PR middle-end/32003
37528 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
37529 removed in a prior commit.
37531 2017-02-15 Bin Cheng <bin.cheng@arm.com>
37533 PR tree-optimization/79347
37534 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
37535 counters during peeling.
37537 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
37539 * Makefile.in (site.exp): Remove "set ISLVER".
37541 2017-02-15 Jakub Jelinek <jakub@redhat.com>
37544 * real.c (real_from_integer): Call real_convert even for decimal.
37546 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
37549 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
37551 2017-02-14 Andrew Pinski <apinski@cavium.com>
37553 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
37554 cores and change the partno/implementer to be correct.
37555 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
37556 the 'B" as the implementer.
37557 * config/aarch64/aarch64-tune.md: Regenerate.
37559 2017-02-14 Carl Love <cel@us.ibm.com>
37561 * config/rs6000/rs6000.c: Add case statement entry to make the
37562 xvcvuxdsp built-in argument unsigned.
37563 * config/rs6000/vsx.md: Fix the source and return operand types so they
37564 match the instruction definitions from the ISA document. Fix typo
37565 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
37568 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
37571 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
37572 member early_clobber_alts.
37573 * lra-lives.c (reg_early_clobber_p): New.
37574 (process_bb_lives): Use it.
37575 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
37576 (debug_operand_data): Initialize early_clobber_alts.
37577 (setup_operand_alternative): Set up early_clobber_alts.
37578 (collect_non_operand_hard_regs): Ditto. Pass early clobber
37579 alternatives to new_insn_reg.
37580 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
37582 (lra_update_insn_regno_info): Pass the new arg.
37584 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37586 PR middle-end/79505
37587 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
37588 (new_oacc_loop_raw): Don't clear already cleared fields.
37591 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
37592 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
37593 _mm512_prefetch_i64gather_ps): New inline functions and macros.
37595 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
37598 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
37600 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
37603 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
37604 the extra instruction to the right place to store 128-bit constant
37607 2017-02-14 Martin Sebor <msebor@redhat.com>
37609 PR middle-end/79448
37610 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
37611 warning for strings of unknown length.
37613 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
37615 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
37617 2017-02-14 Jeff Law <law@redhat.com>
37620 * ira-costs.c (scan_one_insn): Initialize register move costs
37621 for pseudos seen in USE/CLOBBER insns.
37623 PR tree-optimization/79095
37624 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37625 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37626 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37627 if the operands are known to be not equal, then the resulting range
37629 (intersect_ranges): If the new range is ~[0,0] and the old range is
37630 wide, then prefer ~[0,0].
37631 * tree-vrp.c (overflow_comparison_p_1): New function.
37632 (overflow_comparison_p): New function.
37633 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37634 if NAME is used in an overflow test.
37635 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37636 overflow check that can be expressed as an equality test, then adjust
37637 ops to be that equality test.
37639 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37641 * config/s390/s390-builtin-types.def: Remove flags argument.
37642 * config/s390/s390.c (s390_init_builtins): Likewise.
37644 2017-02-14 Martin Liska <mliska@suse.cz>
37646 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37647 vector. Fix trailing white spaces.
37649 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37651 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37654 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37656 PR rtl-optimization/68664
37657 * config/arm/arm.c (arm_sched_can_speculate_insn):
37658 New function. Declare prototype.
37659 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37661 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37663 PR rtl-optimization/68664
37664 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37666 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37668 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37670 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37671 max skip bytes for function, loop and jump.
37673 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37675 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37676 ABS_EXPR for gimple dump.
37678 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37681 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37683 PR tree-optimization/79408
37684 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37685 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37686 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37687 also if rhs1 is INTEGER_CST.
37689 2017-02-14 Richard Biener <rguenther@suse.de>
37691 PR middle-end/79432
37692 * tree-into-ssa.c (insert_phi_nodes): When the function can
37693 have abnormal edges rewrite SSA names with broken use-def
37694 dominance out of SSA and register them for PHI insertion.
37696 2017-02-13 Martin Sebor <msebor@redhat.com>
37698 PR middle-end/79496
37699 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37700 clearing info.nowrite flag when snprintf size argument is a range.
37702 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37704 * cprop.c (cprop_jump): Add missing space in string literal.
37705 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37706 (get_constraint_for_component_ref): Likewise.
37707 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37708 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37709 * lra-constraints.c (process_alt_operands): Likewise.
37710 * ipa-inline.c (inline_small_functions): Likewise.
37711 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37712 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37713 * trans-mem.c (diagnose_tm_1_op): Likewise.
37714 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37715 (grid_parallel_clauses_gridifiable): Likewise.
37717 * config/nvptx/mkoffload.c (process): Add space in between
37720 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37721 "MOD4_SSE_REGS" and "ALL_REGS".
37723 * spellcheck.c (test_data): Add , in between "foo" and "food".
37725 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37728 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37729 boundary crossing check and subsequent code generation agree.
37731 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37733 * config/aarch64/aarch64.c (has_memory_op): Delete.
37734 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37737 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37739 PR rtl-optimization/79388
37740 PR rtl-optimization/79450
37741 * combine.c (distribute_notes): When removing TEM_INSN for which
37742 corresponding dest has last value recorded, invalidate that last
37745 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37747 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37748 of explicit '@'. Add missing assembly comment marker on branch costs
37751 2017-02-13 Nathan Sidwell <nathan@acm.org>
37753 * gengtype-lex.l (<in_struct>): Add '/'.
37755 2017-02-13 Martin Liska <mliska@suse.cz>
37758 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37760 2017-02-13 Richard Biener <rguenther@suse.de>
37762 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37764 * configure: Re-generate.
37765 * config.in: Likewise.
37766 * graphite-dependences.c: Simplify as if
37767 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37768 * graphite-isl-ast-to-gimple.c: Likewise.
37769 * graphite-optimize-isl.c: Likewise.
37770 * graphite-poly.c: Likewise.
37771 * graphite-sese-to-poly.c: Likewise.
37772 * graphite.h: Likewise.
37773 * toplev.c: Include isl/version.h and use isl_version () for
37774 printing the ISL version.
37775 * doc/install.texi: Update ISL requirement.
37777 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37779 * doc/standards.texi (Standards): Update reference to
37782 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37784 * doc/extend.texi (Named Address Spaces): sourceware.org now
37786 * doc/install.texi (Binaries): Ditto.
37789 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37791 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37792 standard terminology "stringize"/"stringizing" throughout.
37793 * doc/cppinternals.texi: Likewise.
37795 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37797 * doc/extend.texi: Fix some spelling mistakes and typos.
37798 * doc/invoke.texi: Likewise.
37800 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37803 * params.def (inline-min-speedup) Change from 10 to 8.
37805 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37807 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37810 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37813 * ipa-inline-analysis.c (get_minimal_bb): New function.
37814 (record_modified): Use it.
37815 (remap_edge_change_prob): Handle also ancestor functions.
37817 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37819 * doc/contrib.texi (Contributors): Remove broken link into
37820 the Mauve CVS repository.
37822 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37824 PR middle-end/79454
37825 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37826 result computation whenever lhs doesn't have vector mode, not
37827 just when it has BLKmode.
37829 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37831 * doc/makefile.texi (profiledbootstrap): Refer to the
37832 installation instructions only in textual form.
37834 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37837 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37839 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37841 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37842 (Specific): Update mingw-w64 reference.
37844 (Specific): Remove broken link to Renesas RX processor.
37846 2017-02-10 Richard Biener <rguenther@suse.de>
37848 * toplev.c (process_options): Do not mention obsolete graphite
37849 options when printing sorry message about missing graphite support.
37850 Mention -floop-nest-optimize.
37852 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37854 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37855 (vtst_p16): Likewise.
37856 (vtstq_p8): Likewise.
37857 (vtstq_p16): Likewise.
37859 (vtstq_p64): Likewise.
37860 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37861 (vset_lane_p64): New.
37862 (vsetq_lane_p64): New.
37864 2017-02-10 Jakub Jelinek <jakub@redhat.com>
37866 PR tree-optimization/79411
37867 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
37868 stmt operands are SSA_NAMEs used in abnormal phis.
37869 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
37872 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37875 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
37878 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37880 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
37882 2017-02-09 Jakub Jelinek <jakub@redhat.com>
37884 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
37888 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
37889 not arbitrary TREE_CONSTANT.
37892 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
37893 "omp declare target link" attribute unless is_global_var.
37894 * omp-offload.c (find_link_var_op): Likewise.
37896 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
37897 Chung-Lin Tang <cltang@codesourcery.com>
37899 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
37901 (gimplify_adjust_omp_clauses): Don't delete TILE.
37902 (gimplify_omp_for): Deal with TILE.
37903 * internal-fn.c (expand_GOACC_TILE): New function.
37904 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
37906 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
37907 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
37909 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
37910 avoid DIV for outermost collapse var.
37911 (expand_oacc_for): Insert tile element loop as needed. Adjust.
37912 Remove out of date comments, fix whitespace.
37913 * omp-general.c (omp_extract_for_data): Deal with tiling.
37914 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
37915 adjust OLF_DIM_BASE value.
37916 (struct omp_for_data): Add tiling field.
37917 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
37918 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
37919 for auto loops. Remove default auto determining, moved to
37920 oacc_loop_fixed_partitions.
37921 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
37922 stmts, add e_mask field.
37923 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
37924 (oacc_thread_numbers): Use oacc_dim_call.
37925 (oacc_xform_tile): New.
37926 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
37927 (finish_oacc_loop): Adjust for ifns vector.
37928 (oacc_loop_discover_walk): Append loop abstraction sites to list,
37929 add case for GOACC_TILE fns.
37930 (oacc_loop_xform_loop): Delete.
37931 (oacc_loop_process): Iterate over call list directly, and add
37932 handling for GOACC_TILE fns.
37933 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
37935 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
37936 vector partitioning to outer loops. Assign 2 partitions to loops
37937 when available. Add TILE handling.
37938 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
37939 (execite_oacc_device_lower): Process GOACC_TILE fns,
37940 ignore unknown specs.
37941 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
37942 * tree.c (omp_clause_num_ops): Adjust TILE ops.
37943 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
37945 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
37947 * configure.ac (ACX_BUGURL): Update.
37948 * configure: Regenerate.
37950 2017-02-09 Richard Biener <rguenther@suse.de>
37952 PR tree-optimization/69823
37953 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
37954 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
37956 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37958 * config/arc/arc-c.def: Add __NPS400__ definition.
37959 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
37960 (TARGET_NPS400): Define.
37962 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37964 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
37966 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
37967 pointer, arch_info.
37968 (arc_cpu_types): Fill the arch_info field with a pointer into the
37969 arc_arch_types table.
37970 (arc_selected_cpu): Declare.
37971 * config/arc/arc.c (arc_selected_cpu): Make global.
37972 (arc_selected_arch): Delete.
37973 (arc_base_cpu): Delete.
37974 (arc_override_options): Remove references to deleted variables,
37975 update access to arch information.
37976 (ARC_OPT): Update access to arch information.
37977 (ARC_OPTX): Likewise.
37978 * config/arc/arc.h (arc_base_cpu): Remove declaration.
37979 (TARGET_ARC600): Update access to arch information.
37980 (TARGET_ARC601): Likewise.
37981 (TARGET_ARC700): Likewise.
37982 (TARGET_EM): Likewise.
37983 (TARGET_HS): Likewise.
37984 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
37987 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
37990 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
37991 condition/operands for integer GE/LE/GEU/LEU operations.
37993 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
37995 PR translation/79397
37996 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
37999 2017-02-08 Martin Jambor <mjambor@suse.cz>
38002 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38003 whether allocation happened.
38004 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38005 nothing was allocated.
38007 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38009 PR tree-optimization/79408
38010 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38011 constant, but SSA_NAME with a known integer range, use the minimum
38012 of that range instead of op1 to determine if modulo can be replaced
38013 with its first operand.
38015 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38017 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38019 2017-02-08 Richard Biener <rguenther@suse.de>
38021 PR tree-optimization/71824
38022 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38023 Check all loops contained in the merged region.
38025 2017-02-07 Andrew Pinski <apinski@cavium.com>
38027 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38029 2017-02-07 Andrew Pinski <apinski@cavium.com>
38031 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38032 (thunderxt88): Likewise.
38033 (thunderxt81): Disable LSE and change v8.1 to v8.
38034 (thunderxt83): Likewise.
38036 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38037 Richard Biener <rguenther@suse.de>
38039 PR middle-end/79399
38040 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38041 type from int to size_t.
38042 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38044 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38046 PR rtl-optimization/79386
38047 * cprop.c (bypass_conditional_jumps): Initialize
38048 bypass_last_basic_block already before splitting bbs after
38049 unconditional traps...
38050 (bypass_conditional_jumps): ... rather than here.
38053 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38054 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38055 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38056 fix -masm=intel patterns.
38058 2017-02-07 Richard Biener <rguenther@suse.de>
38060 PR tree-optimization/79256
38061 PR middle-end/79278
38062 * builtins.c (get_object_alignment_2): Use min_align_of_type
38063 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38064 and ADJUST_FIELD_ALIGN.
38066 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38068 * doc/tm.texi: Regenerate.
38069 * stor-layout.c (layout_decl): Adjust.
38070 (update_alignment_for_field): Likewise.
38071 (place_field): Likewise.
38072 (min_align_of_type): Likewise.
38073 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38074 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38075 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38076 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38077 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38078 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38079 * config/i386/i386.c (x86_field_alignment): Likewise.
38080 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38081 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38082 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38083 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38084 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38085 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38089 2017-01-30 Richard Biener <rguenther@suse.de>
38091 PR tree-optimization/79256
38092 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38093 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38096 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38098 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38099 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38100 builtins to SImode and emit a zero-extend, if necessary.
38102 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38104 * docs/invoke.texi (RISC-V Options): Alphabetize.
38106 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38108 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
38111 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38113 * config/riscv/riscv.c: New file.
38114 * common/config/riscv/riscv-common.c: Likewise.
38115 * config.gcc: Likewise.
38116 * config/riscv/constraints.md: Likewise.
38117 * config/riscv/elf.h: Likewise.
38118 * config/riscv/generic.md: Likewise.
38119 * config/riscv/linux.h: Likewise.
38120 * config/riscv/multilib-generator: Likewise.
38121 * config/riscv/peephole.md: Likewise.
38122 * config/riscv/pic.md: Likewise.
38123 * config/riscv/predicates.md: Likewise.
38124 * config/riscv/riscv-builtins.c: Likewise.
38125 * config/riscv/riscv-c.c: Likewise.
38126 * config/riscv/riscv-ftypes.def: Likewise.
38127 * config/riscv/riscv-modes.def: Likewise.
38128 * config/riscv/riscv-opts.h: Likewise.
38129 * config/riscv/riscv-protos.h: Likewise.
38130 * config/riscv/riscv.h: Likewise.
38131 * config/riscv/riscv.md: Likewise.
38132 * config/riscv/riscv.opt: Likewise.
38133 * config/riscv/sync.md: Likewise.
38134 * config/riscv/t-elf-multilib: Likewise.
38135 * config/riscv/t-linux: Likewise.
38136 * config/riscv/t-linux-multilib: Likewise.
38137 * config/riscv/t-riscv: Likewise.
38138 * configure.ac: Likewise.
38139 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
38140 Waterman as RISC-V maintainers.
38141 * doc/install.texi: Add RISC-V entries.
38142 * doc/invoke.texi: Add RISC-V options section.
38143 * doc/md.texi: Add RISC-V constraints section.
38144 * configure: Regenerated.
38146 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
38149 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
38150 false values to be constant vectors with all 0 or all 1 bits set.
38151 (vcondu<mode><mode>): Likewise.
38152 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
38154 (fpmask_comparison_operator): Update comment.
38155 (vecint_comparison_operator): New predicate.
38156 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
38157 vector conditionals when the true and false values are constant
38158 vectors with all 0 bits or all 1 bits set.
38160 2017-02-06 Martin Sebor <msebor@redhat.com>
38162 PR tree-optimization/79376
38163 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
38165 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
38167 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
38168 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
38169 to simplify split condition.
38171 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38173 * omp-expand.c (oxpand_omp_atomic_fetch_op,
38174 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
38177 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
38179 PR rtl-optimization/68664
38180 * target.def (can_speculate_insn): New hook.
38181 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
38182 * doc/tm.texi: Regenerate.
38183 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
38184 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
38185 (rs6000_sched_can_speculate_insn): New function.
38187 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38189 PR tree-optimization/79284
38190 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
38191 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
38192 vectorizable_mask_load_store, vectorizable_operation,
38193 vect_is_simple_cond, get_same_sized_vectype): Use it instead
38194 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
38195 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
38196 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
38197 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
38198 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
38199 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
38200 is_gimple_assign (stmt). Replace another such test with
38201 is_gimple_assign (stmt).
38203 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38206 * config/avr/avr.c (rtl-iter.h): Include it.
38207 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
38208 (avr_legitimate_combined_insn): ...and implementation.
38210 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
38212 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
38213 * config/s390/s390.c (s390_const_operand_ok)
38214 (s390_canonicalize_comparison, s390_extract_part)
38215 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
38216 (s390_contiguous_bitmask_p, s390_rtx_costs)
38217 (legitimize_pic_address): Likewise.
38218 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
38219 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
38220 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
38221 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
38222 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
38224 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38226 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
38227 REGNO($0) == REGNO($1).
38229 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38231 * config/s390/linux.h(SIZE_TYPE): Add comment.
38233 2017-02-06 Julian Brown <julian@codesourcery.com>
38234 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38235 Virendra Pathak <virendra.pathak@broadcom.com>
38237 * config/aarch64/aarch64-cores.def: Change the scheduler
38239 * config/aarch64/aarch64.md: Include thunderx2t99.md.
38240 * config/aarch64/thunderx2t99.md: New file.
38242 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38244 * doc/standards.texi (Go Language): Update link to language
38247 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38249 * tree-eh.c (lower_resx): Sanitize profile.
38250 (cleanup_empty_eh_move_lp): Likewise.
38252 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38255 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
38257 * cfgloopmanip.h (loop_version): Update prototype.
38258 * modulo-sched.c (sms_schedule): Update call of loop_version.
38259 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
38260 * tree-parloops.c (gen_parallel_loop): Likewise.
38261 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
38262 * tree-ssa-loop-split.c (split_loop): Likewise.
38263 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
38264 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
38266 2017-02-05 Martin Liska <mliska@suse.cz>
38269 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
38271 (print_operand_address): Initialize a struct to zero.
38273 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38275 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
38276 garbage collector only in textual form.
38278 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38280 * doc/extend.texi (x86 specific memory model extensions for
38281 transactional memory): Simplify a phrase.
38283 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
38286 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
38287 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
38288 (atomic_storedi_1): Likewise.
38290 2017-02-04 Jakub Jelinek <jakub@redhat.com>
38292 PR tree-optimization/79338
38293 * tree-parloops.c (gather_scalar_reductions): Don't call
38294 vect_analyze_loop_form for loop->inner before destroying loop's
38297 2017-02-03 Martin Sebor <msebor@redhat.com>
38299 PR tree-optimization/79327
38300 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
38301 when precision has resulted in leading zeros.
38302 (format_integer): Adjust the likely counter to assume an unknown
38303 argument that may be zero is non-zero.
38305 2017-02-03 Jason Merrill <jason@redhat.com>
38308 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
38309 avoid copying non-taken branch.
38311 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38313 PR tree-optimization/79340
38314 * tree-vect-loop.c (vectorizable_reduction): Release
38315 vec_defs elements after safe_splicing them into other vectors.
38318 PR tree-optimization/79327
38319 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
38320 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
38322 (format_integer): Use wide_int_to_tree instead of build_int_cst
38323 + to_?hwi. If argmin is NULL, just set argmin and argmax to
38324 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
38325 of shortest and longest sequence.
38327 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
38329 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
38330 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
38332 2017-02-03 Walter Lee <walt@tilera.com>
38335 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
38336 after initial stackframe link reg save.
38337 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
38339 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38342 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
38343 wu for stxssp alternative.
38345 2017-02-03 Martin Sebor <msebor@redhat.com>
38347 PR tree-optimization/79352
38348 * gimple-fold.c (get_range_strlen): Add argument.
38349 (get_range_strlen): Change return type to bool.
38350 (get_maxval_strlen): Pass in a dummy argument.
38351 * gimple-fold.h (get_range_strlen): Change return type to bool.
38352 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
38353 * tree.h (array_at_struct_end_p): Add argument.
38354 * tree.c (array_at_struct_end_p): Handle it.
38356 2017-02-03 Martin Liska <mliska@suse.cz>
38359 * multiple_target.c (create_dispatcher_calls): Redirect edge
38360 from a caller of a dispatcher.
38361 (expand_target_clones): Make the clones local.
38362 (ipa_target_clone): Do both target clones and resolvers.
38363 (ipa_dispatcher_calls): Remove the pass.
38364 (pass_dispatcher_calls::gate): Likewise.
38365 (make_pass_dispatcher_calls): Likewise.
38366 * passes.def (pass_target_clone): Put as very first IPA early
38369 2017-02-03 Martin Liska <mliska@suse.cz>
38371 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
38372 in case of a function with ifunc attribute.
38374 2017-02-03 Martin Liska <mliska@suse.cz>
38376 * cgraph.c (cgraph_node::dump): Dump function version info.
38377 * symtab.c (symtab_node::dump_base): Add missing new line.
38379 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38381 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
38382 (ifcombine_ifandif): Use it.
38384 2017-02-03 Martin Liska <mliska@suse.cz>
38386 * doc/invoke.texi: Document default value for
38387 use-after-scope-direct-emission-threshold.
38389 2017-02-03 Martin Liska <mliska@suse.cz>
38391 PR tree-optimization/79339
38392 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
38393 (format_floating): Likewise.
38395 2017-02-03 Martin Liska <mliska@suse.cz>
38398 * ipa-prop.c (ipa_node_params_t::insert): Remove current
38400 (ipa_node_params_t::remove): Likewise.
38401 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
38402 initialization from removed ipa_node_params_t::insert.
38403 (ipa_node_params::~ipa_node_params): Move from removed
38404 ipa_node_params_t::release.
38405 * symbol-summary.h (symbol_summary::m_released): New member.
38406 Do not release a summary twice. Do not allow to call finalizer
38407 for types of a summary that live in GGC memory.
38409 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38411 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
38414 2017-02-02 Martin Sebor <msebor@redhat.com>
38416 PR middle-end/79275
38417 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
38418 (format_string): Tighten up the range of output for non-constant
38419 strings and correct the expected range for wide non-constant strings.
38421 2017-02-02 Martin Sebor <msebor@redhat.com>
38423 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
38425 PR middle-end/32003
38426 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
38428 (-fdump-tree-@var): Add to index and document how to come up
38429 with pass-specific option and dump file names.
38430 (-fdump-passes): Clarify where to look for output.
38432 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38434 PR middle-end/77445
38435 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
38436 statistics of the analyzed path; allow threading for speed when
38437 any of BBs along the path are optimized for speed.
38439 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
38441 PR middle-end/78468
38442 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
38443 settings of the virtual registers.
38446 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
38448 * explow.c (get_dynamic_stack_size): Take known alignment of stack
38449 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
38452 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38454 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
38455 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
38457 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38459 * config/s390/s390.md: Add missing comments with the expanded
38461 * config/s390/vector.md: Likewise.
38462 * config/s390/vx-builtins.md: Likewise.
38464 2017-02-02 Jakub Jelinek <jakub@redhat.com>
38467 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
38468 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
38469 conditions on a single line.
38471 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38473 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38474 __S390_VX__ to __VX__.
38476 2017-02-01 Andrew Pinski <apinski@cavium.com>
38478 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
38479 stmt_info to record_stmt_cost.
38480 (vect_get_known_peeling_cost): Pass stmt_info if known to
38482 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
38483 cpu_vector_cost field into
38484 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
38485 field into vec_int_stmt_cost and vec_fp_stmt_cost.
38486 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
38487 splitting of scalar_stmt_cost and vec_stmt_cost.
38488 (thunderx_vector_cost): Likewise.
38489 (cortexa57_vector_cost): LIkewise.
38490 (exynosm1_vector_cost): Likewise.
38491 (xgene1_vector_cost): Likewise.
38492 (thunderx2t99_vector_cost): Improve after the splitting of the two
38494 (aarch64_builtin_vectorization_cost): Update for the splitting of
38495 scalar_stmt_cost and vec_stmt_cost.
38497 2017-02-01 Torvald Riegel <triegel@redhat.com>
38498 Richard Henderson <rth@redhat.com>
38500 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
38501 conditional on existance of a fast atomic load.
38502 * optabs-query.c (can_atomic_load_p): New function.
38503 * optabs-query.h (can_atomic_load_p): Declare it.
38504 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
38505 no fast atomic load is available for the particular size of access.
38506 (expand_atomic_compare_and_swap): Likewise.
38507 (expand_atomic_load): Likewise.
38508 (expand_atomic_store): Likewise.
38509 (expand_atomic_fetch_op): Likewise.
38510 * testsuite/lib/target-supports.exp
38511 (check_effective_target_sync_int_128): Remove x86 because it provides
38512 no fast atomic load.
38513 (check_effective_target_sync_int_128_runtime): Likewise.
38515 2017-02-01 Richard Biener <rguenther@suse.de>
38517 * graphite.c: Include tree-vectorizer.h for find_loop_location.
38518 (graphite_transform_loops): Provide opt-info for optimized nests.
38519 * tree-parloop.c (parallelize_loops): Provide opt-info for
38520 parallelized loops.
38522 2017-02-01 Richard Biener <rguenther@suse.de>
38524 PR middle-end/79315
38525 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
38526 was not set before.
38528 2017-02-01 Richard Biener <rguenther@suse.de>
38530 PR tree-optimization/71824
38531 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38532 Verify the loops are valid in the merged SESE region.
38533 (scop_detection::can_represent_loop_1): Check analyzing the
38534 evolution of the number of iterations in the region succeeds.
38536 2017-01-31 Ian Lance Taylor <iant@golang.org>
38538 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
38539 REG_ARGS_SIZE note to 32-bit push insns and call insn.
38541 2017-01-31 David Malcolm <dmalcolm@redhat.com>
38543 PR preprocessor/79210
38544 * input.c (get_substring_ranges_for_loc): Replace line_width
38545 assertion with error-handling.
38547 2017-01-31 Richard Biener <rguenther@suse.de>
38549 PR tree-optimization/77318
38550 * graphite-sese-to-poly.c (extract_affine): Fix assert.
38551 (create_pw_aff_from_tree): Take loop parameter.
38552 (add_condition_to_pbb): Pass loop of the condition to
38553 create_pw_aff_from_tree.
38555 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38557 * config/s390/s390.c (s390_asan_shadow_offset): New function.
38558 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
38560 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
38564 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
38566 (convert_int_to_float128): Likewise.
38567 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
38568 (convert_int_to_float128): Likewise.
38569 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
38570 (UNSPEC_IEEE128_CONVERT): Likewise.
38571 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
38572 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
38573 Use local variables for IBM extended format.
38574 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
38575 (fix_trunc<mode>si2_fprs): Likewise.
38576 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
38577 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
38578 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
38579 to know that we can now have integers of all sizes in vector
38581 (fix<uns>_<mode>di2_hw): Likewise.
38582 (float<uns>_<mode>si2_hw): Likewise.
38583 (fix_<mode>si2_hw): Likewise.
38584 (fixuns_<mode>si2_hw): Likewise.
38585 (float<uns>_<mode>di2_hw): Likewise.
38586 (float_<mode>di2_hw): Likewise.
38587 (float_<mode>si2_hw): Likewise.
38588 (floatuns_<mode>di2_hw): Likewise.
38589 (floatuns_<mode>si2_hw): Likewise.
38590 (xscvqp<su>wz_<mode>): Delete, no longer used.
38591 (xscvqp<su>dz_<mode>): Likewise.
38592 (xscv<su>dqp_<mode>): Likewise.
38593 (ieee128_mfvsrd_64bit): Likewise.
38594 (ieee128_mfvsrd_32bit): Likewise.
38595 (ieee128_mfvsrwz): Likewise.
38596 (ieee128_mtvsrw): Likewise.
38597 (ieee128_mtvsrd_64bit): Likewise.
38598 (ieee128_mtvsrd_32bit): Likewise.
38600 2017-01-31 Martin Liska <mliska@suse.cz>
38603 * ipa-prop.c (ipa_free_all_node_params): Call release method
38604 instead of ~sumbol_summary to not to trigger double times
38607 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
38609 PR tree-optimization/71691
38610 * bitmap.h (class auto_bitmap): New.
38611 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
38612 is_maybe_undefined instead of ssa_undefined_value_p.
38614 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38616 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38617 __S390_ARCH_LEVEL__ to __ARCH__.
38619 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38621 PR tree-optimization/79267
38622 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38623 if should_remove_lhs_p is true.
38625 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38628 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38629 (add_alignment_attribute): New.
38630 (base_type_die): Add alignment attribute.
38631 (subrange_type_die): Likewise.
38632 (modified_type_die): Likewise.
38633 (gen_array_type_die): Likewise.
38634 (gen_descr_array_type_die: Likewise.
38635 (gen_enumeration_type_die): Likewise.
38636 (gen_subprogram_die): Likewise.
38637 (gen_variable_die): Likewise.
38638 (gen_field_die): Likewise.
38639 (gen_ptr_to_mbr_type_die): Likewise.
38640 (gen_struct_or_union_type_die): Likewise.
38641 (gen_subroutine_type_die): Likewise.
38642 (gen_typedef_die): Likewise.
38643 (base_type_cmp): Compare alignment attribute.
38645 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38648 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38649 (setb_unsigned) New pattern for setb with CCUNS.
38650 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38651 subfc./subfe sequence to avoid overflow problems. Generate a
38652 shorter sequence with cmpld/setb for power9.
38653 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38654 for generating subfc. instruction.
38655 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38656 now uses this instruction.
38658 2017-01-30 Ian Lance Taylor <iant@google.com>
38661 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38662 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38664 2017-01-30 Martin Sebor <msebor@redhat.com>
38666 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38667 Move constant to the right of a relational operator.
38668 (get_mpfr_format_length, format_character, format_string): Ditto.
38669 (should_warn_p, maybe_warn): Same.
38671 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38673 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38676 * asan.c (get_translation_unit_decl): Remove function.
38677 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38679 2017-01-30 Martin Liska <mliska@suse.cz>
38681 PR gcov-profile/79259
38682 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38683 -fprofile-generate.
38685 2017-01-30 Martin Liska <mliska@suse.cz>
38688 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38689 Initialize variables with NULL value.
38691 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38694 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38696 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38698 2017-01-30 Richard Biener <rguenther@suse.de>
38700 PR tree-optimization/79276
38701 * tree-vrp.c (process_assert_insertions): Properly adjust common
38702 when removing a duplicate.
38704 2017-01-30 Richard Biener <rguenther@suse.de>
38706 PR tree-optimization/79256
38707 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38708 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38710 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38712 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38715 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38716 ("*r<noxa>sbg_<mode>_sll_bitmask")
38717 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38718 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38719 Use contiguous_bitmask_nowrap_operand.
38721 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38724 * config/rs6000/altivec.h (vec_xl): Revise #define.
38725 (vec_xst): Likewise.
38727 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38729 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38731 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38733 PR rtl-optimization/79194
38734 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38735 traps before call to bypass_conditional_jumps.
38737 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38739 PR tree-optimization/71374
38740 * lra-constraints.c (check_conflict_input_operands): New.
38741 (match_reload): Use it.
38743 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38746 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38747 account to calculate conflict_set.
38749 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38751 PR rtl-optimization/78559
38752 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38753 other_insn in combine.
38755 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38757 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38758 uint16_type_node for BT_UINT16.
38760 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38762 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38763 "RTL Tests" to menu.
38764 (GIMPLE Tests): New node.
38765 (RTL Tests): New node.
38767 2017-01-27 Richard Biener <rguenther@suse.de>
38769 PR tree-optimization/79245
38770 * tree-loop-distribution.c (distribute_loop): Apply cost
38771 modeling also to detected patterns.
38773 2017-01-27 Richard Biener <rguenther@suse.de>
38775 PR tree-optimization/71433
38776 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38777 (compare_assert_loc): New function.
38778 (process_assert_insertions): Sort and optimize assert locations
38779 to remove duplicates and push down identical assertions on
38780 edges to their destination block.
38782 2017-01-27 Richard Biener <rguenther@suse.de>
38784 PR tree-optimization/79244
38785 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38786 out SSA names even if abnormal.
38788 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38790 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38791 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38792 instead of MPFR_RNDN.
38794 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38797 * arm.c (arm_option_override): Don't call build_target_option_node
38798 until after doing all option overrides.
38799 (arm_valid_target_attribute_tree): Likewise.
38801 2017-01-27 Martin Liska <mliska@suse.cz>
38803 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38804 for {cd}tors and C++ {cd}tors.
38806 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38808 * config/s390/s390.md ("*setmem_long_and")
38809 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38811 2017-01-26 Martin Sebor <msebor@redhat.com>
38813 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38816 2017-01-26 Martin Sebor <msebor@redhat.com>
38818 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38819 HAVE_DFmode before using XFmode or DFmode.
38820 (parse_directive): Avoid using the z length modifier to avoid
38821 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38823 PR middle-end/78703
38824 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38825 to accept adjustment as an array.
38826 (get_int_range): New function.
38827 (struct directive): Make width and prec arrays.
38828 (directive::set_width, directive::set_precision): Call get_int_range.
38829 (format_integer, format_floating): Handle width and precision ranges.
38830 (format_string, parse_directive): Same.
38832 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38835 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38836 comdat_type_p set, just clone them, but keep the children in the
38840 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38841 which have direct callers with -fvar-tracking-assignments enabled
38843 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38844 inside of type units.
38846 2017-01-26 Martin Sebor <msebor@redhat.com>
38848 PR middle-end/78703
38849 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38851 (struct format_result): Replace number_chars, number_chars_min,
38852 and number_chars_max with a single member of struct result_range.
38854 (format_result::operator+=): Adjust.
38855 (struct fmtresult): Remove bounded. Handle likely and unlikely
38857 (fmtresult::adjust_for_width_or_precision): New function.
38858 (fmtresult:type_max_digits): New function.
38859 (bytes_remaining): Handle likely and unlikely counters.
38860 (min_bytes_remaining): Remove.
38861 (format_percent): Simplify.
38862 (format_integer, format_floating): Set likely and unlikely counters.
38863 (get_string_length, format_character, format_string): Same.
38864 (format_plain, should_warn_p): New function.
38865 (maybe_warn): Call should_warn_p. Update diagnostic messages
38866 and handle those for all directives, including plain strings.
38867 (format_directive): Handle likely and unlikely counters.
38868 Remove unnecessary quoting from diagnostics. Add an informational
38870 (add_bytes): Remove.
38871 (pass_sprintf_length::compute_format_length): Simplify.
38872 (try_substitute_return_value): Handle likely and unlikely counters.
38874 2017-01-26 Carl Love <cel@us.ibm.com>
38876 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
38877 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
38879 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
38882 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
38883 endianess for subregs into account.
38884 * lra-constraints.c (lra_constraints): Do risky transformations
38885 always on the first iteration.
38886 * lra-lives.c (check_pseudos_live_through_calls): Add arg
38887 last_call_used_reg_set.
38888 (process_bb_lives): Define and use last_call_used_reg_set.
38889 * lra.c (lra): Always continue after lra_constraints on the first
38892 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
38894 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
38896 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
38898 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38900 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
38901 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
38902 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
38903 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
38904 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
38905 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
38906 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
38907 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
38908 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
38910 2017-01-26 Marek Polacek <polacek@redhat.com>
38913 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
38914 for the third operand.
38916 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38918 PR middle-end/79236
38919 * omp-low.c (struct omp_context): Add simt_stmt field.
38920 (scan_omp_for): Return omp_context *.
38921 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
38922 context to the _simt_ SIMD stmt.
38923 (lower_omp_for): For combined SIMD with sibling _simt_
38924 SIMD, make sure to use the same decls in _looptemp_
38925 clauses as in the sibling.
38927 2017-01-26 David Sherwood <david.sherwood@arm.com>
38929 PR middle-end/79212
38930 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
38933 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38936 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
38937 emit fld b; fld a; if possible.
38939 * brig-builtins.def: Update copyright years.
38940 * config/arm/arm_acle_builtins.def: Update copyright years.
38942 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
38945 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
38946 constraint instead of o for the stxsd instruction.
38948 2017-01-25 Carl Love <cel@us.ibm.com>
38950 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
38951 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
38953 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
38955 * doc/invoke.texi (C++ Dialect Options): Fix typo.
38957 2017-01-25 Richard Biener <rguenther@suse.de>
38959 PR tree-optimization/69264
38960 * target.def (vector_alignment_reachable): Improve documentation.
38961 * doc/tm.texi: Regenerate.
38962 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
38964 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
38965 earlier changes with respect to TYPE_USER_ALIGN.
38966 (vector_alignment_reachable_p): Likewise. Improve dumping.
38968 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38971 * config/arm/arm.md (xordi3): Force constant operand into a register
38974 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38976 * doc/invoke.texi (-fstore-merging): Correct default optimization
38977 levels at which it is enabled.
38978 (-O): Move -fstore-merging from list to...
38981 2017-01-25 Richard Biener <rguenther@suse.de>
38984 * omp-expand.c: Include debug.h.
38985 (expand_omp_taskreg): Make sure to generate early debug before
38986 outlining anything from a function.
38987 (expand_omp_target): Likewise.
38988 (grid_expand_target_grid_body): Likewise.
38990 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
38993 * asan.c (get_translation_unit_decl): New function.
38994 (asan_add_global): Extract modules file name from globals
38995 TRANSLATION_UNIT_DECL name.
38997 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39000 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39001 for long calls with APCS frame and VFP.
39003 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39005 * cfg.c (original_copy_tables_initialized_p): New function.
39006 * cfg.h (original_copy_tables_initialized_p): New decl.
39007 * cfgrtl.c (relink_block_chain): Guard the call to
39008 free_original_copy_tables with a call to
39009 original_copy_tables_initialized_p.
39010 * cgraph.h (symtab_node::native_rtl_p): New decl.
39011 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39012 (symtab_node::needed_p): Don't assert for early assembly output
39013 for __RTL functions.
39014 (cgraph_node::finalize_function): Set "force_output" for __RTL
39016 (cgraph_node::analyze): Bail out early for __RTL functions.
39017 (analyze_functions): Update assertion to support __RTL functions.
39018 (cgraph_node::expand): Bail out early for __RTL functions.
39019 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39021 * function.h (struct function): Update comment for field
39023 * gimple-expr.c: Include "tree-pass.h".
39024 (gimple_has_body_p): Return false for __RTL functions.
39025 * Makefile.in (OBJS): Add run-rtl-passes.o.
39026 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39028 (gcc::pass_manager::get_clean_slate): New accessor.
39029 * passes.c: Include "insn-addr.h".
39030 (should_skip_pass_p): Add logging. Update logic for running
39031 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39032 property-provider override so it is only done for gimple passes.
39034 (skip_pass): New function.
39035 (execute_one_pass): Call skip_pass when skipping passes.
39036 * read-md.c (md_reader::read_char): Support filtering
39037 the input to a subset of line numbers.
39038 (md_reader::md_reader): Initialize fields
39039 m_first_line and m_last_line.
39040 (md_reader::read_file_fragment): New function.
39041 * read-md.h (md_reader::read_file_fragment): New decl.
39042 (md_reader::m_first_line): New field.
39043 (md_reader::m_last_line): New field.
39044 * read-rtl-function.c (function_reader::create_function): Only
39045 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39046 curr_properties. Set DECL_INITIAL to a dummy block.
39047 (read_rtl_function_body_from_file_range): New function.
39048 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39050 * run-rtl-passes.c: New file.
39051 * run-rtl-passes.h: New file.
39053 2017-01-24 Jeff Law <law@redhat.com>
39055 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39058 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39060 PR tree-optimization/79159
39061 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39062 (record_nonwrapping_iv): Improve boundary using above function if no
39063 value range information.
39065 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39066 Martin Jambor <mjambor@suse.cz>
39068 * brig-builtins.def: New file.
39069 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39070 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39071 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39072 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39073 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39074 * builtin-types.def (BT_INT8): New.
39075 (BT_INT16): Likewise.
39076 (BT_UINT8): Likewise.
39077 (BT_UINT16): Likewise.
39078 (BT_FN_ULONG): Likewise.
39079 (BT_FN_UINT_INT): Likewise.
39080 (BT_FN_UINT_ULONG): Likewise.
39081 (BT_FN_UINT_LONG): Likewise.
39082 (BT_FN_UINT_PTR): Likewise.
39083 (BT_FN_ULONG_PTR): Likewise.
39084 (BT_FN_INT8_FLOAT): Likewise.
39085 (BT_FN_INT16_FLOAT): Likewise.
39086 (BT_FN_UINT32_FLOAT): Likewise.
39087 (BT_FN_UINT16_FLOAT): Likewise.
39088 (BT_FN_UINT8_FLOAT): Likewise.
39089 (BT_FN_UINT64_FLOAT): Likewise.
39090 (BT_FN_UINT16_UINT32): Likewise.
39091 (BT_FN_UINT32_UINT16): Likewise.
39092 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39093 (BT_FN_INT_PTR_INT): Likewise.
39094 (BT_FN_UINT_PTR_UINT): Likewise.
39095 (BT_FN_LONG_PTR_LONG): Likewise.
39096 (BT_FN_ULONG_PTR_ULONG): Likewise.
39097 (BT_FN_VOID_UINT64_UINT64): Likewise.
39098 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39099 (BT_FN_INT8_INT8_INT8): Likewise.
39100 (BT_FN_INT16_INT16_INT16): Likewise.
39101 (BT_FN_INT_INT_INT): Likewise.
39102 (BT_FN_UINT_FLOAT_UINT): Likewise.
39103 (BT_FN_FLOAT_UINT_UINT): Likewise.
39104 (BT_FN_ULONG_UINT_UINT): Likewise.
39105 (BT_FN_ULONG_UINT_PTR): Likewise.
39106 (BT_FN_ULONG_ULONG_ULONG): Likewise.
39107 (BT_FN_UINT_UINT_UINT): Likewise.
39108 (BT_FN_VOID_UINT_PTR): Likewise.
39109 (BT_FN_UINT_UINT_PTR: Likewise.
39110 (BT_FN_UINT32_UINT64_PTR): Likewise.
39111 (BT_FN_INT_INT_UINT_UINT): Likewise.
39112 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
39113 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
39114 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
39115 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
39116 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
39117 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
39118 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
39119 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
39120 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
39121 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
39122 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
39123 * doc/frontends.texi: List BRIG FE.
39124 * doc/install.texi (Testing): Add BRIG tesring requirements.
39125 * doc/invoke.texi (Overall Options): Mention BRIG.
39126 * doc/standards.texi (Standards): Doucment BRIG HSA version.
39128 2017-01-24 Richard Biener <rguenther@suse.de>
39130 PR translation/79208
39131 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
39133 2017-01-24 Martin Jambor <mjambor@suse.cz>
39136 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
39137 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
39138 and known_contexts.
39140 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
39142 PR middle-end/79123
39143 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
39144 casts from signed to unsigned really don't have a range.
39146 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
39148 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
39149 GMP_RNDx for compatiblity.
39151 2017-01-24 Martin Liska <mliska@suse.cz>
39154 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
39155 that would prevent us to call alloca with -1 as argument.
39157 2017-01-24 Jakub Jelinek <jakub@redhat.com>
39159 * dwarf2out.c (output_compilation_unit_header, output_file_names):
39160 Avoid -Wformat-security warning.
39162 2017-01-23 Andrew Pinski <apinski@cavium.com>
39164 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
39167 2017-01-23 Martin Sebor <msebor@redhat.com>
39169 PR middle-end/78703
39170 * gimple-ssa-sprintf.c (warn_level): New global.
39171 (format_integer): Use it here and throughout the rest of the file.
39172 Use the same switch to compute sign as base.
39173 (maybe_warn): New function.
39174 (format_directive): Factor out warnings into maybe_warn.
39175 Add debugging output. Use warn_level.
39176 (add_bytes): Use warn_level.
39177 (pass_sprintf_length::compute_format_length): Add debugging output.
39178 (try_substitute_return_value): Same.
39179 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
39181 PR middle-end/78703
39182 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
39183 (struct fmtresult, format_integer, format_floating): Adjust.
39184 (fmtresult::fmtresult): Set max correctly in two argument ctor.
39185 (get_string_length, format_string,format_directive): Same.
39186 (pass_sprintf_length::compute_format_length): Same.
39187 (try_substitute_return_value): Simplify slightly.
39189 PR middle-end/78703
39190 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
39191 (fmtresult::operator+=): Outlined.
39192 (struct fmtresult): Add ctors.
39193 (struct conversion_spec): Rename...
39194 (struct directive): ...to this. Add and remove data members.
39195 (directive::set_width, directive::set_precision): New functions.
39196 (format_percent): Use fmtresult ctor.
39197 (get_width_and_precision): Remove.
39198 (format_integer): Make naming changes. Avoid computing width and
39200 (format_floating): Same. Adjust indentation.
39201 (format_character, format_none): New functions.
39202 (format_string): Moved character handling to format_character.
39203 (format_directive): Remove arguments, change return type.
39204 (parse_directive): New function.
39205 (pass_sprintf_length::compute_format_length): Move directive
39206 parsing to parse_directive.
39208 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39210 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
39211 (assign_assembler_name_if_needed): ... this.
39212 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
39213 (assign_assembler_name_if_needed): ... this.
39214 (free_lang_data_in_cgraph): Adjust callers.
39215 * cgraphunit.c (cgraph_node::analyze): Likewise.
39216 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
39219 2017-01-23 Richard Biener <rguenther@suse.de>
39221 PR tree-optimization/79088
39222 PR tree-optimization/79188
39223 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
39224 resetting loop bounds after last path deletion. Reset loop
39225 bounds of the target loop, make code match the comments.
39226 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
39227 Make sure loops need no fixups.
39229 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
39231 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
39232 exponent support with double type for first argument.
39233 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
39234 type returned by __builtin_vec_extract_sig,
39235 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
39236 functions from "vector int" to "vector unsigned int" or from
39237 "vector long long int" to "vector unsigned long long int".
39238 Changed type returned by __builtin_vec_extract_exp,
39239 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
39240 functions from "vector int" to "vector unsigned int" or from
39241 "vector long long int" to "vector unsigned long long int".
39242 Changed return type of __builtin_vec_test_data_class,
39243 __builtin_vec_test_data_class_sp, and
39244 __builtin_vec_test_data_class_dp from "vector int" to
39245 "vector bool int" or from "vector long long int" to "vector bool
39246 long long int" and changed second argument type from "unsigned
39247 int" to "int". Added new overloaded function forms "vector float
39248 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
39249 "vector float __builtin_vec_insert_exp_sp (vector float, vector
39250 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
39251 double, vector unsigned long long int)" and "vector double
39252 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
39253 long int)". Changed return type of
39254 __builtin_scalar_test_data_class and
39255 __builtin_scalar_test_data_class_sp and
39256 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
39257 int" and changed second argument from "unsigned int" to "int".
39258 Changed type returned by __builtin_scalar_test_neg,
39259 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
39260 from "int" to "bool int". Added new overloaded function form
39261 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
39262 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
39263 exponent double-precision with floating point first argument.
39264 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
39265 documentation of scalar_test_data_class, scalar_test_neg,
39266 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
39267 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
39268 vec_test_data_class built-in functions to reflect refinements in
39269 their type signatures.
39271 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
39273 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
39275 (aarch64_elf_asm_destructor): Likewise.
39277 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
39279 PR rtl-optimization/78634
39280 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
39281 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
39282 * ifcvt.c (noce_try_cmove): Add missing cost check.
39284 PR rtl-optimization/71724
39285 * combine.c (if_then_else_cond): Look for situations where it is
39286 beneficial to undo the work of one of the recursive calls.
39288 2017-01-23 Bin Cheng <bin.cheng@arm.com>
39290 PR tree-optimization/70754
39291 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
39292 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
39293 combined stmt before it if not NULL.
39294 (combine_chains): Process refs reversely and compute dominance point
39297 2017-01-23 Martin Liska <mliska@suse.cz>
39299 PR tree-optimization/79196
39300 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
39301 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
39303 (strlen_optimize_stmt): Call the renamed function.
39305 2017-01-23 Michael Matz <matz@suse.de>
39307 PR tree-optimization/78384
39308 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
39310 2017-01-23 Richard Biener <rguenther@suse.de>
39312 PR tree-optimization/79186
39313 * tree-vrp.c (register_new_assert_for): Make sure we've seen
39314 both incoming edges before moving an assert.
39316 2017-01-23 Martin Jambor <mjambor@suse.cz>
39318 * ipa-prop.c (load_from_param_1): Removed.
39319 (load_from_unmodified_param): Bits from load_from_param_1 put back
39321 (load_from_param): Removed.
39322 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
39323 with stmt. Reverted back to use of load_from_unmodified_param.
39325 2017-01-23 Martin Jambor <mjambor@suse.cz>
39328 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
39329 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
39330 field a pointer to garbage collected vector, mark lattices and
39331 ipcp_orig_node with GTY((skip)).
39332 (ipa_get_param_count): Adjust to descriptors being a pointer.
39333 (ipa_get_param): Likewise.
39334 (ipa_get_type): Likewise.
39335 (ipa_get_param_move_cost): Likewise.
39336 (ipa_set_param_used): Likewise.
39337 (ipa_get_controlled_uses): Likewise.
39338 (ipa_set_controlled_uses): Likewise.
39339 (ipa_is_param_used): Likewise.
39340 (ipa_node_params_t): Move into garbage collector. New methods insert
39342 (ipa_node_params_sum): Annotate wth GTY(()).
39343 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
39345 (ipa_load_from_parm_agg): Adjust declaration.
39346 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
39347 * ipa-profile.c (ipa_profile): Likewise.
39348 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
39349 (ipa_populate_param_decls): Make descriptors parameter garbage
39351 (ipa_dump_param): Adjust to descriptors being a pointer.
39352 (ipa_alloc_node_params): Likewise.
39353 (ipa_initialize_node_params): Likewise.
39354 (load_from_param_1): Make descriptors parameter garbage collected.
39355 (load_from_unmodified_param): Likewise.
39356 (load_from_param): Likewise.
39357 (ipa_load_from_parm_agg): Likewise.
39358 (ipa_node_params::~ipa_node_params): Removed.
39359 (ipa_free_all_node_params): Remove call to delete operator.
39360 (ipa_node_params_t::insert): New.
39361 (ipa_node_params_t::remove): Likewise.
39362 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
39363 copy known_csts and known_contexts vectors.
39364 (ipa_read_node_info): Adjust to descriptors being a pointer.
39365 (ipcp_modif_dom_walker): Make m_descriptors field garbage
39367 (ipcp_transform_function): Make descriptors variable garbage
39370 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
39372 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
39373 * config/i386/avx512dqintrin.h: Ditto.
39374 * config/i386/avx512fintrin.h: Ditto.
39375 * config/i386/i386.c: Handle new builtins.
39376 * config/i386/i386-builtin.def: Add new builtins.
39377 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
39378 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
39380 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39381 Martin Liska <mliska@suse.cz>
39383 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
39384 * asan.c (asan_expand_poison_ifn): Support stores and use
39385 appropriate ASAN report function.
39386 * internal-fn.c (expand_ASAN_POISON_USE): New function.
39387 * internal-fn.def (ASAN_POISON_USE): Declare.
39388 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
39389 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
39390 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
39391 ASAN_POISON calls w/o LHS.
39392 * tree-ssa.c (execute_update_addresses_taken): Create clobber
39393 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
39394 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
39395 * gimplify.c (asan_poison_variables): Add attribute
39396 use_after_scope_memory to variables that really needs to live
39398 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
39399 having the attribute.
39401 2017-01-23 Martin Liska <mliska@suse.cz>
39403 * asan.c (create_asan_shadow_var): New function.
39404 (asan_expand_poison_ifn): Likewise.
39405 * asan.h (asan_expand_poison_ifn): New declaration.
39406 * internal-fn.c (expand_ASAN_POISON): Likewise.
39407 * internal-fn.def (ASAN_POISON): New builtin.
39408 * sanopt.c (pass_sanopt::execute): Expand
39409 asan_expand_poison_ifn.
39410 * tree-inline.c (copy_decl_for_dup_finish): Make function
39412 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
39413 * tree-ssa.c (is_asan_mark_p): New function.
39414 (execute_update_addresses_taken): Rewrite local variables
39415 (identified just by use-after-scope as addressable) into SSA.
39417 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39419 * doc/install.texi (Specific): opensource.apple.com uses https
39420 now. Remove trailing slash.
39422 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39424 * README.Portability: Remove note on an Irix compatibility issue.
39426 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
39428 * gcov.c (INCLUDE_ALGORITHM): Define.
39429 (INCLUDE_VECTOR): Define.
39430 No longer include <vector> and <algorithm> directly.
39432 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
39434 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
39436 * doc/invoke.texi (Code Gen Options): Ditto.
39438 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
39441 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
39443 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
39445 rtl-optimization/79125
39446 * cprop.c (local_cprop_pass): Handle cases where we make an
39447 unconditional trap.
39449 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
39453 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
39454 read from, for big endian.
39456 2017-01-20 Jiong Wang <jiong.wang@arm.com>
39458 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
39459 register pauth builtins for LP64 only.
39461 2017-01-20 Marek Polacek <polacek@redhat.com>
39464 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
39467 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39469 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
39471 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
39472 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
39473 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
39475 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39478 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
39479 in big-endian mode when they are not a single duplicated value.
39481 2017-01-20 Richard Biener <rguenther@suse.de>
39483 * BASE-VER: Bump to 7.0.1.
39485 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39487 * omp-low.c (omplow_simd_context): New struct. Use it...
39488 (lower_rec_simd_input_clauses): ...here and...
39489 (lower_rec_input_clauses): ...here to hold common data. Adjust all
39490 references to idx, lane, max_vf, is_simt.
39492 2017-01-20 Graham Markall <graham.markall@embecosm.com>
39494 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
39497 2017-01-20 Martin Jambor <mjambor@suse.cz>
39499 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
39500 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
39502 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
39503 (GTFILES): Rename hsa.c to hsa-common.c.
39504 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
39505 * hsa-dump.c: Likewise.
39506 * hsa-gen.c: Likewise.
39507 * hsa-regalloc.c: Likewise.
39508 * ipa-hsa.c: Likewise.
39509 * omp-expand.c: Likewise.
39510 * omp-low.c: Likewise.
39511 * toplev.c: Likewise.
39513 2017-01-20 Marek Polacek <polacek@redhat.com>
39516 * doc/invoke.texi: Document -Wduplicated-branches.
39517 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
39518 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
39519 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
39520 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
39521 return 0 only when not OEP_LEXICOGRAPHIC.
39522 (fold_build_cleanup_point_expr): Use the expression
39523 location when building CLEANUP_POINT_EXPR.
39524 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
39525 * tree.c (add_expr): Handle error_mark_node.
39527 2017-01-20 Martin Liska <mliska@suse.cz>
39530 * tree-profile.c (init_ic_make_global_vars): Do not call
39532 (gimple_init_gcov_profiler): Likewise.
39534 2017-01-20 Martin Liska <mliska@suse.cz>
39537 * cgraph.h (maybe_create_reference): Remove argument and
39539 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
39541 * ipa-cp.c (create_specialized_node): Likewise.
39542 * symtab.c (symtab_node::maybe_create_reference): Handle
39543 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
39545 2017-01-20 Martin Liska <mliska@suse.cz>
39547 * read-rtl-function.c (function_reader::create_function): Use
39548 build_decl instread of build_decl_stat.
39550 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
39552 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
39553 * config/i386/avx512dqintrin.h: Ditto.
39554 * config/i386/avx512fintrin.h: Ditto.
39555 * config/i386/i386-builtin-types.def: Add new types.
39556 * config/i386/i386.c: Handle new types.
39557 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
39558 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
39559 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
39560 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
39561 (__builtin_ia32_kshiftridi): New.
39562 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
39564 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
39568 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
39569 define to rs6000_init_stack_protect_guard.
39570 (rs6000_init_stack_protect_guard): New function.
39572 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39573 Yunqiang Su <yunqiang.su@imgtec.com>
39575 * config.gcc (supported_defaults): Add madd4.
39576 (with_madd4): Add validation.
39577 (all_defaults): Add madd4.
39578 * config/mips/mips.opt (mmadd4): New option.
39579 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39581 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39583 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
39584 (ISA_HAS_FUSED_MADD4): Likewise.
39585 * doc/invoke.texi (-mmadd4): Document the new option.
39586 * doc/install.texi (--with-madd4): Document the new option.
39588 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39590 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
39591 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
39592 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
39593 (aarch64_init_pauth_hint_builtins): New.
39594 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
39595 (aarch64_expand_builtin): Expand new builtins.
39597 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39599 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
39600 * combine-stack-adj.c (no_unhandled_cfa): Handle
39601 REG_CFA_TOGGLE_RA_MANGLE.
39602 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
39603 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
39604 info for return address signing.
39605 (aarch64_expand_epilogue): Likewise.
39607 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39609 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
39610 * config/aarch64/aarch64-protos.h
39611 (aarch64_return_address_signing_enabled): New declaration.
39612 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
39614 (aarch64_expand_prologue): Sign return address before it's pushed onto
39616 (aarch64_expand_epilogue): Authenticate return address fetched from
39618 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39619 (aarch64_attributes): New function attributes for "sign-return-address".
39620 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39621 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39622 ("*do_return"): Generate combined instructions according to key index.
39623 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39624 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39626 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39627 * config/aarch64/aarch64.opt (msign-return-address=): New.
39628 * doc/extend.texi (AArch64 Function Attributes): Documents
39629 "sign-return-address=".
39630 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39632 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39634 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39635 overall option summary.
39637 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39639 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39640 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39641 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39642 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39644 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39646 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39647 -mpower9-minmax by default for -mcpu=power9.
39648 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39649 128-bit floating point.
39651 2017-01-20 Alan Modra <amodra@gmail.com>
39653 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39654 optimizing for size.
39656 2017-01-20 Alan Modra <amodra@gmail.com>
39659 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39660 for strcmp and strncmp from corresponding builtin decl.
39662 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39664 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39665 instead of i386/rtems-64.h.
39666 * config/i386/rtems-64.h: Remove.
39668 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39672 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39674 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39676 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39678 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39679 Change int to HOST_WIDE_INT.
39680 * config/aarch64/aarch64-protos.h
39681 (aarch64_simd_gen_const_vector_dup): Likewise.
39682 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39684 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39686 * langhooks-def.h (lhd_type_for_size): New decl.
39687 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39688 * langhooks.c (lhd_type_for_size): New function, taken from
39691 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39693 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39694 define_bypass for CR latency.
39695 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39696 (power9-alu2): Add define_bypass for CR latency.
39698 (power9-mul): Update insn latency.
39699 (power9-mul-compare): Update insn latency, bypass latency and remove
39702 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39704 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39706 * config/aarch64/aarch64.md
39707 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39708 aarch64_nopcrelative_literal_loads.
39709 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39711 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39713 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39714 TARGET_LOONGSON_3A.
39715 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39717 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39720 * config.gcc (supported_defaults): Add lxc1-sxc1.
39721 (with_lxc1_sxc1): Add validation.
39722 (all_defaults): Add lxc1-sxc1.
39723 * config/mips/mips.opt (mlxc1-sxc1): New option.
39724 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39726 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39727 __mips_no_lxc1_sxc1.
39728 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39729 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39730 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39732 2017-01-19 Richard Biener <rguenther@suse.de>
39734 PR tree-optimization/72488
39735 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39736 sure to restore SSA info.
39737 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39739 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39741 PR rtl-optimization/79121
39742 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39743 of the inner type when shifting an extended value.
39745 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39748 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39749 interposable aliases.
39751 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39754 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39755 Use the evmergelohi instruction.
39756 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39757 (mov_sitf_e500_subreg8_2_be): Likewise.
39758 (mov_sitf_e500_subreg12_2_le): Likewise.
39759 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39760 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39761 (mov_sitf_e500_subreg8_2_le): Likewise.
39762 (mov_sitf_e500_subreg12_2_be): Likewise.
39764 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39766 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39767 attribute from vecsimple to vecperm.
39768 (altivec_vbpermq2): Likewise.
39770 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39773 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39775 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39776 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39777 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39778 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39779 case where N arg is SIZE_MAX.
39780 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39781 (cmpstrsi): Add pattern.
39783 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39785 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39786 __builtin_vec_revb builtins.
39787 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39788 built-in functions to support generation of the ISA 3.0 XXBR<x>
39789 vector byte reverse instructions.
39790 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39791 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39792 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39793 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39794 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39795 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39796 (P9V_BUILTIN_VEC_REVB): Likewise.
39797 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39798 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39799 (p9_xxbrq_v16qi): Likewise.
39800 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39801 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39802 (p9_xxbrh_v8hi): Likewise.
39803 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39804 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39805 vec_revb built-in functions.
39807 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39809 PR rtl-optimization/78952
39810 * config/i386/i386.md (any_extract): New code iterator.
39811 (*insvqi_2): Use any_extract for source operand.
39812 (*insvqi_3): Use any_shiftrt for source operand.
39814 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39816 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39818 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39820 2017-01-18 Matthias Klose <doko@ubuntu.com>
39822 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39824 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39826 * config/rs6000/altivec.h (vec_bperm): Change #define.
39827 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39828 (altivec_vbpermq2): New define_insn.
39829 (altivec_vbpermd): Likewise.
39830 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39831 function interface.
39832 (VBPERMD): Likewise.
39833 (VBPERM): New polymorphic function interface.
39834 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39835 Add entries for P9V_BUILTIN_VEC_VBPERM.
39836 * doc/extend.texi: Add interfaces for vec_bperm.
39838 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39840 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39841 first letter of error messages.
39842 (s390_resolve_overloaded_builtin): Likewise.
39843 * config/s390/s390.c (s390_expand_builtin): Likewise.
39844 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39845 (s390_valid_target_attribute_inner_p): Likewise.
39846 * config/s390/s390.md ("tabort"): Likewise.
39848 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39850 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39851 (ISA_AVOID_DIV_HILO): New macro.
39852 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39853 (ISA_HAS_DDIV): Likewise.
39855 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39857 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39859 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39861 * doc/invoke.texi (fabi-version): Spelling fix.
39863 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39866 * doc/invoke.texi (fabi-version): Mention mangling fix for
39869 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39872 * doc/invoke.texi (fabi-version): Document discriminator mangling.
39874 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
39877 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
39878 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
39880 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
39882 (stack_protect_test): Ditto.
39883 * config/rs6000/rs6000.opt (mstack-protector-guard=,
39884 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
39886 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
39887 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
39888 -mstack-protector-guard-offset=.
39889 (RS/6000 and PowerPC Options): Ditto.
39891 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39893 * config/i386/i386.h (MASK_CLASS_P): New define.
39894 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
39895 there are no registers from different register sets also when
39896 mask registers are used. Update function comment.
39897 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
39898 to (*k/*r) and (*k/*km) alternatives.
39900 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
39902 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
39903 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
39904 (EH_RETURN_HANDLER_RTX): New define.
39905 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
39906 Force frame pointer in EH return functions.
39907 (aarch64_expand_epilogue): Add barrier for eh_return.
39908 (aarch64_final_eh_return_addr): Remove.
39909 (aarch64_eh_return_handler_rtx): New function.
39910 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
39912 (aarch64_eh_return_handler_rtx): New prototype.
39914 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39916 * config/rs6000/altivec.h (vec_rlmi): New #define.
39917 (vec_vrlnm): Likewise.
39918 (vec_rlnm): Likewise.
39919 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
39920 (UNSPEC_VRLNM): Likewise.
39921 (VIlong): New mode iterator.
39922 (altivec_vrl<VI_char>mi): New define_insn.
39923 (altivec_vrl<VI_char>nm): Likewise.
39924 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
39926 (VRLDNM): Likewise.
39927 (RLNM): New polymorphic function entry.
39928 (VRLWMI): New monomorphic function entry.
39929 (VRLDMI): Likewise.
39930 (RLMI): New polymorphic function entry.
39931 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
39932 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
39933 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
39936 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39939 * dwarf2out.c (field_byte_offset): Restore the
39940 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
39941 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
39942 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
39945 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39948 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
39950 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39953 * dwarf2out.c (add_data_member_location_attribute): For constant
39954 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
39955 instead of DW_AT_data_member_location, DW_AT_bit_offset and
39956 DW_AT_byte_size attributes.
39958 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39960 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
39961 after forcing to constant memory when the code model is medium.
39963 2017-01-17 Julia Koval <julia.koval@intel.com>
39966 * config/i386/avx512fintrin.h
39967 (_mm512_i32gather_ps): Change __addr type to void const*.
39968 (_mm512_mask_i32gather_ps): Ditto.
39969 (_mm512_i32gather_pd): Ditto.
39970 (_mm512_mask_i32gather_pd): Ditto.
39971 (_mm512_i64gather_ps): Ditto.
39972 (_mm512_mask_i64gather_ps): Ditto.
39973 (_mm512_i64gather_pd): Ditto.
39974 (_mm512_mask_i64gather_pd): Ditto.
39975 (_mm512_i32gather_epi32): Ditto.
39976 (_mm512_mask_i32gather_epi32): Ditto.
39977 (_mm512_i32gather_epi64): Ditto.
39978 (_mm512_mask_i32gather_epi64): Ditto.
39979 (_mm512_i64gather_epi32): Ditto.
39980 (_mm512_mask_i64gather_epi32): Ditto.
39981 (_mm512_i64gather_epi64): Ditto.
39982 (_mm512_mask_i64gather_epi64): Ditto.
39983 (_mm512_i32scatter_ps): Change __addr type to void*.
39984 (_mm512_mask_i32scatter_ps): Ditto.
39985 (_mm512_i32scatter_pd): Ditto.
39986 (_mm512_mask_i32scatter_pd): Ditto.
39987 (_mm512_i64scatter_ps): Ditto.
39988 (_mm512_mask_i64scatter_ps): Ditto.
39989 (_mm512_i64scatter_pd): Ditto.
39990 (_mm512_mask_i64scatter_pd): Ditto.
39991 (_mm512_i32scatter_epi32): Ditto.
39992 (_mm512_mask_i32scatter_epi32): Ditto.
39993 (_mm512_i32scatter_epi64): Ditto.
39994 (_mm512_mask_i32scatter_epi64): Ditto.
39995 (_mm512_i64scatter_epi32): Ditto.
39996 (_mm512_mask_i64scatter_epi32): Ditto.
39997 (_mm512_i64scatter_epi64): Ditto.
39998 (_mm512_mask_i64scatter_epi64): Ditto.
39999 * config/i386/avx512pfintrin.h
40000 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40001 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40002 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40003 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40004 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40005 (_mm512_prefetch_i32scatter_ps): Ditto.
40006 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40007 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40008 (_mm512_prefetch_i64scatter_pd): Ditto.
40009 (_mm512_prefetch_i64scatter_ps): Ditto.
40010 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40011 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40012 * config/i386/avx512vlintrin.h
40013 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40014 (_mm_mmask_i32gather_ps): Ditto.
40015 (_mm256_mmask_i32gather_pd): Ditto.
40016 (_mm_mmask_i32gather_pd): Ditto.
40017 (_mm256_mmask_i64gather_ps): Ditto.
40018 (_mm_mmask_i64gather_ps): Ditto.
40019 (_mm256_mmask_i64gather_pd): Ditto.
40020 (_mm_mmask_i64gather_pd): Ditto.
40021 (_mm256_mmask_i32gather_epi32): Ditto.
40022 (_mm_mmask_i32gather_epi32): Ditto.
40023 (_mm256_mmask_i32gather_epi64): Ditto.
40024 (_mm_mmask_i32gather_epi64): Ditto.
40025 (_mm256_mmask_i64gather_epi32): Ditto.
40026 (_mm_mmask_i64gather_epi32): Ditto.
40027 (_mm256_mmask_i64gather_epi64): Ditto.
40028 (_mm_mmask_i64gather_epi64): Ditto.
40029 (_mm256_i32scatter_ps): Change __addr type to void*.
40030 (_mm256_mask_i32scatter_ps): Ditto.
40031 (_mm_i32scatter_ps): Ditto.
40032 (_mm_mask_i32scatter_ps): Ditto.
40033 (_mm256_i32scatter_pd): Ditto.
40034 (_mm256_mask_i32scatter_pd): Ditto.
40035 (_mm_i32scatter_pd): Ditto.
40036 (_mm_mask_i32scatter_pd): Ditto.
40037 (_mm256_i64scatter_ps): Ditto.
40038 (_mm256_mask_i64scatter_ps): Ditto.
40039 (_mm_i64scatter_ps): Ditto.
40040 (_mm_mask_i64scatter_ps): Ditto.
40041 (_mm256_i64scatter_pd): Ditto.
40042 (_mm256_mask_i64scatter_pd): Ditto.
40043 (_mm_i64scatter_pd): Ditto.
40044 (_mm_mask_i64scatter_pd): Ditto.
40045 (_mm256_i32scatter_epi32): Ditto.
40046 (_mm256_mask_i32scatter_epi32): Ditto.
40047 (_mm_i32scatter_epi32): Ditto.
40048 (_mm_mask_i32scatter_epi32): Ditto.
40049 (_mm256_i32scatter_epi64): Ditto.
40050 (_mm256_mask_i32scatter_epi64): Ditto.
40051 (_mm_i32scatter_epi64): Ditto.
40052 (_mm_mask_i32scatter_epi64): Ditto.
40053 (_mm256_i64scatter_epi32): Ditto.
40054 (_mm256_mask_i64scatter_epi32): Ditto.
40055 (_mm_i64scatter_epi32): Ditto.
40056 (_mm_mask_i64scatter_epi32): Ditto.
40057 (_mm256_i64scatter_epi64): Ditto.
40058 (_mm256_mask_i64scatter_epi64): Ditto.
40059 (_mm_i64scatter_epi64): Ditto.
40060 (_mm_mask_i64scatter_epi64): Ditto.
40061 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40062 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40063 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40064 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40065 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40066 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40067 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40068 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40069 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40070 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40071 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40072 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40073 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40074 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40075 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40076 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40077 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40078 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40079 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40080 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40081 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40082 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40083 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40084 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40085 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40086 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40087 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40088 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40089 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40090 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40091 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40092 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40093 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40094 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40095 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40096 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40097 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40098 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40099 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40100 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40101 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40102 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40103 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40104 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40105 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
40106 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
40107 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
40108 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
40109 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
40110 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
40111 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
40112 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
40113 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
40114 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
40115 definitions accordingly.
40117 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
40118 Kuan-Lin Chen <kuanlinchentw@gmail.com>
40121 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
40124 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
40127 * ira-conflicts.c (ira_build_conflicts): Update total conflict
40128 hard regs for inner regno.
40130 2017-01-17 Martin Liska <mliska@suse.cz>
40133 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
40134 assumption and add comment.
40136 2017-01-17 Nathan Sidwell <nathan@acm.org>
40138 * ipa-visibility.c (localize_node): New function, broken out of ...
40139 (function_and_variable_visibility): ... here. Call it.
40141 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40143 PR middle-end/77445
40144 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
40145 correctly set frequency of oudgoing edge.
40146 (duplicate_thread_path): Fix profile updating.
40148 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40151 * configure.ac: Add GCC_BASE_VER.
40152 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
40153 version from BASE-VER file.
40154 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
40155 (gcc.o): Depend on $(BASEVER).
40156 * common.opt (dumpfullversion): New option.
40157 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
40158 * doc/invoke.texi: Document -dumpfullversion.
40159 * doc/install.texi: Document --with-gcc-major-version-only.
40160 * configure: Regenerated.
40162 2017-01-17 Richard Biener <rguenther@suse.de>
40164 PR tree-optimization/71433
40165 * tree-vrp.c (register_new_assert_for): Merge same asserts
40166 on all incoming edges.
40167 (process_assert_insertions_for): Handle insertions at the
40170 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
40172 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
40173 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
40175 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
40178 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
40181 2017-01-17 Alan Modra <amodra@gmail.com>
40184 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
40185 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
40186 symbolic stack limit when pic.
40188 2017-01-16 Martin Sebor <msebor@redhat.com>
40190 PR tree-optimization/78608
40191 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
40193 2017-01-16 Jeff Law <law@redhat.com>
40196 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
40197 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
40198 for several include directories that may be relative to sysroot.
40199 * config/i386/x-mingw32 (gplus_includedir): Define.
40200 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
40201 (native_system_includedir): Likewise.
40202 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
40203 override if TARGET_SYSTEM_ROOT is defined.
40204 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
40206 PR tree-optimization/79090
40207 PR tree-optimization/33562
40208 PR tree-optimization/61912
40209 PR tree-optimization/77485
40210 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
40211 and computed trims into the dump file.
40213 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40215 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
40217 2017-01-16 Jakub Jelinek <jakub@redhat.com>
40220 * gimplify.c (gimplify_init_constructor): If want_value and
40221 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
40225 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
40226 sequence. Formatting fixes.
40227 (doloop_optimize): Formatting fixes.
40230 * gcc.c (debug_level_greater_than_spec_func): New function.
40231 (static_spec_functions): Add debug-level-gt spec function.
40232 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
40234 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40235 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40236 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
40237 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
40238 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
40239 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
40241 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
40243 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
40244 QImode fixups to general and mask registers only.
40246 2017-01-16 Carl Love <cel@us.ibm.com>
40248 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
40249 for built-in functions
40250 vector signed char vec_nabs (vector signed char)
40251 vector signed short vec_nabs (vector signed short)
40252 vector signed int vec_nabs (vector signed int)
40253 vector signed long long vec_nabs (vector signed long long)
40254 vector float vec_nabs (vector float)
40255 vector double vec_nabs (vector double)
40256 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
40258 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
40259 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
40260 * doc/extend.texi: Update the documentation file for the new built-in
40263 2017-01-16 Martin Sebor <msebor@redhat.com>
40265 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
40268 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40270 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
40271 UNSPEC_VSX__XXSPLTD to require special splat handling.
40273 2017-01-16 David Malcolm <dmalcolm@redhat.com>
40276 * system.h: Poison strndup.
40278 2017-01-16 Alan Modra <amodra@gmail.com>
40281 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
40284 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
40286 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
40288 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
40290 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
40291 call recog here. Assert that INSN_CODE (insn) is non-negative.
40293 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40296 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
40298 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
40299 in the currently scheduled RTL fragment.
40301 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40303 PR rtl-optimization/78751
40304 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
40307 2017-01-14 Jeff Law <law@redhat.com>
40309 PR tree-optimization/79090
40310 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
40311 variable length stores.
40312 (compute_trims): Delete dead assignment to *trim_tail.
40313 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
40316 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
40318 PR rtl-optimization/78626
40319 PR rtl-optimization/78727
40320 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
40321 of a block, and split such blocks after everything else is finished.
40323 2017-01-14 Alan Modra <amodra@gmail.com>
40326 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
40327 target legitimate_combined_insn.
40328 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
40329 (rs6000_legitimate_combined_insn): New function.
40330 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
40332 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
40333 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
40334 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
40336 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
40338 * doc/frontends.texi (G++ and GCC): Remove references to Java.
40340 2017-01-13 Jeff Law <law@redhat.com>
40342 PR tree-optimization/33562
40343 PR tree-optimization/61912
40344 PR tree-optimization/77485
40345 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
40347 (delete_dead_assignment): Likewise.
40348 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
40349 statement to delete_dead_call and delete_dead_assignment.
40351 2017-01-13 David Malcolm <dmalcolm@redhat.com>
40354 * substring-locations.c (format_warning_va): Strengthen case 1 so
40355 that both endpoints of the substring must be within the format
40356 range for just the substring to be printed.
40358 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
40360 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
40361 * config/i386/i386.c (ix86_target_string): Add missing options
40362 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
40363 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
40364 flags_other and ix86_target_other to flags2_other. Display unknown
40366 (ix86_valid_target_attribute_inner_p): Add missing options and
40367 reorder options by implied ISAs, as in ix86_target_string.
40369 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40371 * hash-table.h (hash_table::too_empty_p): New function.
40372 (hash_table::expand): Use it.
40373 (hash_table::traverse): Likewise.
40374 (hash_table::empty_slot): Use sizeof (value_type) instead of
40375 sizeof (PTR) to convert bytes to elements. Shrink the table
40376 if the current size is excessive for the current number of
40379 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40381 * ira-costs.c (record_reg_classes): Break from the inner loop
40382 early once alt_fail is known to be true. Update outer loop
40383 handling accordingly.
40385 2017-01-13 Jeff Law <law@redhat.com>
40387 * tree-ssa-dse.c (decrement_count): New function.
40388 (increment_start_addr, maybe_trim_memstar_call): Likewise.
40389 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
40390 when we know the partially dead statement is a mem* function.
40392 PR tree-optimization/61912
40393 PR tree-optimization/77485
40394 * tree-ssa-dse.c: Include expr.h.
40395 (maybe_trim_constructor_store): New function.
40396 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
40398 PR tree-optimization/33562
40399 PR tree-optimization/61912
40400 PR tree-optimization/77485
40401 * doc/invoke.texi: Document new dse-max-object-size param.
40402 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
40403 * tree-ssa-dse.c: Include params.h.
40404 (dse_store_status): New enum.
40405 (initialize_ao_ref_for_dse): New, partially extracted from
40407 (valid_ao_ref_for_dse, normalize_ref): New.
40408 (setup_live_bytes_from_ref, compute_trims): Likewise.
40409 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
40410 (maybe_trim_partially_dead_store): Likewise.
40411 (maybe_trim_complex_store): Likewise.
40412 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
40413 Track what bytes live from the original store. Return tri-state
40414 for dead, partially dead or live.
40415 (dse_dom_walker): Add constructor, destructor and new private members.
40416 (delete_dead_call, delete_dead_assignment): New extracted from
40418 (dse_optimize_stmt): Make a member of dse_dom_walker.
40419 Use initialize_ao_ref_for_dse.
40421 PR tree-optimization/33562
40422 PR tree-optimization/61912
40423 PR tree-optimization/77485
40424 * sbitmap.h (bitmap_count_bits): Prototype.
40425 (bitmap_clear_range, bitmap_set_range): Likewise.
40426 * sbitmap.c (bitmap_clear_range): New function.
40427 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
40429 2017-01-13 Martin Liska <mliska@suse.cz>
40432 * function.c (set_cfun): Add new argument force.
40433 * function.h (set_cfun): Likewise.
40434 * ipa-inline-transform.c (inline_call): Use the function when
40435 strict alising from is dropped for function we inline to.
40437 2017-01-13 Richard Biener <rguenther@suse.de>
40439 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
40440 for dumping GIMPLE INTEGER_CSTs.
40442 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
40444 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
40445 to 201112L since C++17.
40447 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
40450 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
40451 if -fsanitize=kernel-address is present.
40453 2017-01-13 Richard Biener <rguenther@suse.de>
40455 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
40456 as _Literal ( type ) number in case usual suffixes do not
40457 preserve all information.
40459 2017-01-13 Richard Biener <rguenther@suse.de>
40461 PR tree-optimization/77283
40462 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
40463 and ssa-iterators.h.
40464 (is_feasible_trace): Implement a cost model based on joiner
40467 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
40470 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
40471 char or short to __float128/_Float128 directly.
40473 2017-01-12 Martin Sebor <msebor@redhat.com>
40475 to -Wformat-overflow.
40476 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
40477 (min_bytes_remaining): Same.
40478 (get_string_length): Same.
40479 (format_string): Same.
40480 (format_directive): Same.
40482 (pass_sprintf_length::handle_gimple_call): Same.
40484 2017-01-12 Jakub Jelinek <jakub@redhat.com>
40486 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
40487 info.nowrite calls with no lhs that can't throw. Return bool
40488 whether gsi_remove has been called or not.
40489 (pass_sprintf_length::handle_gimple_call): Return bool whether
40490 try_substitute_return_value called gsi_remove. Formatting fix.
40491 (pass_sprintf_length::execute): Don't use gsi_remove if
40492 handle_gimple_call returned true.
40495 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
40496 be removed due to side-effects, don't remove following barrier nor
40497 turn the successor edge into fallthru edge.
40499 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40502 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
40503 element-reversing loads and stores as not swappable.
40505 2017-01-12 Nathan Sidwell <nathan@acm.org>
40506 Nicolai Stange <nicstange@gmail.com>
40508 * combine.c (try_combine): Don't ignore result of overlap checking
40509 loop. Combine overlap & asm check into single loop.
40511 2017-01-12 Richard Biener <rguenther@suse.de>
40513 * tree-pretty-print.c (dump_generic_node): Provide -gimple
40514 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
40516 2017-01-12 Richard Biener <rguenther@suse.de>
40518 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
40519 and TS_TARGET_OPTION directly derive from TS_BASE.
40520 * tree-core.h (tree_optimization_option): Derive from tree_base.
40521 (tree_target_option): Likewise.
40523 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
40525 * config/i386/i386.c (memory_address_length): Increase len
40526 only when rip_relative_addr_p returns false.
40528 2017-01-11 Julia Koval <julia.koval@intel.com>
40530 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
40531 (OPTION_MASK_ISA_SGX_SET): New.
40532 (ix86_handle_option): Handle OPT_msgx.
40533 * config.gcc: Added sgxintrin.h.
40534 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
40535 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
40536 * config/i386/i386.c (ix86_target_string): Add -msgx.
40538 (ix86_option_override_internal): Handle new options.
40539 (ix86_valid_target_attribute_inner_p): Add sgx.
40540 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
40541 * config/i386/i386.opt: Add msgx.
40542 * config/i386/sgxintrin.h: New file.
40543 * config/i386/x86intrin.h: Add sgxintrin.h.
40545 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40548 * fold-const.c (maybe_nonzero_address): Return 1 for function
40550 (tree_single_nonzero_warnv_p): Don't handle function local objects
40554 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
40557 2017-01-11 David Malcolm <dmalcolm@redhat.com>
40560 * opts.c: Include "spellcheck.h"
40561 (struct string_fragment): New struct.
40562 (struct edit_distance_traits<const string_fragment &>): New
40564 (get_closest_sanitizer_option): New function.
40565 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
40567 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40569 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
40571 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
40572 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
40573 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
40574 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
40575 for initial die_offset if dwarf_split_debug_info.
40576 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
40577 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
40579 (output_skeleton_debug_sections): Formatting fix. Use
40580 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
40581 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
40583 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
40585 * config/arm/cortex-a53.md: Add bypasses for
40586 cortex_a53_r2f_cvt.
40587 (cortex_a53_r2f): Only use for transfers.
40588 (cortex_a53_f2r): Likewise.
40589 (cortex_a53_r2f_cvt): Add reservation for conversions.
40590 (cortex_a53_f2r_cvt): Likewise.
40592 2017-01-11 Tamar Christina <tamar.christina@arm.com>
40594 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
40595 to all inlined functions, change static to extern.
40597 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
40600 * config/arm/arm.c (legitimize_pic_address): Handle reference to
40602 (arm_assemble_integer): Likewise.
40604 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
40606 * config.gcc: Use new awk script to check CPU, FPU and architecture
40607 parameters for --with-... options.
40608 * config/arm/parsecpu.awk: New file
40609 * config/arm/arm-cpus.in: New file.
40610 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
40612 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
40614 * config/arm/t-arm: Update dependency rules.
40615 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
40616 of processing .def files.
40617 * config/arm/genopt.sh: Deleted.
40618 * config/arm/gentune.sh: Deleted.
40619 * config/arm/arm-cores.def: Deleted.
40620 * config/arm/arm-arches.def: Deleted.
40621 * config/arm/arm-fpus.def: Deleted.
40622 * config/arm/arm-tune.md: Regenerated.
40623 * config/arm/arm-tables.opt: Regenerated.
40624 * config/arm/arm-cpu.h: New generated file.
40625 * config/arm/arm-cpu-data.h: New generated file.
40626 * config/arm/arm-cpu-cdata.h: New generated file.
40628 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40631 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40633 (input_varpool_node): Unpack dynamically_initialized bit.
40635 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40637 PR rtl-optimization/79032
40638 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40639 the alignment of the adjusted memory reference against that of MODE,
40640 instead of the alignment of the original memory reference.
40642 2017-01-11 Martin Jambor <mjambor@suse.cz>
40644 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40646 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40647 decorated functions.
40649 2017-01-11 Richard Biener <rguenther@suse.de>
40651 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40652 set range/nonnull info for PHI results. Do not set it on
40653 stmts marked for removal.
40655 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40657 * expr.c (store_field): In the bitfield case, fetch the return value
40658 from the registers before applying a single big-endian adjustment.
40659 Always do a final load for a BLKmode value not larger than a word.
40661 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40664 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40665 that we correctly handle column numbers greater than
40666 LINE_MAP_MAX_COLUMN_NUMBER.
40668 2017-01-10 Martin Sebor <msebor@redhat.com>
40670 PR middle-end/78245
40671 * gimple-ssa-sprintf.c (get_destination_size): Call
40672 {init,fini}object_sizes.
40673 * tree-object-size.c (addr_object_size): Adjust.
40674 (pass_through_call): Adjust.
40675 (pass_object_sizes::execute): Adjust.
40676 * tree-object-size.h (fini_object_sizes): Declare.
40678 2017-01-10 Martin Sebor <msebor@redhat.com>
40680 PR tree-optimization/78775
40681 * builtins.c (get_size_range): Move...
40682 * calls.c: ...to here.
40683 (alloc_max_size): Accept zero argument.
40684 (operand_signed_p): Remove.
40685 (maybe_warn_alloc_args_overflow): Call get_size_range.
40686 * calls.h (get_size_range): Declare.
40688 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40690 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40691 from TI's devices.csv file as of September 2016.
40692 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40694 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40696 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40697 * doc/invoke.texi: Likewise.
40698 * doc/md.texi: Likewise.
40699 * doc/objc.texi: Likewise.
40701 2017-01-10 Joshua Conner <joshconner@google.com>
40703 * config/arm/fuchsia-elf.h: New file.
40704 * config/fuchsia.h: New file.
40705 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40706 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40708 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40710 2016-01-10 Richard Biener <rguenther@suse.de>
40712 PR tree-optimization/79034
40713 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40714 Propagate out degenerate PHIs in the joiner.
40716 2017-01-10 Martin Liska <mliska@suse.cz>
40718 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40719 (sort_congruence_classes_by_decl_uid): Likewise.
40720 (sort_congruence_class_groups_by_decl_uid): Likewise.
40721 (sem_item_optimizer::merge_classes): Sort class, groups in these
40722 classes and members in the groups by DECL_UID of declarations.
40723 This would make merge operations stable.
40725 2017-01-10 Martin Liska <mliska@suse.cz>
40727 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40728 usage of m_classes_vec.
40729 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40730 (sem_item_optimizer::get_group_by_hash): Likewise.
40731 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40732 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40733 (sem_item_optimizer::verify_classes): Likewise.
40734 (sem_item_optimizer::process_cong_reduction): Likewise.
40735 (sem_item_optimizer::dump_cong_classes): Likewise.
40736 (sem_item_optimizer::merge_classes): Likewise.
40737 * ipa-icf.h (congruence_class_hash): Rename from
40738 congruence_class_group_hash. Remove declaration of m_classes_vec.
40740 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40742 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40743 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40744 * config.gcc: Add avx512vpopcntdqintrin.h.
40745 * config/i386/avx512vpopcntdqintrin.h: New.
40746 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40747 * config/i386/i386-builtin-types.def: Add new types.
40748 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40749 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40750 __builtin_ia32_vpopcountq_v8di_mask): New.
40751 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40752 __AVX512VPOPCNTDQ__.
40753 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40754 (PTA_AVX512VPOPCNTDQ): Define.
40755 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40756 TARGET_AVX512VPOPCNTDQ_P): Define.
40757 * config/i386/i386.opt: Add mavx512vpopcntdq.
40758 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40759 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40761 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40763 PR middle-end/77484
40764 * predict.def (PRED_CALL): Set to 67.
40766 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40768 * expr.c (store_field): In the bitfield case, if the value comes from
40769 a function call and is of an aggregate type returned in registers, do
40770 not modify the field mode; extract the value in all cases if the mode
40771 is BLKmode and the size is not larger than a word.
40773 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40776 * config/i386/cpuid.h: Fix undefined behavior.
40778 2017-01-04 Jeff Law <law@redhat.com>
40780 PR tree-optimization/79007
40781 PR tree-optimization/67955
40782 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40783 conservative for pt.null when flag_non_call_exceptions is on.
40785 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40787 PR translation/79019
40788 PR translation/79020
40789 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40790 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40791 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40793 * config/avr/avr.opt (maccumulate-args): Likewise.
40794 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40795 * common.opt (freport-bug): Likewise.
40796 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40797 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40798 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40799 translatable string.
40800 * config/i386/i386.c (function_value_32): Likewise.
40801 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40802 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40804 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40805 * common/config/msp430/msp430-common.c (msp430_handle_option):
40807 * symtab.c (symtab_node::verify_base): Likewise.
40808 * opts.c (set_debug_level): Likewise.
40809 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40810 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40811 missing whitespace to translatable strings.
40812 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40813 * config/sh/superh.h: Likewise.
40814 * config/i386/xopintrin.h: Likewise.
40815 * config/i386/znver1.md: Likewise.
40816 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40817 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40818 * double-int.h (struct double_int): Likewise.
40819 * double-int.c (div_and_round_double): Likewise.
40820 * wide-int.cc: Likewise.
40821 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40822 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40823 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40824 (crossjumps_occurred): ... this.
40825 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40828 PR tree-optimization/78899
40829 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40830 returning bool return struct loop *, NULL for failure and the new
40832 (versionable_outer_loop_p): Don't version outer loop if it has
40833 dont_vectorized bit set.
40834 (tree_if_conversion): When versioning outer loop, ensure
40835 tree_if_conversion is performed also on the inner loop of the
40836 non-vectorizable outer loop copy.
40837 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40838 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40839 prevent vectorization of it.
40840 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40841 the outer loop vectorization of the non-scalar version is attempted
40842 before vectorization of the inner loop in scalar version. If
40843 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40844 vectorization of its inner loop.
40845 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40846 has 2 inner loops, rename also on edges from bb whose single pred
40847 is outer_loop->header. Fix typo in function comment.
40849 2017-01-09 Martin Sebor <msebor@redhat.com>
40852 * asan.c (asan_emit_stack_protection): Increase local buffer size
40853 to avoid snprintf truncation warning.
40855 2017-01-09 Andrew Pinski <apinski@cavium.com>
40857 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40858 to reference thunderx2t99 for the tuning structure
40859 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40861 (thunderx2t99_extra_costs): This.
40862 * config/aarch64/aarch64-tune.md: Regenerate.
40863 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
40864 (vulcan_addrcost_table): This.
40865 (vulcan_regmove_cost): Rename to ...
40866 (thunderx2t99_regmove_cost): This.
40867 (vulcan_vector_cost): Rename to ...
40868 (thunderx2t99_vector_cost): this.
40869 (vulcan_branch_cost): Rename to ...
40870 (thunderx2t99_branch_cost): This.
40871 (vulcan_tunings): Rename to ...
40872 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
40873 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
40875 2017-01-09 Martin Jambor <mjambor@suse.cz>
40879 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
40880 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
40881 (propagate_vr_accross_jump_function): Use the above function for all
40882 value range computations for pass-through jump functions and type
40883 converasion from explicit value range values.
40884 (ipcp_propagate_stage): Do not attempt to deduce types of formal
40885 parameters from TYPE_ARG_TYPES.
40886 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
40887 (ipa_write_node_info): Stream type of the actual argument.
40888 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
40890 2017-01-09 Martin Liska <mliska@suse.cz>
40893 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
40894 (lookup_compiler): Do not show error message with have_E.
40896 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40898 PR tree-optimization/78938
40899 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
40900 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
40901 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
40902 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
40905 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40907 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
40910 2017-01-09 Richard Biener <rguenther@suse.de>
40912 PR tree-optimization/78997
40913 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
40914 name condition properly.
40916 2017-01-09 Richard Biener <rguenther@suse.de>
40919 * dwarf2out.c (is_cxx): New overload with context.
40920 (is_naming_typedef_decl): Use it.
40922 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40924 * invoke.texi (Option Summary): Correct spacing in option lists
40925 and add line breaks to fix over-long lines.
40927 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40929 PR middle-end/17660
40931 * extend.texi (Common Variable Attributes): Add xref to GCC
40932 Internals manual to explain mode attribute keywords.
40934 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40937 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
40938 and Preprocessor Options.
40939 (Options for Linking): Document -pthread here....
40940 (RS/6000 and PowerPC Options): ...not here.
40941 (Solaris 2 Options): ...or here.
40942 * doc/cppopts.texi: Document -pthread.
40944 2017-01-08 Martin Sebor <msebor@redhat.com>
40946 PR middle-end/77708
40947 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
40948 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
40949 New member functions.
40950 (format_directive): Used them.
40952 (pass_sprintf_length::handle_gimple_call): Same.
40953 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
40954 to avoid truncation for any argument.
40955 (extract_affine_mul): Same.
40956 * tree.c (get_file_function_name): Same.
40958 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40960 PR middle-end/77484
40961 * predict.def (PRED_INDIR_CALL): Set to 86.
40963 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40965 PR preprocessor/54124
40966 * doc/cppopts.texi: Reformat -d subtable to list the full name
40967 of the options. Add cross-reference to the docs for the general
40968 compiler -d options.
40969 * doc/invoke.texi (Developer Options): Add cross-reference to the
40970 preprocessor-specific -d option documentation.
40972 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40974 PR preprocessor/13498
40975 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
40976 redudant material, and reflect new command-line options.
40977 (System Headers): Likewise.
40979 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40981 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
40982 -isystem, and -idirafter. Copy-edit.
40983 * doc/cppopts.texi: Copy-edit. Remove contradiction about
40984 default for -ftrack-macro-expansion. Delete obsolete and
40985 badly-formatted implementation details about -fdebug-cpp output.
40986 * doc/cppwarnopts.texi: Copy-edit.
40988 2017-01-07 David Malcolm <dmalcolm@redhat.com>
40991 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40992 that the transition from a max line width >= 1<<10 to narrower
40993 lines works correctly.
40995 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
40997 * doc/options.texi (PerFunction): New.
40998 * opt-functions.awk (switch_flags): Map both Optimization and
40999 PerFunction to CL_OPTIMIZATION.
41000 * opth-gen.awk: Test for PerFunction flag along with
41002 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41003 it only when the latter is present. Skip those that don't in
41004 the hash function generator.
41005 * common.opt (fvar-tracking): Mark as PerFunction instead of
41007 (fvar-tracking-assignments): Likewise.
41008 (fvar-tracking-assignments-toggle): Likewise.
41009 (fvar-tracking-uninit): Likewise.
41011 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41013 PR translation/79018
41014 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41017 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41020 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41021 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41022 TARGET_LONG_JUMP_TABLE_OFFSETS.
41023 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41024 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41025 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41026 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41027 * config/m68k/m68k.md (tablejump expander): Likewise.
41028 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41029 TARGET_LONG_JUMP_TABLE_OFFSETS.
41030 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41031 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41033 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41034 David Holsgrove <david.holsgrove@xilinx.com>
41036 * common/config/microblaze/microblaze-common.c
41037 (TARGET_EXCEPT_UNWIND_INFO): Remove.
41038 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
41040 * config/microblaze/microblaze.c (microblaze_must_save_register)
41041 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41043 (microblaze_eh_return): New function.
41044 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41045 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41046 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41047 * config/microblaze/microblaze.md (eh_return): New pattern.
41049 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41051 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41052 GCC_DIAGNOSTIC_STRINGIFY): Define.
41054 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41056 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41058 * config/arm/arm.md (<mcrr>): New.
41060 * config/arm/arm.c (arm_arch5te): New.
41061 (arm_option_override): Set arm_arch5te.
41062 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41064 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41065 (arm_mcrr_qualifiers): ... this. New.
41066 (MRRC_QUALIFIERS): Define to...
41067 (arm_mrrc_qualifiers): ... this. New.
41068 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41070 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41071 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41072 (MRRCI, mrrc, MRRC): New.
41073 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41074 VUNSPEC_MRRC2): New.
41076 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41078 * config/arm/arm.md (<mcr>): New.
41080 * config/arm/arm.c (arm_coproc_builtin_available): Add
41081 support for mcr, mrc, mcr2 and mrc2.
41082 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41083 (arm_mcr_qualifiers): ... this. New.
41084 (MRC_QUALIFIERS): Define to ...
41085 (arm_mrc_qualifiers): ... this. New.
41086 (MCR_QUALIFIERS): Define to ...
41087 (arm_mcr_qualifiers): ... this. New.
41088 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41090 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41091 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41092 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41093 VUNSPEC_MRC2): New.
41095 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41097 * config/arm/arm.md (*ldc): New.
41101 * config/arm/arm.c (arm_coproc_builtin_available): Add
41102 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41103 (arm_coproc_ldc_stc_legitimate_address): New.
41104 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41105 'qualifier_const_pointer'.
41106 (LDC_QUALIFIERS): Define to...
41107 (arm_ldc_qualifiers): ... this. New.
41108 (STC_QUALIFIERS): Define to...
41109 (arm_stc_qualifiers): ... this. New.
41110 * config/arm/arm-protos.h
41111 (arm_coproc_ldc_stc_legitimate_address): New.
41112 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
41113 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
41114 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
41115 stc2, stcl, stc2l): New.
41116 * config/arm/constraints.md (Uz): New.
41117 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
41118 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
41119 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
41120 VUNSPEC_STC2L): New.
41122 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41124 * config/arm/arm.md (<cdp>): New.
41125 * config/arm/arm.c (neon_const_bounds): Rename this ...
41126 (arm_const_bounds): ... this.
41127 (arm_coproc_builtin_available): New.
41128 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
41129 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
41130 (CDP_QUALIFIERS): Define to...
41131 (arm_cdp_qualifiers): ... this. New.
41133 (arm_expand_builtin_args): Add case for 6 arguments.
41134 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
41135 (arm_const_bounds): ... this.
41136 (arm_coproc_builtin_available): New.
41137 * config/arm/arm_acle.h (__arm_cdp): New.
41139 * config/arm/arm_acle_builtins.def (cdp): New.
41141 * config/arm/iterators.md (CDPI,CDP,cdp): New.
41142 * config/arm/neon.md: Rename all 'neon_const_bounds' to
41143 'arm_const_bounds'.
41144 * config/arm/types.md (coproc): New.
41145 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
41146 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
41147 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
41148 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
41150 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41152 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
41153 (UBINOP_QUALIFIERS): New.
41155 (acle_builtin_data): New. Change comment.
41156 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
41157 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
41158 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
41159 arm_acle_builtins.def.
41160 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
41161 (arm_init_acle_builtins): New.
41162 (CRC32_BUILTIN): Remove.
41163 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
41164 crc32cb, crc32ch and crc32cw.
41165 (arm_init_crc32_builtins): Remove.
41166 (arm_init_builtins): Use arm_init_acle_builtins rather
41167 than arm_init_crc32_builtins.
41168 (arm_expand_acle_builtin): New.
41169 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
41170 * config/arm/arm_acle_builtins.def: New.
41172 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41174 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
41175 (arm_builtin_datum): ... this.
41176 (arm_init_neon_builtin): Rename to ...
41177 (arm_init_builtin): ... this. Add a new parameters PREFIX
41178 and USE_SIG_IN_NAME.
41179 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
41180 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
41181 'arm_builtin_datum'.
41182 (arm_init_vfp_builtins): Likewise.
41183 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
41184 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
41185 (arm_expand_neon_args): Rename to ...
41186 (arm_expand_builtin_args): ... this. Rename builtin_arg
41187 enum values and differentiate between ARG_BUILTIN_MEMORY
41188 and ARG_BUILTIN_NEON_MEMORY.
41189 (arm_expand_neon_builtin_1): Rename to ...
41190 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
41191 values, arm_expand_builtin_args and add bool parameter NEON.
41192 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
41193 (arm_expand_vfp_builtin): Likewise.
41194 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
41196 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41198 PR middle-end/77484
41199 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
41200 * predict.c (tree_estimate_probability_bb): Reverse direction of
41201 polymorphic call predictor.
41203 2017-01-06 David Malcolm <dmalcolm@redhat.com>
41205 * passes.c (execute_one_pass): Split out pass-skipping logic into...
41206 (determine_pass_name_match): ...this new function and...
41207 (should_skip_pass_p): ...this new function.
41209 2017-01-06 Nathan Sidwell <nathan@acm.org>
41211 * ipa-visibility.c (function_and_variable_visibility): Reformat
41212 comments and long lines. Remove extrneous if.
41213 * symtab.c (symtab_node::make_decl_local): Fix code format.
41214 (symtab_node::set_section_for_node): Fix comment typo.
41216 2017-01-06 Martin Liska <mliska@suse.cz>
41219 * lra-constraints.c: Rename invariant to lra_invariant.
41220 * predict.c (set_even_probabilities): Initialize e to NULL.
41222 2017-01-05 Martin Sebor <msebor@redhat.com>
41224 PR tree-optimization/78910
41225 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
41226 (format_integer): Correct off-by-one error in the handling
41227 of precision with negative numbers in signed conversions..
41229 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
41231 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
41233 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41235 PR tree-optimization/71016
41236 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
41237 factor_out_conditional_conversion. Formatting fix.
41238 (factor_out_conditional_conversion): Add cond_stmt argument.
41239 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
41240 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
41243 2017-01-05 David Malcolm <dmalcolm@redhat.com>
41245 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
41246 read-rtl-function.o, and selftest-rtl.o.
41247 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
41248 (selftest::aarch64_test_loading_full_dump): New function.
41249 (selftest::aarch64_run_selftests): New function.
41250 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
41251 selftest::aarch64_run_selftests.
41252 * config/i386/i386.c
41253 (selftest::ix86_test_loading_dump_fragment_1): New function.
41254 (selftest::ix86_test_loading_call_insn): New function.
41255 (selftest::ix86_test_loading_full_dump): New function.
41256 (selftest::ix86_test_loading_unspec): New function.
41257 (selftest::ix86_run_selftests): Call the new functions.
41258 * emit-rtl.c (maybe_set_max_label_num): New function.
41259 * emit-rtl.h (maybe_set_max_label_num): New decl.
41260 * function.c (instantiate_decls): Guard call to
41261 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
41262 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
41264 * gensupport.c (gen_reader::gen_reader): Pass "false"
41265 for new "compact" param of rtx_reader.
41266 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
41267 rather than an empty string for NULL strings.
41268 * read-md.c: Potentially include config.h rather than bconfig.h.
41269 Wrap include of errors.h with #ifdef GENERATOR_FILE.
41270 (have_error): New global, copied from errors.c.
41271 (md_reader::read_name): Rename to...
41272 (md_reader::read_name_1): ...this, adding "out_loc" param,
41273 and converting "missing name or number" to returning false, rather
41275 (md_reader::read_name): Reimplement in terms of read_name_1.
41276 (md_reader::read_name_or_nil): New function.
41277 (md_reader::read_string): Handle "(nil)" by returning NULL.
41278 (md_reader::md_reader): Add new param "compact".
41279 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
41280 (md_reader::read_file): New method.
41281 * read-md.h (md_reader::md_reader): Add new param "compact".
41282 (md_reader::read_file): New method.
41283 (md_reader::is_compact): New accessor.
41284 (md_reader::read_name): Convert return type from void to file_location.
41285 (md_reader::read_name_or_nil): New decl.
41286 (md_reader::read_name_1): New decl.
41287 (md_reader::m_compact): New field.
41288 (noop_reader::noop_reader): Pass "false" for new "compact" param
41290 (rtx_reader::rtx_reader): Add new "compact" param.
41291 (rtx_reader::read_rtx_operand): Make virtual and convert return
41292 type from void to rtx.
41293 (rtx_reader::read_until): New decl.
41294 (rtx_reader::handle_any_trailing_information): New virtual function.
41295 (rtx_reader::postprocess): New virtual function.
41296 (rtx_reader::finalize_string): New virtual function.
41297 (rtx_reader::m_in_call_function_usage): New field.
41298 (rtx_reader::m_reuse_rtx_by_id): New field.
41299 * read-rtl-function.c: New file.
41300 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
41301 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
41302 (selftest::verify_three_block_rtl_cfg): New decl.
41303 * read-rtl-function.h: New file.
41304 * read-rtl.c: Potentially include config.h rather than bconfig.h.
41305 For host, include function.h, memmodel.h, and emit-rtl.h.
41306 (one_time_initialization): New function.
41307 (struct compact_insn_name): New struct.
41308 (compact_insn_names): New array.
41309 (find_code): Handle insn codes in compact dumps.
41310 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
41311 (bind_subst_iter_and_attr): Likewise.
41312 (add_condition_to_string): Likewise.
41313 (add_condition_to_rtx): Likewise.
41314 (apply_attribute_uses): Likewise.
41315 (add_current_iterators): Likewise.
41316 (apply_iterators): Likewise.
41317 (initialize_iterators): Guard usage of apply_subst_iterator with
41318 #ifdef GENERATOR_FILE.
41319 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
41320 (md_reader::read_mapping): Likewise.
41321 (add_define_attr_for_define_subst): Likewise.
41322 (add_define_subst_attr): Likewise.
41323 (read_subst_mapping): Likewise.
41324 (check_code_iterator): Likewise.
41325 (rtx_reader::read_rtx): Likewise. Move one-time initialization
41327 (one_time_initialization): New function.
41328 (rtx_reader::read_until): New method.
41329 (read_flags): New function.
41330 (parse_reg_note_name): New function.
41331 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
41332 Handle reuse_rtx ids.
41333 Wrap iterator lookup within #ifdef GENERATOR_FILE.
41334 Add parsing support for RTL dumps, mirroring the special-cases in
41335 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
41336 values, and calling handle_any_trailing_information.
41337 (rtx_reader::read_rtx_operand): Convert return type from void
41338 to rtx, returning return_rtx. Handle case 'e'. Call
41339 finalize_string on XSTR and XTMPL fields.
41340 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
41341 "(nil)" values were omitted. Call the postprocess vfunc on the
41343 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
41344 class ctor. Initialize m_in_call_function_usage. Call
41345 one_time_initialization.
41346 * rtl-tests.c (selftest::test_uncond_jump): Call
41347 set_new_first_and_last_insn.
41348 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
41349 * selftest-rtl.c: New file.
41350 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
41351 (selftest::get_insn_by_uid): New decl.
41352 * selftest-run-tests.c (selftest::run_tests): Call
41353 read_rtl_function_c_tests.
41354 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
41355 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
41358 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
41360 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
41361 operands in a special way. Assert that pos+len <= mode precision.
41363 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41365 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
41366 3 argument Alias with unlimited for the negative form.
41367 (fno-vect-cost-model): Removed.
41369 2017-01-05 Martin Liska <mliska@suse.cz>
41371 * hsa-gen.c (gen_hsa_divmod): New function.
41372 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
41374 2017-01-05 Martin Liska <mliska@suse.cz>
41377 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
41380 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41382 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
41383 small constant length operands.
41385 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41387 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
41388 between loop iterations.
41390 2017-01-05 Martin Liska <mliska@suse.cz>
41393 * gimplify.c (gimplify_decl_expr): Compare to
41394 asan_poisoned_variables instread of checking flags.
41395 (gimplify_target_expr): Likewise.
41396 (gimplify_expr): Likewise.
41397 (gimplify_function_tree): Conditionally initialize
41398 asan_poisoned_variables.
41400 2017-01-04 Jeff Law <law@redhat.com>
41402 PR tree-optimizatin/78812
41403 * rtl.h (contains_mem_rtx_p): Prototype.
41404 * ifcvt.c (containts_mem_rtx_p): Move from here to...
41405 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
41406 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
41407 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
41408 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
41410 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41412 * input.c (assert_char_at_range): Default-initialize actual_range.
41414 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41416 * df-scan.c (df_ref_create_structure): Make regno unsigned,
41417 to match the caller.
41419 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41421 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
41422 insns after final jump in test to emit dummy move.
41424 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41426 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
41427 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
41429 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41431 * multiple_target.c (create_dispatcher_calls): Init e_next.
41432 * tree-ssa-loop-split.c (split_loop): Init border.
41433 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
41436 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
41441 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
41442 (altivec_register_operand): Do not return true if the operand
41443 contains a SUBREG mixing SImode and SFmode.
41444 (vsx_register_operand): Likewise.
41445 (vsx_reg_sfsubreg_ok): New predicate.
41446 (vfloat_operand): Do not return true if the operand contains a
41447 SUBREG mixing SImode and SFmode.
41448 (vint_operand): Likewise.
41449 (vlogical_operand): Likewise.
41450 (gpc_reg_operand): Likewise.
41451 (int_reg_operand): Likewise.
41452 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
41453 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
41454 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
41456 (rs6000_emit_move_si_sf_subreg): New helper function.
41457 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
41458 fixup SUBREGs involving SImode and SFmode.
41459 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
41460 numbers for the new peephole2 optimization.
41461 (peephole2 for SFmode unions): New peephole2 to optimize cases in
41462 the GLIBC math library that do AND/IOR/XOR operations on single
41463 precision floating point.
41464 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
41465 target macros to say whether we need to avoid SUBREGs mixing
41467 (TARGET_ALLOW_SF_SUBREG): Likewise.
41468 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
41469 (UNSPEC_SI_FROM_SF): Likewise.
41470 (iorxor): Change spacing.
41471 (and_ior_xor): New iterator for AND, IOR, and XOR.
41472 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
41473 (movdi_from_sf_zero_ext): Likewise.
41474 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
41475 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
41476 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
41477 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
41478 (fms<mode>4): Likewise.
41479 (fnma<mode>4): Likewise.
41480 (fnms<mode>4): Likewise.
41481 (nfma<mode>4): Likewise.
41482 (nfms<mode>4): Likewise.
41484 2017-01-04 Marek Polacek <polacek@redhat.com>
41487 * doc/invoke.texi: Document -Wpointer-compare.
41489 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41491 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
41494 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
41495 descriptions for -gdwarf-5 and emit them as uleb128 instead of
41498 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
41501 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
41502 documentation of the powerpc_popcntb_ok attribute.
41503 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
41504 code to issue warning messages if a requested CPU configuration is
41505 not supported by the binary (assembler and loader) toolchain.
41506 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
41507 made to define a built-in function that has been disabled.
41508 (paired_init_builtins): Add assertion to prevent ICE if attempt is
41509 made to define a built-in function that has been disabled.
41510 (altivec_init_builtins): Add comment explaining why definition
41511 of the DST built-in functions is not preceded by an assertion
41512 check. Add assertions to prevent ICE if attempts are made to
41513 define an altivec predicate or an abs* built-in function that has
41515 (htm_init_builtins): Add comment explaining why definition of the
41516 htm built-in functions is not preceded by an assertion check.
41518 2017-01-04 Jeff Law <law@redhat.com>
41520 PR tree-optimizatin/67955
41521 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
41522 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
41523 the points-to solution does not include pt_null. Use DECL_PT_UID
41526 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
41528 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
41529 Use gen_int_mode instead of gen_lopwart for const_int operands.
41531 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41533 PR tree-optimization/71563
41534 * match.pd: Simplify X << Y into X if Y is known to be 0 or
41535 out of range value - has low bits known to be zero.
41537 2017-01-04 Alan Modra <amodra@gmail.com>
41539 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
41540 * configure: Regenerate.
41541 * config.in: Regenerate.
41543 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41546 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
41547 a substring of the message, but strcmp with the whole message. Ifdef
41548 ENABLE_NLS, translate the message first using dgettext.
41550 2017-01-03 Jeff Law <law@redhat.com>
41552 PR tree-optimizatin/78856
41553 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
41554 (mark_threaded_blocks): Remove code to truncate thread paths that
41555 cross multiple loop headers. Instead invalidate the cached loop
41556 iteration information and handle case of a thread path walking
41557 into an irreducible region.
41559 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
41562 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
41563 assertions. Add support for doing the signbit if the IEEE 128-bit
41564 floating point value is in a GPR.
41565 * config/rs6000/rs6000.md (Fsignbit): Delete.
41566 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
41567 Update the length attribute if the value is in a GPR.
41568 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
41569 the sign or zero extension instruction, since the value is always 0/1.
41570 (signbit<mode>2_dm2): Delete using <Fsignbit>.
41573 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
41574 extracting SImode to a GPR register so that we can generate a
41575 store, limit the vector to be in a traditional Altivec register
41576 for the vextuwrx instruction.
41578 2017-01-03 Ian Lance Taylor <iant@google.com>
41580 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
41582 2017-01-03 Martin Sebor <msebor@redhat.com>
41584 PR tree-optimization/78696
41585 * gimple-ssa-sprintf.c (format_floating): Correct handling of
41586 precision. Use MPFR for %f for greater fidelity. Correct handling
41588 (pass_sprintf_length::compute_format_length): Set width and precision
41589 specified by asrerisk to void_node for vararg functions.
41590 (try_substitute_return_value): Adjust dump output.
41592 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
41594 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
41596 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
41598 * doc/invoke.texi (SPARC options): Document -mlra as the default.
41599 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
41600 -mlra/-mno-lra was passed to the compiler.
41602 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
41604 PR rtl-optimization/65618
41605 * emit-rtl.c (try_split): Move initialization of "before" and
41606 "after" to just before the call to emit_insn_after_setloc.
41608 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
41610 * doc/md.texi (Standard Names): Remove reference to Java frontend.
41612 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
41614 * dwarf2out.c (gen_enumeration_type_die): When
41615 -gno-strict-dwarf, add a DW_AT_encoding attribute.
41617 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41619 PR tree-optimization/78965
41620 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41621 Change first argument from const call_info & to call_info &. For %n
41622 set info.nowrite to false.
41624 PR middle-end/78901
41625 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41626 possibly throwing calls.
41628 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41629 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41630 and fns handling, rather than in a separate case SSA_NAME.
41632 2017-01-02 Jeff Law <law@redhat.com>
41634 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41635 fixes for first_period and second_period variables.
41637 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41640 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41641 (*insvqi_1): New insn pattern.
41642 (*insvqi_1_mem_rex64): Ditto.
41643 (*insvqi_2): Ditto.
41644 (*insvqi_3): Rename from *insvqi.
41646 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41648 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41650 * doc/cfg.texi (Edges): Remove reference to Java.
41651 (Maintaining the CFG): Ditto.
41653 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41655 PR middle-end/77674
41656 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41657 transparent aliases.
41659 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41661 PR middle-end/77484
41662 * predict.def (PRED_CALL): Update hitrate.
41663 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41664 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41665 into direct/indirect/polymorphic variants.
41667 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41669 Update copyright years.
41671 * gcc.c (process_command): Update copyright notice dates.
41672 * gcov-dump.c (print_version): Ditto.
41673 * gcov.c (print_version): Ditto.
41674 * gcov-tool.c (print_version): Ditto.
41675 * gengtype.c (create_file): Ditto.
41676 * doc/cpp.texi: Bump @copying's copyright year.
41677 * doc/cppinternals.texi: Ditto.
41678 * doc/gcc.texi: Ditto.
41679 * doc/gccint.texi: Ditto.
41680 * doc/gcov.texi: Ditto.
41681 * doc/install.texi: Ditto.
41682 * doc/invoke.texi: Ditto.
41684 Copyright (C) 2017 Free Software Foundation, Inc.
41686 Copying and distribution of this file, with or without modification,
41687 are permitted in any medium without royalty provided the copyright
41688 notice and this notice are preserved.