1 2017-12-25 Jakub Jelinek <jakub@redhat.com>
4 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
5 ix86_isa_flags2 to ix86_isa_flags.
6 * config/i386/i386-c.c (ix86_target_macros_internal): Test
7 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
8 isa_flags rather than isa_flags2.
9 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
10 and -mavx512bitalg from isa2_opts to isa_opts.
11 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
12 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
13 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
15 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
16 OPTION_MASK_ISA_AVX512F ored with another option similarly to
17 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
18 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
19 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
20 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
21 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
22 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
23 builtins from bdesc_args2 to bdesc_args section.
24 (__builtin_ia32_compressstoreuqi512_mask,
25 __builtin_ia32_compressstoreuhi512_mask,
26 __builtin_ia32_compressstoreuqi256_mask,
27 __builtin_ia32_expandloadqi512_mask,
28 __builtin_ia32_expandloadqi512_maskz,
29 __builtin_ia32_expandloadhi512_mask,
30 __builtin_ia32_expandloadhi512_maskz,
31 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
32 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
33 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
34 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
35 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
36 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
37 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
38 __builtin_ia32_vpshldv_v32hi_maskz,
39 __builtin_ia32_vpopcountb_v64qi_mask,
40 __builtin_ia32_vpopcountw_v32hi_mask,
41 __builtin_ia32_vpshufbitqmb512_mask,
42 __builtin_ia32_vpshufbitqmb256_mask): Add
43 " | OPTION_MASK_ISA_AVX512BW".
44 (__builtin_ia32_expandloadqi256_mask,
45 __builtin_ia32_expandloadqi256_maskz,
46 __builtin_ia32_vpopcountb_v32qi_mask): Add
47 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
48 (__builtin_ia32_expandloadhi256_mask,
49 __builtin_ia32_expandloadhi256_maskz,
50 __builtin_ia32_expandloadqi128_mask,
51 __builtin_ia32_expandloadqi128_maskz,
52 __builtin_ia32_expandloadhi128_mask,
53 __builtin_ia32_expandloadhi128_maskz,
54 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
55 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
56 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
57 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
58 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
59 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
60 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
61 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
62 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
63 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
64 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
65 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
66 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
67 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
68 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
69 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
70 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
71 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
72 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
73 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
74 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
75 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
76 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
77 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
78 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
79 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
80 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
81 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
82 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
83 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
84 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
85 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
86 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
87 __builtin_ia32_vpopcountw_v8hi_mask): Add
88 " | OPTION_MASK_ISA_AVX512VL".
89 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
90 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
91 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
92 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
93 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
94 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
95 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
96 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
97 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
98 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
99 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
100 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
101 _mm_mask_bitshuffle_epi64_mask): Likewise.
102 * common/config/i386/i386-common.c
103 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
104 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
105 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
106 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
107 OPTION_MASK_ISA_AVX512BITALG_UNSET.
108 (OPTION_MASK_ISA2_AVX512F_UNSET,
109 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
110 (ix86_handle_option): For -mno-general-regs-only, clear from
111 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
112 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
113 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
114 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
115 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
116 for moving from ix86_isa_flags2 to ix86_isa_flags.
118 2017-12-23 Jakub Jelinek <jakub@redhat.com>
121 * fold-const.c (struct contains_label_data): New type.
122 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
123 inside of a SWITCH_BODY seen during the walk.
124 (contains_label_p): Use walk_tree instead of
125 walk_tree_without_duplicates, prepare data for contains_label_1 and
128 2017-12-22 Jakub Jelinek <jakub@redhat.com>
131 * tree-iterator.c (alloc_stmt_list): Start with cleared
132 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
136 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
137 enable avx512f explicitly in #pragma GCC target.
138 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
139 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
140 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
141 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
142 __builtin_ia32_vpdpbusds_v8si_mask,
143 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
144 __builtin_ia32_vpdpbusds_v4si_mask,
145 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
146 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
147 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
148 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
149 __builtin_ia32_vpdpwssds_v8si_mask,
150 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
151 __builtin_ia32_vpdpwssds_v4si_mask,
152 __builtin_ia32_vpdpwssds_v4si_maskz): Use
153 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
154 just OPTION_MASK_ISA_AVX512VNNI.
156 2017-12-22 Martin Jambor <mjambor@suse.cz>
159 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
162 2017-12-22 Julia Koval <julia.koval@intel.com>
163 Sebastian Peryt <sebastian.peryt@intel.com>
165 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
166 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
167 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
168 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
169 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
170 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
171 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
172 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
173 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
174 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
175 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
176 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
177 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
178 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
179 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
180 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
181 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
182 _mm256_maskz_popcnt_epi64): New intrinsics.
183 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
184 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
185 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
186 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
187 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
188 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
189 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
190 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
191 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
192 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
193 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
194 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
195 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
196 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
197 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
198 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
199 __builtin_ia32_vpshufbitqmb256_mask,
200 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
201 * config/i386/i386-c.c (__AVX512BITALG__): New.
202 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
203 (ix86_valid_target_attribute_inner_p): Ditto.
204 (ix86_expand_args_builtin): Handle new types.
205 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
206 * config/i386/i386.opt: Add -mavx512bitalg.
207 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
208 avx512bitalgintrin.h.
209 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
210 (vpopcount<mode><mask_name>): Add more types.
211 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
212 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
214 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
216 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
217 Or in OPTION_MASK_ISA_AVX512F_SET.
218 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
219 OPTION_MASK_ISA_AVX512VNNI_UNSET.
220 (ix86_handle_option): Adjust for
221 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
222 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
224 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
225 isa_flag instead of isa_flag2.
226 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
227 isa_opts2 to isa_opts.
228 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
231 2017-12-22 Mike Stump <mikestump@comcast.net>
232 Eric Botcazou <ebotcazou@adacore.com>
234 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
236 2017-12-21 Jakub Jelinek <jakub@redhat.com>
239 * config/i386/i386.c (ix86_function_arg_boundary): Return
240 PARM_BOUNDARY for TYPE_EMPTY_P types.
243 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
244 if navail is >= dir.len.
246 2017-12-21 Steve Ellcey <sellcey@cavium.com>
248 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
251 2017-12-21 Jakub Jelinek <jakub@redhat.com>
253 PR rtl-optimization/80747
254 PR rtl-optimization/83512
255 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
256 succ edge from ENTRY, copy partition from e->dest to the newly
258 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
259 ENTRY, use BB_PARTITION of its successor block as current_partition.
260 Don't copy partition when splitting succ edge from ENTRY.
262 PR tree-optimization/83523
263 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
264 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
265 (convert_mult_to_fma): Likewise.
267 PR tree-optimization/83521
268 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
269 gimple_build_assign without code on result of
270 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
273 2017-12-21 Andrew Pinski <apinski@cavium.com>
274 Steve Ellcey <sellcey@cavium.com>
276 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
277 multi-arch for ilp32.
279 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
282 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
283 constraints to operand 2.
284 (*ashl<mode>3_mask_1): Ditto.
285 (*<shift_insn><mode>3_mask): Ditto.
286 (*<shift_insn><mode>3_mask_1): Ditto.
287 (*<rotate_insn><mode>3_mask): Ditto.
288 (*<rotate_insn><mode>3_mask_1): Ditto.
290 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
292 * reorg.c (make_return_insns): Reemit each insn with its own location.
294 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
296 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
297 generation for cases where splatting a value is not useful.
298 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
299 across a vec_duplicate and a paradoxical subreg forming a vector
300 mode to a vec_concat.
302 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
304 * combine.c (simplify_set): Do not transform subregs to zero_extends
305 if the destination is not a scalar int mode.
307 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
310 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
311 shared zero if the input has overflowed.
313 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
315 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
316 support for Cortex-A55 and Cortex-A75.
318 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
320 * common/config/arm/arm-common.c (compare_opt_names): Add function
321 comment. Use strcmp instead of manual loop.
323 2017-12-21 Martin Liska <mliska@suse.cz>
325 PR gcov-profile/83509
326 * gcov-dump.c (dump_gcov_file): Do not read info about
327 support_unexecuted_blocks for gcda files.
329 2017-12-21 Jakub Jelinek <jakub@redhat.com>
331 PR rtl-optimization/82973
332 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
333 (valid_for_const_vector_p): ... this.
334 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
335 (valid_for_const_vector_p): ... this. Adjust function comment.
336 (gen_vec_duplicate): Adjust caller.
337 * optabs.c (expand_vector_broadcast): Likewise.
338 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
339 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
341 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
342 instead of GET_CODE == CONST_FIXED.
343 (simplify_subreg): Use CONST_FIXED_P macro instead of
344 GET_CODE == CONST_FIXED.
347 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
348 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
349 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
350 (ix86_option_override_internal): Adjust for
351 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
352 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
353 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
354 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
355 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
356 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
357 def_builtin2 for CET builtins.
358 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
359 up formatting in IX86_BUILTIN_RDPID code.
360 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
361 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
362 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
363 ix86_isa_flags2 to ix86_isa_flags.
364 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
366 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
367 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
368 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
370 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
371 Or in OPTION_MASK_ISA_AVX512F_SET.
372 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
373 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
374 (ix86_handle_option): Adjust for
375 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
376 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
378 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
379 Alan Hayward <alan.hayward@arm.com>
380 David Sherwood <david.sherwood@arm.com>
382 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
383 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
384 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
385 Track polynomial offsets.
387 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
388 Alan Hayward <alan.hayward@arm.com>
389 David Sherwood <david.sherwood@arm.com>
391 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
392 Treat drb->init as a poly_int. Fail if its misalignment wrt
393 vector_alignment isn't known.
395 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
396 Alan Hayward <alan.hayward@arm.com>
397 David Sherwood <david.sherwood@arm.com>
399 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
401 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
402 (vect_loop_versioning): Take the loop versioning threshold as a
404 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
405 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
406 versioning_threshold.
407 (vect_analyze_loop_2): Compute the loop versioning threshold
408 whenever loop versioning is needed, and store it in the new
409 field rather than combining it with the cost model threshold.
410 (vect_transform_loop): Update call to vect_loop_versioning.
411 Try to combine the loop versioning and cost thresholds here.
413 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
414 Alan Hayward <alan.hayward@arm.com>
415 David Sherwood <david.sherwood@arm.com>
417 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
418 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
419 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
420 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
421 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
422 (group_compare_offset): Update accordingly.
423 (split_small_address_groups_p): Likewise.
424 (record_use): Take addr_offset as a poly_uint64 rather than
425 an unsigned HOST_WIDE_INT.
426 (strip_offset): Return the offset as a poly_uint64 rather than
427 an unsigned HOST_WIDE_INT.
428 (record_group_use, split_address_groups): Track polynomial offsets.
429 (add_iv_candidate_for_use): Likewise.
430 (addr_offset_valid_p): Take the offset as a poly_int64 rather
431 than a HOST_WIDE_INT.
432 (strip_offset_1): Return the offset as a poly_int64 rather than
435 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
436 Alan Hayward <alan.hayward@arm.com>
437 David Sherwood <david.sherwood@arm.com>
439 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
440 rather than a HOST_WIDE_INT.
441 * tree.c (get_binfo_at_offset): Likewise.
443 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
444 Alan Hayward <alan.hayward@arm.com>
445 David Sherwood <david.sherwood@arm.com>
447 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
448 rather than a HOST_WIDE_INT.
449 * tree-sra.c (build_ref_for_offset): Likewise.
451 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
452 Alan Hayward <alan.hayward@arm.com>
453 David Sherwood <david.sherwood@arm.com>
455 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
457 * tree.c (mem_ref_offset): Likewise.
458 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
459 * builtins.c (get_object_alignment_2): Likewise.
460 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
461 * gimple-fold.c (get_base_constructor): Likewise.
462 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
463 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
465 * ipa-polymorphic-call.c
466 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
467 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
468 (get_ancestor_addr_info): Likewise.
469 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
470 * match.pd: Likewise.
471 * tree-data-ref.c (dr_analyze_innermost): Likewise.
472 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
473 * tree-eh.c (tree_could_trap_p): Likewise.
474 * tree-object-size.c (addr_object_size): Likewise.
475 * tree-ssa-address.c (copy_ref_info): Likewise.
476 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
477 (indirect_refs_may_alias_p): Likewise.
478 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
479 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
480 (non_rewritable_mem_ref_base): Likewise.
481 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
482 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
483 * varasm.c (decode_addr_const): Likewise.
485 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
486 Alan Hayward <alan.hayward@arm.com>
487 David Sherwood <david.sherwood@arm.com>
489 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
490 offsets for MEM_REFs.
492 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
493 Alan Hayward <alan.hayward@arm.com>
494 David Sherwood <david.sherwood@arm.com>
496 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
497 as a poly_uint64 rather than an unsigned int.
498 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
500 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
501 Alan Hayward <alan.hayward@arm.com>
502 David Sherwood <david.sherwood@arm.com>
504 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
506 (decode_addr_const): Update accordingly.
508 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
509 Alan Hayward <alan.hayward@arm.com>
510 David Sherwood <david.sherwood@arm.com>
512 * tree.h (bit_field_size, bit_field_offset): New functions.
513 * hsa-gen.c (gen_hsa_addr): Use them.
514 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
515 (simplify_vector_constructor): Likewise.
516 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
517 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
518 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
519 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
522 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
523 Alan Hayward <alan.hayward@arm.com>
524 David Sherwood <david.sherwood@arm.com>
526 * expr.h (emit_group_load, emit_group_load_into_temps)
527 (emit_group_store): Take the size as a poly_int64 rather than an int.
528 * expr.c (emit_group_load_1, emit_group_load): Likewise.
529 (emit_group_load_into_temp, emit_group_store): Likewise.
531 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
532 Alan Hayward <alan.hayward@arm.com>
533 David Sherwood <david.sherwood@arm.com>
535 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
536 an unsigned int to a poly_uint64.
537 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
538 sizes as poly_uint64s rather than unsigned ints.
539 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
542 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
543 Alan Hayward <alan.hayward@arm.com>
544 David Sherwood <david.sherwood@arm.com>
546 * lra-constraints.c (emit_inc): Change inc_amount from an int
549 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
550 Alan Hayward <alan.hayward@arm.com>
551 David Sherwood <david.sherwood@arm.com>
553 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
555 (add_stack_var, stack_var_cmp, partition_stack_vars)
556 (dump_stack_var_partition): Update accordingly.
557 (alloc_stack_frame_space): Take the size as a poly_int64 rather
558 than a HOST_WIDE_INT.
559 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
560 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
561 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
562 rather than a HOST_WIDE_INT.
564 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
565 Alan Hayward <alan.hayward@arm.com>
566 David Sherwood <david.sherwood@arm.com>
568 * target.def (return_pops_args): Treat both the input and output
569 sizes as poly_int64s rather than HOST_WIDE_INTS.
570 * targhooks.h (default_return_pops_args): Update accordingly.
571 * targhooks.c (default_return_pops_args): Likewise.
572 * doc/tm.texi: Regenerate.
573 * emit-rtl.h (incoming_args): Change pops_args, size and
574 outgoing_args_size from int to poly_int64_pod.
575 * function.h (expr_status): Change x_pending_stack_adjust and
576 x_stack_pointer_delta from int to poly_int64.
577 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
578 (ARGS_SIZE_RTX): Update accordingly.
579 * calls.c (highest_outgoing_arg_in_use): Change from int to
581 (stack_usage_watermark, stored_args_watermark): New variables.
582 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
583 (emit_call_1): Change the stack_size and rounded_stack_size
584 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
586 (save_fixed_argument_area): Check stack_usage_watermark.
587 (initialize_argument_information): Change old_pending_adj from
588 a HOST_WIDE_INT * to a poly_int64_pod *.
589 (compute_argument_block_size): Return the size as a poly_int64
591 (finalize_must_preallocate): Track polynomial argument sizes.
592 (compute_argument_addresses): Likewise.
593 (internal_arg_pointer_based_exp): Track polynomial offsets.
594 (mem_overlaps_already_clobbered_arg_p): Rename to...
595 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
596 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
597 Check stored_args_used_watermark.
598 (load_register_parameters): Update accordingly.
599 (check_sibcall_argument_overlap_1): Likewise.
600 (combine_pending_stack_adjustment_and_call): Take the unadjusted
601 args size as a poly_int64 rather than an int. Return a bool
602 indicating whether the optimization was possible and return
603 the new adjustment by reference.
604 (check_sibcall_argument_overlap): Track polynomail argument sizes.
605 Update stored_args_watermark.
606 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
607 (expand_call): Likewise. Maintain stack_usage_watermark and
608 stored_args_watermark. Update calls to
609 combine_pending_stack_adjustment_and_call.
610 (emit_library_call_value_1): Handle polynomial argument sizes.
611 Call stack_region_maybe_used_p and mark_stack_region_used.
612 Maintain stack_usage_watermark.
613 (store_one_arg): Likewise. Update call to
614 mem_overlaps_already_clobbered_arg_p.
615 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
617 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
618 * config/microblaze/microblaze.c (microblaze_function_prologue):
620 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
621 TARGET_RETURN_POPS_ARGS interface.
622 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
624 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
625 * config/i386/i386.c (ix86_return_pops_args): Update for new
626 TARGET_RETURN_POPS_ARGS interface.
627 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
628 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
629 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
630 TARGET_RETURN_POPS_ARGS interface.
631 * config/vax/vax.c (vax_return_pops_args): Likewise.
632 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
633 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
634 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
635 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
636 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
637 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
638 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
639 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
640 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
641 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
642 and x_stack_pointer_delta from int to poly_int64.
643 * dojump.c (do_pending_stack_adjust): Update accordingly.
644 * explow.c (allocate_dynamic_stack_space): Handle polynomial
645 stack_pointer_deltas.
646 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
647 (pad_to_arg_alignment): Track polynomial offsets.
648 (assign_parm_find_stack_rtl): Likewise.
649 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
650 * toplev.c (output_stack_usage): Update reference to
651 current_function_pushed_stack_size.
653 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
654 Alan Hayward <alan.hayward@arm.com>
655 David Sherwood <david.sherwood@arm.com>
657 * function.c (in_arg_offset, var_offset, dynamic_offset)
658 (out_arg_offset, cfa_offset): Change from int to poly_int64.
659 (instantiate_new_reg): Return the new offset as a poly_int64_pod
660 rather than a HOST_WIDE_INT.
661 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
662 (instantiate_virtual_regs_in_insn): Likewise.
664 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
665 Alan Hayward <alan.hayward@arm.com>
666 David Sherwood <david.sherwood@arm.com>
668 * rtl.h (get_args_size, add_args_size_note): New functions.
669 (find_args_size_adjust): Return a poly_int64 rather than a
671 (fixup_args_size_notes): Likewise. Make the same change to the
672 end_args_size parameter.
673 * rtlanal.c (get_args_size, add_args_size_note): New functions.
674 * builtins.c (expand_builtin_trap): Use add_args_size_note.
675 * calls.c (emit_call_1): Likewise.
676 * explow.c (adjust_stack_1): Likewise.
677 * cfgcleanup.c (old_insns_match_p): Update use of
678 find_args_size_adjust.
679 * combine.c (distribute_notes): Track polynomial arg sizes.
680 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
681 end_true_args_size, beg_delay_args_size and end_delay_args_size
682 from HOST_WIDE_INT to poly_int64.
683 (add_cfi_args_size): Take the args_size as a poly_int64 rather
684 than a HOST_WIDE_INT.
685 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
686 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
687 polynomial arg sizes.
688 * emit-rtl.c (try_split): Use get_args_size.
689 * recog.c (peep2_attempt): Likewise.
690 * reload1.c (reload_as_needed): Likewise.
691 * expr.c (find_args_size_adjust): Return the adjustment as a
692 poly_int64 rather than a HOST_WIDE_INT.
693 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
694 to a poly_int64 and change the return type in the same way.
695 (emit_single_push_insn): Track polynomial arg sizes.
697 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
698 Alan Hayward <alan.hayward@arm.com>
699 David Sherwood <david.sherwood@arm.com>
701 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
702 from HOST_WIDE_INT to poly_int64.
703 * expr.c (push_block, emit_push_insn): Likewise.
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.h (frame_space): Change start and length from HOST_WIDE_INT
711 (get_frame_size): Return the size as a poly_int64 rather than a
713 (frame_offset_overflow): Take the offset as a poly_int64 rather
714 than a HOST_WIDE_INT.
715 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
716 (assign_stack_temp): Likewise for the size.
717 * function.c (get_frame_size): Return a poly_int64 rather than
719 (frame_offset_overflow): Take the offset as a poly_int64 rather
720 than a HOST_WIDE_INT.
721 (try_fit_stack_local): Take the start, length and size as poly_int64s
722 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
723 rather than a HOST_WIDE_INT.
724 (add_frame_space): Take the start and end as poly_int64s rather than
726 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
727 (assign_stack_temp): Likewise for the size.
728 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
730 (find_temp_slot_from_address): Handle polynomial offsets.
731 (combine_temp_slots): Likewise.
732 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
734 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
735 poly_int64 rather than a HOST_WIDE_INT.
736 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
737 than a HOST_WIDE_INT.
738 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
739 polynomial frame offsets.
740 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
741 as a poly_int64 rather than an int.
742 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
743 * config/v850/v850-protos.h (compute_frame_size): Likewise.
744 * config/v850/v850.c (compute_frame_size): Likewise.
745 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
746 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
747 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
748 * config/pa/pa.c (pa_compute_frame_size): Likewise.
749 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
750 rather than a HOST_WIDE_INT.
751 * explow.c (get_dynamic_stack_base): Likewise.
752 * final.c (final_start_function): Use the constant lower bound
753 of the frame size for -Wframe-larger-than.
754 * ira.c (do_reload): Adjust for new get_frame_size return type.
755 * lra.c (lra): Likewise.
756 * reload1.c (reload): Likewise.
757 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
758 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
759 * rtlanal.c (get_initial_register_offset): Return the offset as
760 a poly_int64 rather than a HOST_WIDE_INT.
762 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
763 Alan Hayward <alan.hayward@arm.com>
764 David Sherwood <david.sherwood@arm.com>
766 * reload1.c (elim_table): Change initial_offset, offset and
767 previous_offset from HOST_WIDE_INT to poly_int64_pod.
768 (offsets_at): Change the target array's element type from
769 HOST_WIDE_INT to poly_int64_pod.
770 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
771 (elimination_costs_in_insn, update_eliminable_offsets)
772 (verify_initial_elim_offsets, set_offsets_for_label)
773 (init_eliminable_invariants): Update after above changes.
775 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
776 Alan Hayward <alan.hayward@arm.com>
777 David Sherwood <david.sherwood@arm.com>
779 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
780 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
781 (decomposition): Change start and end from HOST_WIDE_INT
783 (decompose, immune_p): Update accordingly.
784 (find_inc_amount): Return a poly_int64 rather than an int.
785 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
788 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
789 Alan Hayward <alan.hayward@arm.com>
790 David Sherwood <david.sherwood@arm.com>
792 * tree.h (get_inner_reference): Return the bitsize and bitpos
793 as poly_int64_pods rather than HOST_WIDE_INT.
794 * fold-const.h (ptr_difference_const): Return the pointer difference
795 as a poly_int64_pod rather than a HOST_WIDE_INT.
796 * expr.c (get_inner_reference): Return the bitsize and bitpos
797 as poly_int64_pods rather than HOST_WIDE_INT.
798 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
800 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
801 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
802 (optimize_bit_field_compare): Update call to get_inner_reference.
803 (decode_field_reference): Likewise.
804 (fold_unary_loc): Track polynomial offsets and sizes.
805 (split_address_to_core_and_offset): Return the bitpos as a
806 poly_int64_pod rather than a HOST_WIDE_INT.
807 (ptr_difference_const): Likewise for the pointer difference.
808 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
809 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
810 * dbxout.c (dbxout_expand_expr): Likewise.
811 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
812 (loc_list_from_tree_1, fortran_common): Likewise.
813 * gimple-laddress.c (pass_laddress::execute): Likewise.
814 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
815 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
816 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
817 * tree-affine.c (tree_to_aff_combination): Likewise.
818 (get_inner_reference_aff): Likewise.
819 * tree-data-ref.c (split_constant_offset_1): Likewise.
820 (dr_analyze_innermost): Likewise.
821 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
822 * tree-sra.c (ipa_sra_check_caller): Likewise.
823 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
824 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
825 (instrument_bool_enum_load, instrument_object_size): Likewise.
826 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
827 to get_inner_reference.
828 * hsa-gen.c (gen_hsa_addr): Likewise.
829 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
830 * tsan.c (instrument_expr): Likewise.
831 * match.pd: Update call to ptr_difference_const.
833 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
834 Alan Hayward <alan.hayward@arm.com>
835 David Sherwood <david.sherwood@arm.com>
837 * fold-const.c (fold_comparison): Track sizes and offsets as
838 poly_int64s rather than HOST_WIDE_INTs when folding address
841 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
842 Alan Hayward <alan.hayward@arm.com>
843 David Sherwood <david.sherwood@arm.com>
845 * expr.h (get_bit_range): Return the bitstart and bitend as
846 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
847 as a poly_int64 rather than a HOST_WIDE_INT.
848 * expr.c (get_bit_range): Likewise.
849 (expand_assignment): Update call accordingly.
850 * fold-const.c (optimize_bit_field_compare): Likewise.
852 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
853 Alan Hayward <alan.hayward@arm.com>
854 David Sherwood <david.sherwood@arm.com>
856 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
857 (aarch64_legitimate_address_p): Use it instead of an rtx code,
858 as an optional final parameter.
859 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
860 (aarch64_legitimate_address_p): Likewise.
861 (aarch64_print_address_internal): Take an aarch64_addr_query_type
862 instead of an rtx code.
863 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
864 (aarch64_legitimate_address_hook_p): Likewise.
865 (aarch64_print_ldpstp_address): Likewise.
866 (aarch64_print_operand_address): Likewise.
867 (aarch64_address_cost): Likewise.
868 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
869 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
870 (aarch64_mem_pair_lanes_operand): Likewise.
872 2017-12-20 Richard Biener <rguenther@suse.de>
874 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
875 update_stmt_if_modified.
877 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
879 PR tree-optimization/83491
880 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
881 before walking uses. Improve coding style and comments.
883 2017-12-20 Tom de Vries <tom@codesourcery.com>
885 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
887 2017-12-20 Jakub Jelinek <jakub@redhat.com>
890 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
892 (pass_ipa_free_fn_summary): Add small_p private data member,
893 initialize to false in the ctor.
894 (pass_ipa_free_fn_summary::clone,
895 pass_ipa_free_fn_summary::set_pass_param,
896 pass_ipa_free_fn_summary::gate): New methods.
897 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
898 | TODO_dump_symtab if small_p.
899 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
900 entry and add another instance of the pass with false parm after
902 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
903 ipa_free_fn_summary here.
905 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
907 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
909 2017-12-20 Martin Sebor <msebor@redhat.com>
912 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
913 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
914 (expand_builtin_strncmp): Same.
916 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
919 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
920 allowed debug stmts before labels.
921 (expand_gimple_basic_block): Likewise.
922 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
923 * gimple-iterator.h (gsi_after_labels): Likewise.
924 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
925 rename reused variable, and simplify using gsi_move_before.
926 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
927 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
928 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
929 (gimple_verify_flow_info, gimple_block_label): Likewise.
930 (make_blocks): Move debug markers after adjacent labels.
931 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
932 allowed debug insns outside blocks.
933 * df-scan.c (df_insn_delete): Likewise.
934 * lra-constraints.c (update_ebb_live_info): Likewise.
935 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
936 (vt_initialize, delete_vta_debug_insns): Likewise.
937 (reemit_marker_as_note): Drop BB parm. Adjust callers.
939 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
940 Alan Hayward <alan.hayward@arm.com>
941 David Sherwood <david.sherwood@arm.com>
943 * poly-int-types.h (round_down_to_byte_boundary): New macro.
944 (round_up_to_byte_boundary): Likewise.
945 * expr.h (get_bit_range): Add temporary shim.
946 * gimple-ssa-store-merging.c (store_operand_info): Change the
947 bitsize, bitpos, bitregion_start and bitregion_end fields from
948 unsigned HOST_WIDE_INT to poly_uint64.
949 (merged_store_group): Likewise load_align_base.
950 (compatible_load_p, compatible_load_p): Update accordingly.
951 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
952 (split_group, imm_store_chain_info::output_merged_store): Likewise.
953 (mem_valid_for_store_merging): Return the bitsize, bitpos,
954 bitregion_start and bitregion_end as poly_uint64s rather than
955 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
956 (handled_load): Take the bitsize, bitpos,
957 bitregion_start and bitregion_end as poly_uint64s rather than
958 unsigned HOST_WIDE_INTs.
959 (pass_store_merging::process_store): Update call to
960 mem_valid_for_store_merging.
962 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
963 Alan Hayward <alan.hayward@arm.com>
964 David Sherwood <david.sherwood@arm.com>
966 * builtins.c (get_object_alignment_2): Track polynomial offsets
967 and sizes. Update the alignment handling.
969 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
970 Alan Hayward <alan.hayward@arm.com>
971 David Sherwood <david.sherwood@arm.com>
973 * tree.h (get_inner_reference): Add a version that returns the
974 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
975 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
976 the case in which bitpos is not associated with the first byte.
978 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
979 Alan Hayward <alan.hayward@arm.com>
980 David Sherwood <david.sherwood@arm.com>
982 * tree-affine.h (get_inner_reference_aff): Return the size as a
984 * tree-affine.c (get_inner_reference_aff): Likewise.
985 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
986 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
988 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
989 Alan Hayward <alan.hayward@arm.com>
990 David Sherwood <david.sherwood@arm.com>
992 * fold-const.c (pointer_may_wrap_p): Take the offset as a
993 HOST_WIDE_INT rather than a poly_int64.
995 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
996 Alan Hayward <alan.hayward@arm.com>
997 David Sherwood <david.sherwood@arm.com>
999 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1000 HOST_WIDE_INT to poly_int64_pod.
1001 (perform_symbolic_merge): Update accordingly.
1002 (bswap_replace): Likewise.
1004 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1005 Alan Hayward <alan.hayward@arm.com>
1006 David Sherwood <david.sherwood@arm.com>
1008 * tree-affine.h (aff_tree::offset): Change from widest_int
1010 (wide_int_ext_for_comb): Delete.
1011 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1012 constants as poly_widest_int rather than widest_int.
1013 (aff_combination_constant_multiple_p): Return the multiplier
1014 as a poly_widest_int.
1015 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1017 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1018 version static and add an overload for poly_widest_int.
1019 (aff_combination_const, aff_combination_add_cst)
1020 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1021 the constants as poly_widest_int rather than widest_int.
1022 (tree_to_aff_combination): Generalize INTEGER_CST case to
1024 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1025 (aff_combination_add_product, aff_combination_mult): Handle
1027 (aff_combination_constant_multiple_p): Return the multiplier
1028 as a poly_widest_int.
1029 * tree-predcom.c (determine_offset): Return the offset as a
1031 (split_data_refs_to_components, suitable_component_p): Update
1033 (valid_initializer_p): Update call to
1034 aff_combination_constant_multiple_p.
1035 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1036 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1037 as a poly_int64 rather than a HOST_WIDE_INT.
1038 (get_address_cost): Handle polynomial offsets.
1039 (iv_elimination_compare_lt): Likewise.
1040 (rewrite_use_nonlinear_expr): Likewise.
1042 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1043 Alan Hayward <alan.hayward@arm.com>
1044 David Sherwood <david.sherwood@arm.com>
1046 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1047 as a poly_int64_pod rather than a HOST_WIDE_INT.
1048 (get_addr_base_and_unit_offset): Likewise.
1049 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1050 (get_addr_base_and_unit_offset): Likewise.
1051 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1052 to poly_int64 in example.
1053 * fold-const.c (fold_binary_loc): Update call to
1054 get_addr_base_and_unit_offset.
1055 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1056 (maybe_canonicalize_mem_ref_addr): Likewise.
1057 (gimple_fold_stmt_to_constant_1): Likewise.
1058 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1060 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
1061 * match.pd: Likewise.
1062 * omp-low.c (lower_omp_target): Likewise.
1063 * tree-sra.c (build_ref_for_offset): Likewise.
1064 (build_debug_ref_for_model): Likewise.
1065 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
1066 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
1067 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
1068 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1069 (constant_pointer_difference): Likewise.
1070 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
1071 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
1072 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
1073 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1074 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
1075 (set_ssa_val_to): Likewise.
1076 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
1077 (maybe_diag_stxncpy_trunc): Likewise.
1078 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1079 * tree.c (build_simple_mem_ref_loc): Likewise.
1080 (array_at_struct_end_p): Likewise.
1082 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1083 Alan Hayward <alan.hayward@arm.com>
1084 David Sherwood <david.sherwood@arm.com>
1086 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
1087 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1088 (get_ref_base_and_extent_hwi): Declare.
1089 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
1090 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1091 (get_ref_base_and_extent_hwi): New function.
1092 * cfgexpand.c (expand_debug_expr): Update call to
1093 get_ref_base_and_extent.
1094 * dwarf2out.c (add_var_loc_to_decl): Likewise.
1095 * gimple-fold.c (get_base_constructor): Return the offset as a
1096 poly_int64_pod rather than a HOST_WIDE_INT.
1097 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
1098 * ipa-polymorphic-call.c
1099 (ipa_polymorphic_call_context::set_by_invariant)
1100 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
1101 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
1102 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
1103 rather than get_ref_base_and_extent.
1104 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
1105 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
1106 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
1108 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
1109 call to get_ref_base_and_extent.
1110 * tree-sra.c (create_access, get_access_for_expr): Likewise.
1111 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
1112 (stmt_kills_ref_p): Likewise.
1113 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1114 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
1116 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
1117 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
1118 when calling native_encode_expr.
1119 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
1120 call to get_ref_base_and_extent.
1121 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
1122 get_ref_base_and_extent.
1123 * var-tracking.c (track_expr_p): Likewise.
1125 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1126 Alan Hayward <alan.hayward@arm.com>
1127 David Sherwood <david.sherwood@arm.com>
1129 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
1130 HOST_WIDE_INT to poly_int64_pod.
1131 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
1132 polynomail parameter offsets.
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 * gengtype.c (main): Handle poly_int64_pod.
1139 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
1140 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
1141 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
1142 from HOST_WIDE_INT to poly_int64_pod.
1143 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
1144 (copy_cfa): New function.
1145 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
1146 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
1147 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
1148 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
1149 (initial_return_save): Treat offsets as poly_ints.
1150 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
1151 offset is nonconstant.
1152 (reg_save): Take the offset as a poly_int64. Fall back to
1153 DW_CFA_expression for nonconstant offsets.
1154 (queue_reg_save): Take the offset as a poly_int64.
1155 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
1157 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1158 Alan Hayward <alan.hayward@arm.com>
1159 David Sherwood <david.sherwood@arm.com>
1161 * rtl.h (operand_subword, operand_subword_force): Take the offset
1162 as a poly_uint64 an unsigned int.
1163 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
1165 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1166 Alan Hayward <alan.hayward@arm.com>
1167 David Sherwood <david.sherwood@arm.com>
1169 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
1170 'p' format code. Use INT_LIST rather than SUBREG as the example of
1171 a code with an XINT and an XEXP. Remove the implication that
1172 accessing an rtx field using XINT is expected to work.
1173 * rtl.def (SUBREG): Change format from "ei" to "ep".
1174 * rtl.h (rtunion::rt_subreg): New field.
1175 (XCSUBREG): New macro.
1176 (SUBREG_BYTE): Use it.
1177 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
1178 Update constructor accordingly.
1179 (subreg_shape::operator ==): Update accordingly.
1180 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
1181 than an unsigned int.
1182 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
1183 a poly_uint64 rather than an unsigned int.
1184 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
1185 than an unsigned int.
1186 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
1187 (subreg_size_highpart_offset): Return a poly_uint64 rather than
1188 an unsigned int. Take the sizes as poly_uint64s.
1189 (subreg_offset_from_lsb): Return a poly_uint64 rather than
1190 an unsigned int. Take the shift as a poly_uint64 rather than
1192 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
1193 as a poly_uint64 rather than an unsigned int.
1194 (simplify_subreg_regno): Likewise.
1195 (byte_lowpart_offset): Return the memory offset as a poly_int64
1197 (subreg_memory_offset): Likewise. Take the subreg offset as a
1198 poly_uint64 rather than an unsigned int.
1199 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
1200 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
1201 poly_uint64 rather than an unsigned int.
1202 * rtl.c (rtx_format): Describe 'p' in comment.
1203 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
1204 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
1205 offset as a poly_uint64 rather than an unsigned int.
1206 (byte_lowpart_offset): Return the memory offset as a poly_int64
1208 (subreg_memory_offset): Likewise. Take the subreg offset as a
1209 poly_uint64 rather than an unsigned int.
1210 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
1211 mode sizes as poly_uint64s rather than unsigned ints. Return a
1212 poly_uint64 rather than an unsigned int.
1213 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
1214 (copy_insn_1): Handle 'p'.
1215 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
1216 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
1217 an unsigned int. Return the shift in the same way.
1218 (subreg_lsb): Return the shift as a poly_uint64 rather than an
1220 (subreg_size_offset_from_lsb): Take the sizes and shift as
1221 poly_uint64s rather than unsigned ints. Return the offset as
1223 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
1224 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
1226 * rtlhash.c (add_rtx): Handle 'p'.
1227 * genemit.c (gen_exp): Likewise.
1228 * gengenrtl.c (type_from_format, gendef): Likewise.
1229 * gensupport.c (subst_pattern_match, get_alternatives_number)
1230 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
1231 (subst_dup): Likewise.
1232 * gengtype.c (adjust_field_rtx_def): Likewise.
1233 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
1234 (match_pattern_2): Likewise.
1235 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
1236 (rtx_test::subreg_field): New function.
1237 (operator ==, safe_to_hoist_p, transition_parameter_type)
1238 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
1239 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
1240 * genpeep.c (match_rtx): Likewise.
1241 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
1242 (rtx_writer::print_rtx_operand): Handle 'p'.
1243 (print_value): Handle SUBREG.
1244 * read-rtl.c (apply_int_iterator): Likewise.
1245 (rtx_reader::read_rtx_operand): Handle 'p'.
1246 * alias.c (rtx_equal_for_memref_p): Likewise.
1247 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
1248 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
1250 * calls.c (expand_call): Likewise.
1251 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
1252 (make_extraction, gen_lowpart_for_combine): Likewise.
1253 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
1255 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
1256 rather than an unsigned int. Treat subreg offsets as poly_ints.
1257 (exp_equiv_p): Handle 'p'.
1258 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
1259 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
1260 * dse.c (find_shift_sequence): Likewise.
1261 * dwarf2out.c (rtl_for_decl_location): Likewise.
1262 * expmed.c (extract_low_bits): Likewise.
1263 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
1264 (expand_expr_real_2): Likewise.
1265 * final.c (alter_subreg): Likewise.
1266 (leaf_renumber_regs_insn): Handle 'p'.
1267 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
1268 Treat subreg offsets as poly_ints.
1269 * fwprop.c (forward_propagate_and_simplify): Likewise.
1270 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
1271 * ira.c (get_subreg_tracking_sizes): Likewise.
1272 * ira-conflicts.c (go_through_subreg): Likewise.
1273 * ira-lives.c (process_single_reg_class_operands): Likewise.
1274 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
1275 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
1276 as a poly_uint64 rather than an unsigned int.
1277 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
1278 subreg offsets as poly_ints.
1279 * lra-constraints.c (operands_match_p): Handle 'p'.
1280 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
1281 * lra-spills.c (assign_mem_slot): Likewise.
1282 * postreload.c (move2add_valid_value_p): Likewise.
1283 * recog.c (general_operand, indirect_operand): Likewise.
1284 * regcprop.c (copy_value, maybe_mode_change): Likewise.
1285 (copyprop_hardreg_forward_1): Likewise.
1286 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
1287 (record_subregs_of_mode): Likewise.
1288 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
1289 * reload.c (operands_match_p): Handle 'p'.
1290 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
1291 * reload1.c (alter_reg, choose_reload_regs): Likewise.
1292 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
1293 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
1294 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
1295 (simplify_const_poly_int_tests<N>::run): Likewise.
1296 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
1297 a poly_uint64 rather than an unsigned int.
1298 * valtrack.c (debug_lowpart_subreg): Likewise.
1299 * var-tracking.c (var_lowpart): Likewise.
1300 (loc_cmp): Handle 'p'.
1302 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1303 Alan Hayward <alan.hayward@arm.com>
1304 David Sherwood <david.sherwood@arm.com>
1306 * ira.c (get_subreg_tracking_sizes): New function.
1307 (init_live_subregs): Take an integer size rather than a register.
1308 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
1309 to init_live_subregs.
1311 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1312 Alan Hayward <alan.hayward@arm.com>
1313 David Sherwood <david.sherwood@arm.com>
1315 * expr.c (store_constructor_field): Change bitsize from a
1316 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
1317 HOST_WIDE_INT to a poly_int64.
1318 (store_constructor): Change size from a HOST_WIDE_INT to
1320 (store_field): Likewise bitsize and bitpos.
1322 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1323 Alan Hayward <alan.hayward@arm.com>
1324 David Sherwood <david.sherwood@arm.com>
1326 * expmed.h (store_bit_field): Change bitregion_start and
1327 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
1328 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
1329 (store_bit_field_1, store_integral_bit_field, store_bit_field)
1330 (store_fixed_bit_field, store_split_bit_field): Likewise.
1331 * expr.c (store_constructor_field, store_field): Likewise.
1332 (optimize_bitfield_assignment_op): Likewise. Make the same change
1333 to bitsize and bitpos.
1334 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
1335 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
1336 same change in the constructor arguments.
1337 (get_best_mode): Change bitregion_start and bitregion_end from
1338 unsigned HOST_WIDE_INT to poly_uint64.
1339 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1340 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
1342 (bit_field_mode_iterator::next_mode): Update for new types
1343 of m_bitregion_start and m_bitregion_end.
1344 (get_best_mode): Change bitregion_start and bitregion_end from
1345 unsigned HOST_WIDE_INT to poly_uint64.
1347 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1348 Alan Hayward <alan.hayward@arm.com>
1349 David Sherwood <david.sherwood@arm.com>
1351 * rtl.h (simplify_gen_subreg): Add a temporary overload that
1352 accepts poly_uint64 offsets.
1353 * expmed.h (extract_bit_field): Take bitsize and bitnum as
1354 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1355 * expmed.c (lowpart_bit_field_p): Likewise.
1356 (extract_bit_field_as_subreg): New function, split out from...
1357 (extract_bit_field_1): ...here. Take bitsize and bitnum as
1358 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
1359 extractions, check that BITSIZE matches the size of the extracted
1360 value and that BITNUM is an exact multiple of that size.
1361 If all else fails, try forcing the value into memory if
1362 BITNUM is variable, and adjusting the address so that the
1363 offset is constant. Split the part that can only handle constant
1364 bitsize and bitnum out into...
1365 (extract_integral_bit_field): ...this new function.
1366 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
1367 rather than unsigned HOST_WIDE_INTs.
1369 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1370 Alan Hayward <alan.hayward@arm.com>
1371 David Sherwood <david.sherwood@arm.com>
1373 * expmed.h (store_bit_field): Take bitsize and bitnum as
1374 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1375 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
1376 that returns the byte size.
1377 (store_bit_field_1): Take bitsize and bitnum as
1378 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
1379 to simple_mem_bitfield_p. Split the part that can only handle
1380 constant bitsize and bitnum out into...
1381 (store_integral_bit_field): ...this new function.
1382 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
1383 than unsigned HOST_WIDE_INTs.
1384 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
1386 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1387 Alan Hayward <alan.hayward@arm.com>
1388 David Sherwood <david.sherwood@arm.com>
1390 * lra-int.h (lra_reg): Change offset from int to poly_int64.
1391 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
1393 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1394 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
1395 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
1396 offset as a poly_int64 rather than an int.
1397 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
1398 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1399 * lra-constraints.c (equiv_address_substitution): Track offsets
1401 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
1402 (curr_insn_transform): Handle the new form of sp_offset.
1403 * lra-eliminations.c (lra_elim_table): Change previous_offset
1404 and offset from HOST_WIDE_INT to poly_int64.
1405 (print_elim_table, update_reg_eliminate): Update accordingly.
1406 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
1407 (get_elimination): Update accordingly.
1408 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
1409 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1410 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
1411 poly_int64 offsets generally.
1412 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
1413 (mark_not_eliminable, init_elimination): Update accordingly.
1414 (remove_reg_equal_offset_note): Return a bool and pass the new
1415 offset back by pointer as a poly_int64.
1416 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
1417 rather than a HOST_WIDE_INT.
1418 (do_remat): Track offsets poly_int64s.
1419 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
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 * rtl.h (mem_attrs): Add a default constructor. Change size and
1426 offset from HOST_WIDE_INT to poly_int64.
1427 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
1428 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
1429 (widen_memory_access): Take the sizes and offsets as poly_int64s
1430 rather than HOST_WIDE_INTs.
1431 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
1432 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
1434 (adjust_offset_for_component_ref): Change the offset from a
1435 HOST_WIDE_INT to a poly_int64.
1436 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
1437 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
1438 * dce.c (find_call_stack_args): Likewise.
1439 * dse.c (record_store): Likewise.
1440 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
1441 * print-rtl.c (rtx_writer::print_rtx): Likewise.
1442 * read-rtl-function.c (test_loading_mem): Likewise.
1443 * rtlanal.c (may_trap_p_1): Likewise.
1444 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1445 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
1446 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
1447 (mem_attrs::mem_attrs): New function.
1448 (set_mem_attributes_minus_bitpos): Change bitpos from a
1449 HOST_WIDE_INT to poly_int64.
1450 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
1451 (clear_mem_offset, clear_mem_size, change_address)
1452 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
1453 initialize mem_attrs.
1454 (set_mem_offset, set_mem_size, adjust_address_1)
1455 (adjust_automodify_address_1, offset_address, widen_memory_access):
1456 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
1458 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1459 Alan Hayward <alan.hayward@arm.com>
1460 David Sherwood <david.sherwood@arm.com>
1462 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
1463 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
1464 rather than 0 to represent an unknown size. Assert that the size
1465 is known when the mode isn't BLKmode.
1466 (may_trap_p_1): Use -1 for unknown sizes.
1467 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
1469 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1470 Alan Hayward <alan.hayward@arm.com>
1471 David Sherwood <david.sherwood@arm.com>
1473 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
1474 to poly_int64. Update commentary for positions_needed.large.
1475 (read_info_type): Change offset and width from HOST_WIDE_INT
1477 (set_usage_bits): Likewise.
1478 (canon_address): Return the offset as a poly_int64 rather than
1479 a HOST_WIDE_INT. Use strip_offset_and_add.
1480 (set_all_positions_unneeded, any_positions_needed_p): Use
1481 positions_needed.large to track stores with non-constant widths.
1482 (all_positions_needed_p): Likewise. Take the offset and width
1483 as poly_int64s rather than ints. Assert that rhs is nonnull.
1484 (record_store): Cope with non-constant offsets and widths.
1485 Nullify the rhs of an earlier store if we can't tell which bytes
1487 (find_shift_sequence): Take the access_size and shift as poly_int64s
1489 (get_stored_val): Take the read_offset and read_width as poly_int64s
1490 rather than HOST_WIDE_INTs.
1491 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
1492 non-constant offsets and widths.
1494 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1495 Alan Hayward <alan.hayward@arm.com>
1496 David Sherwood <david.sherwood@arm.com>
1498 * inchash.h (inchash::hash::add_poly_int): New function.
1499 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1500 Use poly_int64 rather than HOST_WIDE_INT.
1501 (ao_ref::max_size_known_p): New function.
1502 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1503 rather than HOST_WIDE_INT.
1504 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
1505 to temporaries until its interface is adjusted to match.
1506 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
1507 (aliasing_component_refs_p, decl_refs_may_alias_p)
1508 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
1509 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
1510 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
1512 * alias.c (ao_ref_from_mem): Likewise.
1513 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1514 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
1515 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
1516 (maybe_trim_complex_store, maybe_trim_constructor_store)
1517 (live_bytes_read, dse_classify_store): Likewise.
1518 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
1519 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
1520 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
1521 (vn_reference_lookup_3): Likewise.
1522 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
1524 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1525 Alan Hayward <alan.hayward@arm.com>
1526 David Sherwood <david.sherwood@arm.com>
1528 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
1529 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
1530 instead of ranges_overlap_p.
1532 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1533 Alan Hayward <alan.hayward@arm.com>
1534 David Sherwood <david.sherwood@arm.com>
1536 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
1537 sizes as poly_int64s rather than HOST_WIDE_INTs.
1539 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1540 Alan Hayward <alan.hayward@arm.com>
1541 David Sherwood <david.sherwood@arm.com>
1543 * gimple-fold.h (fold_ctor_reference): Take the offset and size
1544 as poly_uint64 rather than unsigned HOST_WIDE_INT.
1545 * gimple-fold.c (fold_ctor_reference): Likewise.
1547 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1548 Alan Hayward <alan.hayward@arm.com>
1549 David Sherwood <david.sherwood@arm.com>
1551 * target.def (dwarf_poly_indeterminate_value): New hook.
1552 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
1553 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
1554 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
1555 * doc/tm.texi: Regenerate.
1556 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
1557 offset as a poly_int64.
1558 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
1559 offset as a poly_int64.
1560 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
1561 Take the offset as a poly_int64.
1562 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1563 (frame_pointer_fb_offset): Change to a poly_int64.
1564 (int_loc_descriptor): Take the offset as a poly_int64. Use
1565 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1566 (based_loc_descr): Take the offset as a poly_int64.
1567 Use strip_offset_and_add to handle (plus X (const)).
1568 Use new_reg_loc_descr instead of an open-coded version of the
1569 previous implementation.
1570 (mem_loc_descriptor): Handle CONST_POLY_INT.
1571 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1572 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1574 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1575 Alan Hayward <alan.hayward@arm.com>
1576 David Sherwood <david.sherwood@arm.com>
1578 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1579 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1580 * inchash.h (inchash::hash::add_poly_hwi): New function.
1581 * gengtype.c (main): Register poly_int64.
1582 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1583 offset as a poly_int.
1584 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1585 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1586 offset as a poly_int64.
1587 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1588 * print-rtl.c (print_poly_int): New function.
1589 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1591 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1593 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1594 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1595 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1596 Enforce track_offset_p for parts of a PARALLEL.
1597 (vt_add_function_parameter): Use const_offset for the final
1598 offset to track. Use get_tracked_reg_offset for the parts
1601 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1602 Alan Hayward <alan.hayward@arm.com>
1603 David Sherwood <david.sherwood@arm.com>
1605 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1606 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1607 * doc/tm.texi: Regenerate.
1608 * hooks.h (hook_bool_uint_uint_true): Delete.
1609 (hook_bool_puint64_puint64_true): Declare.
1610 * hooks.c (hook_bool_uint_uint_true): Delete.
1611 (hook_bool_puint64_puint64_true): New function.
1612 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1613 instead of unsigned ints.
1614 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1615 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1617 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1618 Alan Hayward <alan.hayward@arm.com>
1619 David Sherwood <david.sherwood@arm.com>
1621 * optabs.h (expand_operand): Add an int_value field.
1622 (create_expand_operand): Add an int_value parameter and use it
1623 to initialize the new expand_operand field.
1624 (create_integer_operand): Replace with a declaration of a function
1625 that accepts poly_int64s. Move the implementation to...
1626 * optabs.c (create_integer_operand): ...here.
1627 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1628 the mode preserves the value of int_value, instead of calling
1629 const_int_operand on the rtx. Use gen_int_mode to generate
1632 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1633 Alan Hayward <alan.hayward@arm.com>
1634 David Sherwood <david.sherwood@arm.com>
1636 * dumpfile.h (dump_dec): Declare.
1637 * dumpfile.c (dump_dec): New function.
1638 * pretty-print.h (pp_wide_integer): Turn into a function and
1639 declare a poly_int version.
1640 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1642 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1643 Alan Hayward <alan.hayward@arm.com>
1644 David Sherwood <david.sherwood@arm.com>
1646 * doc/generic.texi (POLY_INT_CST): Document.
1647 * tree.def (POLY_INT_CST): New tree code.
1648 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1649 * tree-core.h (tree_poly_int_cst): New struct.
1650 (tree_node): Add a poly_int_cst field.
1651 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1652 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1653 instead of a wide_int_ref.
1654 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1656 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1657 instead of an unsigned HOST_WIDE_INT.
1658 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1659 (ptrdiff_tree_p): Declare.
1660 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1661 extern inline implementations if the target doesn't use POLY_INT_CST.
1662 (poly_int_tree_p): New function.
1663 (wi::unextended_tree): New class.
1664 (wi::int_traits <unextended_tree>): New override.
1665 (wi::extended_tree): Add a default constructor.
1666 (wi::extended_tree::get_tree): New function.
1667 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1668 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1669 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1670 (wi::tree_to_poly_wide_ref): New typedefs.
1671 (wi::ints_for): Provide overloads for extended_tree and
1673 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1674 (wi::to_wide): New functions.
1675 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1676 * tree.c (poly_int_cst_hasher): New struct.
1677 (poly_int_cst_hash_table): New variable.
1678 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1679 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1681 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1682 (init_ttree): Initialize poly_int_cst_hash_table.
1683 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1684 a poly_int64 instead of a HOST_WIDE_INT.
1685 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1686 instead of an unsigned HOST_WIDE_INT.
1687 (wide_int_to_tree): Rename to...
1688 (wide_int_to_tree_1): ...this.
1689 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1690 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1691 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1692 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1693 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1694 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1696 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1697 (streamer_read_tree_body): Likewise.
1698 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1699 (streamer_write_tree_body): Likewise.
1700 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1701 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1702 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1703 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1704 * gimple-expr.h (is_gimple_constant): Likewise.
1705 * gimplify.c (maybe_with_size_expr): Likewise.
1706 * print-tree.c (print_node): Likewise.
1707 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1708 * tree-pretty-print.c (dump_generic_node): Likewise.
1709 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1710 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1711 * tree-vrp.c (compare_values_warnv): Likewise.
1712 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1713 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1714 (force_expr_to_var_cost): Likewise.
1715 * tree-ssa-loop.c (for_each_index): Likewise.
1716 * fold-const.h (build_invariant_address, size_int_kind): Take a
1717 poly_int64 instead of a HOST_WIDE_INT.
1718 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1719 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1721 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1722 (int_const_binop_2): New function, split out from...
1723 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1724 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1725 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1726 * gimple-ssa-strength-reduction.c (slsr_process_add)
1727 (slsr_process_mul): Check for INTEGER_CSTs before using them
1729 * stor-layout.c (bits_from_bytes): New function.
1730 (bit_from_pos): Use it.
1731 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1732 by BITS_PER_UNIT to get the TYPE_SIZE.
1733 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1734 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1736 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1737 Alan Hayward <alan.hayward@arm.com>
1738 David Sherwood <david.sherwood@arm.com>
1740 * doc/rtl.texi (const_poly_int): Document. Also document the
1741 rtl sharing behavior.
1742 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1743 * rtl.h (const_poly_int_def): New struct.
1744 (rtx_def::u): Add a cpi field.
1745 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1746 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1747 (wi::rtx_to_poly_wide_ref): New typedef
1748 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1749 (poly_int_rtx_p): New functions.
1750 (trunc_int_for_mode): Declare a poly_int64 version.
1751 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1752 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1754 (strip_offset): Declare.
1755 (strip_offset_and_add): New function.
1756 * rtl.def (CONST_POLY_INT): New rtx code.
1757 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1758 (shared_const_p): Use poly_int_rtx_p.
1759 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1761 (gen_int_shift_amount): Likewise.
1762 * emit-rtl.c (const_poly_int_hasher): New class.
1763 (const_poly_int_htab): New variable.
1764 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1765 (const_poly_int_hasher::hash): New function.
1766 (const_poly_int_hasher::equal): Likewise.
1767 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1768 (immed_wide_int_const): Rename to...
1769 (immed_wide_int_const_1): ...this and make static.
1770 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1771 instead of a wide_int_ref.
1772 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1773 (gen_lowpart_common): Handle CONST_POLY_INT.
1774 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1775 * cselib.c (cselib_hash_rtx): Likewise.
1776 * dwarf2out.c (const_ok_for_output_1): Likewise.
1777 * expr.c (convert_modes): Likewise.
1778 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1779 * rtlhash.c (add_rtx): Likewise.
1780 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1781 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1782 Handle existing CONST_POLY_INT rtxes.
1783 * expmed.h (expand_shift): Take a poly_int64 instead of a
1785 * expmed.c (expand_shift): Likewise.
1786 * rtlanal.c (strip_offset): New function.
1787 (commutative_operand_precedence): Give CONST_POLY_INT the same
1788 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1790 * rtl-tests.c (const_poly_int_tests): New struct.
1791 (rtl_tests_c_tests): Use it.
1792 * simplify-rtx.c (simplify_const_unary_operation): Handle
1794 (simplify_const_binary_operation): Likewise.
1795 (simplify_binary_operation_1): Fold additions of symbolic constants
1796 and CONST_POLY_INTs.
1797 (simplify_subreg): Handle extensions and truncations of
1799 (simplify_const_poly_int_tests): New struct.
1800 (simplify_rtx_c_tests): Use it.
1801 * wide-int.h (storage_ref): Add default constructor.
1802 (wide_int_ref_storage): Likewise.
1803 (trailing_wide_ints): Use GTY((user)).
1804 (trailing_wide_ints::operator[]): Add a const version.
1805 (trailing_wide_ints::get_precision): New function.
1806 (trailing_wide_ints::extra_size): Likewise.
1808 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1809 Alan Hayward <alan.hayward@arm.com>
1810 David Sherwood <david.sherwood@arm.com>
1812 * emit-rtl.h (gen_int_shift_amount): Declare.
1813 * emit-rtl.c (gen_int_shift_amount): New function.
1814 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
1816 * calls.c (shift_return_value): Likewise.
1817 * cse.c (fold_rtx): Likewise.
1818 * dse.c (find_shift_sequence): Likewise.
1819 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
1820 (expand_shift, expand_smod_pow2): Likewise.
1821 * lower-subreg.c (shift_cost): Likewise.
1822 * optabs.c (expand_superword_shift, expand_doubleword_mult)
1823 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
1824 (expand_vec_perm_var): Likewise.
1825 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1826 (simplify_binary_operation_1): Likewise.
1827 * combine.c (try_combine, find_split_point, force_int_to_mode)
1828 (simplify_shift_const_1, simplify_shift_const): Likewise.
1829 (change_zero_ext): Likewise. Use simplify_gen_binary.
1831 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1833 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
1835 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1837 * doc/invoke.texi (ARM Options): Document accepted extension options
1838 for -march=armv8.3-a.
1840 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
1843 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
1844 or --with-float=softfp, set the default CPU to arm10e.
1846 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
1848 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
1849 * config/visium/predicates.md (const_shift_operand): Likewise.
1850 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
1851 (visium_legitimize_reload_address): Likewise.
1853 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1856 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
1858 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1861 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
1863 2017-12-20 Julia Koval <julia.koval@intel.com>
1865 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
1866 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
1867 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
1868 * config.gcc: Include vpclmulqdqintrin.h.
1869 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
1870 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
1872 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
1873 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
1874 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
1875 * config/i386/i386.c (isa2_opts): Add -mcx16.
1876 (isa_opts): Add -mpclmulqdq, remove -mcx16.
1877 (ix86_option_override_internal): Move mcx16 to flags2.
1878 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
1879 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
1880 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
1881 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
1882 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
1883 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
1884 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
1885 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
1886 * doc/invoke.texi: Add -mvpclmulqdq.
1888 2017-12-20 Tom de Vries <tom@codesourcery.com>
1891 * config/i386/i386.c (ix86_static_chain): Move
1892 DECL_STATIC_CHAIN test ...
1893 * calls.c (rtx_for_static_chain): ... here. New function.
1894 * calls.h (rtx_for_static_chain): Declare.
1895 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
1896 instead of targetm.calls.static_chain.
1897 * df-scan.c (df_get_entry_block_def_set): Same.
1899 2017-12-19 Tom de Vries <tom@codesourcery.com>
1901 PR tree-optimization/83493
1902 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
1905 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1907 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
1908 inform with hardcoded english plural handling.
1910 2017-12-18 Jeff Law <law@redhat.com>
1912 PR tree-optimization/83477
1913 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
1914 a non-virtual PHI, always push a new range.
1916 2017-12-19 Martin Sebor <msebor@redhat.com>
1919 * builtins.c (compute_objsize): Handle non-constant offsets.
1921 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1923 PR tree-optimization/83444
1924 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
1925 character load case, if get_stridx on MEM_REF's operand doesn't
1926 look usable, retry with get_addr_stridx.
1928 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
1931 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
1932 (variable_tracking_main_1): Keep markers even when VTA fails.
1935 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
1936 even if there are markers before it.
1937 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
1939 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1941 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
1942 typical order conditions.
1943 * sel-sched.c (extract_new_fences_from): Likewise.
1944 * config/visium/constraints.md (J, K, L): Likewise.
1945 * config/visium/predicates.md (const_shift_operand): Likewise.
1946 * config/visium/visium.c (visium_legitimize_address,
1947 visium_legitimize_reload_address): Likewise.
1948 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
1949 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
1950 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
1951 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
1952 SET_DUMP_DETAIL): Likewise.
1953 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
1954 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
1955 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
1956 avr_set_core_architecture, avr_set_current_function,
1957 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
1958 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
1959 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
1960 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
1961 avr_map_decompose, avr_fold_builtin): Likewise.
1962 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
1963 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
1964 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
1965 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
1966 * config/m32c/m32c.c (m32c_conditional_register_usage,
1967 m32c_address_cost): Likewise.
1968 * config/m32c/predicates.md (shiftcount_operand,
1969 longshiftcount_operand): Likewise.
1970 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
1971 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
1972 can_use_cdx_ldstw): Likewise.
1973 * config/nios2/nios2.h (CDX_REG_P): Likewise.
1974 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
1976 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
1977 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
1978 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
1979 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
1980 * config/vax/vax.c (adjacent_operands_p): Likewise.
1981 * config/ft32/constraints.md (L, b, KA): Likewise.
1982 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
1984 * cfgexpand.c (expand_stack_alignment): Likewise.
1985 * gcse.c (insert_expr_in_table): Likewise.
1986 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
1987 * cgraphunit.c (cgraph_node::expand): Likewise.
1988 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
1989 * emit-rtl.c (add_insn): Likewise.
1990 * input.c (dump_location_info): Likewise.
1991 * passes.c (NEXT_PASS): Likewise.
1992 * read-rtl-function.c (parse_note_insn_name,
1993 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
1995 * sched-rgn.c (sched_rgn_init): Likewise.
1996 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
1997 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
1998 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2000 * explow.c (eliminate_constant_term): Likewise.
2001 * final.c (leaf_renumber_regs_insn): Likewise.
2002 * cfgrtl.c (print_rtl_with_bb): Likewise.
2003 * genhooks.c (emit_init_macros): Likewise.
2004 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2005 * tree-data-ref.c (conflict_fn): Likewise.
2006 * selftest.c (assert_streq): Likewise.
2007 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2008 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2009 fold_binary_loc, multiple_of_p): Likewise.
2010 * reload.c (push_reload, find_equiv_reg): Likewise.
2011 * et-forest.c (et_nca, et_below): Likewise.
2012 * dbxout.c (dbxout_symbol_location): Likewise.
2013 * reorg.c (relax_delay_slots): Likewise.
2014 * dojump.c (do_compare_rtx_and_jump): Likewise.
2015 * gengtype-parse.c (type): Likewise.
2016 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2017 simplify_const_relational_operation): Likewise.
2018 * reload1.c (do_output_reload): Likewise.
2019 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2020 * gengtype.c (type_for_name): Likewise.
2021 * gimple-ssa-sprintf.c (format_directive): Likewise.
2023 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2026 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2027 accessing it. Adjust comment.
2029 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2032 * predict.c (zero_one_minusone): New function.
2033 (apply_return_prediction): Avoid return prediction for functions
2034 returning only -1, 0 and 1 values, unless they only return -1 and 0
2037 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2039 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2040 fall-through warning.
2041 (arc_compute_frame_size): Remove unused variables.
2042 (arc_print_operand): Fix fprintif format.
2043 (arc_can_follow_jump): Clean fall-through warning.
2045 2017-12-19 Marek Polacek <polacek@redhat.com>
2048 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2051 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2053 * config/arc/arc.c (overriderregs): New variable.
2054 (arc_override_options): Track fixed/call saved/call options.
2055 (arc_conditional_register_usage): Check against overrideregs
2056 variable whenever we change register properties.
2058 2017-12-19 Nathan Sidwell <nathan@acm.org>
2060 * opts.c (finish_options): Don't prefix dump_base_name if it
2061 already contains directories.
2063 2017-12-19 Martin Liska <mliska@suse.cz>
2065 PR rtl-optimization/82675
2066 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
2067 more element in sbitmap.
2069 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
2071 * gimplify.c (gimplify_expr): Use error_operand_p.
2073 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
2076 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
2079 2017-12-19 Daniel Cederman <cederman@gaisler.com>
2081 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
2082 the jump is to a label.
2084 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2086 PR tree-optimization/83444
2087 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
2092 * ipa-inline.c (flatten_remove_node_hook): New function.
2093 (ipa_inline): Keep only nodes with flatten attribute at the end of
2094 the array in the order from ipa_reverse_postorder, only walk that
2095 portion of array for flattening, if there is more than one such
2096 node, temporarily register a removal hook and ignore removed nodes.
2098 PR tree-optimization/80631
2099 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
2100 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
2101 IFN_REDUC_MAX or IFN_REDUC_MIN.
2103 2017-12-18 Jeff Law <law@redhat.com>
2105 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
2106 record symbolic equivalences from backedges in the CFG.
2109 2017-11-19 Jeff Law <law@redhat.com>
2111 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2112 of degenerates resulting from ignoring an edge.
2114 2017-12-18 Martin Sebor <msebor@redhat.com>
2117 PR tree-optimization/78450
2118 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
2119 (handle_builtin_strlen): Call it.
2121 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
2123 PR rtl-optimization/83424
2124 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
2126 2017-12-18 Marek Polacek <polacek@redhat.com>
2129 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
2130 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
2133 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2135 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
2136 (umaddsidi4, umaddsidi_split): Likewise.
2138 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2140 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
2143 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2146 * config/s390/s390.c (s390_option_override): Avoid strncpy.
2148 2017-12-18 Richard Biener <rguenther@suse.de>
2150 PR tree-optimization/81877
2151 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
2152 (outermost_indep_loop): Adjust.
2153 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
2154 (can_sm_ref_p): Adjust.
2156 2017-12-18 Richard Biener <rguenther@suse.de>
2159 * tree.c (array_at_struct_end_p): Return true if the underlying
2160 object has space for at least one element in excess of what
2161 the array domain specifies.
2163 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2165 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
2166 example to avoid overfull hbox.
2167 * doc/invoke.texi (Option Summary): Add missing @gol.
2168 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
2171 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2172 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2174 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
2175 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
2176 (AArch64 Options): Document -mverbose-cost-dump.
2177 (ARM Options): Likewise, plus -mflip-thumb.
2179 2017-12-17 Martin Sebor <msebor@redhat.com>
2182 * gimple-ssa-warn-restrict.c
2183 (builtin_memref::offset_out_of_bounds): Correct the handling of
2186 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2188 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
2190 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
2192 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
2194 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
2196 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
2199 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
2201 * doc/invoke.texi: Fix some typos.
2203 2017-12-16 Martin Sebor <msebor@redhat.com>
2205 PR tree-optimization/78918
2206 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
2207 * builtins.c (check_sizes): Rename...
2208 (check_access): ...to this. Rename function arguments for clarity.
2209 (check_memop_sizes): Adjust names.
2210 (expand_builtin_memchr, expand_builtin_memcpy): Same.
2211 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
2212 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
2213 (check_strncat_sizes, expand_builtin_strncat): Same.
2214 (expand_builtin_strncpy, expand_builtin_memset): Same.
2215 (expand_builtin_bzero, expand_builtin_memcmp): Same.
2216 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
2217 (maybe_emit_sprintf_chk_warning): Same.
2218 (expand_builtin_strcpy): Adjust.
2219 (expand_builtin_stpcpy): Same.
2220 (expand_builtin_with_bounds): Detect out-of-bounds accesses
2221 in pointer-checking forms of memcpy, memmove, and mempcpy.
2222 (gcall_to_tree_minimal, max_object_size): Define new functions.
2223 * builtins.h (max_object_size): Declare.
2224 * calls.c (alloc_max_size): Call max_object_size instead of
2225 hardcoding ssizetype limit.
2226 (get_size_range): Handle new argument.
2227 * calls.h (get_size_range): Add a new argument.
2228 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
2229 * doc/invoke.texi (-Wrestrict): Adjust, add example.
2230 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
2232 (gimple_fold_builtin_memory_chk): Same.
2233 (gimple_fold_builtin_stxcpy_chk): New function.
2234 * gimple-ssa-warn-restrict.c: New source.
2235 * gimple-ssa-warn-restrict.h: New header.
2236 * gimple.c (gimple_build_call_from_tree): Propagate location.
2237 * passes.def (pass_warn_restrict): Add new pass.
2238 * tree-pass.h (make_pass_warn_restrict): Declare.
2239 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
2241 (handle_builtin_strcat): Same.
2242 (strlen_optimize_stmt): Rename...
2243 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
2244 stpncpy, strncpy, and their checking forms.
2246 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
2248 PR rtl-optimization/82849
2249 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
2250 and get_max_loop_iterations_int.
2252 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2253 Alan Hayward <alan.hayward@arm.com>
2254 David Sherwood <david.sherwood@arm.com>
2256 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
2257 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
2258 size as a poly_uint64.
2259 (mode_for_vector, mode_for_int_vector): Take the number of vector
2260 elements as a poly_uint64.
2261 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
2262 size as a poly_uint64.
2263 (mode_for_vector, mode_for_int_vector): Take the number of vector
2264 elements as a poly_uint64.
2266 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2267 Alan Hayward <alan.hayward@arm.com>
2268 David Sherwood <david.sherwood@arm.com>
2270 * machmode.h (MACRO_MODE): New macro.
2271 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
2272 * targhooks.c (default_libcall_value, default_secondary_reload)
2273 (default_memory_move_cost, default_register_move_cost)
2274 (default_class_max_nregs): Likewise.
2276 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2277 Alan Hayward <alan.hayward@arm.com>
2278 David Sherwood <david.sherwood@arm.com>
2280 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
2281 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
2283 * genautomata.c (main): Likewise.
2284 * genconditions.c (write_header): Likewise.
2285 * genemit.c (main): Likewise.
2286 * genextract.c (print_header): Likewise.
2287 * genopinit.c (main): Likewise.
2288 * genoutput.c (output_prologue): Likewise.
2289 * genpeep.c (main): Likewise.
2290 * genpreds.c (write_insn_preds_c): Likewise.
2291 * genrecog.c (writer_header): Likewise.
2292 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
2293 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
2294 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
2295 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
2296 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
2297 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
2298 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
2299 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
2300 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
2301 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
2302 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
2303 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
2304 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
2305 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
2306 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
2307 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
2308 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
2309 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
2310 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
2311 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
2312 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
2313 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
2314 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
2315 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
2316 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
2317 * config/darwin.c (IN_TARGET_CODE): Likewise.
2318 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
2319 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
2320 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
2321 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
2322 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
2323 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
2324 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
2325 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
2326 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
2327 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
2328 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
2329 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
2330 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
2331 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
2332 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
2333 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
2334 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
2335 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
2336 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
2337 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
2338 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
2339 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
2340 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
2341 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
2342 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
2343 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
2344 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
2345 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
2346 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
2347 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
2348 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
2349 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
2350 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
2351 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
2352 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
2353 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
2354 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
2355 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
2356 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
2357 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
2358 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
2359 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
2360 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
2361 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
2362 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
2363 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
2364 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
2365 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
2366 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
2367 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
2368 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
2369 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
2370 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
2371 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
2372 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
2373 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
2374 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
2375 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
2376 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
2377 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
2378 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2379 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
2380 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
2381 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
2382 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
2383 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
2384 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
2385 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
2386 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
2387 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
2388 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
2389 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2390 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
2391 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
2392 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
2393 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
2394 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
2395 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
2396 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
2397 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
2398 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
2399 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
2400 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
2401 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
2402 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
2403 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
2404 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
2405 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
2406 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
2407 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
2408 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
2409 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
2410 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
2411 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
2412 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
2413 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
2414 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
2415 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
2416 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
2417 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
2418 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
2419 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
2420 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
2421 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
2422 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
2423 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
2425 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2426 Alan Hayward <alan.hayward@arm.com>
2427 David Sherwood <david.sherwood@arm.com>
2429 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
2432 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2433 Alan Hayward <alan.hayward@arm.com>
2434 David Sherwood <david.sherwood@arm.com>
2436 * doc/generic.texi (VEC_SERIES_EXPR): Document.
2437 * doc/md.texi (vec_series@var{m}): Document.
2438 * tree.def (VEC_SERIES_EXPR): New tree code.
2439 * tree.h (build_vec_series): Declare.
2440 * tree.c (build_vec_series): New function.
2441 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
2442 * tree-pretty-print.c (dump_generic_node): Likewise.
2443 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
2444 * tree-inline.c (estimate_operator_cost): Likewise.
2445 * expr.c (expand_expr_real_2): Likewise.
2446 * optabs-tree.c (optab_for_tree_code): Likewise.
2447 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
2448 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
2449 * expmed.c (make_tree): Handle VEC_SERIES.
2450 * optabs.def (vec_series_optab): New optab.
2451 * optabs.h (expand_vec_series_expr): Declare.
2452 * optabs.c (expand_vec_series_expr): New function.
2453 * tree-vect-generic.c (expand_vector_operations_1): Check that
2454 the operands also have vector type.
2456 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2457 Alan Hayward <alan.hawyard@arm.com>
2458 David Sherwood <david.sherwood@arm.com>
2460 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
2461 (VEC_COND_EXPR): Add missing @tindex.
2462 * doc/md.texi (vec_duplicate@var{m}): Document.
2463 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
2464 * tree.c (build_vector_from_val): Add stubbed-out handling of
2465 variable-length vectors, using VEC_DUPLICATE_EXPR.
2466 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
2467 * cfgexpand.c (expand_debug_expr): Likewise.
2468 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2469 * tree-inline.c (estimate_operator_cost): Likewise.
2470 * tree-pretty-print.c (dump_generic_node): Likewise.
2471 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
2472 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
2473 (test_vec_duplicate_folding): New function.
2474 (fold_const_c_tests): Call it.
2475 * optabs.def (vec_duplicate_optab): New optab.
2476 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
2477 * optabs.h (expand_vector_broadcast): Declare.
2478 * optabs.c (expand_vector_broadcast): Make non-static. Try using
2479 vec_duplicate_optab.
2480 * expr.c (store_constructor): Try using vec_duplicate_optab for
2482 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
2484 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
2487 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
2488 div/mod latencies a bit.
2490 2017-12-15 Jeff Law <law@redhat.com>
2492 PR tree-optimization/36550
2493 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
2494 (mark_threaded_blocks): Rewrite code to avoid block copying when
2495 optimizing for size. Don't pessimize blocks which will be
2496 copied, but all the statements will be dead.
2498 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
2500 PR tree-optimization/81165
2501 * tree-ssa-threadupdate.c (uses_in_bb): New.
2502 (estimate_threading_killed_stmts): New.
2503 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
2504 * tree-ssa-threadedge.c
2505 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
2508 2017-12-15 Jeff Law <law@redhat.com>
2510 PR tree-optimization/83410
2511 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
2512 threads when parallelizing loops.
2514 2017-12-15 Jakub Jelinek <jakub@redhat.com>
2516 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
2518 * config/alpha/alpha.c (vms_attribute_table): Swap
2519 affects_type_identity and handler fields, adjust comments.
2520 * config/mips/mips.c (mips_attribute_table): Likewise.
2521 * config/visium/visium.c (visium_attribute_table): Likewise.
2522 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
2523 * config/microblaze/microblaze.c (microblaze_attribute_table):
2525 * config/spu/spu.c (spu_attribute_table): Likewise.
2526 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
2527 * config/arc/arc.c (arc_attribute_table): Likewise.
2528 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
2529 * config/v850/v850.c (v850_handle_interrupt_attribute,
2530 v850_handle_data_area_attribute): Formatting fixes.
2531 (v850_attribute_table): Swap affects_type_identity and handler
2532 fields, adjust comments.
2533 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
2534 * config/arm/arm.c (arm_attribute_table): Likewise.
2535 * config/avr/avr.c (avr_attribute_table): Likewise.
2536 * config/s390/s390.c (s390_attribute_table): Likewise.
2537 * config/sh/sh.c (sh_attribute_table): Likewise.
2538 * config/i386/i386.c (ix86_handle_cconv_attribute,
2539 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
2540 (ix86_attribute_table): Swap affects_type_identity and handler
2541 fields, adjust comments.
2542 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2543 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
2544 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
2545 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
2546 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
2547 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2548 * config/rx/rx.c (rx_attribute_table): Likewise.
2549 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
2550 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
2551 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
2552 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
2553 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2554 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
2555 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
2556 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
2557 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2558 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2559 * doc/plugins.texi (user_attr): Likewise. Add NULL for
2561 * attribs.c (empty_attribute_table): Swap affects_type_identity and
2563 (register_scoped_attributes, decl_attributes): Formatting fixes.
2565 PR tree-optimization/83269
2566 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2567 subtraction in arg0's type if type is signed and arg0 is unsigned.
2571 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2572 convert? on @0 instead of convert. Check type of @1, not @0.
2573 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2574 convert? on @0 instead of convert. Check type of @1, not @0.
2575 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2576 only optimize if either both @1 and @2 types are narrower
2577 precision, or both are wider or equal precision, and in the former
2578 case only if both have undefined overflow.
2580 2017-12-15 Richard Biener <rguenther@suse.de>
2583 * internal-fn.def (IFN_NOP): Add.
2584 * internal-fn.c (expand_NOP): Do nothing.
2585 * lto-streamer-in.c (input_function): Instead of removing
2586 sanitizer calls replace them with IFN_NOP calls.
2588 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2589 Alan Hayward <alan.hayward@arm.com>
2590 David Sherwood <david.sherwood@arm.com>
2592 * dse.c (store_info, read_info_type): Replace begin and end with
2594 (print_range): New function.
2595 (set_all_positions_unneeded, any_positions_needed_p)
2596 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2598 (record_store): Likewise. Optimize the case in which all positions
2600 (get_stored_val): Replace read_begin and read_end with read_offset
2602 (replace_read): Update call accordingly.
2604 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2606 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2607 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2608 (loop_cand::loop_cand): Initialize above members.
2609 (loop_cand::supported_operations): Delete.
2610 (loop_cand::can_interchange_p): Inline above function.
2611 (loop_cand::classify_simple_reduction): Record number of constant
2612 initialized simple reductions.
2613 (should_interchange_loops): New parameters. Check stmt cost of loops
2615 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2616 Update call to should_interchange_loops.
2617 (should_interchange_loop_nest): Update call to
2618 should_interchange_loops.
2620 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2623 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2624 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2625 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2627 2017-12-15 Julia Koval <julia.koval@intel.com>
2629 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2630 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
2632 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2633 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2634 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2636 2017-12-15 Julia Koval <julia.koval@intel.com>
2638 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2639 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2640 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2641 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2642 _mm_aesenc_epi128): New intrinsics.
2644 2017-12-15 Julia Koval <julia.koval@intel.com>
2646 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2647 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
2649 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2650 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2651 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2653 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2655 * gimple-ssa-strength-reduction.c (analyze_increments):
2656 Distinguish replacement costs for constant strides from those for
2659 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2661 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2663 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2665 * doc/invoke.texi: Document -Wcast-function-type.
2666 * recog.h (stored_funcptr): Change signature.
2667 * tree-dump.c (dump_node): Avoid warning.
2668 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2670 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2673 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2676 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2678 PR tree-optimization/83312
2679 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2680 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2681 "val" param, and to cope with arbitrary basic blocks.
2682 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2683 handle NULL_TREE for "val", dropping "bb" param.
2684 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2685 drop "bb" param. Handle NULL_TREE for "val".
2686 (find_case_label_for_value): Make "switch_stmt" param const.
2687 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2689 (vrp_prop::check_all_array_refs): Reimplement as...
2690 (check_array_bounds_dom_walker::before_dom_children): ...this new
2691 vfunc. Replace linear search through BB block list, excluding
2692 those with non-executable in-edges via dominator walk.
2694 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2696 * config/arm/arm.opt (mverbose-cost-dump): New option.
2697 * config/arm/arm.c (arm_rtx_costs): Use it.
2699 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2702 * reload1.c (emit_input_reload_insns): Skip debug markers.
2704 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2706 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2710 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2711 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2712 * config/sh/sh.c: Skip debug insns besides notes.
2713 * config/sh/sh.md: Likewise.
2714 * config/sh/sh_treg_combine.cc: Likewise.
2715 * config/sh/sync.md: Likewise.
2717 2017-12-14 Tom de Vries <tom@codesourcery.com>
2719 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2720 Add item for weak_undefined.
2722 2017-12-14 Richard Biener <rguenther@suse.de>
2724 PR tree-optimization/67842
2725 * sese.h (bb_in_region): Remove #if 0'ed code.
2727 2017-12-14 Richard Biener <rguenther@suse.de>
2729 PR tree-optimization/83326
2730 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2731 may_be_zero parameter and handle it by not marking the first
2732 peeled copy as not exiting the loop.
2733 (try_peel_loop): Likewise.
2734 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2735 to handle the case of constant or zero iterations and perform
2736 loop header copying on-the-fly.
2738 2017-12-14 Richard Biener <rguenther@suse.de>
2740 PR tree-optimization/83418
2741 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2742 Instead of asserting we don't get unfolded comparisons deal with them.
2744 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2747 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2748 basic blocks. Assert debug bind insns don't appear outside of bbs,
2749 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2752 PR tree-optimization/83198
2753 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2754 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2755 value if arg is a REAL_CST with incompatible type.
2757 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2758 Bin Cheng <bin.cheng@arm.com>
2761 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2763 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2766 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2768 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2769 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2771 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2772 Alan Hayward <alan.hayward@arm.com>
2773 David Sherwood <david.sherwood@arm.com>
2775 * poly-int.h: New file.
2776 * poly-int-types.h: Likewise.
2777 * coretypes.h: Include them.
2778 (POLY_INT_CONVERSION): Define.
2779 * target.def (estimated_poly_value): New hook.
2780 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2781 * doc/tm.texi: Regenerate.
2782 * doc/poly-int.texi: New file.
2783 * doc/gccint.texi: Include it.
2784 * doc/rtl.texi: Describe restrictions on subreg modes.
2785 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2786 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2787 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2788 * targhooks.h (default_estimated_poly_value): Declare.
2789 * targhooks.c (default_estimated_poly_value): New function.
2790 * target.h (estimated_poly_value): Likewise.
2791 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2792 (wi::unary_traits): Delete.
2793 (wi::binary_traits::signed_shift_result_type): Define for
2794 offset_int << HOST_WIDE_INT, etc.
2795 (generic_wide_int::operator <<=): Define for all types and use
2796 wi::lshift instead of <<.
2797 (wi::hwi_with_prec): Add a default constructor.
2798 (wi::ints_for): New class.
2799 (operator <<): Define for all wide-int types.
2800 (operator /): New function.
2801 (operator %): Likewise.
2802 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2803 (ASSERT_MAYBE_NE_AT): New macros.
2805 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2806 Dominik Vogt <vogt@linux.vnet.ibm.com>
2809 * emit-rtl.c (init_emit): Remove ??? comment.
2810 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2811 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
2812 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
2813 alignment of 3 virtual registers to BITS_PER_WORD.
2815 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
2817 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
2819 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
2820 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
2822 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2825 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
2827 2017-12-13 Jakub Jelinek <jakub@redhat.com>
2829 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
2830 with the exception of debug begin stmt markers appear before
2834 * final.c (rest_of_handle_final): Call variable_tracking_main only
2835 if !flag_var_tracking.
2837 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2838 Jakub Jelinek <jakub@redhat.com>
2842 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
2843 labels debug stmts that can only appear after labels.
2845 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
2847 PR rtl-optimization/82398
2848 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
2849 EXPR_USEFULNESS in priority comparison.
2851 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
2853 PR rtl-optimization/83393
2854 * combine.c (move_deaths): If reg_stat points to a too new insn in
2855 last_death, do not use it: find the proper insn instead.
2857 2017-12-12 Jeff Law <law@redhat.com>
2859 PR tree-optimization/83298
2860 PR tree-optimization/83362
2861 PR tree-optimization/83383
2862 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
2863 push_value_range a public interface. Add new argument to
2864 record_ranges_from_stmt.
2865 * gimple-ssa-evrp-analyze.c
2866 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
2867 Update comments. Handle recording temporary equivalences.
2868 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
2869 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
2870 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
2871 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
2872 gimple-ssa-evrp-analyze.h.
2873 (record_temporary_equivalences_from_phis): Add new argument. When
2874 the PHI arg is an SSA_NAME, set the result's range to the range
2876 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
2877 from statements too.
2878 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
2879 Pass it down to children as needed.
2880 (thread_outgoing_edges): Likewise.
2881 (thread_across_edge): Likewise. Push/pop range state as needed.
2882 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
2884 2017-12-12 Julia Koval <julia.koval@intel.com>
2886 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
2887 (PTA_CANNONLAKE): Remove PTA_CLWB.
2889 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2892 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
2893 (vcondv2div2df): Likewise.
2894 (vconduv2dfv2di): Likewise.
2896 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2898 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
2899 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
2900 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
2901 ATTR_MATHFN_FPROUNDING_ERRNO.
2903 2017-12-12 Richard Biener <rguenther@suse.de>
2905 PR tree-optimization/83385
2906 * graphite-scop-detection.c (get_order, order): Remove.
2907 (bb_to_rpo): New global.
2909 (build_scops): Sort pbbs in RPO order.
2911 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
2913 * combine.c (simplify_set): Do not transform subregs to zero_extends
2914 if the destination mode is a vector mode.
2916 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2918 PR tree-optimization/83359
2919 * tree-cfg.h (fold_loop_internal_call): Declare.
2920 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
2921 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
2922 (find_loop_dist_alias): New function.
2923 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
2924 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
2927 PR tree-optimization/80631
2928 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
2929 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
2930 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
2931 hardcoding zero as the value if COND_EXPR is never true. For
2932 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
2933 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
2934 hardcoding MAX_EXPR as the reduction operation.
2935 (is_nonwrapping_integer_induction): Allow negative step.
2936 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
2937 vect_create_epilog_for_reduction, if no value is suitable, don't
2938 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
2940 2017-12-12 Richard Biener <rguenther@suse.de>
2942 PR tree-optimization/81889
2943 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
2944 range info from the non-wrapping IV instead of just the range
2947 2017-12-12 Julia Koval <julia.koval@intel.com>
2949 * config.gcc: Add vaesintrin.h.
2950 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2952 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2953 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2955 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2956 * config/i386/immintrin.h: Include vaesintrin.h.
2957 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2958 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2959 _mm_aesdec_epi128): New intrinsics.
2961 2017-12-12 Julia Koval <julia.koval@intel.com>
2963 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
2964 OPTION_MASK_ISA_VAES_UNSET): New.
2965 (ix86_handle_option): Handle -mvaes.
2966 * config/i386/cpuid.h: Define bit_VAES.
2967 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
2968 * config/i386/i386-c.c (__VAES__): New.
2969 * config/i386/i386.c (ix86_target_string): Add -mvaes.
2970 (ix86_valid_target_attribute_inner_p): Ditto.
2971 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
2972 * config/i386/i386.opt: Add -mvaes.
2973 * doc/invoke.texi: Ditto.
2975 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
2977 * debug.h (gcc_debug_hooks): Add inline_entry.
2978 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
2979 * debug.c (do_nothing_debug_hooks): Likewise.
2980 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2981 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
2982 (dwarf2_lineno_debug_hooks): Likewise.
2984 * common.opt (gstatement-frontiers): New, setting
2985 debug_nonbind_markers_p.
2986 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
2987 * toplev.c (process_options): Autodetect value for debug statement
2989 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
2990 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
2992 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
2993 markers. Integrate source bind into debug stmt expand loop.
2994 (pass_expand::execute): Check debug marker limit. Avoid deep
2995 TER and expand debug locations for debug bind insns only.
2996 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
2998 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
2999 * final.c (reemit_insn_block_notes): Take current block from
3000 nonbind markers. Declare note where it's first set.
3001 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3002 begin stmt markers if enabled.
3003 (notice_source_line): Handle nonbind markers. Fail if their
3004 location is unknown or that of builtins.
3005 (rest_of_handle_final): Convert begin stmt markers to notes if
3006 var-tracking didn't run.
3007 (rest_of_clean_state): Skip begin stmt markers.
3008 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3010 * function.c (allocate_struct_function): Set begin_stmt_markers.
3011 * function.h (struct function): Add debug_marker_count counter
3012 and debug_nonbind_markers flag.
3013 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3014 * gimple-low.c (lower_function_body): Adjust
3015 debug_nonbind_markers.
3016 (lower_stmt): Drop or skip gimple debug stmts.
3017 (lower_try_catch): Skip debug stmts.
3018 * gimple.c (gimple_build_debug_begin_stmt): New.
3019 (gimple_copy): Increment debug_marker_count if copying one.
3020 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3021 * gimplify.c (rexpr_location): New.
3022 (rexpr_has_location): New.
3023 (warn_switch_unreachable_r): Handle gimple debug stmts.
3024 (shortcut_cond_r): Call expr_location.
3026 (find_goto_label): New.
3027 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3029 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3031 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3032 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3033 (LANG_HOOKS_INITIALIZER): ... this.
3034 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3035 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3037 (update_ebb_live_info): Skip debug insn markers.
3038 * lra.c (debug_insn_static_data): Rename to...
3039 (debug_bind_static_data): ... this.
3040 (debug_marker_static_data): New.
3041 (lra_set_insn_recog_data): Select one of the above depending
3043 (lra_update_isn_regno_info): Don't assume debug insns have
3045 (push_insns): Skip debug insns.
3046 * lto-streamer-in.c (input_function): Drop debug stmts
3047 depending on active options. Adjust debug_nonbind_markers.
3048 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3049 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3050 begin stmt marker notes.
3051 (print_insn): Likewise.
3052 * recog.c (extract_insn): Recognize rtl for debug markers.
3053 * rtl.def (DEBUG_MARKER): New.
3054 * tree-inline.c: Include params.h.
3055 (remap_gimple_stmt): Handle nonbind markers.
3056 (maybe_move_debug_stmts_to_successors): Likewise.
3057 (copy_debug_stmt): Likewise.
3058 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
3059 markers regardless of no side effects.
3060 (tsi_link_before): Don't update container's side effects when adding
3061 a begin stmt marker.
3062 (tsi_link_after): Likewise.
3063 (expr_first): Skip begin stmt markers.
3064 (expr_last): Likewise.
3065 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
3066 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3067 Disregard nonbind markers.
3068 * tree.c (make_node_stat): Don't set side effects for begin stmt
3070 (build1_stat): Likewise.
3071 * tree.def (DEBUG_BEGIN_STMT): New.
3072 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
3073 * var-tracking.c (delete_debug_insns): Renamed to...
3074 (delete_vta_debug_insns): ... this.
3075 (reemit_marker_as_note): New.
3076 (vt_initialize): Reemit markers.
3077 (delete_vta_debug_insns): Likewise.
3078 (vt_debug_insns_local): Reemit or delete markers.
3079 (variable_tracking_main_1): Likewise.
3080 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
3081 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
3082 (gimple_debug_nonbind_marker_p): New.
3083 (gimple_build_debug_bind): Adjust.
3084 (gimple_build_debug_begin_stmt): New.
3085 * doc/invoke.texi (max-debug-marker-count): New param.
3086 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
3087 (debug_parameter_ref, debug_marker): New.
3088 (NOTE_INSN_BEGIN_STMT): New.
3089 (DEBUG_INSN): Describe begin stmt markers.
3091 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
3092 without debug insns, we wouldn't, but clean up debug insns
3093 after a control flow insn nevertheless.
3095 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
3096 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
3097 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
3098 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
3099 (rtl_tidy_fallthru_edge): Likewise.
3100 (rtl_verify_fallthru): Likewise.
3101 (rtl_verify_bb_layout): Likewise.
3102 (skip_insns_after_block): Likewise.
3103 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
3104 * dwarf2out.c: Include print-rtl.h.
3105 (dwarf2out_next_real_insn): New.
3106 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
3107 Dump debug binds in asm comments.
3108 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
3109 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
3110 callers to use gsi_start_nondebug_bb instead.
3111 (gsi_after_labels): Skip gimple debug stmts.
3112 (gsi_start_nondebug): New.
3113 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
3114 (proper_loop_form_for_interchange): Adjust.
3115 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
3116 * gimple.h (gimple_seq_last_nondebug_stmt): New.
3117 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
3118 (collect_fallthrough_labels): Likewise.
3119 (should_warn_for_implicit_fallthrough): Likewise.
3120 (warn_implicit_fallthrough_r): Likewise.
3121 (expand_FALLTHROUGH_r): Likewise.
3122 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
3123 (graphite_copy_stmts_from_block): Skip nonbind markers.
3124 * haifa-sched.c (sched_extend_bb): Skip debug insns.
3125 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
3126 * jump.c (clean_barriers): Skip debug insns.
3127 * omp-expand.c (expand_parallel_call): Skip debug insns.
3128 (expand_task_call): Likewise.
3129 (remove_exit_barrier): Likewise.
3130 (expand_omp_taskreg): Likewise.
3131 (expand_omp_for_init_counts): Likewise.
3132 (expand_omp_for_generic): Likewise.
3133 (expand_omp_for_static_nochunk): Likewise.
3134 (expand_omp_for_static_chunk): Likewise.
3135 (expand_omp_simd): Likewise.
3136 (expand_omp_taskloop_for_outer): Likewise.
3137 (expand_omp_taskloop_for_inner): Likewise.
3138 (expand_oacc_for): Likewise.
3139 (expand_omp_sections): Likewise.
3140 (expand_omp_single): Likewise.
3141 (expand_omp_synch): Likewise.
3142 (expand_omp_atomic_load): Likewise.
3143 (expand_omp_atomic_store): Likewise.
3144 (expand_omp_atomic_fetch_op): Likewise.
3145 (expand_omp_atomic_pipeline): Likewise.
3146 (expand_omp_atomic_mutex): Likewise.
3147 (expand_omp_target): Likewise.
3148 (grid_expand_omp_for_loop): Likewise.
3149 (grid_expand_target_grid_body): Likewise.
3150 (build_omp_regions_1): Likewise.
3151 * omp-low.c (check_combined_parallel): Skip debug stmts.
3152 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
3153 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
3154 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
3156 * tree-cfg.c (make_blobs_1): Skip debug stmts.
3157 (make_edges): Likewise.
3158 (cleanup_dead_labels): Likewise.
3159 (gimple_can_merge_blocks_p): Likewise.
3160 (stmt_starts_bb_p): Likewise.
3161 (gimple_block_label): Likewise.
3162 (gimple_redirect_edge_and_branch): Likewise.
3163 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
3165 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
3167 * tree-pretty-print (print_declaration): Omit initializer in slim
3169 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
3171 (eliminate_unnecessary_stmts): Stabilize block removal order.
3172 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
3173 * var-tracking.c (get_first_insn): New.
3174 (vt_emit_notes): Call it.
3175 (vt_initialize): Walk any insns before the first BB.
3176 (delete_debug_insns): Likewise.
3178 * gimple.h (enum gimple_debug_subcode): Add
3179 GIMPLE_DEBUG_BEGIN_STMT.
3180 (gimple_debug_begin_stmt_p): New.
3181 (gimple_debug_nonbind_marker_p): New.
3182 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
3183 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
3184 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
3185 * insn-notes.def (BEGIN_STMT): New.
3186 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
3187 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
3188 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
3189 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
3190 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
3191 (INSN_DEBUG_MARKER_KIND): New.
3192 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
3193 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
3194 (INSN_VAR_LOCATION_PTR): New.
3195 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
3197 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
3198 deps only if debug bind insns are enabled.
3199 (pass_expand::execute): Avoid deep TER and expand
3200 debug locations for debug bind insns only.
3201 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
3202 debug stmts special handling down to debug bind stmts.
3203 * combine.c (try_combine): Narrow debug insns special handling
3204 down to debug bind insns.
3205 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
3206 Narrow debug insns preexisting special handling down to debug
3208 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
3209 handling down to debug bind insns.
3210 * function.c (instantiate_virtual_regs): Skip debug markers,
3211 adjust handling of debug binds.
3212 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
3213 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
3214 * haifa-sched.c (schedule_insn): Narrow special handling of debug
3215 insns to debug bind insns.
3216 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
3217 special handling of debug stmts to debug bind stmts.
3218 * ipa-split.c (split_function): Likewise.
3219 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
3220 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
3222 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
3223 * regrename.c (build_def_use): Likewise.
3224 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
3225 (pass_cprop_hardreg): Narrow special casing of debug insns to
3227 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
3228 * reload1.c (reload): Likewise.
3229 * sese.c (sese_insert_phis_for_liveouts): Narrow special
3230 casing of debug stmts to debug bind stmts.
3231 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
3232 * ssa-iterators.h (num_imm_uses): Likewise.
3233 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
3234 debug stmts to debug bind stmts.
3235 * tree-inline.c (tree_function_versioning): Narrow special casing
3236 of debug stmts to debug bind stmts.
3237 * tree-loop-distribution.c (generate_loops_for_partition):
3238 Narrow special casing of debug stmts to debug bind stmts.
3239 * tree-sra.c (analyze_access_subtree): Narrow special casing
3240 of debug stmts to debug bind stmts.
3241 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
3242 stmts to debug bind stmts.
3243 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
3244 casing of debug stmts to debug bind stmts.
3245 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
3246 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
3247 casing of debug stmts to debug bind stmts.
3248 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3250 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
3251 debug stmts to debug bind stmts.
3252 (gimple_replace_ssa_lhs): Likewise.
3253 (insert_debug_temp_for_var_def): Likewise.
3254 (insert_debug_temps_for_defs): Likewise.
3255 (reset_debug_uses): Likewise.
3256 * tree-ssanames.c (release_ssa_name_fn): Likewise.
3257 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
3258 (adjust_debug_stmts): Likewise.
3259 (adjust_phi_and_debug_stmts): Likewise.
3260 (vect_do_peeling): Likewise.
3261 * tree-vect-loop.c (vect_transform_loop): Likewise.
3262 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
3263 * var-tracking.c (adjust_mems): Narrow special casing of debug
3264 insns to debug bind insns.
3265 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
3266 (compute_bb_dataflow, vt_find_locations): Likewise.
3267 (vt_expand_loc, emit_notes_for_changes): Likewise.
3268 (vt_init_cfa_base): Likewise.
3269 (vt_emit_notes): Likewise.
3270 (vt_initialize): Likewise.
3271 (vt_finalize): Likewise.
3273 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
3274 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
3275 (next_nonnote_nondebug_insn_bb): New.
3276 (prev_nonnote_nondebug_insn_bb): New.
3277 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
3278 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
3279 (prev_nonnote_nondebug_insn_bb): Declare.
3280 (next_nonnote_nondebug_insn_bb): Declare.
3281 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
3282 * cfgrtl.c (get_last_bb_insn): Likewise.
3283 * lra.c (push_insns): Likewise.
3285 2017-12-11 David Malcolm <dmalcolm@redhat.com>
3288 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
3290 * selftest.c (selftest::test_runner::test_runner): New ctor.
3291 (selftest::test_runner::~test_runner): New dtor.
3292 * selftest.h (class selftest::test_runner): New class.
3294 2017-12-11 Carl Love <cel@us.ibm.com>
3296 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
3297 vec_extract_fp32_from_shortl]): Add #defines.
3298 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
3299 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
3300 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
3301 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
3302 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
3303 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
3304 * doc/extend.texi: Add documentation for the added builtins.
3306 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
3308 PR rtl-optimization/80693
3309 PR rtl-optimization/81019
3310 PR rtl-optimization/81020
3311 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
3312 are not mentioned in i3. Place the REG_UNUSED note on i2,
3313 possibly modified to REG_DEAD, if it did not originate in i3.
3315 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3317 * recog.c (store_data_bypass_p_1): New function.
3318 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
3319 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
3321 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
3323 PR rtl-optimization/83361
3324 * ifcvt.c (if_convert): Call fixup_partitions.
3326 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
3328 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3329 early folding of splat_u{8,16,32}.
3331 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3333 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
3334 output_operand_lossage first argument with capital letter.
3335 (aarch64_override_options): Don't start error and sorry first argument
3336 with capital letter.
3338 2017-12-11 Andi Kleen <ak@linux.intel.com>
3340 PR gcov-profile/83355
3341 * auto-profile.c (string_table::get_index_by_decl): Don't
3342 recurse when abstract origin points to itself.
3344 2017-12-11 Bin Cheng <bin.cheng@arm.com>
3346 PR tree-optimization/83320
3347 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
3348 (prune_datarefs_not_in_loop): Ditto.
3350 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
3352 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
3354 2017-12-10 Jakub Jelinek <jakub@redhat.com>
3356 PR tree-optimization/83337
3357 * gimple-loop-interchange.cc (compute_access_stride): Handle
3358 bitfield DRs properly.
3360 2017-12-09 Jakub Jelinek <jakub@redhat.com>
3362 PR tree-optimization/83338
3363 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
3364 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
3365 vector of unsigned integers to vector of signed integers.
3367 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
3369 PR rtl-optimization/83317
3370 * lra-constraints.c (process_address_1): Add insn code check.
3372 2017-12-08 Michael Matz <matz@suse.de>
3374 Fix PR tree-optimization/83323
3375 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
3376 head-controlled loops and loop BBs.
3377 * common.opt (funroll-and-jam): Remove, instead ...
3378 (floop-unroll-and-jam): ... reuse this option.
3379 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
3380 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
3381 (-floop-unroll-and-jam): ... this option.
3383 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3385 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
3386 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
3387 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
3389 2017-12-08 Richard Biener <rguenther@suse.de>
3392 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
3393 handle accesses outside of zero-sized vars.
3395 2017-12-08 Martin Jambor <mjambor@suse.cz>
3397 PR tree-optimization/83141
3398 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
3399 test for MEM_REFs implicitely changing types with padding. Remove
3401 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
3403 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3405 * config/arc/arc.c (arc_attribute_table): Add exclusions to
3407 * config/avr/avr.c (avr_attribute_table): Likewise.
3408 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3409 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3410 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3411 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
3412 of struct attribute_spec.
3413 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3415 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
3418 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
3420 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
3423 * profile-count.c (profile_count::from_gcov_type): Move from
3424 profile-count.h; handle overflow.
3425 * profile-count.h (profile_count::from_gcov_type): Move offline.
3427 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
3429 PR rtl-optimization/83304
3430 * combine.c (move_deaths): If we do not know where a register died,
3433 2017-12-08 Richard Biener <rguenther@suse.de>
3435 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
3436 Provide -fopt-info-loop feedback when we interchange in a nest.
3438 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3440 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
3441 for armv6 ARM CPU IDs.
3443 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3445 * common/config/arm/arm-common.c: Include <algorithm>.
3446 (INCLUDE_VECTOR): Define.
3447 (compare_opt_names): New function.
3448 (arm_rewrite_selected_arch): Only strip out extensions that can be
3449 expressed through -mfpu. Sort the remaining extensions
3452 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3454 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
3455 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
3456 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
3457 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
3459 2017-06-08 Tristan Gingold <gindold@adacore.com>
3462 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
3463 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
3465 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3467 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
3468 of struct attribute_spec.
3470 2017-12-08 Julia Koval <julia.koval@intel.com>
3472 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
3473 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
3474 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
3475 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
3476 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
3477 _mm_maskz_dpwssds_epi32): Ditto.
3479 2017-12-08 Richard Biener <rguenther@suse.de>
3481 PR tree-optimization/81303
3482 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
3483 conditions try to create a comparison vector type matching
3484 the data vector type.
3485 (vectorizable_condition): Adjust.
3486 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3487 Leave invariant conditions alone in case we can vectorize those.
3489 2017-12-08 Julia Koval <julia.koval@intel.com>
3491 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
3492 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
3493 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
3494 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
3495 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
3497 2017-12-08 Julia Koval <julia.koval@intel.com>
3499 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
3500 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
3501 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
3502 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
3503 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
3504 _mm_maskz_dpbusds_epi32): New intrinsics.
3506 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3508 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
3510 (ld<bh>io_signed): Likewise.
3511 (st<bhw>io): Likewise.
3512 * config/nios2/predicates.md (ldstio_memory_operand): Allow
3513 SMALL_INT12 constant integer operand.
3515 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3517 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
3519 (nios2_large_constant_allowed): ...this. Adjust uses.
3520 (nios2_plus_symbolic_constant_p): Rename to...
3521 (nios2_plus_large_constant_p): ...this. Adjust uses.
3522 (nios2_legitimate_address_p): Correct CONST_INT handling.
3523 (nios2_symbolic_memory_operand_p): Rename to...
3524 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
3525 (nios2_large_constant_p): Check for large constant integers too.
3526 (nios2_split_large_constant): Handle constant integers.
3527 (nios2_split_symbolic_memory_operand): Rename to...
3528 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
3529 (nios2_legitimize_constant_address): Handle constant integers.
3530 (r0rel_constant_p): Handle small constant integers.
3531 (nios2_print_operand_address): Handle r0-relative integer addresses.
3532 * config/nios2/nios2-protos.h: Adjust for renamed functions.
3533 * config/nios2/nios2.md: Adjust for renamed functions.
3535 2017-12-07 Andrew Waterman <andrew@sifive.com>
3537 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
3538 (TARGET_HAVE_SRODATA_SECTION): New define.
3539 (riscv_select_section): New function.
3541 2017-12-08 Joseph Myers <joseph@codesourcery.com>
3542 Alexander Monakov <amonakov@ispras.ru>
3543 Jakub Jelinek <jakub@redhat.com>
3546 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
3548 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3550 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
3551 a bool success value. Don't call output_operand_lossage here.
3552 (aarch64_print_ldpstp_address): Return a bool success value.
3553 (aarch64_print_operand_address): Call output_addr_const if
3554 aarch64_print_address_internal fails.
3555 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
3556 'y'; call output_operand_lossage instead. Call output_operand_lossage
3557 if aarch64_print_ldpstp_address fails.
3559 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3561 * tree-vector-builder.h
3562 (tree_vector_builder::binary_encoded_nelts): Declare.
3563 * tree-vector-builder.c
3564 (tree_vector_builder::binary_encoded_nelts): New function.
3565 * fold-const.c (negate_expr_p): Likewise.
3566 (operand_equal_p, fold_checksum_tree): Likewise.
3567 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3568 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3569 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3570 (uniform_vector_p): Likewise.
3571 * varasm.c (const_hash_1, compare_constant): Likewise.
3572 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3573 (valid_lattice_transition): Operate directly on the VECTOR_CST
3575 * ipa-icf.c: Include tree-vector-builder.h.
3576 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3577 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3579 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3581 * tree.c (build_vector): Delete.
3582 * tree.h (build_vector): Make static and move into the self-testing
3585 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3587 * vector-builder.h (vector_builder::derived): New const overload.
3588 (vector_builder::elt): New function.
3589 * tree-vector-builder.h (tree_vector_builder::type): New function.
3590 (tree_vector_builder::apply_step): Declare.
3591 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3593 * gimple-fold.h (tree_vector_builder): Declare.
3594 (gimple_build_vector): Take a tree_vector_builder instead of a
3595 type and vector of elements.
3596 * gimple-fold.c (gimple_build_vector): Likewise.
3597 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3599 (get_initial_defs_for_reduction): Likewise.
3600 (vectorizable_induction): Likewise.
3602 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3604 * tree-vector-builder.h
3605 (tree_vector_builder::new_binary_operation): Declare.
3606 * tree-vector-builder.c
3607 (tree_vector_builder::new_binary_operation): New function.
3608 * fold-const.c (fold_relational_const): Use it.
3609 (const_binop): Likewise. Check that both input vectors have
3610 the same number of elements, thus excluding things like WIDEN_SUM.
3611 Check whether it is possible to operate directly on the encodings
3614 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3616 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3617 new_unary_operation, operating only on the encoded elements.
3618 (const_unop): Likewise.
3619 (exact_inverse): Likewise.
3620 (distributes_over_addition_p): New function.
3621 (const_binop): Use tree_vector_builder and new_unary_operation
3622 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3623 on the encoded elements unless the encoding is strided and the
3624 operation does not distribute over addition.
3625 (fold_convert_const): Use tree_vector_builder and
3626 new_unary_operation. Operate only on the encoded elements
3627 for truncating integer conversions, or for non-stepped encodings.
3629 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3631 * config/sparc/sparc.c: Include tree-vector-builder.h.
3632 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3633 * expmed.c: Include tree-vector-builder.h.
3634 (make_tree): Use tree_vector_builder instead of build_vector.
3635 * fold-const.c: Include tree-vector-builder.h.
3636 (const_binop): Use tree_vector_builder instead of build_vector.
3637 (const_unop): Likewise.
3638 (native_interpret_vector): Likewise.
3639 (fold_vec_perm): Likewise.
3640 (fold_ternary_loc): Likewise.
3641 * gimple-fold.c: Include tree-vector-builder.h.
3642 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3644 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3645 (simplify_vector_constructor): Use tree_vector_builder instead
3647 * tree-vect-generic.c: Include tree-vector-builder.h.
3648 (add_rshift): Use tree_vector_builder instead of build_vector.
3649 (expand_vector_divmod): Likewise.
3650 (optimize_vector_constructor): Likewise.
3651 * tree-vect-loop.c: Include tree-vector-builder.h.
3652 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3653 of build_vector. Explicitly use a stepped encoding for
3655 * tree-vect-slp.c: Include tree-vector-builder.h.
3656 (vect_get_constant_vectors): Use tree_vector_builder instead
3658 (vect_transform_slp_perm_load): Likewise.
3659 (vect_schedule_slp_instance): Likewise.
3660 * tree-vect-stmts.c: Include tree-vector-builder.h.
3661 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3662 (vect_gen_perm_mask_any): Likewise.
3663 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3664 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3666 (build_vector_from_val): Likewise. Explicitly use a duplicate
3669 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3671 * doc/generic.texi (VECTOR_CST): Describe new representation of
3673 * vector-builder.h: New file.
3674 * tree-vector-builder.h: Likewise.
3675 * tree-vector-builder.c: Likewise.
3676 * Makefile.in (OBJS): Add tree-vector-builder.o.
3677 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3678 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3679 (tree_vector): Change the number of elements to
3680 vector_cst_encoded_nelts.
3681 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3682 (VECTOR_CST_ELTS): Delete.
3683 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3684 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3685 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3686 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3687 (VECTOR_CST_ENCODED_ELT): Likewise.
3688 (vector_cst_encoded_nelts): New function.
3689 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3690 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3691 (vector_cst_int_elt, vector_cst_elt): Declare.
3692 * tree.c: Include tree-vector-builder.h.
3693 (tree_code_size): Abort if passed VECTOR_CST.
3694 (tree_size): Update for new VECTOR_CST layout.
3695 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3696 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3697 (build_vector): Use tree_vector_builder.
3698 (vector_cst_int_elt, vector_cst_elt): New functions.
3699 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3700 encoded elements and then create the vector in the canonical form.
3701 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3702 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3703 (tree_c_tests): Call test_vector_cst_patterns.
3704 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3706 (hash_tree): Likewise.
3707 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3708 (streamer_write_tree_header): Likewise.
3709 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3710 (streamer_alloc_tree): Likewise. Update call to make_vector.
3711 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3713 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3715 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3716 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3717 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3719 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3720 Richard Biener <rguenther@suse.de>
3722 PR tree-optimization/81303
3723 * Makefile.in (gimple-loop-interchange.o): New object file.
3724 * common.opt (floop-interchange): Reuse the option from graphite.
3725 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3726 -floop-interchange and mention it for -O3.
3727 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3728 * gimple-loop-interchange.cc: New file.
3729 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3730 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3731 * passes.def (pass_linterchange): New pass.
3732 * timevar.def (TV_LINTERCHANGE): New time var.
3733 * tree-pass.h (make_pass_linterchange): New declaration.
3734 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3735 interchange. Record IV before/after increment in new parameters.
3736 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3737 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3739 (check_reduction_path): ...New function here.
3740 * tree-vectorizer.h (check_reduction_path): New declaration.
3742 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3745 PR rtl-optimization/80818
3746 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3747 always early clobbered.
3748 * lra-lives.c (process_bb_lives): Check input hard regs for early
3749 clobbered non-operand hard reg.
3751 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3754 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3755 types_compatible_p, just that TYPE_MODE is the same.
3757 2017-12-07 Martin Sebor <msebor@redhat.com>
3760 * attribs.c (empty_attribute_table): Initialize new member of
3761 struct attribute_spec.
3762 (decl_attributes): Add argument. Handle mutually exclusive
3763 combinations of attributes.
3764 (selftests::test_attribute_exclusions): New function.
3765 (selftests::attribute_c_tests): Ditto.
3766 * attribs.h (decl_attributes): Add default argument.
3767 * selftest.h (attribute_c_tests): Declare.
3768 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3769 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3771 * doc/extend.texi (Common Function Attributes): Update const and pure.
3772 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3773 of struct attribute_spec.
3774 * config/arc/arc.c (arc_attribute_table): Same.
3775 * config/arm/arm.c (arm_attribute_table): Same.
3776 * config/avr/avr.c ( avr_attribute_table): Same.
3777 * config/bfin/bfin.c (bfin_attribute_table): Same.
3778 * config/cr16/cr16.c (cr16_attribute_table): Same.
3779 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3780 * config/h8300/h8300.c (h8300_attribute_table): Same.
3781 * config/i386/i386.c (ix86_attribute_table): Same.
3782 * config/ia64/ia64.c (ia64_attribute_table): Same.
3783 * config/m32c/m32c.c (m32c_attribute_table): Same.
3784 * config/m32r/m32r.c (m32r_attribute_table): Same.
3785 * config/m68k/m68k.c (m68k_attribute_table): Same.
3786 * config/mcore/mcore.c (mcore_attribute_table): Same.
3787 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3788 * config/mips/mips.c (mips_attribute_table): Same.
3789 * config/msp430/msp430.c (msp430_attribute_table): Same.
3790 * config/nds32/nds32.c (nds32_attribute_table): Same.
3791 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3792 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3793 * config/rl78/rl78.c (rl78__attribute_table): Same.
3794 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3795 * onfig/rx/rx.c (rx_attribute_table): Same.
3796 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3797 * config/sh/sh.c (sh_attribute_table): Same.
3798 * config/sparc/sparc.c (sparc_attribute_table): Same.
3799 * config/spu/spu.c (spu_attribute_table): Same.
3800 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3801 * config/v850/v850.c (v850_attribute_table): Same.
3802 * config/visium/visium.c (visium_attribute_table): Same.
3804 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3807 * config/arm/arm.c (INCLUDE_STRING): Define.
3808 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3809 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
3812 2017-12-07 Michael Matz <matz@suse.de>
3814 Add unroll and jam pass
3816 * gimple-loop-jam.c: New file.
3817 * Makefile.in (OBJS): Add gimple-loop-jam.o.
3818 * common.opt (funroll-and-jam): New option.
3819 * opts.c (default_options_table): Add unroll-and-jam at -O3.
3820 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
3821 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
3822 * passes.def: Add pass_loop_jam.
3823 * timevar.def (TV_LOOP_JAM): Add.
3824 * tree-pass.h (make_pass_loop_jam): Declare.
3825 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
3826 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
3827 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
3828 to flow_loop_tree_node_add.
3829 (duplicate_subloops, copy_loops_to): Append to sibling list.
3830 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
3831 * doc/invoke.texi (-funroll-and-jam): Document new option.
3832 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
3834 2017-12-07 Richard Biener <rguenther@suse.de>
3836 PR tree-optimization/83296
3837 PR tree-optimization/67769
3838 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
3839 flow sensitive info in an unrelated BB.
3840 (value_replacement): Use reset_flow_sensitive_info.
3841 (minmax_replacement): Reset flow sensitive info on the def
3842 we move. Do not reset flow sensitive info in the whole BB
3843 we move the stmt to.
3844 (abs_replacement): Likewise.
3846 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3849 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3850 rs6000_cpu to the given -mcpu=, or to the default processor.
3852 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3854 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
3855 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
3856 instead of rs6000_cpu_attr.
3857 (rs6000_adjust_cost): Ditto.
3858 (is_microcoded_insn): Ditto.
3859 (rs6000_adjust_priority): Ditto.
3860 (rs6000_issue_rate): Ditto.
3861 (rs6000_use_sched_lookahead): Ditto.
3862 (rs6000_use_sched_lookahead_guard): Ditto.
3863 (rs6000_sched_reorder): Ditto.
3864 (force_new_group): Ditto.
3865 * config/rs6000/rs6000.md (cpu attribute): Ditto.
3866 (group_ending_nop): Ditto.
3868 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3870 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
3871 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
3872 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
3873 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
3875 (rs6000_reassociation_width): Ditto.
3876 (rs6000_emit_epilogue): Ditto.
3877 (rs6000_adjust_cost): Ditto.
3878 (is_microcoded_insn): Ditto.
3879 (is_cracked_insn): Ditto.
3880 (rs6000_adjust_priority): Ditto.
3881 (rs6000_sched_reorder): Ditto.
3882 (rs6000_sched_reorder2): Ditto.
3883 (insn_must_be_first_in_group): Ditto.
3884 (insn_must_be_last_in_group): Ditto.
3885 (rs6000_register_move_cost): Ditto.
3886 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
3889 2017-12-07 Julia Koval <julia.koval@intel.com>
3891 * config.gcc: Add vaesintrin.h.
3892 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3894 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3895 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3897 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3898 * config/i386/immintrin.h: Include vaesintrin.h.
3899 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3900 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3901 _mm_aesdec_epi128): New intrinsics.
3903 2017-12-06 David Malcolm <dmalcolm@redhat.com>
3905 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
3906 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
3907 c-family/c-spellcheck.cc.
3908 (best_macro_match::best_macro_match): Likewise.
3910 (struct edit_distance_traits<cpp_hashnode *>): Move to
3911 c-family/c-spellcheck.h.
3912 (class best_macro_match): Likewise.
3914 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3916 PR tree-optimization/83293
3917 * gimple-ssa-strength-reduction.c (insert_initializers): Use
3918 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
3919 might insert into empty bb.
3922 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
3923 simplify for plus with :c added, and pointer_plus without that.
3924 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
3925 with undefined overflow and the conversion is not widening,
3926 perform negation in utype and only convert to type afterwards.
3927 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
3928 simplify for plus with :c added, and pointer_plus without that.
3929 If type is integral with undefined overflow and the conversion is
3930 not widening, perform minus in utype and only convert to type
3931 afterwards. Move the last pointer_diff_expr simplify into the
3934 2017-12-06 Martin Sebor <msebor@redhat.com>
3936 PR tree-optimization/82646
3937 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
3938 strncpy, not maxlen.
3940 2017-12-06 Martin Sebor <msebor@redhat.com>
3942 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
3945 PR tree-optimization/83075
3946 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
3947 strncat/strncpy don't change length of source string.
3949 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
3952 2017-11-29 Martin Aberg <maberg@gaisler.com>
3954 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3955 to prevent b2bst errata sequence.
3956 (sqrtdf2_fix): Likewise.
3958 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3960 PR tree-optimization/81945
3961 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
3962 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
3963 to dest_cfun has orig_loop_num set, either remap it to the new
3964 loop number if the loop got moved too, or clear it.
3966 2017-12-05 Steve Ellcey <sellcey@cavium.com>
3968 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
3970 (thunderx2t99_nothing): New insn reservation.
3971 (thunderx2t99_mrs): New insn reservation.
3972 (thunderx2t99_multiple): New insn reservation.
3973 (thunderx2t99_alu_basi): Add bfx to reservation.
3974 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
3976 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3979 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
3981 2017-12-05 Bin Cheng <bin.cheng@arm.com>
3983 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
3984 tree-ssa-pre.c::remove_dead_inserted_code.
3985 * tree-ssa-dce.h: New file.
3986 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
3987 (remove_dead_inserted_code): Move and rename to function
3988 tree-ssa-dce.c::simple_dce_from_worklist.
3989 (pass_pre::execute): Update use.
3991 2017-12-05 Richard Biener <rguenther@suse.de>
3993 PR tree-optimization/83277
3994 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
3995 to code-gen liveout vars.
3997 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
3999 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4000 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4001 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4004 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4006 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4008 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4009 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4012 2017-12-05 Richard Biener <rguenther@suse.de>
4014 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4015 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4016 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4017 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4019 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4021 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4023 2017-12-05 Martin Liska <mliska@suse.cz>
4024 Jakub Jelinek <jakub@redhat.com>
4026 * doc/invoke.texi: Document the options.
4027 * flag-types.h (enum sanitize_code): Add
4028 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4029 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4030 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4031 * opts.c: Define new sanitizer options.
4032 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4033 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4035 2017-12-05 Julia Koval <julia.koval@intel.com>
4037 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4038 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4039 (ix86_handle_option): Handle -mavx512vnni.
4040 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4041 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4042 * config/i386/i386-c (__AVX512VNNI__): New.
4043 * config/i386/i386.c (ix86_target_string): Handle new option.
4044 (ix86_valid_target_attribute_inner_p): Handle new option.
4045 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4046 * config/i386/i386.opt (mavx512vnni): New option.
4048 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4051 * athlon.md: Disable for generic.
4052 * haswell.md: Enable for generic.
4053 * i386.c (ix86_sched_init_global): Add core hooks for generic.
4054 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
4056 (ix86_adjust_cost): Move generic to haswell path.
4058 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
4060 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
4061 instead of MEM_P in a couple more places. Fix formatting issues.
4063 2017-12-04 Jim Wilson <jimw@sifive.com>
4065 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
4066 instead of GP_REG_LAST-1.
4067 (riscv_adjust_libcall_cfi_prologue): Likewise.
4068 (riscv_adjust_libcall_cri_epilogue): Likewise.
4069 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
4072 2017-12-04 Luis Machado <luis.machado@linaro.org>
4074 * ipa-pure-const.c (check_decl): Add missing newline.
4075 (state_from_flags): Likewise.
4077 2017-12-04 Jeff Law <law@redhat.com>
4079 PR tree-optimizatin/78496
4080 * gimple-ssa-evrp-analyze.h
4081 (evrp_range_analyzer::get_vr_values): Simplify.
4082 * gimple-ssa-evrp-analyze.c: Corresponding changes.
4083 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
4084 and gimple-ssa-evrp-analyze.h.
4085 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
4086 (simplify_stmt_for_jump_threading): Copy a blob of code from
4087 tree-vrp.c to use ranges to simplify statements.
4088 (dom_opt_dom_walker::before_dom_children): Call
4089 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
4090 (dom_opt_dom_walker::after_dom_children): Similarly for
4091 evrp_range_analyzer::leave.
4092 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
4095 * gimple-ssa-evrp-analyze.c
4096 (evrp_range_analyzer::extract_range_from_stmt): Always use
4097 vr_values::update_value_range so preexisting range info is
4098 medged with new range info, even if the new range is VR_VARYING.
4100 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4102 * combine.c: Adjust comment.
4103 (use_crosses_set_p): Delete.
4104 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
4105 (try_combine): Ditto.
4107 2017-12-04 Richard Biener <rguenther@suse.de>
4109 PR tree-optimization/83255
4110 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
4111 Re-add zero-iteration check.
4113 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4115 PR rtl-optimization/83245
4116 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
4117 hard registers as earlyclobber, also if not in an asm.
4119 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4123 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4126 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4127 rs6000_cpu based on cpu_index, not tune_index.
4129 2017-12-04 Richard Biener <rguenther@suse.de>
4131 PR tree-optimization/83238
4132 * graphite-scop-detection.c (scop_detection::merge_sese): Make
4133 code match comment, rejecting invalid SESE regions.
4135 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
4137 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
4138 require base operand is a REG_POINTER prior to reload on targets
4139 with non-equivalent space registers.
4141 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4143 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
4144 (update_specialized_profile): Fix updating of counts.
4145 (perhaps_add_new_callers): Likewise.
4147 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4150 * x86-tune.def: Remove obsolette FIXMEs.
4151 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
4152 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
4153 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
4155 (X86_TUNE_PAD_RETURNS): Disable for generic.
4157 2017-12-02 Jakub Jelinek <jakub@redhat.com>
4159 PR tree-optimization/83170
4160 PR tree-optimization/83241
4161 * gimple-ssa-store-merging.c
4162 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
4163 gimple_vuse (ins_stmt) in case it has changed.
4164 (imm_store_chain_info::output_merged_store): Likewise.
4166 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
4170 * tree-cfg.c (pass_warn_function_return::execute): Handle
4171 __builtin_ubsan_handle_missing_return like __builtin_unreachable
4172 with BUILTINS_LOCATION.
4176 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
4177 is BLKmode for vector field with vector raw mode, use TYPE_MODE
4178 instead of DECL_MODE.
4180 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
4181 last argument to rtx pointer.
4182 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
4183 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
4184 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
4185 depending on the chosen ISAs.
4186 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
4187 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
4188 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
4190 * config/i386/sse.md (mov<mode>_internal): Likewise.
4191 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
4193 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4195 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
4197 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4199 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
4200 parameters from prototype.
4201 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
4202 parameters. Don't print a comment.
4203 (emit_fusion_gpr_load): Adjust.
4204 (emit_fusion_load_store): Adjust.
4205 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
4206 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
4207 comment on the second line.
4209 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4212 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4213 rs6000_cpu based on cpu_index, not tune_index.
4215 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4217 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
4218 which_alternative instead of which_alternative + 1.
4219 (output_asm_insn): Print an extra tab if the template is short.
4221 2017-12-01 Jim Wilson <jimw@sifive.com>
4223 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
4225 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
4226 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
4228 * doc/tm.texi: Regenerate.
4230 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
4233 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
4234 whether we can allocate pseudos before trying to fix an address.
4235 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
4236 memory address is indexed or indirect.
4237 (floatuns_<mode>si2_hw2): Likewise.
4239 2017-12-01 Jason Merrill <jason@redhat.com>
4241 * Makefile.in (TAGS): Add c-family/*.cc.
4243 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
4245 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
4246 (call_value_insn): Likewise.
4247 (sibcall_insn): Likewise.
4248 (sibcall_value_insn): Likewise.
4249 (movsi_aarch64): Likewise.
4250 (movdi_aarch64): Likewise.
4251 (add_losym_): Likewise.
4252 (ldr_got_small_): Likewise.
4253 (ldr_got_small_sidi): Likewise.
4254 (ldr_got_small_28k_): Likewise.
4255 (ldr_got_small_28k_sidi): Likewise.
4256 * config/aarch64/aarch64.c (aarch64_print_address_internal):
4257 Move output_addr_const to symbolic case. Add error check.
4259 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4261 * config/s390/predicates.md (plus16_Q_operand): New predicate.
4262 * config/s390/s390.md: Disable MVC merging peephole if it would
4263 disable operand forwarding.
4264 (new peephole2): Split MVCs if it would turn them into up to 2
4267 2017-12-01 Richard Biener <rguenther@suse.de>
4269 PR tree-optimization/83232
4270 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
4271 detection of same access. Instead of breaking the group here
4272 do not consider the duplicate. Add comment explaining real fix.
4274 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
4276 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
4278 2017-12-01 Sudakshina Das <sudi.das@arm.com>
4280 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
4282 2017-12-01 Jakub Jelinek <jakub@redhat.com>
4284 * function.h (struct function): Remove cilk_frame_decl,
4285 is_cilk_function and calls_cilk_spawn fields.
4286 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
4288 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
4289 cilk_elemental field.
4290 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
4291 * target.def: Adjust comment.
4292 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
4293 Don't test cilk_elemental.
4295 PR tree-optimization/83233
4296 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
4297 bswap_stat name for the struct.
4300 * tree.h (SWITCH_BREAK_LABEL_P): Define.
4301 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
4302 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
4303 SWITCH_BREAK_LABEL_P set on the label.
4304 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
4305 added for default case if it was missing and not all cases covered.
4306 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
4307 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
4309 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
4310 to avoid -Wimplicit-fallthrough warning.
4311 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
4312 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
4314 PR tree-optimization/83221
4315 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
4317 (init_reassoc): Formatting fix.
4320 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
4321 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
4325 * tree-cfg.c: Include asan.h.
4326 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
4328 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4330 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
4332 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4334 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
4335 vpcmpeqd instruction.
4337 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4339 * config/i386/i386.c (standard_sse_constant_opcode): Fix
4340 registers type for 128bit mode.
4342 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4344 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
4346 * gengtype-state.c (read_state_pair): Likewise.
4347 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
4348 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
4349 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
4350 * ifcvt.c (noce_try_store_flag_constants): Likewise.
4351 * tree-ssa-ccp.c (ccp_finalize): Likewise.
4352 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
4353 * builtins.c (fold_builtin_3): Likewise.
4354 * graphite-scop-detection.c
4355 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
4356 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
4359 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
4362 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
4363 parameter. Return true if ignore_decl_rtl_set_p is true and other
4364 conditions are satisfied.
4365 * asan.h (asan_protect_global): Add new parameter.
4366 * varasm.c (categorize_decl_for_section): Pass true as second parameter
4367 to asan_protect_global calls.
4369 2017-11-30 Jim Wilson <jimw@sifive.com>
4371 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
4372 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
4373 -mstrict-align, add info on default value. Delete redundant lines for
4374 -mabi. Add missing -mexplicit-relocs docs.
4376 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4378 * config/arc/arc.md (trap): New pattern.
4380 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4382 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
4383 instruction to end into a delay slot.
4384 * config/arc/arc.md (cond_delay_insn): Check if the instruction
4385 can be placed into a delay slot against reg_note.
4387 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4389 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
4390 labels number of usages.
4392 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4394 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
4396 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
4398 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4401 * internal-fn.c (expand_mul_overflow): Optimize unsigned
4402 multiplication by power of 2 constant into two shifts + comparison.
4404 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
4407 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
4409 2017-11-30 Richard Biener <rguenther@suse.de>
4411 PR tree-optimization/83202
4412 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
4413 allow_peel argument and guard peeling.
4414 (canonicalize_loop_induction_variables): Likewise.
4415 (canonicalize_induction_variables): Pass false.
4416 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
4417 peeling from cunrolli.
4419 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4421 * combine.c (try_combine): Print a message to dump file whenever
4422 I0, I1, or I2 cannot be combined into I3.
4424 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4426 PR rtl-optimization/83156
4427 PR rtl-optimization/82621
4428 * combine.c (try_combine): Don't split an I2 if one of the dests is
4429 set again before I3. Allow unused dests.
4431 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4433 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
4435 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
4437 PR rtl-optimization/80818
4438 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
4439 recursively. Use insn code for clobber.
4440 (lra_set_insn_recog_data): Pass the new arg to
4441 collect_non_operand_hard_regs.
4442 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
4444 (lra_update_insn_regno_info): Pass insn to
4445 add_regs_to_insn_regno_info.
4447 2017-11-29 Jim Wilson <jimw@sifive.com>
4448 Andrew Waterman <andrew@sifive.com>
4450 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
4451 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
4452 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
4453 use SINGLE_SHIFT_COST.
4454 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
4455 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
4457 2017-11-29 Julia Koval <julia.koval@intel.com>
4459 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
4460 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
4461 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
4462 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
4463 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
4464 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
4465 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
4466 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
4467 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
4468 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
4469 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
4470 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
4471 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
4472 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
4473 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
4474 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
4475 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
4476 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
4477 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
4478 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
4479 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
4480 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
4481 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
4482 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
4483 __builtin_ia32_vpshldv_v2di_mask,
4484 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
4485 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
4486 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
4488 2017-11-29 Julia Koval <julia.koval@intel.com>
4490 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
4491 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
4492 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
4493 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
4494 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
4495 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
4496 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
4497 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
4498 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
4499 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
4500 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
4501 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
4502 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
4503 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
4504 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
4505 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
4506 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
4507 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
4508 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4509 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
4510 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
4512 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4514 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
4515 movsi_pic_gotdata_op instruction as a load for the UT699 errata
4518 2017-11-29 Martin Aberg <maberg@gaisler.com>
4520 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4521 to prevent b2bst errata sequence.
4522 (sqrtdf2_fix): Likewise.
4524 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4526 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
4527 (div_sqrt_insn_p): New function.
4528 (sparc_do_work_around_errata): Insert NOP instructions to
4529 prevent sequences that could trigger the TN-0013 errata for
4530 certain LEON3 processors.
4531 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
4532 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
4533 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
4534 (in_branch_delay): Prevent div and sqrt in delay slot if
4536 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
4538 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4540 * config/sparc/sparc.c (atomic_insn_p): New function.
4541 (sparc_do_work_around_errata): Insert NOP instructions to
4542 prevent sequences that could trigger the TN-0010 errata for
4544 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
4545 instruction referable in atomic_insns_p.
4547 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4549 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
4550 (atomic_compare_and_swap_leon3_1): Likewise.
4553 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4555 * config/sparc/sparc.c (fpop_insn_p): New function.
4556 (sparc_do_work_around_errata): Insert NOP instructions to
4557 prevent sequences that could trigger the TN-0012 errata for
4559 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
4560 * config/sparc/sparc.md (fix_gr712rc): New attribute.
4561 (in_branch_annul_delay): Prevent floating-point instructions
4562 in delay slot of annulled integer branch.
4564 2017-11-29 Richard Biener <rguenther@suse.de>
4566 PR tree-optimization/83202
4567 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4569 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4570 nodes vectorized to the same stmts multiple times.
4571 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4572 (vect_analyze_slp_instance): Adjust.
4573 (scalar_stmts_to_slp_tree_map_t): New typedef.
4574 (vect_schedule_slp_instance): Add a map recording the SLP node
4575 representing the vectorized stmts for a set of scalar stmts.
4576 Avoid code-generating redundancies.
4577 (vect_schedule_slp): Allocate map and pass it down.
4579 2017-11-29 Nathan Sidwell <nathan@acm.org>
4582 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4585 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4587 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4588 cases for printing LDP/STP memory addresses.
4589 (aarch64_print_address_internal): Renamed from
4590 aarch64_print_operand_address, added parameter, add Pmode check.
4591 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4592 (aarch64_print_operand_address): Indirect to
4593 aarch64_print_address_internal.
4594 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4595 'y' operand output specifier.
4597 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4600 * tree.c (build_simple_mem_ref_loc): Handle
4601 get_addr_base_and_unit_offset returning a MEM_REF.
4604 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4607 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4610 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4612 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4614 (layout::move_to_column): Likewise.
4616 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4619 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4620 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4622 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4623 Martin Jambor <mjambor@suse.cz>
4626 * tree.h (expr_type_first_operand_type_p): Declare
4627 * tree.c (expr_type_first_operand_type_p): New function.
4628 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4629 (ipa_value_from_jfunc): Adjust declaration.
4630 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4631 Use it as result type for arithmetics, unless it is NULL in which case
4632 be more conservative.
4633 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4634 ipa_get_jf_pass_through_result.
4635 (propagate_vals_across_pass_through): Likewise.
4636 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4637 is to propagate_vals_across_pass_through.
4638 (propagate_constants_across_call): Pass PARM_TYPE to
4639 propagate_scalar_across_jump_function.
4640 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4641 ipa_value_from_jfunc.
4642 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4643 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4644 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4645 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4646 target_type, pass it to ipa_value_from_jfunc.
4647 (update_indirect_edges_after_inlining): Pass parameter type to
4648 try_make_edge_direct_simple_call.
4650 2017-11-28 Jeff Law <law@redhat.com>
4652 * gimple-ssa-evrp-analyze.c
4653 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4654 refine ranges if scev_initialized_p returns true.
4655 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4657 2017-11-28 Julia Koval <julia.koval@intel.com>
4659 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4660 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4661 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4662 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4663 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4664 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4665 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4666 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4667 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4668 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4669 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4670 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4671 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4672 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4673 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4674 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4675 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4676 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4677 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4678 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4679 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4680 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4682 2017-11-28 Julia Koval <julia.koval@intel.com>
4684 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4685 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4686 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4687 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4688 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4689 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4690 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4691 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4692 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4693 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4694 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4695 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4696 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4697 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4698 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4699 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4700 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4701 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4702 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4703 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4704 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4705 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4706 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4707 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4708 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4709 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4710 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4711 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4712 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4713 __builtin_ia32_vpshld_v2di_mask): New builtins.
4714 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4715 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4717 2017-11-28 Richard Biener <rguenther@suse.de>
4719 PR tree-optimization/80776
4720 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4722 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4724 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4725 If the incoming edge is an effective fallthru because the other
4726 edge only reaches a __builtin_unreachable () then record ranges
4727 derived from the controlling condition in SSA info.
4728 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4729 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4731 2017-11-28 Olivier Hainque <hainque@adacore.com>
4733 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4734 on mingw build hosts.
4736 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4739 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4741 2017-11-28 Richard Biener <rguenther@suse.de>
4744 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4745 copies generated from memcpy use a character array as reference
4748 2017-11-28 Julia Koval <julia.koval@intel.com>
4749 Sebastian Peryt <sebastian.peryt@intel.com>
4751 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4752 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4753 cilk-common.o, cilk.h, cilk-common.c): Remove.
4755 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4756 * builtins.c (is_builtin_name): Remove cilkplus condition.
4757 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4758 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4759 cilk-builtins.def, cilkplus.def): Remove.
4760 * cif-code.def (CILK_SPAWN): Remove.
4761 * cilk-builtins.def: Delete.
4762 * cilk-common.c: Ditto.
4764 * cilkplus.def: Ditto.
4765 * config/darwin.h (fcilkplus): Delete.
4766 * cppbuiltin.c: Ditto.
4767 * doc/extend.texi: Remove cilkplus doc.
4768 * doc/generic.texi: Ditto.
4769 * doc/invoke.texi: Ditto.
4770 * doc/passes.texi: Ditto.
4771 * gcc.c (fcilkplus): Remove.
4772 * gengtype.c (cilk.h): Remove.
4773 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4775 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4777 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4779 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4780 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4781 cilkplus conditions.
4782 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4783 inline_read_section): Ditto.
4784 * ipa-inline-analysis.c (cilk.h): Remove.
4785 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4786 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4787 append_linker_options): Remove condition for fcilkplus.
4788 * lto/lto-lang.c (cilk.h): Remove.
4789 (lto_init): Remove condition for fcilkplus.
4790 * omp-expand.c (expand_cilk_for_call): Delete.
4791 (expand_omp_taskreg, expand_omp_for_static_chunk,
4792 expand_omp_for): Remove cilkplus
4794 (expand_cilk_for): Delete.
4795 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4796 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4797 execute_lower_omp, diagnose_sb_0): Ditto.
4798 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4799 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4800 * tree-nested.c: Ditto.
4801 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4802 (dump_generic_node): Ditto.
4803 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4804 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4806 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4808 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4810 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
4811 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
4812 for vector mode and !TARGET_SIMD.
4814 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4816 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
4818 * tree.h (SWITCH_LABELS): Remove.
4819 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
4820 assert SWITCH_BODY is non-NULL.
4821 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
4823 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
4825 PR tree-optimization/80788
4826 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
4827 has TREE_OVERFLOW set, call drop_tree_overflow.
4829 2017-11-28 Richard Biener <rguenther@suse.de>
4831 PR tree-optimization/83158
4832 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
4834 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
4837 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
4838 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
4840 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
4842 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
4843 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
4845 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4848 * varasm.c (compare_constant): Take the mode of the constants into
4849 account when comparing floating point constants.
4851 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
4853 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
4854 from explicit instantiation of debug_helper.
4855 * vec.h (DEFINE_DEBUG_VEC): Ditto.
4857 2017-11-27 Richard Biener <rguenther@suse.de>
4859 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
4862 2017-11-27 Richard Biener <rguenther@suse.de>
4864 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
4865 (type_hash_canon): Also clear int_cst_hash_table entry for
4867 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
4869 2017-11-27 Tamar Christina <tamar.christina@arm.com>
4871 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4872 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4873 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4875 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4877 * hash-map.h (gt_cleare_cache): Avoid UB.
4879 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
4881 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
4882 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
4883 (decide_unroll_runtime_iterations): Likewise.
4884 (decide_unroll_stupid): Likewise.
4886 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4889 * config/i386/i386.md: Add a loop with incssp.
4891 2017-11-27 Martin Jambor <mjambor@suse.cz>
4893 PR tree-optimization/81248
4894 * tree-sra.c (splice_param_accesses): Remove size check.
4895 (decide_one_param_reduction): Fix size check.
4896 * gimple-pretty-print.c (dump_profile): Silence warning.
4897 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
4899 2017-11-27 Jakub Jelinek <jakub@redhat.com>
4902 * dbxout.c (lastlineno): New variable.
4903 (dbx_debug_hooks): Use dbxout_switch_text_section as
4904 switch_text_section debug hook.
4905 (dbxout_function_end): Switch to current_function_section
4906 rather than function_section. If crtl->has_bb_partition,
4907 output just one N_FUN, depending on in_cold_section_p.
4908 (dbxout_source_line): Remember last lineno in lastlineno.
4909 (dbxout_switch_text_section): New function.
4910 (dbxout_function_decl): Adjust dbxout_block caller.
4911 (dbx_block_with_cold_children): New function.
4912 (dbxout_block): Return true if any LBRAC/RBRAC have been
4913 emitted. Use dbx_block_with_cold_children at depth == 0
4914 in second partition. Add PARENT_BLOCKNUM argument, pass
4915 it optionally adjusted to children. Output LBRAC/RBRAC
4916 around recursive call only if the block is in the current
4917 partition, if not and anything was output, emit empty
4919 * final.c (final_scan_insn): Compute cold_function_name
4920 before calling switch_text_section debug hook. Call
4921 that hook even if dwarf2out_do_frame if not emitting
4925 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
4926 TREE_READONLY decls.
4928 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4930 PR rtl-optimization/82488
4931 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
4933 2017-11-26 Julia Koval <julia.koval@intel.com>
4935 * config/i386/i386.c (processor_target_table): Add skylake_cost for
4937 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
4940 2017-11-26 Julia Koval <julia.koval@intel.com>
4942 * config/i386/driver-i386.c (host_detect_local_cpu):
4943 Detect skylake-avx512.
4945 2017-11-26 Julia Koval <julia.koval@intel.com>
4947 * config.gcc: Add -march=cannonlake.
4948 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
4949 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
4950 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
4951 (PTA_CANNONLAKE): New.
4952 (processor_target_table): Add cannonlake.
4953 (ix86_option_override_internal): Ditto.
4954 (fold_builtin_cpu): Ditto.
4955 (get_builtin_code_for_version): Handle cannonlake.
4956 (M_INTEL_COREI7_CANNONLAKE): New.
4957 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
4958 * doc/invoke.texi: Add -march=cannonlake.
4960 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
4962 * plugin.c (add_new_plugin): Use platform-specific library extensions.
4963 (try_init_one_plugin): Alternative implementation for MinGW.
4964 * Makefile.in (plugin_implib): New.
4965 (gengtype-lex.c): Fix broken AIX workaround.
4966 * configure: Regenerate.
4967 * doc/plugins.texi: Document support for MinGW.
4969 2017-11-25 Jakub Jelinek <jakub@redhat.com>
4971 PR rtl-optimization/81553
4972 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
4973 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
4974 is a shift where C1 has different mode than the whole shift, use C1's
4975 mode for MULT rather than the shift's mode.
4978 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
4979 builtins not enabled in the currently selected ISA.
4981 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
4983 PR tree-optimization/71026
4984 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
4985 (insert_reciprocals): Change to insert reciprocals before a division
4986 by a square and to insert the square of a reciprocal.
4987 (execute_cse_reciprocals_1): Change to consider division by a square.
4988 (register_division_in): Add importance parameter.
4990 2017-11-24 Richard Biener <rguenther@suse.de>
4992 PR tree-optimization/82402
4993 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
4994 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
4996 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
4998 * match.pd (0-ptr): New transformation.
5000 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5003 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5004 when skipping checking.
5006 2017-11-24 Richard Biener <rguenther@suse.de>
5008 PR tree-optimization/83128
5009 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5010 (vn_reference_lookup_3): Likewise.
5012 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5015 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5016 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5019 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5020 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5021 check for the latter with gimple_call_builtin_p. Do not handle
5022 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5024 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5026 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5029 2017-11-23 Julia Koval <julia.koval@intel.com>
5031 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5032 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5033 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5034 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5035 _mm512_maskz_expandloadu_epi16): New intrinsics.
5036 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5037 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5038 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5039 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5040 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5041 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5042 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5043 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5044 _mm256_maskz_expandloadu_epi8): New intrinsics.
5045 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5046 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5047 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5048 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5049 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5050 * config/i386/sse.md (VI248_VLBW): New iterator.
5051 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5053 2017-11-23 Julia Koval <julia.koval@intel.com>
5055 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
5056 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
5057 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
5058 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
5059 _mm512_mask_compressstoreu_epi16): New.
5060 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
5061 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
5062 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
5063 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
5064 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
5065 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
5066 _mm256_mask_compressstoreu_epi8): New.
5067 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
5068 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
5069 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
5070 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
5071 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
5072 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
5073 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
5074 __builtin_ia32_compresshi128_mask,
5075 __builtin_ia32_compressstoreuqi512_mask,
5076 __builtin_ia32_compressstoreuhi512_mask,
5077 __builtin_ia32_compressstoreuqi256_mask,
5078 __builtin_ia32_compressstoreuqi128_mask,
5079 __builtin_ia32_compressstoreuhi256_mask,
5080 __builtin_ia32_compressstoreuhi128_mask): New builtins.
5081 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
5083 (ix86_expand_special_args_builtin): Handle new types.
5084 (s4fma_expand): Handle new builtin array.
5085 * config/i386/immintrin.h: Include new headers.
5086 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
5087 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
5089 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5091 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
5092 fixes. Declare temp and g variables at the top in order to avoid
5093 {} in most of the cases.
5095 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
5097 * match.pd (ptr-0): New transformation.
5099 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5101 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
5102 (struct addr_mode_cost_table): New.
5103 (struct tune_params): Add field addr_mode_costs.
5104 * config/arm/arm.c (generic_addr_mode_costs): New.
5105 (arm_slowmul_tune): Initialise addr_mode_costs field.
5106 (arm_fastmul_tune): Likewise.
5107 (arm_strongarm_tune): Likewise.
5108 (arm_xscale_tune): Likewise.
5109 (arm_9e_tune): Likewise.
5110 (arm_marvell_pj4_tune): Likewise.
5111 (arm_v6t2_tune): Likewise.
5112 (arm_cortex_tune): Likewise.
5113 (arm_cortex_a8_tune): Likewise.
5114 (arm_cortex_a7_tune): Likewise.
5115 (arm_cortex_a15_tune): Likewise.
5116 (arm_cortex_a35_tune): Likewise.
5117 (arm_cortex_a53_tune): Likewise.
5118 (arm_cortex_a57_tune): Likewise.
5119 (arm_exynosm1_tune): Likewise.
5120 (arm_xgene1_tune): Likewise.
5121 (arm_cortex_a5_tune): Likewise.
5122 (arm_cortex_a9_tune): Likewise.
5123 (arm_cortex_a12_tune): Likewise.
5124 (arm_cortex_a73_tune): Likewise.
5125 (arm_v7m_tune): Likewise.
5126 (arm_cortex_m7_tune): Likewise.
5127 (arm_v6m_tune): Likewise.
5128 (arm_fa726te_tune): Likewise.
5129 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
5131 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5133 * config/arm/arm.c (arm_mem_costs): New function.
5134 (arm_rtx_costs_internal): Use arm_mem_costs.
5136 2017-11-23 Mark Wielaard <mark@klomp.org>
5138 * dwarf2out.c (init_sections_and_labels): Use generation to create
5139 unique ranges_section_label and ranges_base_label. Return generation.
5140 (output_rnglists): Add generation argument. Use generation to create
5141 unique ranges labels.
5142 (dwarf2out_finish): Get generation from init_sections_and_labels
5143 and pass generation to output_rnglists.
5145 2017-11-23 Mike Stump <mikestump@comcast.net>
5146 Eric Botcazou <ebotcazou@adacore.com>
5148 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
5149 * cfgloop.h (struct loop): Add unroll field.
5150 * function.h (struct function): Add has_unroll bitfield.
5151 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
5152 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
5153 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
5154 (pass_rtl_unroll_loops::gate): Likewise.
5155 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
5156 for which loop->unroll==1.
5157 (decide_unroll_constant_iterations): Use note for consistency and
5158 take loop->unroll into account. Return early if loop->unroll is set.
5159 Fix thinko in existing test.
5160 (decide_unroll_runtime_iterations): Use note for consistency and
5161 take loop->unroll into account.
5162 (decide_unroll_stupid): Likewise.
5163 * lto-streamer-in.c (input_cfg): Read loop->unroll.
5164 * lto-streamer-out.c (output_cfg): Write loop->unroll.
5165 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
5167 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
5168 (print_loop): Print loop->unroll if set.
5169 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
5170 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
5171 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
5173 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
5174 loop->unroll is set and smaller than the trip count. Otherwise bypass
5175 entirely the heuristics if loop->unroll is set. Remove dead note.
5176 Fix off-by-one bug in other note.
5177 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
5178 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
5180 (tree_unroll_loops_completely): Make static.
5181 (pass_complete_unroll::execute): Use correct type for variable.
5182 (pass_complete_unrolli::execute): Fix formatting.
5183 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
5185 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
5187 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
5188 enable when TARGET_PREFER_AVX128 is set.
5190 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5192 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
5194 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
5195 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
5197 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5199 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
5200 frequencies. Fix estimation of aggregate parameters.
5202 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5204 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
5207 2017-11-23 Tom de Vries <tom@codesourcery.com>
5209 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
5211 2017-11-23 Tom de Vries <tom@codesourcery.com>
5213 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
5216 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
5219 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
5220 sibcall_value_pcrel_fdpic): Use local variable instead of
5222 (calli_tbr_rel): Add missing operand 2.
5223 (call_valuei_tbr_rel): Add missing operand 3.
5225 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5228 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
5229 bitpos/bitsize covering the whole destination, use store_expr only if
5230 the complex mode is the same. Otherwise, use expand_normal and if
5231 it returns CONCAT, subreg each part separately instead of trying to
5232 subreg the whole result.
5234 2017-11-23 Richard Biener <rguenther@suse.de>
5236 PR tree-optimization/23094
5237 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
5238 come here from walking over backedges in the first iteration.
5239 (vn_reference_lookup_3): Skip clobbers that store the same value.
5241 2017-11-23 Richard Biener <rguenther@suse.de>
5243 PR tree-optimization/81403
5244 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
5245 a block we need a leader relative to.
5246 (phi_translate_1): For nary processing require a leader from
5247 get_representative_for given we run expression simplification
5248 using match-and-simplify. Remove previous fix.
5250 2017-11-22 Jeff Law <law@redhat.com>
5252 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
5253 Use new method allocate_value_range rather than accessing the
5254 vrp_value_range_pool data member directly.
5255 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
5256 to use extract_range_from_stmt method to avoid need for
5257 extract_range_from_assignment method.
5258 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
5259 method rather than setting values_propgated data member directly.
5260 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
5261 and values propagated data members and extract_range_from_assignment
5262 method. Reorder private data members to conform to standards.
5263 Add new methods set_lattice_propagation_complete and
5264 allocate_value_range.
5266 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
5268 PR rtl-optimization/83030
5269 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
5270 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
5271 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
5272 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
5273 CROSSING_JUMP_P flag is already set before setting it.
5274 * cfgrtl.c (fixup_partition_crossing): Likewise.
5275 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
5278 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5280 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
5281 Handle the case where both arguments are using gen_const_vec_series.
5283 2017-11-22 David Malcolm <dmalcolm@redhat.com>
5286 * pretty-print.c (pp_format): Move quoting implementation to
5287 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
5288 to pass address of "quote" local.
5289 (pp_begin_quote): New function.
5290 (pp_end_quote): New function.
5291 * pretty-print.h (printer_fn): Convert penultimate param from bool
5293 (pp_begin_quote): New decl.
5294 (pp_end_quote): New decl.
5295 * tree-diagnostic.c (default_tree_printer): Convert penultimate
5296 param from bool to bool *.
5297 * tree-diagnostic.h (default_tree_printer): Likewise.
5299 2017-11-22 Jeff Law <law@redhat.com>
5301 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
5302 Initialize vr_values.
5303 (evrp_range_analyzer::try_find_new_range): Call methods attached to
5304 vr_values via vr_values class instance rather than delegators.
5305 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
5306 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5307 (evrp_range_analyzer::push_value_range): Likewise.
5308 (evrp_range_analyzer::pop_value_range): Likewise.
5309 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
5310 most delegators. Those remaining are exposed as public interfaces.
5311 Make vr_values a pointer and private.
5312 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
5314 (evrp_range_analyzer::get_vr_value): New method.
5315 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
5316 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
5317 (class evrp_dom_walker): Attach evrp_folder class, initialize
5318 it in the ctor. Remove temporary delegators.
5319 (evrp_dom_walker::before_dom_children): Call methods in attached
5320 evrp_range_analyzer class via class instance pointer. Use
5321 free value_range_constant_singleton to remove need for
5322 op_with_constant_singleton_value delegator method. Do not
5323 create a vrp_prop class instance for every call! Narrow
5324 scope of a couple variables.
5325 (evrp_dom_walker::cleanup): Call methods in attached
5326 evrp_range_analyzer class via class instance pointer.
5327 * vr-values.h (class vr_values): Privatize many methods and
5330 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5332 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
5333 * cfgexpand.c (expand_debug_expr): Remove handling for them.
5334 * expr.c (expand_expr_real_2): Likewise.
5335 * fold-const.c (const_unop): Likewise.
5336 * optabs-tree.c (optab_for_tree_code): Likewise.
5337 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
5338 * tree-inline.c (estimate_operator_cost): Likewise.
5339 * tree-pretty-print.c (dump_generic_node): Likewise.
5340 (op_code_prio): Likewise.
5341 (op_symbol_code): Likewise.
5342 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
5343 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
5344 * internal-fn.c (direct_internal_fn_optab): New function.
5345 (direct_internal_fn_array, direct_internal_fn_supported_p
5346 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
5347 * fold-const-call.c (fold_const_reduction): New function.
5348 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
5350 * tree-vect-loop.c: Include internal-fn.h.
5351 (reduction_code_for_scalar_code): Rename to...
5352 (reduction_fn_for_scalar_code): ...this and return an internal
5354 (vect_model_reduction_cost): Take an internal_fn rather than
5356 (vect_create_epilog_for_reduction): Likewise. Build calls rather
5358 (vectorizable_reduction): Use internal functions rather than tree
5359 codes for the reduction operation. Update calls to the functions
5361 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
5362 Use calls to internal functions rather than REDUC tree codes.
5363 * config/aarch64/aarch64-simd.md: Update comment accordingly.
5365 2017-11-22 Olivier Hainque <hainque@adacore.com>
5367 * config/vxworks.c (vxworks_override_options): Pick default
5368 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
5369 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
5370 DWARF_GNAT_ENCODINGS_DEFAULT.
5371 * config/vxworksae.h: Likewise.
5373 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
5375 PR tree-optimization/83104
5376 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
5379 2017-11-22 Marek Polacek <polacek@redhat.com>
5380 H.J. Lu <hongjiu.lu@intel.com>
5381 Jason Merrill <jason@redhat.com>
5386 * calls.c (initialize_argument_information): Call
5387 warn_parameter_passing_abi target hook.
5388 (store_one_arg): Use 0 for empty record size. Don't push 0 size
5389 argument onto stack.
5390 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
5391 * common.opt: Update -fabi-version description.
5392 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
5393 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
5395 (ix86_is_empty_record): New function.
5396 (ix86_warn_parameter_passing_abi): New function.
5397 (TARGET_EMPTY_RECORD_P): Redefine.
5398 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
5399 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
5400 * doc/tm.texi: Regenerated.
5401 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
5402 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
5403 * dwarf2out.c (get_ultimate_context): Move to tree.c.
5404 * explow.c (hard_function_value): Call arg_int_size_in_bytes
5405 instead of int_size_in_bytes.
5406 * expr.c (copy_blkmode_to_reg): Likewise.
5407 * function.c (aggregate_value_p): Return 0 for empty types.
5408 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
5410 (locate_and_pad_parm): Call arg size_in_bytes instead
5412 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
5413 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
5414 * target.def (empty_record_p, warn_parameter_passing_abi): New target
5416 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
5417 (std_gimplify_va_arg_expr): Skip empty records. Call
5418 arg_size_in_bytes instead size_in_bytes.
5419 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
5420 * tree-core.h (tree_type_common): Add empty_flag.
5421 (tree_decl_common): Update comments.
5422 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
5424 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5425 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
5427 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5428 * tree.c (default_is_empty_type): New function.
5429 (default_is_empty_record): New function.
5430 (arg_int_size_in_bytes): New function.
5431 (arg_size_in_bytes): New function.
5432 (get_ultimate_context): New function.
5433 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
5434 TRANSLATION_UNIT_WARN_EMPTY_P.
5435 (default_is_empty_record, arg_int_size_in_bytes,
5436 arg_size_in_bytes, get_ultimate_context): Declare.
5438 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5440 * config/arm/arm.c (cmse_clear_registers): New function.
5441 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
5442 code by call to cmse_clear_registers.
5443 (cmse_nonsecure_entry_clear_before_return): Likewise.
5445 2017-11-22 Tamar Christina <tamar.christina@arm.com>
5447 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
5448 (vdot_s32, vdotq_s32): New.
5449 (vdot_lane_u32, vdotq_lane_u32): New.
5450 (vdot_lane_s32, vdotq_lane_s32): New.
5453 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5456 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
5457 for unsigned values with fewer HWIs than the precision.
5458 (test_overflow): New function.
5459 (wide_int_cc_tests): Call it.
5461 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5463 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
5464 has the same mode class as Pmode.
5466 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5468 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
5469 padding_bits_to_clear_ptr.
5470 (cmse_nonsecure_entry_clear_before_return): Likewise.
5472 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5474 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
5475 auto_sbitap instead of integer bitfield to control register needing
5478 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5480 PR tree-optimization/83044
5481 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
5482 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
5483 that need the upper bound. Subtract offset from
5484 get_addr_base_and_unit_offset only if positive and subtract it
5485 before division by eltsize rather than after it.
5488 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
5489 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
5490 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
5494 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
5495 expand_binop with *mul_widen_optab, make sure at least one of the
5496 operands doesn't have VOIDmode.
5499 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
5501 PR rtl-optimization/82044
5502 PR tree-optimization/82042
5503 * dse.c (record_store): Check for overflow.
5504 (check_mem_read_rtx): Properly check for overflow if width == -1, call
5505 add_wild_read instead of clear_rhs_from_active_local_stores on
5506 overflow and log it into dump_file.
5508 2017-11-22 Richard Biener <rguenther@suse.de>
5510 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
5511 fake edges to exit when looking for a place to insert.
5512 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
5514 (insert_into_preds_of_block): Commit edge insertion immediately,
5515 assert that doesn't require new BBs.
5516 (fini_pre): Release expressions.
5517 (pass_pre::execute): Shuffle things around a bit, if the fn
5518 is too large do not compute AVAIL either as this is really the
5521 2017-11-22 Richard Biener <rguenther@suse.de>
5523 PR tree-optimization/83089
5524 * tree-if-conv.c (pass_if_conversion::execute): If anything
5525 changed reset SCEV and free the number of iteration estimates.
5527 2017-11-21 Martin Sebor <msebor@redhat.com>
5529 PR tree-optimization/82945
5530 * calls.h (warn_nonstring_bound): Remove unused function.
5532 2017-11-21 Martin Sebor <msebor@redhat.com>
5534 PR tree-optimization/82945
5535 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
5536 * calls.h (maybe_warn_nonstring_arg): Declare new function.
5537 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
5539 (initialize_argument_information): Call maybe_warn_nonstring_arg.
5540 * calls.h (get_attr_nonstring_decl): Declare new function.
5541 * doc/extend.texi (attribute nonstring): Update.
5542 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
5543 get_attr_nonstring_decl and handle it.
5544 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
5545 detection of nul-termination.
5546 (strlen_to_stridx): Change to a pointer.
5547 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
5548 (pass_strlen::execute): Same.
5550 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
5552 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
5553 for the new option -mprefer-vector-width=[none|128|256|512].
5554 * config/i386/i386.c (ix86_target_string): remove old style options
5555 -mprefer-avx256 and make -mprefer-avx128 as alias.
5556 (ix86_option_override_internal): Apply defaults for the
5557 -mprefer-vector-width=[128|256] option.
5558 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
5559 Implement macros to work with -mprefer-vector-width=.
5560 * config/i386/i386.opt: Implemented option
5561 -mprefer-vector-width=[none|128|256|512].
5562 * doc/invoke.texi: Documentation for
5563 -mprefer-vector-width=[none|128|256|512].
5565 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5567 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5569 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5571 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5573 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5575 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5577 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5578 * expr.c (expand_expr_real_2): Likewise.
5579 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5580 fold_binary_loc): Likewise.
5581 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5582 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5583 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5584 MINUS_EXPR transformations.
5585 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5586 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5587 * tree-inline.c (estimate_operator_cost): Likewise.
5588 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5589 op_symbol_code): Likewise.
5590 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5591 * vr-values.c (extract_range_from_binary_expr): Likewise.
5592 * varasm.c (initializer_constant_valid_p_1): Likewise.
5593 * tree.def: New tree code POINTER_DIFF_EXPR.
5595 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5597 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5598 integer suffix to movbe mnemonic.
5599 (*bswaphi2_movbe): Ditto.
5600 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5602 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5605 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5606 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5607 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5609 2017-11-21 Martin Liska <mliska@suse.cz>
5611 * tree-inline.c (expand_call_inline): Remove not needed
5614 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5615 Jakub Jelinek <jakub@redhat.com>
5618 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5619 Remove static keyword from f variable.
5621 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5623 PR tree-optimization/83086
5624 * gimple-ssa-store-merging.c
5625 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5626 rather than n.base_addr.
5628 2017-11-21 Martin Liska <mliska@suse.cz>
5630 PR rtl-optimization/82044
5631 PR tree-optimization/82042
5632 * dse.c (check_mem_read_rtx): Check for overflow.
5634 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5636 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5639 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5641 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5642 and its corresponding call arg location note.
5644 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5645 Andrew Burgess <andrew.burgess@embecosm.com>
5647 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5649 (arc_return_slot_offset): Likewise.
5650 (arc_eh_return_address_location): New declaration.
5651 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5652 (MUST_SAVE_REGISTER): Add exception handler case.
5653 (MUST_SAVE_RETURN_ADDR): Likewise.
5654 (arc_frame_pointer_required): Likewise.
5655 (arc_frame_pointer_needed): New function.
5656 (arc_compute_frame_size): Changed.
5657 (arc_expand_prologue): Likewise.
5658 (arc_expand_epilogue): Likewise.
5659 (arc_initial_elimination_offset): Likewise.
5660 (arc_return_slot_offset): Delete.
5661 (arc_eh_return_address_location): New function.
5662 (arc_builtin_setjmp_frame_value): Likewise.
5663 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5664 (EH_RETURN_STACKADJ_RTX): Define.
5665 (EH_RETURN_HANDLER_RTX): Likewise.
5666 * config/arc/arc.md (eh_return): Delete.
5668 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5670 * print-tree.h (debug_vec_tree): Remove prototype.
5671 * gdbinit.in (pvt): Remove macro.
5673 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5675 PR tree-optimization/83047
5676 * gimple-ssa-store-merging.c
5677 (imm_store_chain_info::output_merged_store): If the loads with the
5678 same vuse are in different basic blocks, for load_gsi pick a load
5679 location that is dominated by the other loads.
5682 * config/i386/i386.c (ix86_memmodel_check): Start
5683 -Winvalid-memory-model diagnostics with lowercase letter.
5686 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5687 set in_cold_section_p to the partition containing loc_list->first.
5688 When seeing loc_list->last_before_switch node, update secname and
5689 perform range_across_switch second partition handling only after that.
5692 * run-rtl-passes.c: Include debug.h.
5693 (run_rtl_passes): Call debug_hooks->assembly_start.
5694 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5698 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5699 OPTAB_DIRECT in calls to expand_simple_binop.
5701 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5704 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5706 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5709 * spellcheck.h (best_match::blithely_get_best_candidate): New
5712 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5714 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5715 with lower case letter.
5717 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5719 * config/i386/i386.md (bswaphi2): New expander.
5720 (*bswaphi2_movbe): New insn pattern.
5721 (bswaphi -> rorhi pepehole2): New peephole pattern.
5723 2017-11-20 Jeff Law <law@redhat.com>
5725 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5726 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5727 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5728 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5729 gimple-ssa-evrp-analyze.h.
5731 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5732 set BB_VISITED here.
5733 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5735 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5737 * tree-predcom.c: Add general comment on Store-Store chains.
5738 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5740 (get_chain_last_ref_at): Rename into...
5741 (get_chain_last_write_at): ...this.
5742 (get_chain_last_write_before_load): New function.
5743 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5744 CT_STORE_STORE when write reference is added.
5745 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5746 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5747 (initialize_root_vars_store_elim_1): Ditto.
5748 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5749 definition is created.
5750 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5751 chain by replacing it with dominant stored value.
5753 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5755 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5757 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5759 * vr-values.c (extract_range_from_binary_expr): Use a full range
5762 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5764 * config/arm/arm.md (R4_REGNUM): Define constant.
5765 (nonsecure_call_internal): Remove r4 clobber.
5766 (nonsecure_call_value_internal): Likewise.
5767 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5768 clobber and resequence match_operands.
5769 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5770 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5771 (nonsecure_call_value_reg_thumb2): Likewise.
5773 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5775 PR tree-optimization/78821
5776 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5777 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5778 to the address of the base rather than the base itself.
5779 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5780 (find_bswap_or_nop_finalize): New function.
5781 (find_bswap_or_nop): Use it.
5782 (bswap_replace): Return a tree rather than bool, change first
5783 argument from gimple * to gimple_stmt_iterator, allow inserting
5784 into an empty sequence, allow ins_stmt to be NULL - then emit
5785 all stmts into gsi. Fix up MEM_REF address gimplification.
5786 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5787 (struct store_immediate_info): Add N and INS_STMT non-static
5789 (store_immediate_info::store_immediate_info): Initialize them
5790 from newly added ctor args.
5791 (merged_store_group::apply_stores): Formatting fixes. Sort by
5793 (stmts_may_clobber_ref_p): For stores call also
5794 refs_anti_dependent_p.
5795 (gather_bswap_load_refs): New function.
5796 (imm_store_chain_info::try_coalesce_bswap): New method.
5797 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5798 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5799 (imm_store_chain_info::output_merged_store): Fail if number of
5800 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5801 and NOP_EXPR rhs_code.
5802 (pass_store_merging::process_store): Compute n and ins_stmt, if
5803 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5804 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5806 (pass_store_merging::execute): Calculate dominators.
5808 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5809 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5810 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
5811 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5812 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
5813 class pass_optimize_bswap, bswap_replace,
5814 pass_optimize_bswap::execute): Moved to ...
5815 * gimple-ssa-store-merging.c: ... this file.
5816 Include optabs-tree.h.
5817 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
5818 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5819 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
5820 anonymous namespace, remove static keywords.
5821 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
5822 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
5824 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
5827 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
5829 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
5831 * vec.h (debug_helper): New function.
5832 (DEFINE_DEBUG_VEC): New macro.
5833 * hash-set.h (debug_helper): New function.
5834 (DEFINE_DEBUG_HASH_SET): New macro.
5835 * cfg.c (debug_slim (edge)): New function.
5836 Call DEFINE_DEBUG_VEC for edges.
5837 Call DEFINE_DEBUG_HASH_SET for edges.
5838 * cfghooks.c (debug_slim (basic_block)): New function.
5839 Call DEFINE_DEBUG_VEC for basic blocks.
5840 Call DEFINE_DEBUG_HASH_SET for basic blocks.
5841 * print-tree.c (debug_slim): New function to handle trees.
5842 Call DEFINE_DEBUG_VEC for trees.
5843 Call DEFINE_DEBUG_HASH_SET for trees.
5844 (debug (vec<tree, va_gc>) &): Remove.
5845 (debug (<vec<tree, va_gc>) *): Remove.
5846 * print-rtl.c (debug_slim): New function to handle const_rtx.
5847 Call DEFINE_DEBUG_VEC for rtx_def.
5848 Call DEFINE_DEBUG_VEC for rtx_insn.
5849 Call DEFINE_DEBUG_HASH_SET for rtx_def.
5850 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
5851 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
5852 (debug (vec<rtx_insn *> *ptr): Remove.
5853 (debug_insn_vector): Remove.
5854 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
5856 2017-11-20 Tom de Vries <tom@codesourcery.com>
5858 PR rtl-optimization/82020
5859 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
5860 IF_THEN_ELSE condition.
5862 2017-11-19 Jeff Law <law@redhat.com>
5864 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
5865 of degenerates resulting from ignoring an edge.
5867 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5870 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
5872 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5875 * profile-count.c (profile_count::to_sreal_scale): Fix return value
5876 for uninitialied counts.
5878 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5881 * tree-inline.c (estimate_num_insns): Set to 1 at least.
5883 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5886 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
5888 2017-11-19 Tom de Vries <tom@codesourcery.com>
5890 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
5893 2017-11-19 Tom de Vries <tom@codesourcery.com>
5895 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
5898 2017-11-19 Tom de Vries <tom@codesourcery.com>
5900 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5903 2017-11-19 Tom de Vries <tom@codesourcery.com>
5905 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
5907 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
5909 2017-11-19 Tom de Vries <tom@codesourcery.com>
5911 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
5913 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
5914 semicolon after MCORE_EXPORT_NAME call.
5916 2017-11-19 Tom de Vries <tom@codesourcery.com>
5919 * vmsdbgout.c (vmsdbgout_early_finish): New function.
5920 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
5922 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
5924 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
5925 (cgraph_edge::clone): Cleanup updating of profile.
5926 * ipa-cp.c (update_profiling_info): Likewise.
5927 * ipa-inline-transform.c (inline_transform): Likewise.
5928 * ipa-inline.c (inline_small_functions): Add missing space to dump.
5929 * ipa-split.c (execute_split_functions): Do not split when function
5931 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
5932 * profile-count.c (profile_count::dump): Add global0.
5933 (profile_count::to_cgraph_frequency): Do not ICE when entry is
5935 (profile_count::to_sreal_scale): Likewise.
5936 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
5937 (profile_count::combine_with_ipa_count): New function.
5938 * profile-count.h (profile_guessed_global0adjusted): New.
5939 (profile_count::adjusted_zero): New.
5940 (profile_count::global0adjusted): New.
5941 (profile_count::combine_with_ipa_count): New.
5942 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
5943 correct profile of return block of split functions.
5944 (copy_cfg_body): Remove unused profile_count.
5945 (copy_body): Likewise.
5946 (expand_call_inline): Update.
5947 (tree_function_versioning): Update.
5949 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
5951 * hash-set.h (hash_set::empty): New.
5952 * tree-ssa-threadbackward.h: Delete.
5953 * tree-ssa-threadbackward.c (class thread_jumps): New.
5954 Move max_threaded_paths into class.
5955 (fsm_find_thread_path): Remove arguments that are now in class.
5956 (profitable_jump_thread_path): Rename to...
5957 (thread_jumps::profitable_jump_thread_path): ...this.
5958 (convert_and_register_jump_thread_path): Rename to...
5959 (thread_jumps::convert_and_register_current_path): ...this.
5960 (check_subpath_and_update_thread_path): Rename to...
5961 (thread_jumps::check_subpath_and_update_thread_path): ...this.
5962 (register_jump_thread_path_if_profitable): Rename to...
5963 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
5964 (handle_phi): Rename to...
5965 (thread_jumps::handle_phi): ...this.
5966 (handle_assignment): Rename to...
5967 (thread_jumps::handle_assignment): ...this.
5968 (fsm_find_control_statement_thread_paths): Rename to...
5969 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
5970 (find_jump_threads_backwards): Rename to...
5971 (thread_jumps::find_jump_threads_backwards): ...this.
5972 Initialize path local data.
5973 (pass_thread_jumps::execute): Call find_jump_threads_backwards
5974 from within thread_jumps class.
5975 (pass_early_thread_jumps::execute): Same.
5977 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5979 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
5981 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5983 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
5984 to canonical location.
5986 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5989 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
5991 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
5993 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
5995 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
5996 rcrt1.o%s/grcrt1.o%s for -static-pie.
5998 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6000 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6001 ix86_shift_rotate_cost): Break out from ...
6002 (ix86_rtx_costs): ... here.
6003 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6006 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6008 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6009 when entry block was promoted unlikely.
6010 (estimate_bb_frequencies): Increase frequency scale.
6011 * profile-count.h (profile_count): Export precision info.
6013 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6015 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6016 disturbin profile of entry block.
6018 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6020 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6023 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6025 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6026 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6028 2017-11-17 Jeff Law <law@redhat.com>
6030 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6031 from evrp_dom_walker class. Various methods moved into new class.
6032 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6033 (evrp_range_analyzer::enter): New method.
6034 (evrp_range_analyzer::leave): New method.
6035 (evrp_dom_walker): Remove delegators no longer needed by this class.
6036 Replace vr_values data member with evrp_range_analyzer
6038 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6039 method extracted from evrp_dom_walker::before_dom_children.
6040 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6041 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6043 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6044 Add private copy constructor and move assignment operators.
6045 Privatize methods and class data where trivially possible.
6046 (evrp_dom_walker::cleanup): New function, extracted from
6047 execute_early_vrp. Simplify access to class data.
6049 * vr-values.h (get_output_for_vrp): Prototype.
6050 * vr-values.c (get_output_for_vrp): New function extracted from
6051 vrp_visit_assignment_or_call and extract_range_from_stmt.
6052 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
6054 2017-11-17 Luis Machado <luis.machado@linaro.org>
6056 * config/aarch64/aarch64.c
6057 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
6058 (qdf24xx_tunings) <autoprefetcher_model>: Set to
6059 tune_params::AUTOPREFETCHER_WEAK.
6061 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6064 * config/arm/arm.c (arm_valid_target_attribute_rec):
6065 Parse "arch=" and "+<ext>".
6066 (arm_valid_target_attribute_tree): Re-init global options.
6067 (arm_option_override): Make non-static.
6068 (arm_options_perform_arch_sanity_checks): Make errors fatal.
6069 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
6070 (__ARM_FEATURE_CRC32): Support undef.
6071 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
6072 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
6074 2017-11-17 David Malcolm <dmalcolm@redhat.com>
6076 * gdbinit.in (break-on-diagnostic): New command.
6078 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6080 * config/i386/i386.c (ix86_expand_epilogue): Change simple
6081 return to indirect jump for EH return if control-flow
6082 protection is enabled. Change explicit 'false' argument in
6083 pro_epilogue_adjust_stack with a value of
6085 * config/i386/i386.md (simple_return_indirect_internal):
6086 Remove SImode restriction to support 64-bit.
6088 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6090 * combine.c (added_notes_insn): New.
6091 (try_combine): Handle added_notes_insn like added_links_insn.
6092 Rewrite return value code.
6093 (distribute_notes): Set added_notes_insn to the earliest insn we added
6096 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6098 PR rtl-optimization/82621
6099 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
6100 dest of one of those SETs is unused.
6102 2017-11-17 Richard Biener <rguenther@suse.de>
6105 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
6106 * tree-pretty-print.c (dump_generic_node): Handle
6107 annot_expr_parallel_kind.
6108 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
6109 * gimplify.c (gimple_boolify): Likewise.
6111 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6113 * config.gcc (extra_headers): Add cet.h for x86 targets.
6114 * config/i386/cet.h: New file.
6115 * doc/install.texi: Add --enable-cet/--disable-cet.
6117 2017-11-17 Richard Biener <rguenther@suse.de>
6119 PR tree-optimization/83017
6120 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
6121 (gen_parallel_loop): Properly count iterations.
6122 (parallelize_loops): Handle loop->can_be_parallel independent
6123 of flag_loop_parallelize_all. Make static profitability test match
6125 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
6126 * invoke.texi (parloops-min-per-thread): Document.
6128 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
6130 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
6131 upstreaming review comments.
6133 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6135 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
6136 with fast unaligned access.
6137 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
6139 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
6141 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
6142 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
6144 2017-11-17 Richard Biener <rguenther@suse.de>
6146 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
6147 folding of references.
6149 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
6152 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
6153 of replacing call to strncmp with corresponding call to strcmp when
6156 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
6158 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
6159 option prefer-avx256 for skylake-avx512 configuration.
6160 * config/i386/i386.c (ix86_option_override_internal): Ditto.
6161 (get_builtin_code_for_version): Ditto.
6163 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6164 Monk Chiang <sh.chiang04@gmail.com>
6166 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
6167 (FIXED_REGISTERS): Reserve more register numbers.
6168 (CALL_USED_REGISTERS): Likewise.
6169 (REG_ALLOC_ORDER): Likewise.
6170 (REG_CLASS_CONTENTS): Likewise.
6171 (REGISTER_NAMES): Likewise.
6173 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6174 Kito Cheng <kito.cheng@gmail.com>
6176 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
6178 * config/nds32/iterators.md: Add vector mode iterators and attributes.
6180 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
6182 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
6183 parameter list for vec_splats.
6185 2017-11-16 Joseph Myers <joseph@codesourcery.com>
6187 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
6189 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
6191 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
6194 * doc/install.texi: Document incompatibility of Perl >=5.6.26
6195 with the required version of automake 1.11.6.
6197 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
6199 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
6201 (DU_C2_3_power9): Correct reservation combinations.
6202 (FP_DIV_power9, VEC_DIV_power9): New.
6203 (power9-alu): Split out rotate/shift...
6204 (power9-rot): ...to here, correct dispatch resource.
6205 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
6207 (power9-fp): Correct latency.
6208 (power9-sdiv): Add div/sqrt resource.
6209 (power9-ddiv): Correct latency, add div/sqrt resource.
6210 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
6211 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
6213 (power9-qpdiv, power9-qpmul): Adjust resource usage.
6215 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
6217 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
6218 switch statement mapping KF built-ins to TF built-ins if we don't
6219 have the proper ISA 3.0 assembler support.
6221 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6223 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
6224 (lower_emutls_function_body): Do not compute it.
6226 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6228 * ipa-split.c (split_bb_info): Turn time to sreal.
6229 (split_point): Likewise.
6230 (dump_split_point): Likewise.
6231 (fine_split_points): Likewise.
6232 (execute_split_functions): Only zero split_bbs; turn time to sreals.
6234 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6236 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
6239 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6241 * predict.c (combine_predictions_for_bb): Preserve zero predicted
6243 (expensive_function_p): Remove useless assert.
6244 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
6246 2017-11-16 Martin Sebor <msebor@redhat.com>
6248 PR tree-optimization/82588
6249 PR tree-optimization/82583
6250 * tree-vrp.c (check_array_ref): Handle flexible array members,
6251 string literals, and inner indices.
6252 (search_for_addr_array): Add detail to diagnostics.
6254 2017-11-16 Nathan Sidwell <nathan@acm.org>
6258 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
6259 (SET_DECL_ASSEMBLER_NAME): Forward to
6260 overwrite_decl_assembler_name.
6261 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
6262 (overwrite_decl_assembler_name): Declare.
6263 * tree.c (overwrite_decl_assembler_name): New.
6264 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
6265 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
6266 (LANG_HOOKS_INITIALIZER): Add it.
6267 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
6268 * langhooks.c (lhd_set_decl_assembler_name): Use
6269 SET_DECL_ASSEMBLER_NAME.
6270 (lhd_overwrite_decl_assembler_name): Default implementation.
6272 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6273 Jackson Woodruff <jackson.woodruff@arm.com>
6275 PR tree-optimization/71026
6276 * match.pd: Canonicalize constant multiplies in division.
6278 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6280 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
6281 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
6282 Remove OPT_fomit_frame_pointer entry.
6283 * common/config/arc/arc-common.c: Likewise.
6284 * common/config/arm/arm-common.c: Likewise.
6285 * common/config/avr/avr-common.c: Likewise.
6286 * common/config/c6x/c6x-common.c: Likewise.
6287 * common/config/cr16/cr16-common.c: Likewise.
6288 * common/config/cris/cris-common.c: Likewise.
6289 * common/config/epiphany/epiphany-common.c: Likewise.
6290 * common/config/fr30/fr30-common.c: Likewise.
6291 * common/config/frv/frv-common.c: Likewise.
6292 * common/config/ia64/ia64-common.c: Likewise.
6293 * common/config/iq2000/iq2000-common.c: Likewise.
6294 * common/config/lm32/lm32-common.c: Likewise.
6295 * common/config/m32r/m32r-common.c: Likewise.
6296 * common/config/mcore/mcore-common.c: Likewise.
6297 * common/config/microblaze/microblaze-common.c: Likewise.
6298 * common/config/mips/mips-common.c: Likewise.
6299 * common/config/mmix/mmix-common.c: Likewise.
6300 * common/config/mn10300/mn10300-common.c: Likewise.
6301 * common/config/nios2/nios2-common.c: Likewise.
6302 * common/config/pa/pa-common.c: Likewise.
6303 * common/config/pdp11/pdp11-common.c: Likewise.
6304 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
6305 * common/config/riscv/riscv-common.c: Likewise.
6306 * common/config/rs6000/rs6000-common.c: Likewise.
6307 * common/config/rx/rx-common.c: Likewise.
6308 * common/config/s390/s390-common.c: Likewise.
6309 * common/config/sh/sh-common.c: Likewise.
6310 * common/config/sparc/sparc-common.c: Likewise.
6311 * common/config/tilegx/tilegx-common.c: Likewise.
6312 * common/config/tilepro/tilepro-common.c: Likewise.
6313 * common/config/v850/v850-common.c: Likewise.
6314 * common/config/visium/visium-common.c: Likewise.
6315 * common/config/xstormy16/xstormy16-common.c: Likewise.
6316 * common/config/xtensa/xtensa-common.c: Likewise.
6317 * invoke.texi (-fomit-frame-pointer): Update documentation.
6319 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6321 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
6323 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6325 * cfg.c (scale_bbs_frequencies_int,
6326 cale_bbs_frequencies_gcov_type): Remove.
6327 * cfg.h (scale_bbs_frequencies_int,
6328 cale_bbs_frequencies_gcov_type): Remove.
6330 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6332 * tree-ssa-loop-manip.c
6333 (scale_dominated_blocks_in_loop): Update to profile counts.
6334 (tree_transform_and_unroll_loop): Likewise.
6336 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6338 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
6339 scale_bbs_frequencies_int.
6341 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6343 * final.c (compute_alignments): Use counts rather than frequencies.
6345 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6347 * cfgloopanal.c: Include sreal.h
6348 (average_num_loop_insns): Use counts and sreal for accounting.
6350 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6352 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
6355 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6356 Kito Cheng <kito.cheng@gmail.com>
6358 * config/nds32/constraints.md: Provide more constraints.
6359 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
6360 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
6361 support constraints usage.
6363 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6365 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
6367 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6368 Kito Cheng <kito.cheng@gmail.com>
6370 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
6371 * config/nds32/nds32.opt: Refine the layout.
6372 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
6373 TARGET_EXT_STRING): Support new options.
6374 * config/nds32/nds32.h: Likewise.
6375 * config/nds32/nds32.md: Likewise.
6376 * config/nds32/nds32-predicates.c: Likewise.
6377 * config/nds32/constraints.md: Likewise.
6378 * common/config/nds32/nds32-common.c: Likewise.
6380 2017-11-16 Julia Koval <julia.koval@intel.com>
6383 * config/i386/gfniintrin.h: Add sse check.
6384 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
6386 2017-11-16 Julia Koval <julia.koval@intel.com>
6388 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
6389 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
6390 (ix86_handle_option): Handle -mavx512vbmi2.
6391 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
6392 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
6393 * config/i386/i386-c.c (__AVX512VBMI2__): New.
6394 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
6395 (ix86_valid_target_attribute_inner_p): Ditto.
6396 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
6397 * config/i386/i386.opt (mavx512vbmi2): New option.
6398 * doc/invoke.texi: Add new option.
6400 2017-11-16 Julia Koval <julia.koval@intel.com>
6402 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
6403 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
6404 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
6405 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
6406 _mm512_gf2p8mul_epi8): New intrinsics.
6407 * config/i386/i386-builtin-types.def
6408 (V64QI_FTYPE_V64QI_V64QI): New type.
6409 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
6410 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
6411 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
6412 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
6413 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
6414 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
6416 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
6418 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
6419 explicitly as a stream of bytes.
6421 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6423 * config/rs6000/altivec.h (vec_xst_be): New #define.
6424 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
6425 and externalize from *altivec_vperm_<mode>_internal.
6426 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
6428 (XL_BE_V8HI): Likewise.
6429 (XL_BE_V4SI): Likewise.
6430 (XL_BE_V4SI): Likewise.
6431 (XL_BE_V2DI): Likewise.
6432 (XL_BE_V4SF): Likewise.
6433 (XL_BE_V2DF): Likewise.
6434 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
6435 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
6436 all array entries with these keys: VSX_BUILTIN_VEC_XL,
6437 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
6438 VSX_BUILTIN_VEC_XST_BE.
6439 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
6440 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
6442 (altivec_init_builtins): Replace conditional calls to def_builtin
6443 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
6444 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
6445 with unconditional calls. Remove calls to def_builtin for
6446 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
6447 __builtin_vec_xst_be.
6448 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
6449 to define_expand, and add alternate RTL generation for P8.
6450 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
6451 vsx_ld_elemrev_v8hi.
6452 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
6453 add alternate RTL generation for P8.
6454 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
6455 vsx_ld_elemrev_v16qi.
6456 (vsx_st_elemrev_v8hi): Convert define_insn
6457 to define_expand, and add alternate RTL generation for P8.
6458 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
6459 vsx_st_elemrev_v8hi.
6460 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
6461 add alternate RTL generation for P8.
6462 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
6463 vsx_st_elemrev_v16qi.
6465 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
6468 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
6469 TARGET_AVX512ER check.
6470 (ix86_option_override_internal): Set MASK_VZEROUPPER if
6471 neither -mzeroupper nor -mno-zeroupper is used and
6472 TARGET_EMIT_VZEROUPPER is set.
6473 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
6474 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
6476 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6478 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6479 folding of vector compares.
6480 (fold_build_vec_cmp): New helper function.
6481 (fold_compare_helper): New helper function.
6482 (builtin_function_type): Add compare builtins to the list of functions
6483 having unsigned arguments. Cosmetic updates to comment indentation.
6484 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
6485 the not+eq combination.
6487 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6489 PR tree-optimization/82726
6490 PR tree-optimization/70754
6491 * tree-predcom.c (order_drefs_by_pos): New function.
6492 (combine_chains): Move code setting has_max_use_after to...
6493 (try_combine_chains): ...here. New parameter. Sort combined chains
6494 according to position information.
6495 (tree_predictive_commoning_loop): Update call to above function.
6496 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
6498 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6500 PR tree-optimization/82726
6502 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6504 PR tree-optimization/70754
6505 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6506 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6507 combined stmt before it if not NULL.
6508 (combine_chains): Process refs reversely and compute dominance point
6512 2017-02-23 Bin Cheng <bin.cheng@arm.com>
6514 PR tree-optimization/79663
6515 * tree-predcom.c (combine_chains): Process refs in reverse order
6516 only for ZERO length chains, and add explaining comment.
6518 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6520 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
6521 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
6522 * doc/invoke.texi (ARM Options): Add armv8.3-a.
6524 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6526 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
6528 2017-11-15 Martin Liska <mliska@suse.cz>
6530 * tree-cfg.c (pass_warn_function_return::execute):
6531 Compare warn_return_type for greater than zero.
6533 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
6537 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
6538 to return true on Xeon and not on Xeon Phi.
6539 (ix86_check_avx256_register): Changed to ...
6540 (ix86_check_avx_upper_register): ... this. Add extra check for
6541 VALID_AVX512F_REG_OR_XI_MODE.
6542 (ix86_avx_u128_mode_needed): Changed
6543 ix86_check_avx256_register to ix86_check_avx_upper_register.
6544 (ix86_check_avx256_stores): Changed to ...
6545 (ix86_check_avx_upper_stores): ... this. Changed
6546 ix86_check_avx256_register to ix86_check_avx_upper_register.
6547 (ix86_avx_u128_mode_after): Changed
6548 avx_reg256_found to avx_upper_reg_found. Changed
6549 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
6550 (ix86_avx_u128_mode_entry): Changed
6551 ix86_check_avx256_register to ix86_check_avx_upper_register.
6552 (ix86_avx_u128_mode_exit): Ditto.
6553 * config/i386/i386.h: (host_detect_local_cpu): New define.
6555 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6557 * config/arm/xgene1.md (xgene1): Split into automatons
6558 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
6559 (xgene1_f_load): Adjust reservations and/or types.
6560 (xgene1_f_store): Likewise.
6561 (xgene1_load_pair): Likewise.
6562 (xgene1_store_pair): Likewise.
6563 (xgene1_fp_load1): Likewise.
6564 (xgene1_load1): Likewise.
6565 (xgene1_store1): Likewise.
6566 (xgene1_move): Likewise.
6567 (xgene1_alu): Likewise.
6568 (xgene1_simd): Likewise.
6569 (xgene1_bfm): Likewise.
6570 (xgene1_neon_load1): Likewise.
6571 (xgene1_neon_store1): Likewise.
6572 (xgene1_neon_logic): Likewise.
6573 (xgene1_neon_st1): Likewise.
6574 (xgene1_neon_ld1r): Likewise.
6575 (xgene1_alu_cond): Added.
6576 (xgene1_shift_reg): Likwise.
6577 (xgene1_bfx): Likewise.
6578 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6580 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6583 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6585 (can_widen_mult_without_libcall): New function.
6586 (expand_mul_overflow): If only checking unsigned mul overflow,
6587 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6588 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6589 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6590 (expand_DIVMOD): Formatting fix.
6591 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6592 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6593 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6595 PR tree-optimization/82977
6596 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6597 constructed temporary to strlen_to_stridx.put.
6599 2017-11-15 Martin Liska <mliska@suse.cz>
6601 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6602 * configure: Regenerate.
6604 2017-11-15 Martin Liska <mliska@suse.cz>
6607 * config/sh/sh-mem.cc: Use proper probability for
6610 2017-11-14 Jeff Law <law@redhat.com>
6612 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6613 the red zone for stack_clash_protection_final_dynamic_probe targets
6614 when the total dynamic stack size is zero bytes.
6616 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6617 blocks is post order.
6619 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6621 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6622 * final.c (rest_of_clean_state): Set it for the
6623 -fcompare-debug dump.
6624 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6625 class when TDF_COMPARE_DEBUG is set.
6627 * dwarf2out.c (gen_producer_string): Discard
6630 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6633 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6634 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6635 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6636 (__TGMATH_REAL_2_3): Remove macros.
6637 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6638 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6640 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6642 (remquo): Define using __TGMATH_REAL_3.
6644 2017-11-14 Jeff Law <law@redhat.com>
6646 * vr-values.c: New file with contents extracted from tree-vrp.c.
6647 * Makefile.in (OBJS): Add vr-values.o
6648 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6649 (set_value_range, set_and_canonicalize_value_range): Likewise.
6650 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6651 (value_range_constant_singleton, symbolic_range_p): Likewise.
6652 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6653 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6654 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6655 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6656 (find_case_label_range, find_case_label_index): Likewise.
6657 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6658 (range_int_cst_singleton_p, value_inside_range): Likewise.
6659 (get_single_symbol): Likewise.
6660 (switch_update): Move structure definition here.
6661 (to_remove_edges, to_update_switch_stmts): Provide externs.
6662 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6663 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6664 (vrp_val_is_min, set_value_range): Likewise.
6665 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6666 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6667 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6668 (range_is_nonnull, range_int_cst_p): Likewwise.
6669 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6670 (get_single_symbol, operand_less_p): Likewise
6671 (compare_values_warnv, compare_values): Likewise.
6672 (value_inside_range, value_range_constant_singleton): Likewise.
6673 (zero_nonzero_bitgs_from_vr): Likewise.
6674 (extract_range_from_binary_expr_1): Likewise.
6675 (overflow_comparison_p): Likewise.
6676 (to_remove_edges, to_update_switch_stmts): Likewise.
6677 (find_case_label-index, find_case_label_range): Likewise.
6678 (switch_update, set_value_range_to_nonnegative): Remove.
6679 (set_value_range_to_truthvalue): Likewise.
6680 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6681 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6682 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6683 (find_case_label_ranges, test_for_singularity): Likewise.
6684 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6685 (x_vr_values): Move to its remaining use site.
6687 2017-11-10 Jeff Law <law@redhat.com>
6689 * vr-values.h (VR_INITIALIZER): Move #define here.
6690 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6691 * Makefile.in (OBJS): Add tree-evrp.o
6692 * tree-vrp.h (assert_info): Move structure definition here.
6693 (set_value_range_to_varying): Prototype.
6694 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6695 (infer_value_range, register_edge_assert_for): Likewise.
6696 (stmt_interesting_for_vrp): Likewise.
6697 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6698 (set_value_range_to_varying): No longer static.
6699 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6700 (infer_value_range, register_edge_assert_for): Likewise.
6702 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6704 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6705 generate the XXBRD instruction.
6707 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6708 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6709 as being compatible if -mabi=ieeelongdouble.
6710 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6711 to setup float128 built-ins with hardware support.
6712 (BU_FLOAT128_HW_2): Likewise.
6713 (BU_FLOAT128_HW_3): Likewise.
6714 (BU_FLOAT128_HW_VSX_1): Likewise.
6715 (BU_FLOAT128_HW_VSX_2): Likewise.
6716 (scalar_extract_expq): Change float128 built-in functions to
6717 accommodate having both KFmode and TFmode functions. Use the
6718 KFmode variant as the default.
6719 (scalar_extract_sigq): Likewise.
6720 (scalar_test_neg_qp): Likewise.
6721 (scalar_insert_exp_q): Likewise.
6722 (scalar_insert_exp_qp): Likewise.
6723 (scalar_test_data_class_qp): Likewise.
6724 (sqrtf128_round_to_odd): Delete processing the round to odd
6725 built-in functions as special built-in functions, and define them
6726 as float128 built-ins. Use the KFmode variant as the default.
6727 (truncf128_round_to_odd): Likewise.
6728 (addf128_round_to_odd): Likewise.
6729 (subf128_round_to_odd): Likewise.
6730 (mulf128_round_to_odd): Likewise.
6731 (divf128_round_to_odd): Likewise.
6732 (fmaf128_round_to_odd): Likewise.
6733 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6734 support for KFmode and TFmode xststdcqp calls.
6735 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6736 point, switch the built-in handlers for the get/set float128
6737 exponent, get float128 mantissa, float128 test built-ins, and the
6738 float128 round to odd built-in functions. Eliminate creating the
6739 float128 round to odd built-in functions as special built-ins.
6740 (rs6000_init_builtins): Eliminate special creation of the float128
6741 round to odd built-in functions.
6742 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6743 function insns to support both TFmode and KFmode variants.
6744 (xsxsigqp_<mode>): Likewise.
6745 (xsiexpqpf_<mode>): Likewise.
6746 (xsiexpqp_<mode>): Likewise.
6747 (xststdcqp_<mode>): Likewise.
6748 (xststdcnegqp_<mode>): Likewise.
6749 (xststdcqp_<mode>): Likewise.
6751 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6753 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6754 unused path_in_freq_ptr parameter.
6755 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6757 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6759 * ipa-inline.c (edge_badness): Dump sreal frequency.
6760 (compute_inlined_call_time): Match natural implementaiton ...
6761 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6762 forgotten division by CGRAPH_FREQ_BASE.
6764 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6766 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6767 Solaris 11. Update comment.
6768 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6770 * config/sol2.h (STARTFILE_SPEC): Likewise.
6771 * configure: Regenerate.
6773 2017-11-14 Carl Love <cel@us.ibm.com>
6775 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6776 le_ and be_ prefixes to swap* variables. Remove
6777 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6780 2017-11-14 Jason Merrill <jason@redhat.com>
6782 Support GTY((cache)) on hash_map.
6783 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6784 (ggc_cache_remove): Override it instead of ggc_mx.
6785 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6786 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6787 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6788 (simple_cache_map_traits): Override maybe_mx.
6789 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6790 (hash_map): Friend gt_cleare_cache.
6791 (gt_cleare_cache): New.
6792 * tree.h (tree_cache_traits): New hash_map traits class.
6793 (tree_cache_map): New typedef.
6795 2017-11-14 Richard Biener <rguenther@suse.de>
6797 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6798 paramter and handling.
6799 (cleanup_control_flow_bb): Likewise.
6800 (cleanup_control_flow_pre): New helper performing a DFS walk
6801 to call cleanup_control_flow_bb in PRE order.
6802 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6803 via cleanup_control_flow_pre.
6805 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6807 * config/aarch64/aarch64-simd.md
6808 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6809 (*aarch64_simd_bsl<mode>_alt): Likewise.
6810 (aarch64_simd_bsldi_internal): New.
6811 (aarch64_simd_bsldi_alt): Likewise.
6813 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6815 * tracer.c (better_p): Do not compare frequencies.
6816 * reg-stack.c (better_edge): Likewise.
6817 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
6820 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6822 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
6823 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
6824 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
6825 * lto-streamer-in.c (input_function): Use update_max_bb_count.
6826 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
6827 * predict.c (maybe_hot_frequency_p): Inline to ...
6828 (maybe_hot_count_p): ... here; rewrite to counts.
6829 (counts_to_freqs): Rename to ...
6830 (update_max_bb_count): ... this one.
6831 (expensive_function_p): Use counts.
6832 (estimate_bb_frequencies): Update.
6833 (rebuild_frequencies): Update.
6834 * predict.h (counts_to_freqs): Rename to ...
6835 (update_max_bb_count): ... this one.
6836 * profile.c (compute_branch_probabilities): Add debug info
6837 * tree-inline.c (expand_call_inline): Update debug info.
6838 (optimize_inline_calls): Use update_max_bb_count..
6839 (tree_function_versioning): Use update_max_bb_count..
6840 * value-prof.c (gimple_value_profile_transformations):
6841 Do not use update_max_bb_count.
6843 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6845 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
6846 always use frequencies.
6848 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6850 * bb-reorder.c: Remove frequencies from comments.
6851 (better_edge_p): Use profile counts.
6852 (find_traces): Dump profile counts.
6853 (rotate_loop): Use profile counts.
6854 (find_traces_1_round): Likewise.
6855 (connect_better_edge_p): Use counts instead of probabilities for
6857 (copy_bb_p): Drop early check for non-0 frequency.
6858 (sanitize_hot_paths): Update comments.
6860 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6862 * ipa-split.c (struct split_point): Add count.
6863 (consider_split): Do not compute incoming frequency; compute incoming
6864 count and store it to split_point.
6865 (split_function): Set count of the call to split part correctly.
6867 2017-11-13 Carl Love <cel@us.ibm.com>
6869 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
6871 2017-11-13 Tom Tromey <tom@tromey.com>
6873 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
6875 2017-11-13 Carl Love <cel@us.ibm.com>
6877 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
6878 Add support for builtins:
6879 unsigned int vec_first_{,miss}_match_{,or_eos}index,
6880 vector {un,}signed {char,int,short},
6881 vector {un,}signed {char,int,short}) arguments.
6882 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
6883 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
6884 Add BU_P9V_AV_2 expansions for the builtins.
6885 * config/rs6000/altivec.h (vec_first_match_index,
6886 vec_first_mismatch_index, vec_first_match_or_eos_index,
6887 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
6888 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
6889 new extern declaration.
6890 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
6891 * config/rs6000/vsx.md (first_match_index_<mode>,
6892 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
6893 first_mismatch_or_eos_index_<mode>): Add define expand.
6894 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
6895 * doc/extend.texi: Update the built-in documenation file for the new
6898 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
6900 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
6901 into the min/max operations for _Float<N> and _Float<N>X types.
6903 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6906 * dwarf2out.c (do_eh_frame): New static variable.
6907 (dwarf2out_begin_prologue): Set it.
6908 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
6910 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6912 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
6914 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
6916 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6919 * debug.h (dwarf2out_do_eh_frame): Declare.
6920 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
6921 (dwarf2out_do_frame): Use it.
6922 (dwarf2out_do_cfi_asm): Likewise.
6923 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
6924 (dwarf2out_assembly_start): Likewise.
6925 (dwarf2out_begin_prologue): Fix comment.
6926 * toplev.c (compile_file): Always call dwarf2out_frame_finish
6927 if the target needs either debug or unwind DWARF2 info.
6928 * lto-opts.c (lto_write_options): Do not save -fexceptions,
6929 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
6930 -ftrapping-math, -ftrapv and -fwrapv.
6932 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6934 * cgraph.c (cgraph_edge::sreal_frequency): New function.
6935 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
6936 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
6937 (estimate_edge_size_and_time): Likewise.
6938 (ipa_merge_fn_summary_after_inlining): Likewise.
6939 * ipa-inline.c (cgraph_freq_base_rec): Remove.
6940 (compute_uninlined_call_time): Use sreal_frequency.
6941 (compute_inlined_call_time): Likewise.
6942 (ipa_inline): Do not initialize cgraph_freq_base_rec.
6943 * profile-count.c: Include sreal.h.
6944 (profile_count::to_sreal_scale): New.
6945 * profile-count.h: Forward declare sreal.
6946 (profile_count::to_sreal_scale): Declare.
6948 2017-11-13 Nathan Sidwell <nathan@acm.org>
6950 * diagnostic.c (maybe_line_and_column): New.
6951 (diagnostic_get_location_text): Use it.
6952 (diagnostic_report_current_module): Likewise.
6953 (test_diagnostic_get_location_text): Add tests.
6955 2017-11-13 Luis Machado <luis.machado@linaro.org>
6957 * doc/md.texi (Specifying processor pipeline description): Fix
6958 incorrect latency for the div instruction example.
6960 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6962 PR tree-optimization/78821
6963 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
6964 that bit_not_p is the same.
6965 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
6966 (split_group): Count precisely bit_not_p bits in each statement.
6967 (invert_op): New function.
6968 (imm_store_chain_info::output_merged_store): Use invert_op to
6969 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
6970 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
6972 2017-11-13 Martin Liska <mliska@suse.cz>
6974 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
6975 (struct source_info): Likewise.
6976 (add_branch_counts): Likewise.
6977 (add_line_counts): Likewise.
6978 (function_summary): Likewise.
6979 (output_intermediate_line): Likewise.
6980 (generate_results): Likewise.
6982 2017-11-13 Martin Liska <mliska@suse.cz>
6984 * gcov.c (struct block_info): Remove typedef for block_t.
6985 (struct line_info): Likewise.
6986 (line_info::has_block): Likewise.
6987 (EXIT_BLOCK): Likewise.
6988 (unblock): Likewise.
6989 (circuit): Likewise.
6990 (get_cycles_count): Likewise.
6991 (process_file): Likewise.
6992 (read_graph_file): Likewise.
6993 (solve_flow_graph): Likewise.
6994 (find_exception_blocks): Likewise.
6995 (add_line_counts): Likewise.
6996 (accumulate_line_info): Likewise.
6997 (output_line_details): Likewise.
6999 2017-11-13 Martin Liska <mliska@suse.cz>
7001 * gcov.c (struct arc_info): Remove typedef for arc_t.
7002 (struct line_info): Likewise.
7003 (add_branch_counts): Likewise.
7004 (output_branch_count): Likewise.
7005 (function_info::~function_info): Likewise.
7006 (circuit): Likewise.
7007 (output_intermediate_line): Likewise.
7008 (read_graph_file): Likewise.
7009 (solve_flow_graph): Likewise.
7010 (find_exception_blocks): Likewise.
7011 (add_line_counts): Likewise.
7012 (accumulate_line_info): Likewise.
7013 (output_line_details): Likewise.
7014 (output_function_details): Likewise.
7016 2017-11-13 Martin Liska <mliska@suse.cz>
7018 * gcov.c (struct function_info): Remove typedef for function_t.
7019 (struct source_info): Likewise.
7020 (source_info::get_functions_at_location): Likewise.
7021 (solve_flow_graph): Likewise.
7022 (find_exception_blocks): Likewise.
7023 (add_line_counts): Likewise.
7024 (output_intermediate_file): Likewise.
7025 (process_file): Likewise.
7026 (generate_results): Likewise.
7027 (release_structures): Likewise.
7028 (read_graph_file): Likewise.
7029 (read_count_file): Likewise.
7030 (accumulate_line_counts): Likewise.
7031 (output_lines): Likewise.
7033 2017-11-13 Martin Liska <mliska@suse.cz>
7035 * gcov.c (function_info::function_info): Remove num_counts
7036 and add vector<gcov_type>.
7037 (function_info::~function_info): Use the vector.
7038 (process_file): Likewise.
7039 (read_graph_file): Likewise.
7040 (read_count_file): Likewise.
7041 (solve_flow_graph): Likewise.
7043 2017-11-13 Martin Liska <mliska@suse.cz>
7045 * gcov.c (function_info::is_artificial): New function.
7046 (process_file): Erase all artificial early.
7047 (generate_results): Skip as all artificial are already
7050 2017-11-13 Martin Liska <mliska@suse.cz>
7052 * gcov.c (read_graph_file): Store to global vector of functions.
7053 (read_count_file): Iterate the vector.
7054 (process_file): Likewise.
7055 (generate_results): Likewise.
7056 (release_structures): Likewise.
7058 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7060 PR tree-optimization/82954
7061 * gimple-ssa-store-merging.c
7062 (imm_store_chain_info::coalesce_immediate_stores): If
7063 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
7065 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
7067 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7068 Upddate call to ENDIAN_LANE_N.
7069 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
7070 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
7071 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
7072 and use aarch64_endian_lane_rtx.
7073 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
7075 2017-11-12 Tom de Vries <tom@codesourcery.com>
7077 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
7079 2017-11-12 Tom de Vries <tom@codesourcery.com>
7081 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
7083 2017-11-12 Tom de Vries <tom@codesourcery.com>
7085 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
7087 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7088 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7089 * defaults.h (ASM_OUTPUT_LABELREF): Same.
7091 2017-11-11 Martin Sebor <msebor@redhat.com>
7094 * doc/extend.texi (attribute nonstring): Remove spurious argument.
7097 * prefic.c (translate_name): Replace strncpy with memcpy to
7098 avoid -Wstringop-truncation.
7100 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7102 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
7104 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7106 * predict.c (maybe_hot_frequency_p): Do not use cfun.
7108 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7110 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
7113 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7115 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
7116 enable generating XXBRH if the value is in a vector register.
7117 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
7118 value is in a vector register.
7119 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
7120 register to register bswap64's instead of doing the GPR sequence
7121 used on previous machines.
7122 (bswapdi2_xxbrd): New insn.
7123 (bswapdi2_reg): Disallow on ISA 3.0.
7124 (register to register bswap64 splitter): Do not split the insn on
7125 ISA 3.0 systems that use XXBRD.
7127 2017-11-10 Martin Sebor <msebor@redhat.com>
7130 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
7132 (find_subframework_file): Same.
7134 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7136 * auto-profile.c (afdo_indirect_call): Drop frequency.
7137 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
7138 (cgraph_node::create_edge): Drop frequency argument.
7139 (cgraph_node::create_indirect_edge): Drop frequency argument.
7140 (cgraph_edge::make_speculative): Drop frequency arguments.
7141 (cgraph_edge::resolve_speculation): Do not update frequencies
7142 (cgraph_edge::dump_edge_flags): Do not dump frequency.
7143 (cgraph_node::dump): Check consistency in IPA mode.
7144 (cgraph_edge::maybe_hot_p): Use IPA counter.
7145 (cgraph_edge::verify_count_and_frequency): Rename to ...
7146 (cgraph_edge::verify_count): ... this one; drop frequency checking.
7147 (cgraph_node::verify_node): Update.
7148 * cgraph.h (struct cgraph_edge): Drop frequency.
7149 (cgraph_edge::frequency): New function.
7150 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
7152 (cgraph_edge::rebuild_edges): Likewise.
7153 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
7154 (duplicate_thunk_for_node): Do not pass frequency.
7155 (cgraph_node::create_clone): Scale only counts.
7156 (cgraph_node::create_virtual_clone): Do not pass frequency.
7157 (cgraph_node::create_edge_including_clones): Do not pass frequency.
7158 (cgraph_node::create_version_clone): Do not pass frequency.
7159 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
7160 (cgraph_node::expand_thunk): Do not pass frequency.
7161 (cgraph_node::create_wrapper): Do not pass frequency.
7162 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
7164 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
7165 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
7166 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
7167 (gather_caller_stats): Use frequency function.
7168 (ipcp_cloning_candidate_p): Use frequency function.
7169 (ipcp_propagate_stage): Use frequency function.
7170 (get_info_about_necessary_edges): Use frequency function.
7171 (update_profiling_info): Update only IPA profile.
7172 (update_specialized_profile): Use frequency functoin.
7173 (perhaps_add_new_callers): Update only IPA profile.
7174 * ipa-devirt.c (ipa_devirt): Use IPA profile.
7175 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
7176 (dump_ipa_call_summary): Use frequency function.
7177 (estimate_edge_size_and_time): Use frequency function.
7178 (ipa_merge_fn_summary_after_inlining): Use frequency function.
7179 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
7180 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
7181 (update_noncloned_counts): ... ths one; scale counts only.
7182 (clone_inlined_nodes): Do not scale frequency.
7183 (inline_call): Do not pass frequency.
7184 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
7185 (compute_inlined_call_time): Use IPA profile.
7186 (want_inline_small_function_p): Use IPA profile.
7187 (want_inline_self_recursive_call_p): Use IPA profile.
7188 (edge_badness): Use IPA profile.
7189 (lookup_recursive_calls): Use IPA profile.
7190 (recursive_inlining): Do not pass frequency.
7191 (resolve_noninline_speculation): Do not update frequency.
7192 (inline_small_functions): Collect max of IPA profile.
7193 (dump_overall_stats): Dump IPA porfile.
7194 (dump_inline_stats): Dump IPA porfile.
7195 (ipa_inline): Collect IPA stats.
7196 * ipa-inline.h (clone_inlined_nodes): Update prototype.
7197 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
7198 (ipa_propagate_frequency): Use frequency function.
7199 (ipa_profile): Cleanup.
7200 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
7201 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
7202 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
7203 (input_node): Do not stream frequency.
7204 (input_edge): Do not stream frequency.
7205 (merge_profile_summaries): Scale only IPA profiles.
7206 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
7207 * predict.c (drop_profile): Do not recompute frequency.
7208 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
7209 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
7210 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
7211 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
7212 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
7213 * tree-inline.c (copy_bb): Do not scale frequency.
7214 (expand_call_inline): Do not scale frequency.
7215 (tree_function_versioning): Do not scale frequency.
7216 * ubsan.c (ubsan_create_edge): Do not pass frequency.
7218 2017-11-10 Julia Koval <julia.koval@intel.com>
7220 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
7221 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
7222 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
7223 (_mm256_maskz_gf2p8affine_epi64_epi8)
7224 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
7225 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
7226 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
7227 (__builtin_ia32_vgf2p8affineqb_v32qi)
7228 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
7229 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
7231 2017-11-10 Tamar Christina <tamar.christina@arm.com>
7235 (arm_option_override): Refactor.
7236 (arm_option_reconfigure_globals): New.
7237 (arm_options_perform_arch_sanity_checks): New.
7238 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
7240 (arm_options_perform_arch_sanity_checks): Likewise
7242 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
7244 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
7245 (power9-qpmul): New.
7246 * rs6000/rs6000.md ("type" attr): Add qmul.
7247 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
7248 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
7249 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
7251 2017-11-10 Martin Sebor <msebor@redhat.com>
7254 * builtins.c (compute_objsize): Handle arrays that
7255 compute_builtin_object_size likes to fail for. Make extern.
7256 * builtins.h (compute_objsize): Declare.
7257 (check_strncpy_sizes): New function.
7258 (expand_builtin_strncpy): Call check_strncpy_sizes.
7259 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
7260 -Wstringop-truncation.
7261 (gimple_fold_builtin_strncat): Same.
7262 * gimple.c (gimple_build_call_from_tree): Set call location.
7263 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
7264 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
7265 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
7266 (handle_builtin_strlen): Use strlen_to_stridx.
7267 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
7269 Use strlen_to_stridx.
7270 (pass_strlen::execute): Release strlen_to_stridx.
7271 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
7272 (-Wstringop-truncation): Document new option.
7274 2017-11-10 Martin Liska <mliska@suse.cz>
7276 PR gcov-profile/82702
7277 * gcov.c (main): Handle intermediate files in a different
7279 (get_gcov_intermediate_filename): New function.
7280 (output_gcov_file): Remove support of intermediate files.
7281 (generate_results): Allocate intermediate file.
7282 (release_structures): Clean-up properly fn_end.
7283 (output_intermediate_file): Start iterating with line 1.
7285 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7287 PR tree-optimization/82929
7288 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7289 ops_swapped_p non-static data member.
7290 (store_immediate_info::store_immediate_info): Clear it.
7291 (imm_store_chain_info::coalesce_immediate_stores): If swapping
7292 ops set ops_swapped_p.
7293 (count_multiple_uses): Handle ops_swapped_p.
7295 2017-11-10 Martin Liska <mliska@suse.cz>
7297 * coverage.c (coverage_init): Stream information about
7298 support of has_unexecuted_blocks.
7299 * doc/gcov.texi: Document that.
7300 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
7301 * gcov.c (read_graph_file): Likewise.
7302 (output_line_beginning): Fix a small issue with
7305 2017-11-10 Bin Cheng <bin.cheng@arm.com>
7307 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
7308 reference of trivial component.
7310 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7313 * gimple-ssa-store-merging.c
7314 (pass_store_merging::terminate_all_aliasing_chains): For
7315 gimple_store_p stmts also call refs_output_dependent_p.
7317 PR rtl-optimization/82913
7318 * compare-elim.c (try_merge_compare): Punt if def_insn is not
7321 2017-11-09 Jeff Law <law@redhat.com>
7323 * vr-values.h: New file with vr_values class.
7324 * tree-vrp.c: Include vr-values.h
7325 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
7326 data objects into the vr_values class.
7327 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
7328 (get_value_range): Make it a member function within vr_values class.
7329 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
7330 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
7331 (op_with_constant_singleton_value_range): Likewise.
7332 (extract_range_for_var_from_comparison_expr): Likewise.
7333 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
7334 (extract_range_from_binary_expr): Likewise.
7335 (extract_range_from_unary_expr): Likewise.
7336 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
7337 (check_for_binary_op_overflow, extract_range_basic): Likewise.
7338 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
7339 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
7340 (compare_name_with_value, compare_names): Likewise.
7341 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
7342 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
7343 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
7344 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
7345 (extract_range_from_phi_node): Likewise.
7346 (simplify_truth_ops_using_ranges): Likewise.
7347 (simplify_div_or_mod_using_ranges): Likewise.
7348 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
7349 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
7350 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
7351 (simplify_float_conversion_using_ranges): Likewise.
7352 (simplify_internal_call_using_ranges): Likewise.
7353 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
7354 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
7355 poitner via x_vr_values for calls into gimple folder.
7356 (vrp_initialize_lattice): Make this the vr_values ctor.
7357 (vrp_free_lattice): Make this the vr_values dtor.
7358 (set_vr_value): New function.
7359 (class vrp_prop): Add vr_values data member. Add various member
7360 functions as well as member functions that delegate to vr_values.
7361 (check_array_ref): Make a member function within vrp_prop class.
7362 (search_for_addr_array, vrp_initialize): Likewise.
7363 (vrp_finalize): Likewise. Revamp to avoid direct access to
7364 vr_value, values_propagated, etc.
7365 (check_array_bounds): Extract vrp_prop class instance pointer from
7366 walk info structure. Use it to call member functions.
7367 (check_all_array_refs): Make a member function within vrp_prop class.
7368 Smuggle class instance pointer via walk info structure.
7369 (x_vr_values): New local static.
7370 (vrp_valueize): Use x_vr_values to get class instance.
7371 (vr_valueize_1): Likewise.
7372 (class vrp_folder): Add vr_values data member. Add various member
7373 functions as well as member functions that delegate to vr_values.
7374 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
7375 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
7376 class instance from vr_values. Use it to call member functions.
7377 (vrp_dom_walker): Add vr_values data member.
7378 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
7379 instance via x_vr_values.
7380 (identify_jump_threads): Accept vr_values as argument. Store
7381 it into the walker structure.
7382 (evrp_dom_walker): Add vr_values class data member. Add various
7384 (evrp_dom_walker::try_find_new_range): Use vr_values data
7385 member to access the memory allocator.
7386 (evrp_dom_walker::before_dom_children): Store vr_values class
7387 instance into the vrp_folder class.
7388 (evrp_dom_walker::push_value_range): Rework to avoid direct
7389 access to num_vr_values and vr_value.
7390 (evrp_dom_walker::pop_value_range): Likewise.
7391 (execute_early_vrp): Remove call to vrp_initialize_lattice.
7392 Use vr_values to get to dump_all_value_ranges member function.
7393 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
7394 and simplify_cond_using_ranges_2 via vrp_prop class instance.
7395 Pass vr_values class instance down to identify_jump_threads.
7396 Remove call to vrp_free_lattice.
7397 (debug_all_value_ranges): Remove.
7399 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
7400 (vrp_folder): Likewise.
7402 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
7403 Get it from the existing bitmap instead.
7404 (vrp_intersect_ranges_1): Likewise.
7406 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7408 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7410 (store_immediate_info::store_immediate_info): Add bitnotp argument,
7411 set bit_not_p to it.
7412 (imm_store_chain_info::coalesce_immediate_stores): Break group
7413 if bit_not_p is different.
7414 (count_multiple_uses, split_group,
7415 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
7416 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
7417 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
7418 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
7419 to store_immediate_info ctor.
7421 2017-11-09 Jim Wilson <jimw@sifive.com>
7423 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
7424 (scan_prog_file): Likewise.
7426 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7428 * bb-reorder.c (max_entry_frequency): Remove.
7429 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
7430 connect_traces, push_to_next_round_p): Remove prototypes.
7431 (find_traces_1_round): Use counts only.
7432 (push_to_next_round_p): Likewise.
7433 (find_traces): Likewise.
7434 (rotate_loop): Likewise.
7435 (find_traces_1_round): Likewise.
7436 (connect_traces): Likewise.
7437 (edge_order): Likewise.
7439 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
7441 * config/arm/arm.c (output_return_instruction): Add comments to
7442 indicate requirement for cmse_nonsecure_entry return to account
7443 for the size of clearing instruction output here.
7444 (thumb_exit): Likewise.
7445 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
7446 return in hardfloat mode.
7448 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
7450 * config/rs6000/rs6000.c (machine_function): Add a bool,
7451 "toc_is_wrapped_separately".
7452 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
7453 if it wasn't explicitly set or unset, we are optimizing for speed, and
7454 doing separate shrink-wrapping.
7455 (rs6000_get_separate_components): Enable the TOC component if
7456 saving the TOC register in the prologue.
7457 (rs6000_components_for_bb): Handle the TOC component.
7458 (rs6000_emit_prologue_components): Store the TOC register where needed.
7459 (rs6000_set_handled_components): Mark TOC as handled, if handled.
7460 (rs6000_emit_prologue): Don't save the TOC if that is already done.
7462 2017-11-09 Martin Jambor <mjambor@suse.cz>
7464 * ipa-param-manipulation.c: New file.
7465 * ipa-param-manipulation.h: Likewise.
7466 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
7467 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
7468 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
7469 (ipa_parm_adjustment): Likewise.
7470 (ipa_parm_adjustment_vec): Likewise.
7471 (ipa_get_vector_of_formal_parms): Moved declaration to
7472 ipa-param-manipulation.h.
7473 (ipa_get_vector_of_formal_parm_types): Likewise.
7474 (ipa_modify_formal_parameters): Likewise.
7475 (ipa_modify_call_arguments): Likewise.
7476 (ipa_combine_adjustments): Likewise.
7477 (ipa_dump_param_adjustments): Likewise.
7478 (ipa_modify_expr): Likewise.
7479 (ipa_get_adjustment_candidate): Likewise.
7480 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
7481 ipa-param-manipulation.c.
7482 (ipa_get_vector_of_formal_parm_types): Likewise.
7483 (ipa_modify_formal_parameters): Likewise.
7484 (ipa_modify_call_arguments): Likewise.
7485 (ipa_modify_expr): Likewise.
7486 (get_ssa_base_param): Likewise.
7487 (ipa_get_adjustment_candidate): Likewise.
7488 (index_in_adjustments_multiple_times_p): Likewise.
7489 (ipa_combine_adjustments): Likewise.
7490 (ipa_dump_param_adjustments): Likewise.
7491 * tree-sra.c: Also include ipa-param-manipulation.h
7492 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
7495 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7496 Alan Hayward <alan.hayward@arm.com>
7497 David Sherwood <david.sherwood@arm.com>
7499 * doc/sourcebuild.texi (vect_masked_store): Document.
7501 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7502 Alan Hayward <alan.hayward@arm.com>
7503 David Sherwood <david.sherwood@arm.com>
7505 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
7507 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7508 Alan Hayward <alan.hayward@arm.com>
7509 David Sherwood <david.sherwood@arm.com>
7511 * doc/sourcebuild.texi (vect_variable_length): Document.
7513 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7514 Alan Hayward <alan.hayward@arm.com>
7515 David Sherwood <david.sherwood@arm.com>
7517 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
7519 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7520 Alan Hayward <alan.hayward@arm.com>
7521 David Sherwood <david.sherwood@arm.com>
7523 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
7525 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7526 Alan Hayward <alan.hayward@arm.com>
7527 David Sherwood <david.sherwood@arm.com>
7529 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
7530 previously undocumented selectors.
7531 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
7533 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7535 * doc/rtl.texi (const_vector): Say that elements can be
7536 const_wide_ints too.
7537 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
7538 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
7539 (gen_vec_duplicate): Use it instead of CONSTANT_P.
7540 * optabs.c (expand_vector_broadcast): Likewise.
7542 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7543 Alan Hayward <alan.hayward@arm.com>
7544 David Sherwood <david.sherwood@arm.com>
7546 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
7547 scaled index even if the unscaled address was invalid.
7548 Don't increase the complexity of using a scale in that case.
7550 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7551 Alan Hayward <alan.hayward@arm.com>
7552 David Sherwood <david.sherwood@arm.com>
7554 * doc/rtl.texi: Rewrite the subreg rules so that they partition
7555 the inner register into REGMODE_NATURAL_SIZE bytes rather than
7556 UNITS_PER_WORD bytes.
7557 * emit-rtl.c (validate_subreg): Divide subregs into blocks
7558 based on REGMODE_NATURAL_SIZE of the inner mode.
7559 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
7560 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
7561 * expmed.c (lowpart_bit_field_p): Divide the value up into
7562 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
7563 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
7564 whether something is likely to occupy more than one register.
7566 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7569 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7570 profile_count::adjust_for_ipa_scaling.
7571 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7572 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7574 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7576 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7578 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7579 (split_group): Add total_orig and total_new arguments, estimate the
7580 number of statements related to the store group without store merging
7581 and with store merging.
7582 (imm_store_chain_info::output_merged_store): Adjust split_group
7583 callers, punt if estimated number of statements with store merging
7584 is not smaller than estimated number of statements without it.
7586 (handled_load): Remove has_single_use checks.
7587 (pass_store_merging::process_store): Likewise.
7589 2017-11-09 Richard Biener <rguenther@suse.de>
7591 PR tree-optimization/82902
7592 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7594 2017-11-09 Martin Liska <mliska@suse.cz>
7597 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7600 2017-11-09 Martin Liska <mliska@suse.cz>
7602 PR tree-optimization/82669
7603 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7605 2017-11-09 Martin Liska <mliska@suse.cz>
7607 PR gcov-profile/48463
7608 * coverage.c (coverage_begin_function): Output also end locus
7609 of a function and information whether the function is
7611 * gcov-dump.c (tag_function): Parse and print the information.
7612 * gcov.c (INCLUDE_MAP): Add include.
7613 (INCLUDE_SET): Likewise.
7614 (struct line_info): Move earlier in the source file because
7615 of vector<line_info> in function_info structure.
7616 (line_info::line_info): Likewise.
7617 (line_info::has_block): Likewise.
7618 (struct source_info): Add new member index.
7619 (source_info::get_functions_at_location): New function.
7620 (function_info::group_line_p): New function.
7621 (output_intermediate_line): New function.
7622 (output_intermediate_file): Use the mentioned function.
7623 (struct function_start): New.
7624 (struct function_start_pair_hash): Likewise.
7625 (process_file): Add code that identifies group functions.
7626 Assign lines either to global or function scope.
7627 (generate_results): Skip artificial functions.
7628 (find_source): Assign index for each source file.
7629 (read_graph_file): Read new flag artificial and end_line.
7630 (add_line_counts): Assign it either to global of function scope.
7631 (accumulate_line_counts): Isolate core of the function to
7632 accumulate_line_info and call it for both function and global
7634 (accumulate_line_info): New function.
7635 (output_line_beginning): Fix GNU coding style.
7636 (print_source_line): New function.
7637 (output_line_details): Likewise.
7638 (output_function_details): Likewise.
7639 (output_lines): Iterate both source (global) scope and function
7641 (struct function_line_start_cmp): New class.
7642 * doc/gcov.texi: Reflect changes in documentation.
7644 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7647 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7648 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7649 and similarly for not instead of neg.
7651 2017-11-08 Andi Kleen <ak@linux.intel.com>
7653 * config/i386/i386.opt: Add -mforce-indirect-call.
7654 * config/i386/predicates.md: Check for flag_force_indirect_call.
7655 * doc/invoke.texi: Document -mforce-indirect-call
7657 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7659 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7661 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7662 config/riscv/riscv.c (predict.h): New include.
7663 (riscv_slow_unaligned_access_p): No longer static.
7664 (riscv_block_move_straight): Add require.
7665 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7667 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7670 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7671 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7672 nonimmediate_operand predicate for operand 1 instead of
7675 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7677 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7679 * config/aarch64/constraints.md (Uml): New constraint.
7680 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7683 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7685 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7686 of two vec_duplicates into a vec_concat.
7688 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7690 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7691 Simplify vec_merge of vec_duplicate and vec_concat.
7692 * config/aarch64/constraints.md (Utq): New constraint.
7693 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7696 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7698 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7699 Simplify vec_merge of vec_duplicate and const_vector.
7700 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7702 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7703 mode iterator. Update predicate on operand 1 to
7704 handle non-const_vec constants. Delete constraints.
7705 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7707 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7709 PR tree-optimization/78821
7710 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7712 (store_operand_info::store_operand_info): Initialize it to false.
7713 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7714 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7715 store in the group, and if chain_info is non-NULL, to ignore altogether
7717 (compatible_load_p): Fail if bit_not_p does not match.
7718 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7719 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7720 (pass_store_merging::process_store): Adjust
7721 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7722 call terminate_all_aliasing_chains newly when adding a store into
7723 a chain with non-NULL chain_info.
7725 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7727 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7729 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7731 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7733 (aarch64_layout_frame): Initialise emit_frame_chain.
7734 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7735 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7737 2017-11-08 Martin Liska <mliska@suse.cz>
7739 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7740 of gimple_call_internal_p.
7742 2017-11-07 Tom de Vries <tom@codesourcery.com>
7744 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7746 2017-11-07 Tom de Vries <tom@codesourcery.com>
7748 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7751 2017-11-08 Martin Liska <mliska@suse.cz>
7754 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7756 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7758 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7760 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7763 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7764 (*cmp<mode>_ccz_1): New insn with $k alternative.
7767 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7768 RTX_COMM_COMPARE as commutative as well.
7769 (ix86_binary_operator_ok): Formatting fix.
7770 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7771 *<code><mode>3<mask_name><round_saeonly_name>,
7772 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7773 *<s>mul<mode>3_highpart<mask_name>,
7774 *vec_widen_umult_even_v16si<mask_name>,
7775 *vec_widen_umult_even_v8si<mask_name>,
7776 *vec_widen_umult_even_v4si<mask_name>,
7777 *vec_widen_smult_even_v16si<mask_name>,
7778 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7779 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7780 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7781 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7782 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7783 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7784 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7785 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7786 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7787 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7788 ix86_binary_operator_ok. Formatting fixes.
7789 (*<plusminus_insn><mode>3<mask_name><round_name>,
7790 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7793 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7795 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7796 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7797 eq and ne if TARGET_ISEL.
7798 (cmp): New code_iterator.
7799 (UNS, UNSU_, UNSIK): New code_attrs.
7800 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7801 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7803 ("eq<mode>3"): ... this.
7804 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7806 ("ne<mode>3"): ... this.
7808 2017-11-07 Julia Koval <julia.koval@intel.com>
7811 * common/config/i386/i386-common.c
7812 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
7813 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
7814 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
7815 * config/i386/i386.opt: Ditto.
7816 * config/i386/i386.c (ix86_target_string): Ditto.
7817 (ix86_option_override_internal): Ditto.
7818 (ix86_init_mpx_builtins): Move MPX to args2.
7819 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
7820 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7821 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7822 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7823 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7824 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7825 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
7827 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
7830 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
7831 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
7832 (zero-extendsidi peephole2): Remove peephole.
7834 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7837 * stor-layout.c: Include attribs.h.
7838 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
7839 explicit lookup of "aligned" attribute.
7841 2017-11-07 Andrew Waterman <andrew@sifive.com>
7843 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
7844 (riscv_expand_block_move): Likewise.
7845 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
7847 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
7848 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
7849 * config/riscv/riscv.c (riscv_block_move_straight): New function.
7850 (riscv_adjust_block_mem): Likewise.
7851 (riscv_block_move_loop): Likewise.
7852 (riscv_expand_block_move): Likewise.
7853 * config/riscv/riscv.md (movmemsi): New pattern.
7855 2017-11-07 Michael Clark <michaeljclark@mac.com>
7857 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
7858 (MUSL_DYNAMIC_LINKER): Likewise.
7860 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
7862 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
7863 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
7866 2017-11-07 Richard Biener <rguenther@suse.de>
7868 * match.pd: Fix build.
7870 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
7871 Jackson Woodruff <jackson.woodruff@arm.com>
7873 PR tree-optimization/71026
7874 * match.pd: Canonicalize negate in division.
7876 2017-11-07 Sudakshina Das <sudi.das@arm.com>
7879 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
7881 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7883 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
7884 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
7886 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7888 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
7889 non-scalar integral types.
7890 * match.pd (negate_expr_p): Handle MINUS_EXPR.
7891 (-(A-B), -(~A)): New transformations.
7893 2017-11-07 Tom de Vries <tom@codesourcery.com>
7895 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
7896 semicolon after "do {} while (0)".
7897 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7898 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7899 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7900 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7901 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7903 2017-11-07 Tom de Vries <tom@codesourcery.com>
7905 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
7906 after "do {} while (0)".
7907 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7908 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7909 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7910 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7911 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7913 2017-11-07 Tom de Vries <tom@codesourcery.com>
7916 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
7918 (arm_rtx_costs_internal): Add missing semicolon after
7919 HANDLE_NARROW_SHIFT_ARITH call.
7921 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7923 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7924 disable isel if it was not set explicitly.
7926 2017-11-06 James Bowman <james.bowman@ftdichip.com>
7928 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
7929 (add_type_attribute) likewise.
7931 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
7933 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
7934 of struct ix86_frame.
7935 (ix86_initial_elimination_offset): Likewise.
7936 (ix86_expand_split_stack_prologue): Likewise.
7938 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
7940 * tree-vrp.h (enum value_range_type): Update stale comment.
7942 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7943 Alan Hayward <alan.hayward@arm.com>
7944 David Sherwood <david.sherwood@arm.com>
7946 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
7947 (aarch64_expand_vec_perm_const): Take the number of units too.
7948 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
7949 (aarch64_expand_vec_perm_const): Likewise.
7950 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
7951 (vec_perm<mode>): Update accordingly.
7953 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7954 Alan Hayward <alan.hayward@arm.com>
7955 David Sherwood <david.sherwood@arm.com>
7957 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
7958 Take the number of units too.
7959 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
7960 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
7961 but check for a vector mode before rather than after the call.
7962 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7963 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
7964 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
7965 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
7966 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
7967 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
7968 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
7969 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
7970 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
7971 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
7972 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
7973 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
7974 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
7975 (aarch64_sqdmull2_n<mode>): Update accordingly.
7977 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7978 Alan Hayward <alan.hayward@arm.com>
7979 David Sherwood <david.sherwood@arm.com>
7981 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
7982 the number of units too.
7983 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
7984 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
7985 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
7986 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
7987 (vec_store_lanesxi<mode>): Update accordingly.
7989 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7990 Alan Hayward <alan.hayward@arm.com>
7991 David Sherwood <david.sherwood@arm.com>
7993 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
7994 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
7995 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
7996 of units rather than the mode.
7997 * config/aarch64/iterators.md (nunits): New mode attribute.
7998 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
7999 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8000 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8001 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8002 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8003 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8004 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8005 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8006 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8007 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8008 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8009 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8010 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8011 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8012 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8013 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8014 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8015 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8016 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8017 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8018 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8019 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8020 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8021 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8022 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8023 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8024 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8025 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8026 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8027 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8028 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8029 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8030 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8031 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8032 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8033 (aarch64_simd_vec_setv2di): Likewise.
8035 2017-11-06 Carl Love <cel@us.ibm.com>
8037 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8039 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8040 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8042 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8044 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8045 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8046 Add power 8 macro expansions.
8047 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8048 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8049 power 8 instructions. (VSX_XXBR): Add iterator.
8051 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8053 * config/arm/arm.md (predicable_short_it): Change default to "no",
8054 improve documentation, remove uses that are identical to the default.
8055 (enabled_for_depr_it): Rename to enabled_for_short_it.
8056 * gcc/config/arm/arm-fixed.md (predicable_short_it):
8057 Remove default uses.
8058 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
8059 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
8060 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
8061 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
8063 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
8066 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
8067 float128 helper macros, which are no longer used after deleting
8068 the old 'q' built-in functions, and moving the round to odd
8069 built-in functions to being special built-in functions.
8070 (BU_FLOAT128_2): Likewise.
8071 (BU_FLOAT128_1_HW): Likewise.
8072 (BU_FLOAT128_2_HW): Likewise.
8073 (BU_FLOAT128_3_HW): Likewise.
8074 (FABSQ): Delete old 'q' built-in functions.
8075 (COPYSIGNQ): Likewise.
8076 (SQRTF128_ODD): Move round to odd built-in functions to be
8077 special built-in functions, so that we can handle
8078 -mabi=ieeelongdouble.
8079 (TRUNCF128_ODD): Likewise.
8080 (ADDF128_ODD): Likewise.
8081 (SUBF128_ODD): Likewise.
8082 (MULF128_ODD): Likewise.
8083 (DIVF128_ODD): Likewise.
8084 (FMAF128_ODD): Likewise.
8085 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
8086 built-in names to 'f128'.
8087 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
8088 old 'q' built-in functions, as the machine independent code for
8089 'f128' built-in functions handles this.
8090 (rs6000_expand_builtin): Add expansion for float128 round to odd
8091 functions, keying off on -mabi=ieeelongdouble of whether to use
8092 the KFmode or TFmode variant.
8093 (rs6000_init_builtins): Initialize the _Float128 round to odd
8095 * doc/extend.texi (PowerPC Built-in Functions): Document the old
8096 _Float128 'q' built-in functions are now mapped into the new
8097 'f128' built-in functions.
8099 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
8101 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
8102 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
8104 2017-11-06 David Malcolm <dmalcolm@redhat.com>
8107 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
8108 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
8109 * toplev.c: Include "ipa-fnsummary.h".
8110 (toplev::finalize): Call ipa_fnsummary_c_finalize.
8112 2017-11-06 Jakub Jelinek <jakub@redhat.com>
8114 PR tree-optimization/82838
8115 * gimple-ssa-store-merging.c
8116 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
8117 on a separate gimple_seq which is then appended to seq.
8119 2017-11-06 Jeff Law <law@redhat.com>
8122 * config/i386/i386.c (PROBE_INTERVAL): Remove.
8123 (get_probe_interval): New functions.
8124 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
8125 (ix86_adjust_stack_and_probe): Likewise.
8126 (output_adjust_stack_and_probe): Likewise.
8127 (ix86_emit_probe_stack_range): Likewise.
8128 (ix86_expand_prologue): Likewise.
8130 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8132 PR tree-optimization/82816
8133 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
8134 if the modes of the two types are the same.
8135 (convert_plusminus_to_widen): Likewise.
8137 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8139 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
8141 (usadv16qi): New define_expand.
8142 (usadv8hi): New define_expand.
8144 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
8147 * ipa-inline-transform.c (update_noncloned_frequencies): Always
8149 (inline_transform): Likewise.
8150 * predict.c (counts_to_freqs): Remove useless conditional.
8151 * profile-count.h (profile_count::apply_scale): Move sanity check.
8152 * tree-inline.c (copy_bb): Always scale.
8153 (copy_cfg_body): Likewise.
8155 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
8158 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
8160 (*cmp_ite0): Add enabled_for_depr_it attribute.
8161 (*cmp_ite1): Likewise.
8163 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8165 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
8168 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8170 * tree-vrp.c (vrp_int_const_binop): Return true on success and
8171 return the value by pointer.
8172 (extract_range_from_multiplicative_op_1): Update accordingly.
8173 Return as soon as an operation fails.
8175 2017-11-05 Tom de Vries <tom@codesourcery.com>
8178 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
8179 (DEF_SANITIZER_BUILTIN): ... here.
8180 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
8181 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
8183 2017-11-05 Tom de Vries <tom@codesourcery.com>
8186 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
8188 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
8189 ASM_OUTPUT_BEFORE_CASE_LABEL call.
8190 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
8192 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8193 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8194 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8196 2017-11-05 Tom de Vries <tom@codesourcery.com>
8199 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
8202 2017-11-04 Michael Clark <michaeljclark@mac.com>
8204 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
8205 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
8207 (*addsi3_extended): Likewise.
8208 (*addsi3_extended2): Likewise.
8209 (<optab>si3): Likewise.
8210 (<optab>di3): Likewise.
8211 (<optab><mode>3): Likewise.
8212 (<*optabe>si3_internal): Likewise.
8213 (zero_extendqi<SUPERQI:mode>2): Likewise.
8214 (*add<mode>hi3): Likewise.
8215 (*xor<mode>hi3): Likewise.
8216 (<optab>di3): Likewise.
8217 (*<optab>si3_extend): Likewise.
8218 (*sge<u>_<X:mode><GPR:mode>): Likewise.
8219 (*slt<u>_<X:mode><GPR:mode>): Likewise.
8220 (*sle<u>_<X:mode><GPR:mode>): Likewise.
8222 2017-11-04 Andrew Waterman <andrew@sifive.com>
8224 * config/riscv/riscv.c (riscv_option_override): Conditionally set
8225 TARGET_STRICT_ALIGN based upon -mtune argument.
8227 2017-11-04 Andrew Waterman <andrew@sifive.com>
8229 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
8231 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
8233 * config/i386/i386.c (choose_basereg): Use optional scratch
8234 register and add assertion.
8235 (x86_emit_outlined_ms2sysv_save): Use scratch register when
8236 needed, and don't allocate stack.
8237 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
8238 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
8239 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
8241 2017-11-03 Jeff Law <law@redhat.com>
8243 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
8244 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
8245 to probe at the start of a noreturn function.
8247 2017-11-03 Jakub Jelinek <jakub@redhat.com>
8249 PR tree-optimization/78821
8250 * gimple-ssa-store-merging.c: Update the file comment.
8251 (MAX_STORE_ALIAS_CHECKS): Define.
8252 (struct store_operand_info): New type.
8253 (store_operand_info::store_operand_info): New constructor.
8254 (struct store_immediate_info): Add rhs_code and ops data members.
8255 (store_immediate_info::store_immediate_info): Add rhscode, op0r
8256 and op1r arguments to the ctor, initialize corresponding data members.
8257 (struct merged_store_group): Add load_align_base and load_align
8259 (merged_store_group::merged_store_group): Initialize them.
8260 (merged_store_group::do_merge): Update them.
8261 (merged_store_group::apply_stores): Pick the constant for
8262 encode_tree_to_bitpos from one of the two operands, or skip
8263 encode_tree_to_bitpos if neither operand is a constant.
8264 (class pass_store_merging): Add process_store method decl. Remove
8265 bool argument from terminate_all_aliasing_chains method decl.
8266 (pass_store_merging::terminate_all_aliasing_chains): Remove
8267 var_offset_p argument and corresponding handling.
8268 (stmts_may_clobber_ref_p): New function.
8269 (compatible_load_p): New function.
8270 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
8271 if there is overlap and rhs_code is not INTEGER_CST. For
8272 non-overlapping stores terminate group if rhs is not mergeable.
8273 (get_alias_type_for_stmts): Change first argument from
8274 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
8275 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
8276 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
8278 (get_location_for_stmts): Change first argument from
8279 auto_vec<gimple *> & to vec<gimple *> &.
8280 (struct split_store): Remove orig_stmts data member, add orig_stores.
8281 (split_store::split_store): Create orig_stores rather than orig_stmts.
8282 (find_constituent_stmts): Renamed to ...
8283 (find_constituent_stores): ... this. Change second argument from
8284 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
8285 to info structures rather than the statements.
8286 (split_group): Rename ALLOW_UNALIGNED argument to
8287 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
8288 it. Adjust find_constituent_stores caller.
8289 (imm_store_chain_info::output_merged_store): Handle rhs_code other
8290 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
8291 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
8292 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
8293 (mem_valid_for_store_merging): New function.
8294 (handled_load): New function.
8295 (pass_store_merging::process_store): New method.
8296 (pass_store_merging::execute): Use process_store method. Adjust
8297 terminate_all_aliasing_chains caller.
8299 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8301 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8302 Return true for more constants, symbols and label references.
8303 (aarch64_valid_floating_const): Remove unused function.
8305 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8308 * config/aarch64/aarch64.c (aarch64_layout_frame):
8309 Undo forcing of LR at bottom of frame.
8311 2017-11-03 Jeff Law <law@redhat.com>
8314 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
8315 for int_registers_saved.
8317 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
8318 extracted from tree-ssa-dom.c.
8319 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
8320 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
8321 (record_equivalences_from_incoming_edge): Add additional argument
8322 to single_pred_edge_ignoring_loop_edges call.
8323 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
8324 (uncprop_dom_walker::before_dom_children): Add additional argument
8325 to single_pred_edge_ignoring_loop_edges call.
8326 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
8327 single_pred_edge_ignoring_loop_edges rather than open coding.
8328 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
8330 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
8332 * match.pd (-(-A)): Rewrite.
8334 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8336 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
8337 (rs6000_emit_int_cmove): New declaration.
8338 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
8339 (rs6000_emit_sISEL): Delete.
8340 (rs6000_emit_int_cmove): Make non-static.
8341 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
8342 instead of rs6000_emit_sISEL.
8344 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
8346 * asan.c (create_cond_insert_point): Maintain profile.
8347 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
8349 * basic-block.h (struct basic_block_def): Remove frequency.
8350 (EDGE_FREQUENCY): Use to_frequency
8351 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
8353 (find_traces): Update to use to_frequency.
8354 (find_traces_1_round): Likewise; use only IPA counts.
8355 (bb_to_key): Likewise.
8356 (connect_traces): Use IPA counts only.
8357 (copy_bb_p): Update to use to_frequency.
8358 (fix_up_crossing_landing_pad): Likewise.
8359 (sanitize_hot_paths): Likewise.
8360 * bt-load.c (basic_block_freq): Likewise.
8361 * cfg.c (init_flow): Set count_max to uninitialized.
8362 (check_bb_profile): Remove frequencies; check counts.
8363 (dump_bb_info): Do not dump frequencies.
8364 (update_bb_profile_for_threading): Update counts only.
8365 (scale_bbs_frequencies_int): Likewise.
8366 (MAX_SAFE_MULTIPLIER): Remove.
8367 (scale_bbs_frequencies_gcov_type): Update counts only.
8368 (scale_bbs_frequencies_profile_count): Update counts only.
8369 (scale_bbs_frequencies): Update counts only.
8370 * cfg.h (struct control_flow_graph): Add count-max.
8371 (update_bb_profile_for_threading): Update prototype.
8372 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
8373 (find_many_sub_basic_blocks): Likewise.
8374 * cfgcleanup.c (try_forward_edges): Likewise.
8375 (try_crossjump_to_edge): Likewise.
8376 * cfgexpand.c (expand_gimple_cond): Likewise.
8377 (expand_gimple_tailcall): Likewise.
8378 (construct_init_block): Likewise.
8379 (construct_exit_block): Likewise.
8380 * cfghooks.c (verify_flow_info): Check consistency of counts.
8381 (dump_bb_for_graph): Do not dump frequencies.
8382 (split_block_1): Do not update frequencies.
8383 (split_edge): Do not update frequencies.
8384 (make_forwarder_block): Do not update frequencies.
8385 (duplicate_block): Do not update frequencies.
8386 (account_profile_record): Do not update frequencies.
8387 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
8388 for global heuristics.
8389 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
8390 (expected_loop_iterations_unbounded): Use counts only.
8391 * cfgloopmanip.c (scale_loop_profile): Simplify.
8392 (create_empty_loop_on_edge): Simplify
8394 (duplicate_loop_to_header_edge): Simplify
8395 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
8396 (update_br_prob_note): Take care of removing note when profile
8398 (relink_block_chain): Do not dump frequency.
8399 (rtl_account_profile_record): Use to_frequency.
8400 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
8401 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
8402 (cgraph_update_edges_for_call_stmt_node): Likewise.
8403 (cgraph_edge::verify_count_and_frequency): Update.
8404 (cgraph_node::verify_node): Temporarily disable frequency verification.
8405 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
8406 to_cgraph_frequency.
8407 (cgraph_edge::rebuild_edges): Convert to ipa counts.
8408 * cgraphunit.c (init_lowered_empty_function): Do not initialize
8410 (cgraph_node::expand_thunk): Update profile.
8411 * except.c (dw2_build_landing_pads): Do not update frequency.
8412 * final.c (compute_alignments): Use to_frequency.
8413 (dump_basic_block_info): Do not dump frequency.
8414 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
8415 (dump_gimple_bb_header): Do not dump frequency.
8416 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
8418 * gimple-streamer-in.c (input_bb): Do not stream frequency.
8419 * gimple-streamer-out.c (output_bb): Do not stream frequency.
8420 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
8421 (init_before_recovery): Do not update frequency.
8422 (sched_create_recovery_edges): Do not update frequency.
8423 * hsa-gen.c (convert_switch_statements): Do not update frequency.
8424 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
8425 (ipa_cp_c_finalize): Set max_count to uninitialized.
8426 * ipa-fnsummary.c (get_minimal_bb): Use counts.
8427 (param_change_prob): Use counts.
8428 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
8430 * ipa-split.c (consider_split): Use to_frequency.
8431 (split_function): Use to_frequency.
8432 * ira-build.c (loop_compare_func): Likewise.
8433 (mark_loops_for_removal): Likewise.
8434 (mark_all_loops_for_removal): Likewise.
8435 * loop-doloop.c (doloop_modify): Do not update frequency.
8436 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
8438 * lto-streamer-in.c (input_function): Update count_max.
8439 * omp-expand.c (expand_omp_taskreg): Update count_max.
8440 * omp-simd-clone.c (simd_clone_adjust): Update profile.
8441 * predict.c (maybe_hot_frequency_p): Use to_frequency.
8442 (maybe_hot_count_p): Use ipa counts only.
8443 (maybe_hot_bb_p): Simplify.
8444 (maybe_hot_edge_p): Simplify.
8445 (probably_never_executed): Do not take frequency argument.
8446 (probably_never_executed_bb_p): Do not pass frequency.
8447 (probably_never_executed_edge_p): Likewise.
8448 (combine_predictions_for_bb): Check that profile is nonzero.
8449 (propagate_freq): Do not set frequency.
8450 (drop_profile): Simplify.
8451 (counts_to_freqs): Simplify.
8452 (expensive_function_p): Use to_frequency.
8453 (propagate_unlikely_bbs_forward): Simplify.
8454 (determine_unlikely_bbs): Simplify.
8455 (estimate_bb_frequencies): Add hack to silence graphite issues.
8456 (compute_function_frequency): Use ipa counts.
8457 (pass_profile::execute): Update.
8458 (rebuild_frequencies): Use counts only.
8459 (force_edge_cold): Use counts only.
8460 * profile-count.c (profile_count::dump): Dump new count types.
8461 (profile_count::differs_from_p): Check compatiblity.
8462 (profile_count::to_frequency): New function.
8463 (profile_count::to_cgraph_frequency): New function.
8464 * profile-count.h (struct function): Declare.
8465 (enum profile_quality): Add profile_guessed_local and
8466 profile_guessed_global0.
8467 (class profile_proability): Decrease number of bits to 29;
8468 update from_reg_br_prob_note and to_reg_br_prob_note.
8469 (class profile_count: Update comment; decrease number of bits
8470 to 61. Check compatibility.
8471 (profile_count::compatible_p): New private member function.
8472 (profile_count::ipa_p): New member function.
8473 (profile_count::operator<): Handle global zero correctly.
8474 (profile_count::operator>): Handle global zero correctly.
8475 (profile_count::operator<=): Handle global zero correctly.
8476 (profile_count::operator>=): Handle global zero correctly.
8477 (profile_count::nonzero_p): New member function.
8478 (profile_count::force_nonzero): New member function.
8479 (profile_count::max): New member function.
8480 (profile_count::apply_scale): Handle IPA scalling.
8481 (profile_count::guessed_local): New member function.
8482 (profile_count::global0): New member function.
8483 (profile_count::ipa): New member function.
8484 (profile_count::to_frequency): Declare.
8485 (profile_count::to_cgraph_frequency): Declare.
8486 * profile.c (OVERLAP_BASE): Delete.
8487 (compute_frequency_overlap): Delete.
8488 (compute_branch_probabilities): Do not use compute_frequency_overlap.
8489 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
8490 * sched-ebb.c (rank): Use counts only.
8491 * shrink-wrap.c (handle_simple_exit): Use counts only.
8492 (try_shrink_wrapping): Use counts only.
8493 (place_prologue_for_one_component): Use counts only.
8494 * tracer.c (find_best_predecessor): Use to_frequency.
8495 (find_trace): Use to_frequency.
8496 (tail_duplicate): Use to_frequency.
8497 * trans-mem.c (expand_transaction): Do not update frequency.
8498 * tree-call-cdce.c: Do not update frequency.
8499 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
8500 (gimple_merge_blocks): Likewise.
8501 (gimple_split_edge): Likewise.
8502 (gimple_duplicate_sese_region): Likewise.
8503 (gimple_duplicate_sese_tail): Likewise.
8504 (move_sese_region_to_fn): Likewise.
8505 (gimple_account_profile_record): Likewise.
8506 (insert_cond_bb): Likewise.
8507 * tree-complex.c (expand_complex_div_wide): Likewise.
8508 * tree-eh.c (lower_resx): Update profile.
8509 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
8511 (initialize_cfun): Do not initialize frequencies
8512 (freqs_to_counts): Delete.
8513 (copy_cfg_body): Ignore count parameter.
8514 (copy_body): Update.
8515 (expand_call_inline): Update count_max.
8516 (optimize_inline_calls): Update count_max.
8517 (tree_function_versioning): Update count_max.
8518 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
8519 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
8521 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
8522 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
8523 (try_peel_loop): Likewise.
8524 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
8526 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
8527 (tree_transform_and_unroll_loop): Do not use frequencies
8528 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
8529 Use reliable prediction only.
8530 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
8531 * tree-ssa-sink.c (select_best_block): Use to_frequency.
8532 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
8533 probability scaling.
8534 * tree-ssa-threadupdate.c (create_block_for_threading): Do
8535 not update frequency
8536 (any_remaining_duplicated_blocks): Likewise.
8537 (update_profile): Likewise.
8538 (estimated_freqs_path): Delete.
8539 (freqs_to_counts_path): Delete.
8540 (clear_counts_path): Delete.
8541 (ssa_fix_duplicate_block_edges): Likewise.
8542 (duplicate_thread_path): Likewise.
8543 * tree-switch-conversion.c (gen_inbound_check): Use counts.
8544 * tree-tailcall.c (decrease_profile): Do not update frequency.
8545 (eliminate_tail_call): Likewise.
8546 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
8547 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
8548 (optimize_mask_stores): Likewise.
8549 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8550 * ubsan.c (ubsan_expand_null_ifn): Update profile.
8551 (ubsan_expand_ptr_ifn): Update profile.
8552 * value-prof.c (gimple_ic): Simplify.
8553 * value-prof.h (gimple_ic): Update prototype.
8554 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
8555 * ipa-inline.c (compute_uninlined_call_time): Be sure that
8557 (want_inline_self_recursive_call_p): Likewise.
8558 (resolve_noninline_speculation): Only cummulate defined counts.
8559 (inline_small_functions): Use nonzero_p.
8560 (ipa_inline): Do not access freed node.
8562 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8564 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8565 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8567 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8569 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8570 non-legitimate address.
8572 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8574 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8575 (*lt0_<mode>di, *lt0_<mode>si): New.
8577 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8579 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8580 TARGET_PAIRED_FLOAT.
8582 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8583 Jim Wilson <jim.wilson@linaro.org>
8585 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8586 * config/aarch64/aarch64-tune.md: Regenerated.
8587 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8588 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8590 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8592 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8593 (arc_expand_prologue): Restore blink for millicode.
8594 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8596 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8599 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8600 gen_vec_duplicate after forcing the scalar into a register.
8602 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8604 * combine (try_combine): Print the insns input to try_combine to the
8607 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8610 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8611 Remove second argument from aarch64_process_target_attr call.
8612 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8614 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8616 (aarch64_handle_attr_arch): Remove second argument.
8617 (aarch64_handle_attr_cpu): Ditto.
8618 (aarch64_handle_attr_tune): Ditto.
8619 (aarch64_handle_attr_isa_flags): Ditto.
8620 (aarch64_process_one_target_attr): Ditto.
8621 (aarch64_process_target_attr): Ditto.
8622 (aarch64_option_valid_attribute_p): Remove second argument.
8623 on aarch64_process_target_attr call.
8625 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8627 * diagnostic.c: Include "selftest-diagnostic.h".
8628 (selftest::assert_location_text): New function.
8629 (selftest::test_diagnostic_get_location_text): New function.
8630 (selftest::diagnostic_c_tests): Call it.
8632 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8634 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8635 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8636 (class selftest::test_diagnostic_context): Move to...
8637 * selftest-diagnostic.c: New file.
8638 * selftest-diagnostic.h: New file.
8640 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8642 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8643 offset range for FT32B.
8644 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8645 * config/ft32/ft32.md: Add TARGET_NOPM.
8646 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8647 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8649 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8651 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8653 2017-11-02 Jeff Law <law@redhat.com>
8655 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8656 virtual keyword on FINAL OVERRIDE members.
8658 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8659 virtuals together. Add virtual destructor.
8660 (substitute_and_fold_engine): Similarly.
8662 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8664 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8666 2017-11-02 Richard Biener <rguenther@suse.de>
8668 PR tree-optimization/82795
8669 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8671 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8673 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8675 (gcc_GAS_CHECK_FEATURE): Remove.
8676 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8677 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8678 * configure: Regenerate.
8680 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8682 * config/arc/arc.c (hwloop_optimize): Account for empty
8685 2017-11-02 Richard Biener <rguenther@suse.de>
8688 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8689 Truncate ARRAY_REF index and element size.
8691 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8693 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8695 2017-11-01 Jeff Law <law@redhat.com>
8697 * tree-ssa-ccp.c (ccp_folder): New class derived from
8698 substitute_and_fold_engine.
8699 (ccp_folder::get_value): New member function.
8700 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8701 (ccp_fold_stmt): Remove prototype.
8702 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8703 * tree-ssa-copy.c (copy_folder): New class derived from
8704 substitute_and_fold_engine.
8705 (copy_folder::get_value): Renamed from get_value.
8706 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8707 * tree-vrp.c (vrp_folder): New class derived from
8708 substitute_and_fold_engine.
8709 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8710 (vrp_folder::get_value): New member function.
8711 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8712 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8713 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8714 provide a class interface to folder/substitute routines.
8715 (ssa_prop_fold_stmt_fn): Remove typedef.
8716 (ssa_prop_get_value_fn): Likewise.
8717 (subsitute_and_fold): Remove prototype.
8718 (replace_uses_in): Likewise.
8719 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8720 Renamed from replace_uses_in. Call the virtual member function
8721 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8722 (substitute_and_fold_dom_walker): Remove initialization of
8723 data member entries for calbacks. Add substitute_and_fold_engine
8724 member and initialize it.
8725 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8726 member functions for get_value, replace_phi_args_in c
8727 replace_uses_in, and fold_stmt calls.
8728 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8729 substitute_and_fold. Remove assert. Update ctor call.
8731 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8732 (ssa_prop_visit_phi_fn): Likewise.
8733 (class ssa_propagation_engine): New class to provide an interface
8735 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8737 (ssa_prop_visit_phi): Likewise.
8738 (ssa_propagation_engine::simulate_stmt): Moved into class.
8739 Call visit_phi/visit_stmt from the class rather than via
8740 file scoped static variables.
8741 (ssa_propagation_engine::simulate_block): Moved into class.
8742 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8743 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8744 set file scoped statics for the visit_stmt/visit_phi callbacks.
8745 * tree-complex.c (complex_propagate): New class derived from
8746 ssa_propagation_engine.
8747 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8748 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8749 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8751 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8752 ssa_propagation_engine.
8753 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8754 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8755 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8756 * tree-ssa-copy.c (copy_prop): New class derived from
8757 ssa_propagation_engine.
8758 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8759 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8760 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8761 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8762 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8763 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8764 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8766 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8768 PR rtl-optimization/82778
8769 PR rtl-optimization/82597
8770 * compare-elim.c (struct comparison): Add in_a_setter field.
8771 (find_comparison_dom_walker::before_dom_children): Remove killed
8772 bitmap and df_simulate_find_defs call, instead walk the defs.
8773 Compute last_setter and initialize in_a_setter. Merge definitions
8774 with first initialization for a few variables.
8775 (try_validate_parallel): Use insn_invalid_p instead of
8776 recog_memoized. Return insn rather than just the pattern.
8777 (try_merge_compare): Fix up comment. Don't uselessly test if
8778 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8780 (execute_compare_elim_after_reload): Remove df_chain_add_problem
8783 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8784 Alan Hayward <alan.hayward@arm.com>
8785 David Sherwood <david.sherwood@arm.com>
8787 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
8788 aarch64_hard_regno_nregs to get the number of registers
8791 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8792 Alan Hayward <alan.hayward@arm.com>
8793 David Sherwood <david.sherwood@arm.com>
8795 * config/aarch64/constraints.md (Upl): Rename to...
8797 * config/aarch64/aarch64.md
8798 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
8801 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8802 Alan Hayward <alan.hayward@arm.com>
8803 David Sherwood <david.sherwood@arm.com>
8805 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
8806 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8809 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8810 Alan Hayward <alan.hayward@arm.com>
8811 David Sherwood <david.sherwood@arm.com>
8813 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
8814 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
8815 (aarch64_evpc_dup): Generate rtl direcly, rather than using
8817 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
8818 of a single element.
8819 * config/aarch64/iterators.md: Add a comment above the permute
8820 unspecs to say that they are generated directly by
8821 aarch64_expand_vec_perm_const.
8822 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
8824 2017-11-01 Nathan Sidwell <nathan@acm.org>
8826 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
8828 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8830 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
8831 and medany code models, and describe what they do.
8833 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8835 Revert accidental duplicate:
8837 * combine.c (can_change_dest_mode): Reject changes in
8838 REGMODE_NATURAL_SIZE.
8840 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
8842 PR rtl-optimization/64682
8843 PR rtl-optimization/69567
8844 PR rtl-optimization/69737
8845 PR rtl-optimization/82683
8846 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
8847 register mentioned in the note, drop the note, unless it came from I3,
8848 in which case it should go to I3 again.
8850 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8852 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
8853 and return false if not.
8854 (clear_bytes_written_by, live_bytes_read): Update accordingly.
8856 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8858 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
8859 range is known to be empty.
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 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
8866 and CONST_VECTOR_NUNITS instead of computing the number of units from
8867 the byte sizes of the vector and element.
8868 (simplify_binary_operation_1): Likewise.
8869 (simplify_const_binary_operation): Likewise.
8870 (simplify_ternary_operation): Likewise.
8872 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8873 Alan Hayward <alan.hayward@arm.com>
8874 David Sherwood <david.sherwood@arm.com>
8876 * var-tracking.c (INT_MEM_OFFSET): Replace with...
8877 (int_mem_offset): ...this new function.
8878 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
8879 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
8880 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
8883 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8884 Alan Hayward <alan.hayward@arm.com>
8885 David Sherwood <david.sherwood@arm.com>
8887 * lower-subreg.c (interesting_mode_p): New function.
8888 (compute_costs, find_decomposable_subregs, decompose_register)
8889 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
8890 (resolve_clobber, dump_choices): Use it.
8892 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8893 Alan Hayward <alan.hayward@arm.com>
8894 David Sherwood <david.sherwood@arm.com>
8896 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
8898 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8899 Alan Hayward <alan.hayward@arm.com>
8900 David Sherwood <david.sherwood@arm.com>
8902 * alias.c (find_base_value, find_base_term): Only process integer
8903 truncations. Check the precision rather than the size.
8905 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8906 Alan Hayward <alan.hayward@arm.com>
8907 David Sherwood <david.sherwood@arm.com>
8909 * machmode.h (is_narrower_int_mode): New function
8910 * optabs.c (expand_float, expand_fix): Use it.
8911 * dwarf2out.c (rotate_loc_descriptor): Likewise.
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 * rtl.h (narrower_subreg_mode): New function.
8918 * ira-color.c (update_costs_from_allocno): Use it.
8920 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8921 Alan Hayward <alan.hayward@arm.com>
8922 David Sherwood <david.sherwood@arm.com>
8924 * optabs-query.h (convert_optab_p): New function, split out from...
8925 (convert_optab_handler): ...here.
8926 (widening_optab_handler): Delete.
8927 (find_widening_optab_handler): Remove permit_non_widening parameter.
8928 (find_widening_optab_handler_and_mode): Likewise. Provide an
8929 override that operates on mode class wrappers.
8930 * optabs-query.c (widening_optab_handler): Delete.
8931 (find_widening_optab_handler_and_mode): Remove permit_non_widening
8932 parameter. Assert that the two modes are the same class and that
8933 the "from" mode is narrower than the "to" mode. Use
8934 convert_optab_handler instead of widening_optab_handler.
8935 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
8936 instead of widening_optab_handler.
8937 * expr.c (expand_expr_real_2): Update calls to
8938 find_widening_optab_handler.
8939 * optabs.c (expand_widen_pattern_expr): Likewise.
8940 (expand_binop_directly): Take the insn_code as a parameter.
8941 (expand_binop): Only call find_widening_optab_handler for
8942 conversion optabs; use optab_handler otherwise. Update calls
8943 to find_widening_optab_handler and expand_binop_directly.
8944 Use convert_optab_handler instead of widening_optab_handler.
8945 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
8946 find_widening_optab_handler and use scalar_mode rather than
8948 (convert_plusminus_to_widen): Likewise.
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 * machmode.h (fixed_size_mode): New class.
8955 * rtl.h (get_pool_mode): Return fixed_size_mode.
8956 * gengtype.c (main): Add fixed_size_mode.
8957 * target.def (get_raw_result_mode): Return a fixed_size_mode.
8958 (get_raw_arg_mode): Likewise.
8959 * doc/tm.texi: Regenerate.
8960 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
8961 * targhooks.c (default_get_reg_raw_mode): Likewise.
8962 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
8963 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
8964 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
8965 (msp430_get_raw_result_mode): Likewise.
8966 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
8967 * dbxout.c (dbxout_parms): Require fixed-size modes.
8968 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
8969 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8970 * omp-low.c (lower_oacc_reductions): Likewise.
8971 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
8972 (simplify_subreg): Update accordingly.
8973 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
8974 (force_const_mem): Update accordingly. Return NULL_RTX for modes
8975 that aren't fixed-size.
8976 (get_pool_mode): Return a fixed_size_mode.
8977 (output_constant_pool_2): Take a fixed_size_mode.
8979 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8980 Alan Hayward <alan.hayward@arm.com>
8981 David Sherwood <david.sherwood@arm.com>
8983 * doc/rtl.texi (vec_series): Document.
8984 (const): Say that the operand can be a vec_series.
8985 * rtl.def (VEC_SERIES): New rtx code.
8986 * rtl.h (const_vec_series_p_1): Declare.
8987 (const_vec_series_p): New function.
8988 * emit-rtl.h (gen_const_vec_series): Declare.
8989 (gen_vec_series): Likewise.
8990 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
8991 (gen_vec_series): Likewise.
8992 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
8993 * simplify-rtx.c (simplify_unary_operation): Handle negations
8995 (simplify_binary_operation_series): New function.
8996 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
8997 (test_vector_ops_series): New function.
8998 (test_vector_ops): Call it.
8999 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9000 gen_const_vec_series.
9001 (altivec_lvsr): Likewise.
9002 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9004 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9005 Alan Hayward <alan.hayward@arm.com>
9006 David Sherwood <david.sherwood@arm.com>
9008 * doc/rtl.texi (const): Update description of address constants.
9009 Say that vector constants are allowed too.
9010 * common.md (E, F): Use CONSTANT_P instead of checking for
9012 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9013 checking for CONST_VECTOR.
9014 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9016 * expr.c (expand_expr_real_2): Check for vector modes instead
9017 of checking for CONST_VECTOR.
9018 * rtl.h (const_vec_p): New function.
9019 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9020 (unwrap_const_vec_duplicate): Handle them here too.
9022 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9023 David Malcolm <dmalcolm@redhat.com>
9024 Alan Hayward <alan.hayward@arm.com>
9025 David Sherwood <david.sherwood@arm.com>
9027 * rtl.h (vec_duplicate_p): New function.
9028 * selftest-rtl.c (assert_rtx_eq_at): New function.
9029 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9030 (assert_rtx_eq_at): Declare.
9031 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9032 * selftest-run-tests.c (selftest::run_tests): Call it.
9033 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9034 (simplify_unary_operation_1): Recursively handle vector duplicates.
9035 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9037 (simplify_subreg): Handle subregs of vector duplicates.
9038 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9039 (selftest::simplify_rtx_c_tests): New functions.
9041 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9042 Alan Hayward <alan.hayward@arm.com>
9043 David Sherwood <david.sherwood@arm.com>
9045 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9046 (gen_vec_duplicate): Likewise.
9047 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9049 (gen_const_vector): ...here.
9050 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9051 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9052 whose elements are all equal.
9053 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
9054 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
9055 (simplify_relational_operation): Likewise.
9056 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
9058 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
9059 (aarch64_expand_vector_init): Likewise.
9060 * config/arm/arm.c (neon_vdup_constant): Likewise.
9061 (neon_expand_vector_init): Likewise.
9062 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
9063 (arm_block_set_unaligned_vect): Likewise.
9064 (arm_block_set_aligned_vect): Likewise.
9065 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
9066 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
9067 (expand_vec_perm_even_odd_pack): Likewise.
9068 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
9069 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
9070 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
9071 gen_const_vec_duplicate.
9072 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
9073 * config/mips/mips.c (mips_gen_const_int_vector): Use
9074 gen_const_vec_duplicate.
9075 (mips_expand_vector_init): Use CONST0_RTX.
9076 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
9077 (define_split): Use gen_const_vec_duplicate.
9078 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
9079 (define_split): Use gen_const_vec_duplicate.
9080 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
9081 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
9082 * config/spu/spu.c (spu_const): Likewise.
9084 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9085 Alan Hayward <alan.hayward@arm.com>
9086 David Sherwood <david.sherwood@arm.com>
9088 * combine.c (can_change_dest_mode): Reject changes in
9089 REGMODE_NATURAL_SIZE.
9091 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
9093 * configure.ac (--enable-libssp): New.
9094 (gcc_cv_libc_provides_ssp): Check for explicit setting before
9095 trying to determine target-specific default. Adjust indentation.
9096 * configure: Regenerated.
9097 * doc/install.texi (Configuration): Expand --disable-libssp
9100 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
9102 config/i386/i386.c (ix86_expand_epilogue): Correct stack
9105 2017-10-31 Martin Jambor <mjambor@suse.cz>
9108 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
9110 2017-10-31 David Malcolm <dmalcolm@redhat.com>
9112 * auto-profile.c (autofdo_source_profile::read): Use
9113 UNKNOWN_LOCATION rather than 0.
9114 * diagnostic-core.h (warning_at_rich_loc): Rename to...
9115 (warning_at): ...this overload.
9116 (warning_at_rich_loc_n): Rename to...
9117 (warning_n): ...this overload.
9118 (error_at_rich_loc): Rename to...
9119 (error_at): ...this overload.
9120 (pedwarn_at_rich_loc): Rename to...
9121 (pedwarn): ...this overload.
9122 (permerror_at_rich_loc): Rename to...
9123 (permerror): ...this overload.
9124 (inform_at_rich_loc): Rename to...
9125 (inform): ...this overload.
9126 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
9127 (diagnostic_n_impl_richloc): Rename to...
9128 (diagnostic_n_impl): ...this rich_location *-based decl.
9129 (inform_at_rich_loc): Rename to...
9130 (inform): ...this, and add an assertion.
9131 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
9132 (warning_at_rich_loc): Rename to...
9133 (warning_at): ...this, and add an assertion.
9134 (warning_at_rich_loc_n): Rename to...
9135 (warning_n): ...this, and add an assertion.
9136 (warning_n): Update location_t-based implementation for removal of
9137 location_t-based diagnostic_n_impl.
9138 (pedwarn_at_rich_loc): Rename to...
9139 (pedwarn): ...this, and add an assertion.
9140 (permerror_at_rich_loc): Rename to...
9141 (permerror): ...this, and add an assertion.
9142 (error_n): Update for removal of location_t-based diagnostic_n_impl.
9143 (error_at_rich_loc): Rename to...
9144 (error_at): ...this, and add an assertion.
9145 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
9146 (driver::do_spec_on_infiles): Likewise.
9147 * substring-locations.c (format_warning_va): Update for renaming
9148 of inform_at_rich_loc.
9150 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
9152 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
9153 _Float<N>X built-in functions so that the variant without the
9154 "__builtin_" prefix is only enabled for the GNU C and Objective C
9155 languages when they are in non-strict ANSI/ISO mode.
9156 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
9157 * target.def (floatn_builtin_p): Add a target hook to control
9158 whether _Float<N> and _Float<N>X built-in functions without the
9159 "__builtin_" prefix are enabled, and return true for C and
9160 Objective C in the default hook. Include langhooks.h in
9162 * targhooks.h (default_floatn_builtin_p): Likewise.
9163 * targhooks.c (default_floatn_builtin_p): Likewise.
9164 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
9165 floatn_builtin_p target hook.
9166 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
9168 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
9169 Eric Botcazou <ebotcazou@adacore.com>
9171 PR rtl-optimization/81803
9172 * lra-constraints.c (curr_insn_transform): Also reload the whole
9173 register for a strict subreg no wider than a word if this is for
9174 a WORD_REGISTER_OPERATIONS target.
9176 2017-10-31 Jason Merrill <jason@redhat.com>
9178 * gdbinit.in: Skip over inlines from timevar.h.
9180 2017-10-31 Martin Liska <mliska@suse.cz>
9182 * doc/gcov.texi: Document new option.
9183 * gcov.c (print_usage): Likewise print it.
9184 (process_args): Support the argument.
9185 (format_count): New function.
9186 (format_gcov): Use the function.
9188 2017-10-31 Martin Liska <mliska@suse.cz>
9190 * gcov.c (struct name_map): do not use typedef.
9191 Define operator== and operator<.
9192 (name_search): Remove.
9193 (name_sort): Remove.
9194 (main): Do not allocate names.
9195 (process_file): Add vertical space.
9196 (generate_results): Use std::find.
9197 (release_structures): Do not release memory.
9198 (find_source): Use std::find.
9200 2017-10-31 Martin Liska <mliska@suse.cz>
9202 * gcov.c (struct line_info): Remove it's typedef.
9203 (line_info::line_info): Add proper ctor.
9204 (line_info::has_block): Do not use a typedef.
9205 (struct source_info): Do not use typedef.
9206 (circuit): Likewise.
9207 (get_cycles_count): Likewise.
9208 (output_intermediate_file): Iterate via vector iterator.
9209 (add_line_counts): Use std::vector methods.
9210 (accumulate_line_counts): Likewise.
9211 (output_lines): Likewise.
9213 2017-10-31 Martin Liska <mliska@suse.cz>
9215 * gcov.c (struct source_info): Remove typedef.
9216 (source_info::source_info): Add proper ctor.
9217 (accumulate_line_counts): Use struct, not it's typedef.
9218 (output_gcov_file): Likewise.
9219 (output_lines): Likewise.
9220 (main): Do not allocate an array.
9221 (output_intermediate_file): Use size of vector container.
9222 (process_file): Resize the vector.
9223 (generate_results): Do not preallocate, use newly added vector
9225 (release_structures): Do not release sources.
9226 (find_source): Use vector methods.
9227 (add_line_counts): Do not use typedef.
9229 2017-10-31 Martin Liska <mliska@suse.cz>
9231 * doc/gcov.texi: Document that.
9232 * gcov.c (add_line_counts): Mark lines with a non-executed
9234 (output_line_beginning): Handle such lines.
9235 (output_lines): Pass new argument.
9236 (output_intermediate_file): Print it in intermediate format.
9238 2017-10-31 Martin Liska <mliska@suse.cz>
9240 * color-macros.h: New file.
9241 * diagnostic-color.c: Factor out color related to macros to
9243 * doc/gcov.texi: Document -k option.
9244 * gcov.c (INCLUDE_STRING): Include string.h.
9245 (print_usage): Add -k option.
9246 (process_args): Parse it.
9247 (pad_count_string): New function.
9248 (output_line_beginning): Likewise.
9249 (DEFAULT_LINE_START): New macro.
9250 (output_lines): Support color output.
9252 2017-10-31 Martin Liska <mliska@suse.cz>
9254 PR gcov-profile/82633
9255 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
9256 their interaction with GCOV infrastructure.
9257 * configure.ac: Add -fkeep-{inline,static}-functions to
9259 * configure: Regenerate.
9261 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
9264 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
9266 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
9269 * config/rs6000/rs6000.md (allocate_stack): Force update interval
9270 into a register if it does not fit into an immediate offset field.
9272 2017-10-31 Olivier Hainque <hainque@adacore.com>
9274 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
9276 2017-10-31 Julia Koval <julia.koval@intel.com>
9278 * config.gcc: Add gfniintrin.h.
9279 * config/i386/gfniintrin.h: New.
9280 * config/i386/i386-builtin-types.def
9281 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
9282 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
9283 __builtin_ia32_vgf2p8affineinvqb_v32qi,
9284 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
9285 __builtin_ia32_vgf2p8affineinvqb_v16qi,
9286 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
9287 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
9288 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
9289 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
9290 V64QI_FTYPE_V64QI_V64QI_INT): New types.
9291 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
9292 * config/i386/immintrin.h: Include gfniintrin.h.
9293 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
9295 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
9297 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
9299 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9301 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
9302 (arm_ashldi3_1bit): Remove pattern.
9303 (ashrdi3): Remove shift by 1 expansion.
9304 (arm_ashrdi3_1bit): Remove pattern.
9305 (lshrdi3): Remove shift by 1 expansion.
9306 (arm_lshrdi3_1bit): Remove pattern.
9307 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
9308 cost of ashldi3 by 1.
9309 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
9310 (<shift>di3_neon): Likewise.
9312 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9314 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
9315 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
9316 and (*aarch64_simd_mov<VQ:mode>).
9317 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
9318 pattern alternative.
9319 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
9320 attributes to match pattern alternative.
9322 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
9324 * config.gcc (powerpc*-*-*): Add emmintrin.h.
9325 * config/rs6000/emmintrin.h: New file.
9326 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
9328 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9330 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
9331 * (movdi_vfp_cortexa8): Remove pattern.
9333 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9335 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
9337 (Specific, alpha*-dec-osf5.1): Remove.
9338 (Specific, mips-sgi-irix5): Remove.
9339 (Specific, mips-sgi-irix6): Remove.
9341 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9344 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
9345 arguments to clear_bit_region_be.
9347 2017-10-30 Jim Wilson <wilson@tuliptree.org>
9349 * gimplify.c: Include memmodel.h.
9351 2017-10-30 Martin Jambor <mjambor@suse.cz>
9353 * omp-grid.c (grid_attempt_target_gridification): Also insert a
9354 condition whether loop should be executed at all.
9356 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
9358 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9359 gimple folding of vec_madd() intrinsics.
9360 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
9361 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
9362 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
9364 2017-10-30 Richard Biener <rguenther@suse.de>
9366 PR tree-optimization/82762
9368 2017-10-23 Richard Biener <rguenther@suse.de>
9370 PR tree-optimization/82129
9372 2017-08-01 Richard Biener <rguenther@suse.de>
9374 PR tree-optimization/81181
9375 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9376 (compute_antic): ... end of iteration here.
9378 2017-10-30 Joseph Myers <joseph@codesourcery.com>
9380 * doc/invoke.texi (C Dialect Options): Document -std=c17,
9381 -std=iso9899:2017 and -std=gnu17.
9382 * doc/standards.texi (C Language): Document C17 support.
9383 * doc/cpp.texi (Overview): Mention -std=c17.
9384 (Standard Predefined Macros): Document C11 and C17 values of
9385 __STDC_VERSION__. Do not refer to C99 support as incomplete.
9386 * doc/extend.texi (Inline): Do not list individual options for
9387 standards newer than C99.
9388 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
9390 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
9393 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
9395 * asan.c (asan_finish_file): Align asan globals array by shadow
9398 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9401 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
9402 (struct store_immediate_info): Add bitregion_start and bitregion_end
9404 (store_immediate_info::store_immediate_info): Add brs and bre
9405 arguments and initialize bitregion_{start,end} from those.
9406 (struct merged_store_group): Add bitregion_start, bitregion_end,
9407 align_base and mask fields. Drop unnecessary struct keyword from
9408 struct store_immediate_info. Add do_merge method.
9409 (clear_bit_region_be): Use memset instead of loop storing zeros.
9410 (merged_store_group::do_merge): New method.
9411 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
9412 stores as long as the surrounding bitregions have no gaps.
9413 (merged_store_group::merge_overlapping): Use do_merge.
9414 (merged_store_group::apply_stores): Test that bitregion_{start,end}
9415 is byte aligned, rather than requiring that start and width are
9416 byte aligned. Drop unnecessary struct keyword from
9417 struct store_immediate_info. Allocate and populate also mask array.
9418 Make start of the arrays relative to bitregion_start rather than
9419 start and size them according to bitregion_{end,start} difference.
9420 (struct imm_store_chain_info): Drop unnecessary struct keyword from
9421 struct store_immediate_info.
9422 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
9423 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
9424 struct keyword from struct store_immediate_info.
9425 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
9426 between stores as long as the surrounding bitregions have no gaps.
9428 (struct split_store): Add orig non-static data member.
9429 (split_store::split_store): Initialize orig to false.
9430 (find_constituent_stmts): Return store_immediate_info *, non-NULL
9431 if there is exactly a single original stmt. Change stmts argument
9432 to pointer from reference, if NULL, don't push anything to it. Add
9433 first argument, use it to optimize skipping over orig stmts that
9434 are known to be before bitpos already. Simplify.
9435 (split_group): Return unsigned int count how many stores are or
9436 would be needed rather than a bool. Add allow_unaligned argument.
9437 Change split_stores argument from reference to pointer, if NULL,
9438 only do a dry run computing how many stores would be produced.
9439 Rewritten algorithm to use both alignment and misalign if
9440 !allow_unaligned and handle bitfield stores with gaps.
9441 (imm_store_chain_info::output_merged_store): Set start_byte_pos
9442 from bitregion_start instead of start. Compute allow_unaligned
9443 here, if true, do 2 split_group dry runs to compute which one
9444 produces fewer stores and prefer aligned if equal. Punt if
9445 new count is bigger or equal than original before emitting any
9446 statements, rather than during that. Remove no longer needed
9447 new_ssa_names tracking. Replace num_stmts with
9448 split_stores.length (). Use 32-bit stack allocated entries
9449 in split_stores auto_vec. Try to reuse original store lhs/rhs1
9450 if possible. Handle bitfields with gaps.
9451 (pass_store_merging::execute): Ignore bitsize == 0 stores.
9452 Compute bitregion_{start,end} for the stores and construct
9453 store_immediate_info with that. Formatting fixes.
9455 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
9458 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
9459 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
9461 2017-10-29 Jim Wilson <wilson@tuliptree.org>
9463 * gimplify.c: Include tm_p.h.
9465 * common.opt (gcoff): Re-add as ignored option.
9466 (gcoff1, gcoff2, gcoff3): Likewise.
9468 * Makefile.in (OBJS): Delete sdbout.o.
9469 (GTFILES): Delete $(srcdir)/sdbout.c.
9470 * debug.h: Delete sdb_debug_hooks.
9471 * final.c: Delete sdbout.h include.
9472 (final_scan_insn): Delete SDB_DEBUG check.
9473 (rest_of_clean_state): Likewise.
9474 * output.h: Delete sdb_begin_function_line.
9477 * toplev.c: Delete sdbout.h include.
9478 (process_options): Delete SDB_DEBUG check.
9479 * tree-core.h (tree_type_common): Delete pointer field of
9481 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
9482 TYPE_SYMTAB_POINTER.
9483 * tree.h (TYPE_SYMTAB_POINTER): Delete.
9484 (TYPE_SYMTAB_IS_POINTER): Delete.
9485 (TYPE_SYMTAB_IS_DIE): Renumber.
9486 * xcoffout.c: Refer to former sdbout.c file.
9487 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
9489 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
9490 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
9491 (Debugging Options): Delete -gcoff.
9492 (-gstabs): Delete SDB reference.
9494 (-gcoff@var{level}): Delete.
9495 * doc/passes.texi (Debugging information output): Delete SDB and
9496 sdbout.c references.
9497 * doc/tm.texi: Regenerate.
9498 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
9499 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
9501 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
9502 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
9503 (SDB_DEBUGGING_INFO): Delete.
9504 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
9505 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
9506 * target.def (output_source_filename): Delete COFF reference.
9508 * common.opt (gcoff): Delete.
9509 (gxcoff+): Update Negative chain.
9510 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
9512 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
9513 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
9514 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
9516 (expand_function_start): Change sdb reference to past tense.
9517 (expand_function_end): Change sdb reference to past tense.
9518 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
9519 * opts.c (debug_type_names): Delete coff entry.
9520 (common_handle_option): Delete OPT_gcoff case.
9521 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
9523 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
9524 * config/cris/cris.h: Delete SDB reference in comment.
9525 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
9526 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
9527 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
9528 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
9530 (ix86_expand_prologue): Likewise.
9531 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
9532 * config/ia64/ia64.h: Likewise.
9533 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
9534 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
9536 * config/mmix/mmix.h: Likewise.
9537 * config/nds32/nds32.c: Likewise.
9538 * config/stormy/storym16.h: Likewise.
9539 * config/visium/visium.h: Likewise.
9540 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
9542 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
9544 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
9545 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
9546 FRAME_POINTER_REGNUM point at high end of local var area.
9548 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
9550 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
9551 Move comment around. Do not reset best_edge for a copiable
9552 destination if the copy would cause a partition change.
9553 (better_edge_p): Remove redundant check.
9555 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9557 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
9559 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9561 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
9562 for math functions that have _Float<N> and _Float<N>X variants.
9563 (mathfn_built_in_2): Add support for math functions that have
9564 _Float<N> and _Float<N>X variants.
9565 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9566 (expand_builtin_mathfn_ternary): Add support for fma with
9567 _Float<N> and _Float<N>X variants.
9568 (expand_builtin): Likewise.
9569 (fold_builtin_3): Likewise.
9570 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9571 create math function _Float<N> and _Float<N>X variants as external
9573 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9574 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9575 the __builtin_ prefix and if not strict ansi, without the prefix.
9576 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9577 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9578 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9579 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9580 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9581 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9582 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9583 function signatures for fma _Float<N> and _Float<N>X variants.
9584 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9585 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9586 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9587 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9588 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9589 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9590 * gencfn-macros.c (print_case_cfn): Add support for math functions
9591 that have _Float<N> and _Float<N>X variants.
9592 (print_define_operator_list): Likewise.
9593 (fltfn_suffixes): Likewise.
9595 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9596 for math functions that have _Float<N> and _Float<N>X variants.
9597 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9598 and _Float<N>X variants.
9599 (COPYSIGN): Likewise.
9602 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9603 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9605 (integer_valued_read_call_p): Likewise.
9606 * fold-const-call.c (fold_const_call_ss): Likewise.
9607 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9608 _Float<N> and _Float<N>X variants.
9609 (fold_const_call_ssss): Add support for fma _Float<N> and
9610 _Float<N>X variants.
9611 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9612 support for copysign and fma _Float<N> and _Float<N>X variants.
9613 (backprop::process_builtin_call_use): Likewise.
9614 * tree-call-cdce.c (can_test_argument_range); Add support for
9615 sqrt _Float<N> and _Float<N>X variants.
9616 (edom_only_function): Likewise.
9617 (get_no_error_domain): Likewise.
9618 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9619 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9620 copysign _Float<N> and _Float<N>X variants.
9621 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9622 handled by machine independent code.
9623 (FMAF128): Likewise.
9624 * doc/cpp.texi (Common Predefined Macros): Document defining
9625 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9626 fma _Float<N> and _Float<N>X variants.
9628 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9631 * config/i386/i386-modes.def (CCFPU): Remove definition.
9632 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9633 (ix86_cc_modes_compatible): Ditto.
9634 (ix86_expand_carry_flag_compare): Ditto.
9635 (ix86_expand_int_movcc): Ditto.
9636 (ix86_expand_int_addcc): Ditto.
9637 (ix86_reverse_condition): Ditto.
9638 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9639 Return true/false for unordered/ordered fp comparisons.
9640 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9641 (ix86_prepare_fp_compare_args): Update for rename.
9642 (ix86_expand_fp_compare): Update for rename. Generate unordered
9643 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9644 (ix86_expand_sse_compare_and_jump): Ditto.
9645 * config/i386/predicates.md (fcmov_comparison_operator):
9646 Remove CCFPU mode handling.
9647 (ix86_comparison_operator): Ditto.
9648 (ix86_carry_flag_operator): Ditto.
9649 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9650 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9651 (*cmpu<mode>_cc_i387): Ditto.
9652 (FPCMP): Remove mode iterator.
9653 (unord): Remove mode attribute.
9654 (unord_subst): New define_subst transformation
9655 (unord): New define_subst attribute.
9657 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9658 (*cmpi<unord>xf_i387): Ditto.
9659 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9660 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9661 using unord_subst transformation.
9662 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9663 (round_saeonly): Also handle CCFP mode.
9664 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9665 Remove UNSPEC_SAHF unspec handling.
9667 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9669 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9671 2017-10-27 Jeff Law <law@redhat.com>
9673 * gimple-ssa-sprintf.c: Include domwalk.h.
9674 (class sprintf_dom_walker): New class, derived from dom_walker.
9675 (sprintf_dom_walker::before_dom_children): New function.
9676 (struct call_info): Moved into sprintf_dom_walker class
9677 (compute_formath_length, handle_gimple_call): Likewise.
9678 (sprintf_length::execute): Call the dominator walker rather
9679 than walking the statements.
9681 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9682 gimple statement locations.
9683 (check_array_bounds): Corresponding changes. Get the statement's
9684 location directly from wi->stmt.
9686 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9689 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9691 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9693 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9694 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9696 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9699 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9700 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9701 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9702 maybe_get_pool_constant.
9703 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9706 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9708 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9709 2.26 caveat. Update gas and gld versions.
9710 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9713 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9715 * cgraph.h (set_malloc_flag): Declare.
9716 * cgraph.c (set_malloc_flag_1): New function.
9717 (set_malloc_flag): Likewise.
9718 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9719 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9721 (read_ipa_call_summary): Add support for reading is_return_callee.
9722 (write_ipa_call_summary): Stream is_return_callee.
9723 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9724 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9725 ipa-prop.h, ipa-fnsummary.h.
9726 (pure_const_names): Change to static.
9727 (malloc_state_e): Define.
9728 (malloc_state_names): Define.
9729 (funct_state_d): Add field malloc_state.
9730 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9731 (check_retval_uses): New function.
9732 (malloc_candidate_p): Likewise.
9733 (analyze_function): Add support for malloc attribute.
9734 (pure_const_write_summary): Stream malloc_state.
9735 (pure_const_read_summary): Add support for reading malloc_state.
9736 (dump_malloc_lattice): New function.
9737 (propagate_malloc): New function.
9738 (warn_function_malloc): New function.
9739 (ipa_pure_const::execute): Call propagate_malloc and
9740 ipa_free_fn_summary.
9741 (pass_local_pure_const::execute): Add support for malloc attribute.
9742 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9743 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9745 2017-10-27 Martin Liska <mliska@suse.cz>
9747 PR gcov-profile/82457
9748 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9749 for fork-like functions to be properly instrumented.
9751 2017-10-27 Richard Biener <rguenther@suse.de>
9754 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9755 info when we redirected EH.
9757 2017-10-26 Michael Collison <michael.collison@arm.com>
9759 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9761 (<optab>_trunchf<GPI:mode>2: New pattern.
9762 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9763 * config/aarch64/iterators.md (wv): New mode attribute.
9764 (vf, VF): New mode attributes.
9765 (vgp, VGP): New mode attributes.
9766 (s): Update attribute with SImode and DImode prefixes.
9768 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9770 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9771 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9772 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9773 (nios2_option_overide): Initialize it. Don't allow R0-relative
9774 addressing with PIC.
9775 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9776 (nios2_symbolic_constant_p): Likewise.
9777 (nios2_legitimate_address_p): Likewise.
9778 (nios2_r0rel_section_name_p): New.
9779 (nios2_symbol_ref_in_r0rel_data_p): New.
9780 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9781 (r0rel_constant_p): New.
9782 (nios2_print_operand_address): Handle r0rel_constant_p.
9783 (nios2_cdx_narrow_form_p): Likewise.
9784 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9785 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9786 (Nios II Options): Document -mr0rel-sec.
9788 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9790 * config/nios2/nios2.c: Include xregex.h.
9791 (nios2_gprel_sec_regex): New.
9792 (nios2_option_overide): Initialize it. Don't allow GP-relative
9793 addressing with PIC.
9794 (nios2_small_section_name_p): Check for regex match.
9795 * config/nios2/nios2.opt (mgprel-sec=): New option.
9796 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9797 (Nios II Options): Document -mgprel-sec.
9799 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9801 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9803 2017-10-26 Tom de Vries <tom@codesourcery.com>
9805 PR tree-optimization/82707
9806 * gimple.c (gimple_copy): Fix unsharing of
9807 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9809 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
9811 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
9812 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
9813 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
9814 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
9815 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
9816 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
9817 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
9818 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
9819 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
9820 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
9821 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
9822 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
9823 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
9824 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
9825 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
9826 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
9827 _mm512_mask_cmpunord_ps_mask): New intrinsics.
9829 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9831 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
9833 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
9834 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
9835 warning to rs6000.c. Remove the Undocumented flag, since it has
9837 (-mabi=ibmlongdouble): Likewise.
9838 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
9839 already set, set the default format for long double.
9840 (rs6000_debug_reg_global): Print whether long double is IBM or
9842 (rs6000_option_override_internal): Rework setting long double
9843 format. Only warn if the user is changing the long double default
9844 and they did not use -Wno-psabi.
9845 * doc/invoke.texi (PowerPC options): Update the documentation for
9846 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
9848 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9849 Alan Hayward <alan.hayward@arm.com>
9850 David Sherwood <david.sherwood@arm.com>
9852 * rtl.h (wider_subreg_mode): New function.
9853 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
9854 rather than an unsigned int *.
9855 * ira-color.c (regno_max_ref_width): Replace with...
9856 (regno_max_ref_mode): ...this new variable.
9857 (coalesced_pseudo_reg_slot_compare): Update accordingly.
9858 Use wider_subreg_mode.
9859 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
9860 rather than an unsigned int *.
9861 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
9862 (process_alt_operands): Likewise.
9863 (invariant_p): Likewise.
9864 * lra-spills.c (assign_mem_slot): Likewise.
9865 (add_pseudo_to_slot): Likewise.
9866 * lra.c (collect_non_operand_hard_regs): Likewise.
9867 (add_regs_to_insn_regno_info): Likewise.
9868 * reload1.c (regno_max_ref_width): Replace with...
9869 (regno_max_ref_mode): ...this new variable.
9870 (reload): Update accordingly. Update call to
9871 ira_sort_regnos_for_alter_reg.
9872 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
9873 (init_eliminable_invariants): Update to use regno_max_ref_mode.
9874 (scan_paradoxical_subregs): Likewise.
9876 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9878 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
9879 (aarch64_frame): Add emit_frame_chain boolean.
9880 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9881 Move eh_return case to aarch64_layout_frame.
9882 (aarch64_layout_frame): Initialize emit_frame_chain.
9883 (aarch64_expand_prologue): Use emit_frame_chain.
9885 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9887 * config/aarch64/aarch64.c (aarch64_layout_frame):
9888 Ensure LR is always stored at the bottom of the callee-saves.
9889 Remove rarely used frame layout which saves callee-saves at top of
9890 frame, so the store of LR can be used as a valid probe in all cases.
9892 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9894 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
9895 Improve unaligned TImode/TFmode base/offset split.
9897 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9898 Alan Hayward <alan.hayward@arm.com>
9899 David Sherwood <david.sherwood@arm.com>
9901 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
9902 * combine.c (find_single_use_1): Likewise.
9903 (expand_field_assignment): Likewise.
9904 (move_deaths): Likewise.
9905 * lra-constraints.c (simplify_operand_subreg): Likewise.
9906 (curr_insn_transform): Likewise.
9907 * lra.c (collect_non_operand_hard_regs): Likewise.
9908 (add_regs_to_insn_regno_info): Likewise.
9909 * rtlanal.c (reg_referenced_p): Likewise.
9910 (covers_regno_no_parallel_p): Likewise.
9912 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9914 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
9915 Don't print any bits outside the precision of the value.
9916 * wide-int.cc (test_printing): Add some new tests.
9918 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9920 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
9921 supports -xbrace_comment option.
9922 * configure: Regenerate.
9923 * config.in: Regenerate.
9924 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
9925 (ASM_CPU_SPEC): Use it.
9927 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9929 * target.def (static_rtx_alignment): New hook.
9930 * targhooks.h (default_static_rtx_alignment): Declare.
9931 * targhooks.c (default_static_rtx_alignment): New function.
9932 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
9933 * doc/tm.texi: Regenerate.
9934 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
9935 instead of targetm.constant_alignment. Remove call to
9937 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9938 (cris_preferred_mininum_alignment): New function, split out from...
9939 (cris_constant_alignment): ...here.
9940 (cris_static_rtx_alignment): New function.
9941 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
9943 (ix86_constant_alignment): ...here.
9944 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9945 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9946 (mmix_static_rtx_alignment): New function.
9947 * config/spu/spu.c (spu_static_rtx_alignment): New function.
9948 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9950 2017-10-26 Tamar Christina <tamar.christina@arm.com>
9953 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
9954 Add flag_trapping_math and flag_fp_int_builtin_inexact.
9956 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
9958 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
9959 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
9960 mark as a sign-extending load.
9961 (local_pic_load_u): Define.
9963 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
9966 * fold-const.c (operand_equal_for_comparison_p): Also return true
9967 if ARG0 is a simple variant of ARG1 with narrower precision.
9968 (fold_ternary_loc): Always pass unstripped operands to the predicate.
9970 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
9972 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
9974 * i386.h (processor_costs): Add gather_static, gather_per_elt,
9975 scatter_static, scatter_per_elt.
9976 * x86-tune-costs.h: Add new cost entries.
9978 2017-10-25 Richard Biener <rguenther@suse.de>
9980 * tree-ssa-sccvn.h (vn_eliminate): Declare.
9981 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
9982 class pass_fre): Move to ...
9983 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
9984 class pass_fre): ... here and adjust for statistics.
9986 2017-10-25 Jakub Jelinek <jakub@redhat.com>
9989 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
9990 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
9991 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
9992 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
9995 2017-10-25 Richard Biener <rguenther@suse.de>
9997 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
9998 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
9999 eliminate_push_avail, eliminate_insert): Move inside...
10000 (class eliminate_dom_walker): ... this class in preparation
10002 (fini_eliminate): Remove by merging with ...
10003 (eliminate): ... this function. Adjust for class changes.
10004 (pass_pre::execute): Remove fini_eliminate call.
10005 (pass_fre::execute): Likewise.
10007 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10010 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10011 (VPERMI2, VPERMI2I): New mode iterators.
10012 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10013 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10015 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10016 mode iterator. Remove 3 old define_insn patterns.
10017 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10018 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10019 VPERMI2 mode iterator, remove the other two expanders.
10020 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10021 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10022 instructions, remove the other two patterns.
10023 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10024 mode iterator, remove the other two patterns.
10025 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10026 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10027 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10028 and adjust argument order accordingly.
10029 (ix86_expand_vec_perm): Adjust caller.
10030 (expand_vec_perm_1): Likewise.
10031 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10032 (expand_vec_perm_vpermt2_vpshub2): ... this.
10033 (ix86_expand_vec_perm_const_1): Adjust caller.
10034 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10037 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10038 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10039 (vec_shl_<mode>): Remove unused expander.
10040 (avx512bw_<shift_insn><mode>3): New define_insn.
10041 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10042 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10044 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10047 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10050 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10052 PR rtl-optimization/82396
10053 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
10054 (autopref_multipass_init): Simplify initialization.
10055 (autopref_rank_data): Simplify sort order.
10056 * gcc/sched-int.h (autopref_multipass_data_): Remove
10057 multi_mem_insn_p, min_offset and max_offset.
10059 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10061 PR middle-end/60580
10062 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10063 Check special value of flag_omit_frame_pointer.
10064 (aarch64_can_eliminate): Likewise.
10065 (aarch64_override_options_after_change_1): Simplify handling of
10066 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
10068 2017-10-24 Richard Biener <rguenther@suse.de>
10070 PR tree-optimization/82697
10071 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
10072 zero for conditional load and unconditional store.
10074 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10076 * doc/install.texi: Document bootstrap-cet.
10078 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10081 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
10082 ENDBR instruction at function entrance if function is only
10085 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10088 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
10089 patterns to better describe from which operation the CF is computed.
10090 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
10091 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
10092 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
10093 is 0, use _0 suffixed expanders instead of emitting a comparison
10096 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
10098 * config/i386/i386.md(*movsf_internal, *movdf_internal):
10099 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
10101 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
10103 PR middle-end/82569
10104 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
10105 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
10106 * loop-iv.c (iv_get_reaching_def): Likewise.
10107 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
10108 variable is promoted and the partition contains undefined values.
10110 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10112 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
10114 (nios2_address_cost): Define.
10115 (nios2_legitimize_address): Recognize (exp + constant) directly.
10116 (TARGET_ADDRESS_COST): Define.
10118 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10120 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
10121 (nios2_symbolic_memory_operand_p): Declare.
10122 (nios2_split_large_constant): Declare.
10123 (nios2_split_symbolic_memory_operand): Declare.
10124 * config/nios2/nios2.c: Adjust includes.
10125 (nios2_symbolic_constant_allowed): New.
10126 (nios2_symbolic_constant_p): New.
10127 (nios2_plus_symbolic_constant_p): New.
10128 (nios2_valid_addr_expr_p): Recognize addresses involving
10129 symbolic constants.
10130 (nios2_legitimate_address_p): Likewise, also LO_SUM.
10131 (nios2_symbolic_memory_operand_p): New.
10132 (nios2_large_constant_p): New.
10133 (nios2_split_large_constant): New.
10134 (nios2_split_plus_large_constant): New.
10135 (nios2_split_symbolic_memory_operand): New.
10136 (nios2_legitimize_address): Code refactoring. Handle addresses
10137 involving symbolic constants.
10138 (nios2_emit_move_sequence): Likewise.
10139 (nios2_print_operand): Improve error output.
10140 (nios2_print_operand_address): Handle LO_SUM.
10141 (nios2_cdx_narrow_form_p): Likewise.
10142 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
10143 operands involving symbolic constants.
10144 (movhi_internal, movsi_internal): Likewise.
10145 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
10146 (extendhisi2, extendqi<mode>2): Likewise.
10148 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10150 * tree-pass.h (PROP_rtl_split_insns): Define.
10151 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
10153 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10155 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
10157 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10160 * target.def (const_not_ok_for_debug_p): Default to
10161 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
10162 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
10163 * targhooks.c (default_const_not_ok_for_debug_p): New function.
10164 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
10165 which targetm.const_not_ok_for_debug_p returned true.
10166 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
10168 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
10170 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
10171 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
10172 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
10174 (ix86_const_not_ok_for_debug_p): New function.
10175 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
10176 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
10178 2017-10-23 David Malcolm <dmalcolm@redhat.com>
10181 * system.h: Conditionally include "unique-ptr.h" if
10182 INCLUDE_UNIQUE_PTR is defined.
10183 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
10184 of defining INCLUDE_UNIQUE_PTR before including "system.h".
10186 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
10188 * config/rl78/rl78.md: New define_expand "subdi3".
10190 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
10193 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
10194 DF_REF_INSN if DF_REF_INSN_INFO is false.
10196 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
10198 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
10199 xmm_move instead of sse_move.
10200 (sse_store_index): New function.
10201 (ix86_register_move_cost): Be more sensible about mismatch stall;
10202 model AVX moves correctly; make difference between sse->integer and
10204 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
10205 moves; make difference between SSE and AVX.
10206 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
10207 and zmm_move. Increase size of sse load and store tables;
10208 add unaligned load and store tables; add ssemmx_to_integer.
10209 * x86-tune-costs.h: Update all entries according to real
10210 move latencies from Agner Fog's manual and chip documentation.
10212 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10215 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
10216 * config/i386/constraints.md (Wf): New constraint.
10217 * config/i386/i386.md (UNSPEC_SBB): New unspec.
10218 (cmp<dwi>_doubleword): Removed.
10219 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
10220 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
10221 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
10222 expand with cmp<dwi>_doubleword. For LTU and GEU use
10223 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
10225 * common.opt (gcolumn-info): Enable by default.
10226 * doc/invoke.texi (gcolumn-info): Document new default.
10228 2017-10-23 Richard Biener <rguenther@suse.de>
10230 PR tree-optimization/82672
10231 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
10232 Fold the stmt if we propagated into it.
10234 2017-10-23 Richard Biener <rguenther@suse.de>
10236 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
10237 (bitmap_remove_expr_from_set): ... this. All callers call this
10238 for non-constant values.
10239 (bitmap_set_subtract): Rename to...
10240 (bitmap_set_subtract_expressions): ... this. Adjust and
10242 (bitmap_set_contains_value): Remove superfluous check.
10243 (bitmap_set_replace_value): Inline into single caller ...
10244 (bitmap_value_replace_in_set): ... here and simplify.
10245 (dependent_clean): Merge into ...
10246 (clean): ... this using an overload. Adjust.
10247 (prune_clobbered_mems): Adjust.
10248 (compute_antic_aux): Likewise.
10249 (compute_partial_antic_aux): Likewise.
10251 2017-10-23 Richard Biener <rguenther@suse.de>
10253 PR tree-optimization/82129
10255 2017-08-01 Richard Biener <rguenther@suse.de>
10257 PR tree-optimization/81181
10258 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10259 (compute_antic): ... end of iteration here.
10261 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10263 * target.def (starting_frame_offset): New hook.
10264 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
10265 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
10266 * doc/tm.texi.in: Regenerate.
10267 * hooks.h (hook_hwi_void_0): Declare.
10268 * hooks.c (hook_hwi_void_0): New function.
10269 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
10270 STARTING_FRAME_OFFSET.
10271 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
10272 * reload1.c (reload): Likewise.
10273 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
10274 instead of STARTING_FRAME_OFFSET.
10275 * function.c (try_fit_stack_local): Likewise.
10276 (assign_stack_local_1): Likewise
10277 (instantiate_virtual_regs): Likewise.
10278 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
10279 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
10280 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
10281 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
10282 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
10283 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
10284 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
10285 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
10286 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
10287 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
10288 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
10289 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
10290 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
10291 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
10292 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
10293 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
10294 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
10295 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
10296 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
10297 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
10298 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
10299 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
10300 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
10301 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
10302 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
10303 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
10304 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
10305 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
10306 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
10307 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
10308 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
10309 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
10310 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
10311 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
10312 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
10313 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
10314 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
10315 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
10316 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
10317 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
10318 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
10319 * config/avr/avr.c (avr_starting_frame_offset): Make static and
10320 return a HOST_WIDE_INT.
10321 (avr_builtin_setjmp_frame_value): Use it instead of
10322 STARTING_FRAME_OFFSET.
10323 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10324 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
10325 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
10327 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10328 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
10329 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
10330 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10331 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
10332 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
10333 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10334 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
10335 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
10336 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10337 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
10338 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
10340 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10341 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
10342 * config/mips/mips.c (mips_compute_frame_info): Refer to
10343 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10344 (mips_starting_frame_offset): New function.
10345 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10346 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
10347 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
10348 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
10349 and return a HOST_WIDE_INT.
10350 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10351 (mmix_initial_elimination_offset): Refer to
10352 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10353 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
10354 * config/pa/pa.c (pa_starting_frame_offset): New function.
10355 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
10356 (pa_expand_prologue): Likewise.
10357 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10358 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
10359 !FRAME_GROWS_DOWNWARD handling to...
10360 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10361 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
10362 !FRAME_GROWS_DOWNWARD handling to...
10363 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10364 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
10365 !FRAME_GROWS_DOWNWARD handling to...
10366 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10367 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
10369 (rs6000_starting_frame_offset): New function.
10370 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
10371 !FRAME_GROWS_DOWNWARD handling to...
10372 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10373 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
10374 !FRAME_GROWS_DOWNWARD handling to...
10375 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10376 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
10377 !FRAME_GROWS_DOWNWARD handling to...
10378 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10379 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
10380 (rs6000_starting_frame_offset): New function.
10381 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
10382 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
10383 * config/vax/vax.c (vax_starting_frame_offset): New function.
10384 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
10385 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10386 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
10387 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
10388 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10389 * system.h (STARTING_FRAME_OFFSET): Poison.
10391 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10393 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10394 SCALAR_TYPE_MODE instead of TYPE_MODE.
10396 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10397 Alan Hayward <alan.hayward@arm.com>
10398 David Sherwood <david.sherwood@arm.com>
10400 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
10402 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10403 Alan Hayward <alan.hayward@arm.com>
10404 David Sherwood <david.sherwood@arm.com>
10406 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
10408 2017-10-23 Richard Biener <rguenther@suse.de>
10410 PR tree-optimization/82129
10411 * tree-ssa-pre.c (bitmap_set_and): Remove.
10412 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
10413 canonicalizing expressions in the set to those with lowest
10414 ID rather than taking that from the first edge.
10416 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10418 * combine.c (rtx_equal_for_field_assignment_p): Use
10419 byte_lowpart_offset.
10421 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10422 Alan Hayward <alan.hayward@arm.com>
10423 David Sherwood <david.sherwood@arm.com>
10425 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
10426 to a SUBREG_PROMOTED_VAR.
10428 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10429 Alan Hayward <alan.hayward@arm.com>
10430 David Sherwood <david.sherwood@arm.com>
10432 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
10433 (expand_debug_source_expr): Likewise.
10434 * combine.c (combine_simplify_rtx): Likewise.
10435 * cse.c (fold_rtx): Likewise.
10436 * optabs.c (expand_float): Likewise.
10437 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10438 (simplify_binary_operation_1): Likewise.
10440 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10441 Alan Hayward <alan.hayward@arm.com>
10442 David Sherwood <david.sherwood@arm.com>
10444 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
10445 (record_promoted_value): Likewise.
10446 * expr.c (expand_expr_real_2): Likewise.
10447 * ree.c (update_reg_equal_equiv_notes): Likewise.
10448 (combine_set_extension): Likewise.
10449 * rtlanal.c (low_bitmask_len): Likewise.
10450 * simplify-rtx.c (neg_const_int): Likewise.
10451 (simplify_binary_operation_1): Likewise.
10453 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10454 Alan Hayward <alan.hayward@arm.com>
10455 David Sherwood <david.sherwood@arm.com>
10457 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
10458 * regcprop.c (maybe_mode_change): Likewise.
10459 * reload1.c (alter_reg): Likewise.
10461 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10463 * inchash.h (inchash::hash::add_wide_int): New function.
10464 * lto-streamer-out.c (hash_tree): Use it.
10466 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10468 * inchash.h (inchash::hash::add_wide_int): Rename to...
10469 (inchash::hash::add_hwi): ...this.
10470 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
10471 (polymorphic_call_target_hasher::hash): Likewise.
10472 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
10473 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
10474 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
10475 * lto-streamer-out.c (hash_tree): Likewise.
10476 * optc-save-gen.awk: Likewise.
10477 * tree.c (add_expr): Likewise.
10479 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10482 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
10483 for ordered inequality comparisons even with TARGET_IEEE_FP.
10485 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10488 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
10489 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10490 Expand with cmp<dwi>_doubleword.
10492 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10494 * extend.texi: Add x86 specific to 'nocf_check' attribute.
10495 List CET intrinsics.
10496 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
10497 specific to -fcf-protection option.
10499 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10501 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
10502 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
10503 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
10504 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
10505 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
10506 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
10507 (extra_objs): Add cet.o for Linux/x86 targets.
10508 (tmake_file): Add i386/t-cet for Linux/x86 targets.
10509 * config/i386/cet.c: New file.
10510 * config/i386/cetintrin.h: Likewise.
10511 * config/i386/t-cet: Likewise.
10512 * config/i386/cpuid.h (bit_SHSTK): New.
10513 (bit_IBT): Likewise.
10514 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
10515 pass IBT and SHSTK bits.
10516 * config/i386/i386-builtin-types.def
10517 (VOID_FTYPE_UNSIGNED_PVOID): New.
10518 (VOID_FTYPE_UINT64_PVOID): Likewise.
10519 * config/i386/i386-builtin.def: Add CET intrinsics.
10520 * config/i386/i386-c.c (ix86_target_macros_internal): Add
10521 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
10522 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
10523 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
10525 * config/i386/i386.c (rest_of_insert_endbranch): New.
10526 (pass_data_insert_endbranch): Likewise.
10527 (pass_insert_endbranch): Likewise.
10528 (make_pass_insert_endbranch): Likewise.
10529 (ix86_notrack_prefixed_insn_p): Likewise.
10530 (ix86_target_string): Add -mibt, -mshstk flags.
10531 (ix86_option_override_internal): Add flag_cf_protection
10533 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
10534 (ix86_print_operand): Add 'notrack' prefix output.
10535 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
10536 (ix86_expand_builtin): Expand CET intrinsics.
10537 (x86_output_mi_thunk): Add 'endbranch' instruction.
10538 * config/i386/i386.h (TARGET_IBT): New.
10539 (TARGET_IBT_P): Likewise.
10540 (TARGET_SHSTK): Likewise.
10541 (TARGET_SHSTK_P): Likewise.
10542 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
10543 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
10544 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
10545 (builtin_setjmp_setup): New pattern.
10546 (builtin_longjmp): Likewise.
10547 (rdssp<mode>): Likewise.
10548 (incssp<mode>): Likewise.
10549 (saveprevssp): Likewise.
10550 (rstorssp): Likewise.
10551 (wrss<mode>): Likewise.
10552 (wruss<mode>): Likewise.
10553 (setssbsy): Likewise.
10554 (clrssbsy): Likewise.
10555 (nop_endbr): Likewise.
10556 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
10558 * config/i386/immintrin.h: Include <cetintrin.h>.
10559 * config/i386/linux-common.h
10560 (file_end_indicate_exec_stack_and_cet): New prototype.
10561 (TARGET_ASM_FILE_END): New.
10563 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10565 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10566 latencies instead of having separate table; make difference between
10567 integer and float costs.
10568 * i386.h (processor_costs): Remove scalar_stmt_cost,
10569 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10570 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10572 * x86-tune-costs.h: Remove entries which has been removed in
10573 procesor_costs from all tables; make cond_taken_branch_cost
10574 and cond_not_taken_branch_cost COST_N_INSNS based.
10576 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10578 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10580 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10582 * config/i386/i386.md (isa): Remove fma_avx512f.
10583 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10584 <avx512>_fmadd_<mode>_mask3<round_name>,
10585 <avx512>_fmsub_<mode>_mask<round_name>,
10586 <avx512>_fmsub_<mode>_mask3<round_name>,
10587 <avx512>_fnmadd_<mode>_mask<round_name>,
10588 <avx512>_fnmadd_<mode>_mask3<round_name>,
10589 <avx512>_fnmsub_<mode>_mask<round_name>,
10590 <avx512>_fnmsub_<mode>_mask3<round_name>,
10591 <avx512>_fmaddsub_<mode>_mask<round_name>,
10592 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10593 <avx512>_fmsubadd_<mode>_mask<round_name>,
10594 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10595 (*vec_widen_umult_even_v16si<mask_name>,
10596 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10597 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10599 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10601 * extend.texi: Add 'nocf_check' documentation.
10602 * gimple.texi: Add second parameter to
10603 gimple_build_call_from_tree.
10604 * invoke.texi: Add -fcf-protection documentation.
10605 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10607 2017-10-20 Richard Biener <rguenther@suse.de>
10609 PR tree-optimization/82473
10610 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10611 the largest input type.
10613 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10615 * c-attribs.c (handle_nocf_check_attribute): New function.
10616 (c_common_attribute_table): Add 'nocf_check' handling.
10617 * gimple-parser.c: Add second argument NULL to
10618 gimple_build_call_from_tree.
10619 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10620 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10622 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10623 * common.opt: Add fcf-protection flag.
10624 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10625 * flag-types.h: Add enum cf_protection_level.
10626 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10627 Add 'nocf_check' attribute propagation to gimple call.
10628 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10629 (gimple_build_call_from_tree): Update prototype.
10630 (gimple_call_nocf_check_p): New function.
10631 (gimple_call_set_nocf_check): Likewise.
10632 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10633 * ipa-icf.c: Add nocf_check attribute in statement hash.
10634 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10635 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10636 * toplev.c (process_options): Add flag_cf_protection handling.
10638 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10640 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10642 2017-10-20 Richard Biener <rguenther@suse.de>
10644 PR tree-optimization/82603
10645 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10646 remove false predicated stores.
10648 2017-10-20 Richard Biener <rguenther@suse.de>
10650 * graphite-isl-ast-to-gimple.c
10651 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10652 Remove return value and simplify, dump copied stmt after lhs
10654 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10655 Reduce dump verbosity.
10656 (gsi_insert_earliest): Likewise.
10657 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10658 * graphite.c (print_global_statistics): Adjust dumping.
10659 (print_graphite_scop_statistics): Likewise.
10660 (print_graphite_statistics): Do not dump loops here.
10661 (graphite_transform_loops): But here.
10663 2017-10-20 Nicolas Roche <roche@adacore.com>
10665 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10666 * configure: Regenerate.
10668 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10671 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10672 functions when optimizing replace GIMPLE_RETURN stmts with
10673 calls to __builtin_unreachable ().
10676 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10677 for -fsanitize=thread link of executables.
10678 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10679 link of executables.
10682 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10683 New mode iterators.
10684 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10685 define_insns for logical vector shifts to use VI248_AVX512BW
10686 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10687 condition, useless isa and prefix attributes. Change the first
10688 2 of these define_insns to ...
10689 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10690 define_insn for avx512vl.
10691 (<shift_insn><mode>3): ... and this, new define_insn without
10692 masking for non-avx512vl.
10695 * config/i386/sse.md (*andnot<mode>3,
10696 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10697 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10698 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10699 not applied use empty suffix even for TARGET_AVX512VL.
10700 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10701 is applied, supply evex,evex or evex,evex,evex instead of just
10704 2017-10-20 Julia Koval <julia.koval@intel.com>
10706 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10707 (OPTION_MASK_ISA_GFNI_UNSET): New.
10708 (ix86_handle_option): Handle OPT_mgfni.
10709 * config/i386/cpuid.h (bit_GFNI): New.
10710 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10711 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10712 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10713 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10714 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10715 * config/i386/i386.opt: Add mgfni.
10717 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10719 * config/msp430/msp430.c (msp430_option_override): Disable
10720 -fdelete-null-pointer-checks.
10721 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10723 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10725 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10726 of x87 and SSE instructions.
10728 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10730 * asan.c (create_cond_insert_point): Do not update edge count.
10731 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10732 (afdo_propagate_circuit): Likewise.
10733 (afdo_calculate_branch_prob): Likewise.
10734 (afdo_annotate_cfg): Likewise.
10735 * basic-block.h (struct edge_def): Remove count.
10736 (edge_def::count): New accessor.
10737 * bb-reorder.c (rotate_loop): Update.
10738 (find_traces_1_round): Update.
10739 (connect_traces): Update.
10740 (sanitize_hot_paths): Update.
10741 * cfg.c (unchecked_make_edge): Update.
10742 (make_single_succ_edge): Update.
10743 (check_bb_profile): Update.
10744 (dump_edge_info): Update.
10745 (update_bb_profile_for_threading): Update.
10746 (scale_bbs_frequencies_int): Update.
10747 (scale_bbs_frequencies_gcov_type): Update.
10748 (scale_bbs_frequencies_profile_count): Update.
10749 (scale_bbs_frequencies): Update.
10750 * cfganal.c (connect_infinite_loops_to_exit): Update.
10751 * cfgbuild.c (compute_outgoing_frequencies): Update.
10752 (find_many_sub_basic_blocks): Update.
10753 * cfgcleanup.c (try_forward_edges): Update.
10754 (try_crossjump_to_edge): Update
10755 * cfgexpand.c (expand_gimple_cond): Update
10756 (expand_gimple_tailcall): Update
10757 (construct_exit_block): Update
10758 * cfghooks.c (verify_flow_info): Update
10759 (redirect_edge_succ_nodup): Update
10760 (split_edge): Update
10761 (make_forwarder_block): Update
10762 (duplicate_block): Update
10763 (account_profile_record): Update
10764 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10765 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10766 * cfgloopmanip.c (scale_loop_profile): Update.
10768 (lv_adjust_loop_entry_edge): Update.
10769 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10770 (force_nonfallthru_and_redirect): Update.
10771 (purge_dead_edges): Update.
10772 (rtl_flow_call_edges_add): Update.
10773 * cgraphunit.c (init_lowered_empty_function): Update.
10774 (cgraph_node::expand_thunk): Update.
10775 * gimple-pretty-print.c (dump_probability): Update.
10776 (dump_edge_probability): Update.
10777 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10778 * haifa-sched.c (sched_create_recovery_edges): Update.
10779 * hsa-gen.c (convert_switch_statements): Update.
10780 * ifcvt.c (dead_or_predicable): Update.
10781 * ipa-inline-transform.c (inline_transform): Update.
10782 * ipa-split.c (split_function): Update.
10783 * ipa-utils.c (ipa_merge_profiles): Update.
10784 * loop-doloop.c (add_test): Update.
10785 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10786 * lto-streamer-in.c (input_cfg): Update.
10787 (input_function): Update.
10788 * lto-streamer-out.c (output_cfg): Update.
10789 * modulo-sched.c (sms_schedule): Update.
10790 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10791 * predict.c (maybe_hot_edge_p): Update.
10792 (unlikely_executed_edge_p): Update.
10793 (probably_never_executed_edge_p): Update.
10794 (dump_prediction): Update.
10795 (drop_profile): Update.
10796 (propagate_unlikely_bbs_forward): Update.
10797 (determine_unlikely_bbs): Update.
10798 (force_edge_cold): Update.
10799 * profile.c (compute_branch_probabilities): Update.
10800 * reg-stack.c (better_edge): Update.
10801 * shrink-wrap.c (handle_simple_exit): Update.
10802 * tracer.c (better_p): Update.
10803 * trans-mem.c (expand_transaction): Update.
10804 (split_bb_make_tm_edge): Update.
10805 * tree-call-cdce.c: Update.
10806 * tree-cfg.c (gimple_find_sub_bbs): Update.
10807 (gimple_split_edge): Update.
10808 (gimple_duplicate_sese_region): Update.
10809 (gimple_duplicate_sese_tail): Update.
10810 (gimple_flow_call_edges_add): Update.
10811 (insert_cond_bb): Update.
10812 (execute_fixup_cfg): Update.
10813 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
10814 * tree-complex.c (expand_complex_div_wide): Update.
10815 * tree-eh.c (lower_resx): Update.
10816 (unsplit_eh): Update.
10817 (cleanup_empty_eh_move_lp): Update.
10818 * tree-inline.c (copy_edges_for_bb): Update.
10819 (freqs_to_counts): Update.
10820 (copy_cfg_body): Update.
10821 * tree-ssa-dce.c (remove_dead_stmt): Update.
10822 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10823 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10824 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10825 (unloop_loops): Update.
10826 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10827 * tree-ssa-loop-split.c (connect_loops): Update.
10828 (split_loop): Update.
10829 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10830 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10831 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10832 * tree-ssa-reassoc.c (branch_fixup): Update.
10833 * tree-ssa-tail-merge.c (replace_block_by): Update.
10834 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10835 (compute_path_counts): Update.
10836 (update_profile): Update.
10837 (recompute_probabilities): Update.
10838 (update_joiner_offpath_counts): Update.
10839 (estimated_freqs_path): Update.
10840 (freqs_to_counts_path): Update.
10841 (clear_counts_path): Update.
10842 (ssa_fix_duplicate_block_edges): Update.
10843 (duplicate_thread_path): Update.
10844 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10845 (case_bit_test_cmp): Update.
10846 (collect_switch_conv_info): Update.
10847 (gen_inbound_check): Update.
10848 (do_jump_if_equal): Update.
10849 (emit_cmp_and_jump_insns): Update.
10850 * tree-tailcall.c (decrease_profile): Update.
10851 (eliminate_tail_call): Update.
10852 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10853 (vect_do_peeling): Update.
10854 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10855 * ubsan.c (ubsan_expand_null_ifn): Update.
10856 (ubsan_expand_ptr_ifn): Update.
10857 * value-prof.c (gimple_divmod_fixed_value): Update.
10858 (gimple_mod_pow2): Update.
10859 (gimple_mod_subtract): Update.
10860 (gimple_ic): Update.
10861 (gimple_stringop_fixed_value): Update.
10863 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10866 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
10868 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
10870 PR rtl-optimization/82395
10871 * ira-color.c (allocno_priority_compare_func): Fix comparison step
10872 based on non_spilled_static_chain_regno_p.
10874 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10876 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
10877 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
10878 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
10880 2017-10-19 Martin Sebor <msebor@redhat.com>
10882 PR tree-optimization/82596
10883 * tree.c (array_at_struct_end_p): Handle STRING_CST.
10885 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10887 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
10888 (get_mem_refs_of_builtin_call): Likewise.
10889 * builtins.c (expand_builtin_apply): Adjust call to
10890 allocate_dynamic_stack_space.
10891 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
10892 the third argument to allocate_dynamic_stack_space, otherwise -1.
10893 (expand_builtin): Deal with all alloca variants.
10894 (is_inexpensive_builtin): Likewise.
10895 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
10896 * calls.c (special_function_p): Deal with all alloca variants.
10897 (initialize_argument_information): Adjust call to
10898 allocate_dynamic_stack_space.
10899 (expand_call): Likewise.
10900 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
10901 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
10902 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
10903 use it for the stack usage computation.
10904 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
10905 * function.c (gimplify_parameters): Call build_alloca_call_expr.
10906 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
10907 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
10908 (in_loop_p): Remove first argument and useless check.
10909 (pass_walloca::execute): Remove useless test and adjust call to above.
10910 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
10911 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
10912 (gimplify_call_expr): Deal with all alloca variants.
10913 * hsa-gen.c (gen_hsa_alloca): Likewise.
10914 (gen_hsa_insns_for_call): Likewise.
10915 * ipa-pure-const.c (special_builtin_state): Likewise.
10916 * tree-chkp.c (chkp_build_returned_bound): Likewise.
10917 * tree-object-size.c (alloc_object_size): Likewise.
10918 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
10919 (call_may_clobber_ref_p_1): Likewise.
10920 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
10921 (ccp_fold_stmt): Likewise.
10922 (optimize_stack_restore): Likewise.
10923 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10924 (mark_all_reaching_defs_necessary_1): Likewise.
10925 (propagate_necessity): Likewise.
10926 (eliminate_unnecessary_stmts): Likewise.
10927 * tree.c (build_common_builtin_nodes): Build
10928 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
10929 (build_alloca_call_expr): New function.
10930 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
10931 (CASE_BUILT_IN_ALLOCA): Likewise.
10932 (build_alloca_call_expr): Declare.
10933 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
10935 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10938 * dwarf2out.c (new_die_raw): New static inline function.
10939 (new_die): Use it to create the DIE.
10940 (add_AT_external_die_ref): Likewise.
10941 (clone_die): Likewise.
10942 (clone_as_declaration): Likewise.
10943 (dwarf2out_vms_debug_main_pointer): Likewise.
10944 (base_type_die): Likewise. Remove early return for corner cases.
10945 Do not call add_pubtype on the DIE here.
10946 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
10947 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
10948 typedefs for base types with DW_AT_endianity. Make sure a DIE with
10949 native order exists for base types, attach the DIE manually and call
10950 add_pubtype on it. Do not equate a reverse order DIE to the type.
10952 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
10954 * config/arm/arm.c (align_ok_ldrd_strd): New function.
10955 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
10957 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
10959 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10961 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
10962 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
10963 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
10964 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
10965 * opts.c (sanitizer_opts): Add builtin.
10966 * ubsan.c (instrument_builtin): New function.
10967 (pass_ubsan::execute): Call it.
10968 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
10969 * doc/invoke.texi: Document -fsanitize=builtin.
10971 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
10972 builtins, store max (log2 (align), 0) into uchar field instead of
10973 align into uptr field.
10974 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
10975 store uchar 0 field instead of uptr 0 field.
10976 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
10977 instead of passing one address of struct with 2 locations pass
10978 two addresses of structs with 1 location each.
10979 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
10980 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
10981 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
10982 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
10983 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
10984 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
10985 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
10986 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
10988 2017-10-19 Martin Liska <mliska@suse.cz>
10991 * file-find.c (remove_prefix): Remove.
10992 * file-find.h (remove_prefix): Likewise.
10993 * gcc-ar.c: Remove smartness of lookup.
10995 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
10997 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
10998 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
10999 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11001 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11004 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11005 (setcc + and to xor + setcc): New peephole2.
11007 2017-10-19 Tom de Vries <tom@codesourcery.com>
11009 * doc/sourcebuild.texi (Test Directives, Variants of
11010 dg-require-support): Add dg-require-stack-size.
11012 2017-10-19 Martin Liska <mliska@suse.cz>
11015 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11016 that have a large alignment.
11017 (gimplify_target_expr): Likewise.
11019 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11021 PR rtl-optimization/82602
11022 * ira.c (rtx_moveable_p): Return false for volatile asm.
11024 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11027 * config/i386/i386-modes.def (CCGZ): New CC mode.
11028 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11029 * config/i386/predicates.md (ix86_comparison_operator):
11031 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11032 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11033 with double-word subtraction.
11034 (put_condition_code): Handle CCGZmode.
11036 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11038 * wide-int.cc (debug (const wide_int &)): New.
11039 (debug (const wide_int *)): New.
11040 (debug (const widest_int &)): New.
11041 (debug (const widest_int *)): New.
11043 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11045 PR middle-end/82556
11046 * lra-constraints.c (curr_insn_transform): Use non-input operand
11047 instead of output one for matched reload.
11049 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11051 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11052 (tree-ssa-loop-ivopts.h): New header file.
11053 (struct builtin_info): New fields.
11054 (classify_builtin_1): Compute and record base and offset parts for
11055 memset builtin partition by calling strip_offset.
11056 (offset_cmp, fuse_memset_builtins): New functions.
11057 (finalize_partitions): Fuse adjacent memset partitions by calling
11059 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
11060 Expose the interface.
11061 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
11063 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11065 PR tree-optimization/82574
11066 * tree-loop-distribution.c (find_single_drs): New parameter. Check
11067 that data reference must be executed exactly once per iteration
11068 against the outermost loop in nest.
11069 (classify_partition): Update call to above function.
11071 2017-10-18 Richard Biener <rguenther@suse.de>
11073 PR tree-optimization/82591
11074 * graphite.c (graphite_transform_loops): Move code gen message
11076 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11077 Here. Handle scop_to_isl_ast failing.
11078 (scop_to_isl_ast): Limit the number of ISL operations.
11080 2017-10-18 Richard Biener <rguenther@suse.de>
11082 * graphite-isl-ast-to-gimple.c
11083 (translate_isl_ast_to_gimple::set_rename): Simplify.
11084 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
11085 (graphite_copy_stmts_from_block): ... here.
11086 (copy_bb_and_scalar_dependences): Simplify.
11087 (add_parameters_to_ivs_params): Canonicalize.
11088 (generate_entry_out_of_ssa_copies): Simplify.
11089 * graphite-sese-to-poly.c (extract_affine_name): Simplify
11090 by passing in ISL dimension.
11091 (parameter_index_in_region_1): Rename to ...
11092 (parameter_index_in_region): ... this.
11093 (extract_affine): Adjust assert, pass down parameter index.
11094 (add_param_constraints): Use range-info when available.
11095 (build_scop_context): Adjust.
11096 * sese.c (new_sese_info): Adjust.
11097 (free_sese_info): Likewise.
11098 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
11099 Remove unused typedefs.
11100 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
11102 2017-10-18 Martin Liska <mliska@suse.cz>
11104 * combine.c (simplify_compare_const): Add gcc_fallthrough.
11106 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11108 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
11109 (s390_sched_init): Do not reset s390_sched_state if we entered the
11110 current basic block via a fallthru edge and all others are unlikely.
11112 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11114 * config/s390/s390.c (NUM_SIDES): New variable.
11115 (LONGRUNNING_THRESHOLD): New variable.
11116 (LATENCY_FACTOR): New variable.
11117 (s390_sched_score): Decrease score for long-running instructions on
11119 (s390_sched_variable_issue): Perform bookkeeping for long-running
11122 2017-10-18 Richard Biener <rguenther@suse.de>
11124 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
11125 Simplify with removal of the parameter rename map.
11126 (set_rename): Likewise.
11127 (should_copy_to_new_region): Likewise.
11128 (graphite_copy_stmts_from_block): Likewise.
11129 (copy_bb_and_scalar_dependences): Remove initialization of
11130 unused copied_bb_map.
11131 (copy_def): Remove.
11132 (copy_internal_parameters): Likewise.
11133 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
11134 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
11135 Use INTEGRAL_TYPE_P.
11136 (parameter_index_in_region_1): Rename to ...
11137 (assign_parameter_index_in_region): ... this. Assert we have
11138 a parameter we handle.
11139 (scan_tree_for_params): Adjust.
11140 * sese.h (parameter_rename_map_t): Remove.
11141 (struct sese_info_t): Remove unused parameter_rename_map and
11142 copied_bb_map members.
11143 * sese.c (new_sese_info): Adjust.
11144 (free_sese_info): Likewise.
11146 2017-10-18 Martin Liska <mliska@suse.cz>
11149 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
11150 on an abnormal edge.
11152 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
11154 * doc/invoke.texi (ffunction-sections and fdata-sections):
11157 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11159 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
11160 the use statement can throw internally.
11162 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11164 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
11165 any RTX present on the RHS of a SET.
11166 * compare-elim.c (try_eliminate_compare): Restore comment.
11168 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11170 * langhooks.h (struct lang_hooks): Document that tree_size langhook
11171 may be also called on tcc_type nodes.
11172 * langhooks.c (lhd_tree_size): Likewise.
11174 2017-10-17 David Malcolm <dmalcolm@redhat.com>
11176 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
11177 format_warning_at_substring.
11178 (maybe_warn): Convert source_range * param to a location_t. Pass
11179 UNKNOWN_LOCATION rather than NULL to fmtwarn.
11180 (format_directive): Remove code to extract source_ranges and
11181 source_range * in favor of just a location_t.
11182 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
11184 * substring-locations.c (format_warning_va): Convert
11185 source_range * param to a location_t.
11186 (format_warning_at_substring): Likewise.
11187 * substring-locations.h (format_warning_va): Likewise.
11188 (format_warning_at_substring): Likewise.
11190 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
11192 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
11194 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
11195 and scatter/gather ops.
11197 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
11198 vec_gather_load and vec_scatter_store.
11199 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
11200 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
11201 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
11202 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
11203 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
11204 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
11206 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
11208 * reg-stack.c (compare_for_stack_reg): Add bool argument.
11209 Detect FTST instruction and handle its register pops. Only pop
11210 second operand if can_pop_second_op is true.
11211 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
11212 set can_pop_second_op to false in the compare_for_stack_reg call.
11214 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
11215 output_fp_compare for stack register operands.
11216 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
11217 instructions here. Do not emit stack register pops here. Assert
11218 that FCOMPP pops next to top stack register. Rewrite function.
11220 2017-10-17 Nathan Sidwell <nathan@acm.org>
11222 PR middle-end/82577
11223 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
11224 use DECL_ASSEMBLER_NAME_RAW.
11226 PR middle-end/82546
11227 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
11230 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
11231 Wilco Dijkstra <wilco.dijkstra@arm.com>
11233 * builtins.c (expand_builtin_update_setjmp_buf): Add a
11234 converstion to Pmode from the buf_addr.
11236 2017-10-17 Richard Biener <rguenther@suse.de>
11238 * graphite-dependences.c (scop_get_reads_and_writes): Change
11239 output parameters to references.
11241 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
11243 PR 71026/tree-optimization
11244 * fold-const.c (distribute_real_division): Removed.
11245 (fold_binary_loc): Remove calls to distribute_real_divison.
11247 2017-10-17 Richard Biener <rguenther@suse.de>
11249 * graphite-scop-detection.c
11250 (scop_detection::stmt_has_simple_data_refs_p): Always use
11251 the full nest as region.
11252 (try_generate_gimple_bb): Likewise.
11253 * sese.c (scalar_evolution_in_region): Simplify now that
11254 SCEV can handle instantiation in regions.
11255 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
11256 in the non-loop part of a function if requested.
11258 2017-10-17 Richard Biener <rguenther@suse.de>
11260 PR tree-optimization/82563
11261 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
11263 (graphite_regenerate_ast_isl): Call it.
11264 * graphite-scop-detection.c (build_scops): Remove entry edge split.
11266 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11268 PR tree-optimization/82549
11269 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
11270 Formatting fixes. Instead of calling make_bit_field_ref with negative
11273 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
11275 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
11276 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
11277 _mm_maskz_reduce_ss): New.
11278 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
11279 __builtin_ia32_reducess_mask): Ditto..
11280 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
11281 * config/i386/sse.md (reduces<mode>): Renamed to ...
11282 (reduces<mode><mask_scalar_name>): ... this.
11283 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
11285 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
11286 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
11288 2017-10-16 David Malcolm <dmalcolm@redhat.com>
11290 * Makefile.in (OBJS): Add unique-ptr-tests.o.
11291 * selftest-run-tests.c (selftest::run_tests): Call
11292 selftest::unique_ptr_tests_cc_tests.
11293 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
11294 * unique-ptr-tests.cc: New file.
11296 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
11299 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11301 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11302 (make_hard_regno_born, make_hard_regno_dead): Update
11303 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
11305 2017-10-16 Jeff Law <law@redhat.com>
11307 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11309 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
11311 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
11313 2017-10-16 Olivier Hainque <hainque@adacore.com>
11315 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
11316 with_cpu if we were configured for an e500v2 target cpu name.
11318 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
11320 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
11321 * doc/invoke.texi: Document +nodsp as a valid extension for
11324 2017-10-16 Martin Liska <mliska@suse.cz>
11326 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
11327 (test_set_range): Likewise.
11328 (test_range_functions): Rename to ...
11329 (test_bit_in_range): ... this.
11330 (sbitmap_c_tests): Add new test.
11332 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11334 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
11336 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
11337 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
11339 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11341 * config/aarch64/aarch64-builtins.c
11342 (aarch64_types_quadopu_lane_qualifiers): New.
11343 (TYPES_QUADOPU_LANE): New.
11344 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
11345 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
11346 (aarch64_<sur>dot_laneq<vsi2qi>): New.
11347 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
11348 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
11349 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
11350 (Vdottype, DOTPROD): New.
11351 (sur): Add SDOT and UDOT.
11353 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11355 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
11356 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
11357 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
11358 Add TARGET_DOTPROD.
11359 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
11360 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
11361 Enable TARGET_DOTPROD.
11362 (cortex-a75.cortex-a55): Likewise.
11363 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
11365 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11367 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
11368 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
11370 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
11372 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
11373 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
11374 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
11375 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
11376 * config/arm/types.md (neon_dot, neon_dot_q): New.
11377 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
11379 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11381 * config/arm/arm.h (TARGET_DOTPROD): New.
11382 * config/arm/arm.c (arm_arch_dotprod): New.
11383 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
11384 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
11385 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
11386 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
11387 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
11388 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
11389 * doc/invoke.texi (armv8.2-a): Document dotprod
11391 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11393 * i386.c (ix86_vec_cost): New function.
11394 (ix86_rtx_costs): Handle vector operations better.
11395 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
11396 * x86-tune-costs.h: Add new costs to all tables.
11398 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11400 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
11402 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
11403 divsd, sqrtss and sqrtsd
11404 * x86-tune-costs.h: Add new entries to all costs.
11405 (znver1_cost): Fix to match real instruction latencies.
11407 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11408 Michael Collison <michael.collison@arm.com>
11410 * compare-elim.c: Include emit-rtl.h.
11411 (can_merge_compare_into_arith): New function.
11412 (try_validate_parallel): Likewise.
11413 (try_merge_compare): Likewise.
11414 (try_eliminate_compare): Call the above when no previous clobber
11416 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
11419 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11421 PR middle-end/62263
11422 PR middle-end/82498
11423 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
11424 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
11426 PR middle-end/62263
11427 PR middle-end/82498
11428 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
11429 to be any operand_equal_p operands. For & (B - 1) require
11430 B to be power of 2. Recognize
11431 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
11433 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11436 * optabs.c (expand_memory_blockage): Fix call of
11437 targetm.have_memory_blockage.
11439 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11442 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
11443 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
11444 objects to extra_objs instead of overwriting it.
11446 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11448 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
11449 Use any_fp_register_operand as operand[3] predicate. Simplify
11450 equality test for operands[2] and operands[4] memory location.
11451 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
11452 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
11453 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
11454 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
11455 any_fp_register_operand as operand[1] predicate. Simplify
11456 equality test for operands[0] and operands[3] memory location.
11457 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
11458 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
11459 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
11461 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11463 * target-insns.def: Add memory_blockage.
11464 * optabs.c (expand_memory_blockage): New function.
11465 (expand_asm_memory_barrier): Rename ...
11466 (expand_asm_memory_blockage): ... to this.
11467 (expand_mem_thread_fence): Call expand_memory_blockage
11468 instead of expand_asm_memory_barrier.
11469 (expand_mem_singnal_fence): Ditto.
11470 (expand_atomic_load): Ditto.
11471 (expand_atomic_store): Ditto.
11472 * doc/md.texi (Standard Pattern Names For Generation):
11473 Document memory_blockage instruction pattern.
11475 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
11477 * config/rl78/rl78.c (rl78_emit_libcall): New function.
11478 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
11479 * config/rl78/rl78.md: New define_expand "adddi3".
11481 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11483 * cfghooks.c (verify_flow_info): Disable check that all probabilities
11486 2017-10-13 Jeff Law <law@redhat.com>
11488 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
11490 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11493 * internal-fn.c (expand_mul_overflow): If both operands have
11494 the same highpart of -1 or 0 and the topmost bit of lowpart
11495 is different, overflow is if res <= 0 rather than res < 0.
11497 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
11499 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
11500 TARGET_P9_VECTOR code for unaligned_load case.
11502 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11504 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
11506 2017-10-13 Nathan Sidwell <nathan@acm.org>
11508 * tree-core.h (tree_contains_struct): Make bool.
11509 * tree.c (tree_contains_struct): Likewise.
11510 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11511 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
11512 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
11513 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
11514 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
11516 2017-10-13 Richard Biener <rguenther@suse.de>
11518 * graphite-isl-ast-to-gimple.c
11519 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
11520 parameters and dominance check.
11521 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
11522 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
11523 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
11524 Do not update SSA form here or do intermediate IL verification.
11525 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
11526 (graphite_initialize): Remove check on the number of loops in
11527 the function and inline into graphite_transform_loops.
11528 (graphite_finalize): Inline into graphite_transform_loops.
11529 (graphite_transform_loops): Perform SSA update and IL verification
11531 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
11533 2017-10-13 Richard Biener <rguenther@suse.de>
11535 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
11536 graphite_expression_type_precision): Avoid global constructor
11538 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
11539 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
11540 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
11541 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
11542 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
11543 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
11545 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
11548 * config/i386/i386.h (ix86_red_zone_size): New.
11549 * config/i386/i386.md (push peephole2s): Replace
11550 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
11552 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11553 Alan Hayward <alan.hayward@arm.com>
11554 David Sherwood <david.sherwood@arm.com>
11556 * combine.c (can_change_dest_mode): Reject changes in
11557 REGMODE_NATURAL_SIZE.
11559 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11560 Alan Hayward <alan.hayward@arm.com>
11561 David Sherwood <david.sherwood@arm.com>
11563 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
11564 (expand_debug_source_expr): Likewise.
11565 * combine.c (combine_simplify_rtx): Likewise.
11566 * cse.c (fold_rtx): Likewise.
11567 * fwprop.c (canonicalize_address): Likewise.
11568 * targhooks.c (default_shift_truncation_mask): Likewise.
11570 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11571 Alan Hayward <alan.hayward@arm.com>
11572 David Sherwood <david.sherwood@arm.com>
11574 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11575 (widened_mode): Likewise.
11576 (expand_unop): Likewise.
11577 * ree.c (transform_ifelse): Likewise.
11578 (merge_def_and_ext): Likewise.
11579 (combine_reaching_defs): Likewise.
11580 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11582 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11583 Alan Hayward <alan.hayward@arm.com>
11584 David Sherwood <david.sherwood@arm.com>
11586 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11587 * combine.c (gen_lowpart_for_combine): Likewise.
11588 * dwarf2out.c (rtl_for_decl_location): Likewise.
11589 * final.c (alter_subreg): Likewise.
11590 * rtlhooks.c (gen_lowpart_general): Likewise.
11591 (gen_lowpart_if_possible): Likewise.
11593 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11594 Alan Hayward <alan.hayward@arm.com>
11595 David Sherwood <david.sherwood@arm.com>
11597 * calls.c (expand_call): Use subreg_lowpart_offset.
11598 * cse.c (cse_insn): Likewise.
11599 * regcprop.c (copy_value): Likewise.
11600 (copyprop_hardreg_forward_1): Likewise.
11602 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11605 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11606 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11607 =Q constraints to +Q and into insn condition add check
11608 that operands[0] and operands[1] are equal.
11609 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11610 =Q constraints to +Q and into insn condition add check
11611 that operands[0] is equal to either operands[1] or operands[2].
11614 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11615 instead of handling MINUS_EXPR twice (once for each argument),
11616 canonicalize operand order and handle just once, use rtype where
11617 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11620 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11621 any values of __C while still being pattern recognizable as a simple
11622 rotate instruction.
11624 2017-10-13 Richard Biener <rguenther@suse.de>
11626 PR tree-optimization/82451
11628 2017-10-02 Richard Biener <rguenther@suse.de>
11630 PR tree-optimization/82355
11631 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11632 a mapping for the enclosing loop but avoid generating one for
11633 the loop tree root.
11634 (copy_bb_and_scalar_dependences): Remove premature codegen
11635 error on PHIs in blocks duplicated into multiple places.
11636 * graphite-scop-detection.c
11637 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11638 in the region use it as loop and nest to analyze the DR in.
11639 (try_generate_gimple_bb): Likewise.
11640 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11641 (add_loop_constraints): For blocks in a loop not in the region
11642 create a dimension with a single iteration.
11643 * sese.h (gbb_loop_at_index): Remove assert.
11645 * cfgloop.c (loop_preheader_edge): For the loop tree root
11646 return the single successor of the entry block.
11647 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11648 Reset the SCEV hashtable and niters.
11649 * graphite-scop-detection.c
11650 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11651 assert that we only have POLYNOMIAL_CHREC that vary in loops
11652 contained in the region.
11653 (scop_detection::graphite_can_represent_expr): Adjust.
11654 (scop_detection::stmt_has_simple_data_refs_p): For loops
11655 not in the region set loop to NULL. The nest is now the
11656 entry edge to the region.
11657 (try_generate_gimple_bb): Likewise.
11658 * sese.c (scalar_evolution_in_region): Adjust for
11659 instantiate_scev change.
11660 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11661 Make nest parameter the edge into the region.
11662 (create_data_ref): Likewise.
11663 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11664 entry edge into a region and adjust instantiate_scev calls.
11665 (create_data_ref): Likewise.
11666 (graphite_find_data_references_in_stmt): Likewise.
11667 (find_data_references_in_stmt): Pass the loop preheader edge
11668 from the nest argument.
11669 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11670 parameter the edge into the region.
11671 (instantiate_parameters): Use the loop preheader edge as entry.
11672 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11674 (get_instantiated_value_entry): Make instantiate_below parameter
11675 the edge into the region.
11676 (instantiate_scev_name): Likewise. Adjust dominance checks,
11677 when we cannot use loop-based instantiation instantiate by
11678 walking use-def chains.
11679 (instantiate_scev_poly): Adjust.
11680 (instantiate_scev_binary): Likewise.
11681 (instantiate_scev_convert): Likewise.
11682 (instantiate_scev_not): Likewise.
11683 (instantiate_array_ref): Remove.
11684 (instantiate_scev_3): Likewise.
11685 (instantiate_scev_2): Likewise.
11686 (instantiate_scev_1): Likewise.
11687 (instantiate_scev_r): Do not blindly handle N-operand trees.
11688 Do not instantiate array-refs. Handle all constants and invariants.
11689 (instantiate_scev): Make instantiate_below parameter
11690 the edge into the region.
11691 (resolve_mixers): Use the loop preheader edge for the region
11692 parameter to instantiate_scev_r.
11693 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11695 2017-10-13 Richard Biener <rguenther@suse.de>
11697 PR tree-optimization/82525
11698 * graphite-isl-ast-to-gimple.c
11699 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11701 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11702 Fail code generation when we cannot represent the isl integer.
11703 (binary_op_to_tree): Elide modulo operations that are no-ops
11704 in the type we code generate. Remove now superfluous code
11707 2017-10-13 Richard Biener <rguenther@suse.de>
11709 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11710 (scop_detection::harmful_loop_in_region): Remove premature
11711 IV type restriction.
11712 (scop_detection::graphite_can_represent_scev): We can handle
11713 pointer IVs just fine.
11715 2017-10-13 Alan Modra <amodra@gmail.com>
11717 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11718 "Clobbers and Scratch Registers". Add paragraph on
11719 alternative to clobbers for scratch registers and OpenBLAS
11722 2017-10-13 Alan Modra <amodra@gmail.com>
11724 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11725 example of a memory input for a string of known length. Move
11726 commentary out of table. Add a number of new examples
11727 covering array memory inputs.
11729 2017-10-12 Martin Liska <mliska@suse.cz>
11731 PR tree-optimization/82493
11732 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11733 (test_range_functions): New function.
11734 (sbitmap_c_tests): Likewise.
11735 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11736 * selftest.h (sbitmap_c_tests): New function.
11738 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11740 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11742 * config/rs6000/amo.h: Fix spacing issue.
11744 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11747 * config/i386/i386.md (*ashl<mode>3_mask_1,
11748 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11749 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11752 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11754 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11755 (profile_probability): Set max_probability
11756 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11758 * profile-count.c (profile_probability::differs_from_p): Do not
11759 rely on max_probaiblity == 10000
11761 2017-10-12 Jeff Law <law@redhat.com>
11763 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11766 2017-10-12 Martin Sebor <msebor@redhat.com>
11770 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11771 (handle_alias_pairs): Call it.
11772 * common.opt (-Wattribute-alias): New option.
11773 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11774 * doc/invoke.texi (-Wattribute-alias): Document.
11776 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11779 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11781 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11783 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11784 (make_hard_regno_born, make_hard_regno_dead): Update
11785 bb_killed_pseudos and bb_gen_pseudos.
11787 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11789 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11791 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11793 * config/alpha/alpha.c (alpha_split_conditional_move):
11794 Use std::swap instead of manually swapping.
11795 (alpha_stdarg_optimize_hook): Ditto.
11796 (alpha_canonicalize_comparison): Ditto.
11798 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11800 * tree-loop-distribution.c (struct builtin_info): New struct.
11801 (struct partition): Refactor fields into struct builtin_info.
11802 (partition_free): Free struct builtin_info.
11803 (build_size_arg_loc, build_addr_arg_loc): Delete.
11804 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11805 information from struct builtin_info.
11806 (find_single_drs): New function refactored from classify_partition.
11807 Also moved builtin validity checks to this function.
11808 (compute_access_range, alloc_builtin): New functions.
11809 (classify_builtin_st, classify_builtin_ldst): New functions.
11810 (classify_partition): Refactor code into functions find_single_drs,
11811 classify_builtin_st and classify_builtin_ldst.
11812 (distribute_loop): Don't do runtime alias check when distributing
11814 (find_seed_stmts_for_distribution): New function.
11815 (pass_loop_distribution::execute): Refactor code finding seed
11816 stmts into above function. Support distribution for the innermost
11817 two-level loop nest. Adjust dump information.
11819 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11821 * tree-loop-distribution.c: Adjust the general comment.
11822 (NUM_PARTITION_THRESHOLD): New macro.
11823 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
11824 (classify_partition): Skip builtin pattern of loop nest's inner loop.
11825 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
11826 in call to build_partition_graph.
11827 (finalize_partitions): New parameter. Make loop distribution more
11828 conservative by fusing more partitions.
11829 (distribute_loop): Don't do runtime alias check in case of loop nest
11831 (find_seed_stmts_for_distribution): New function.
11832 (prepare_perfect_loop_nest): New function.
11833 (pass_loop_distribution::execute): Refactor code finding seed stmts
11834 and loop nest into above functions. Support loop nest distribution.
11835 Adjust dump information accordingly.
11837 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11839 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
11840 and set PTYPE_SEQUENTIAL for merged partition.
11842 2017-10-12 Richard Biener <rguenther@suse.de>
11844 PR tree-optimization/69728
11846 2017-09-19 Richard Biener <rguenther@suse.de>
11848 PR tree-optimization/69728
11849 * graphite-sese-to-poly.c (schedule_error): New global.
11850 (add_loop_schedule): Handle empty domain by failing the
11852 (build_original_schedule): Handle schedule_error.
11854 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
11855 domain by returning an unchanged schedule.
11857 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11859 * genrecog.c (validate_pattern): For VEC_SELECT verify that
11860 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
11862 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
11864 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
11867 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11870 * expr.c (store_field): Don't optimize away bitsize == 0 store
11871 from CALL_EXPR with addressable return type.
11873 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11875 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
11876 * config/rs6000/rs6000.md (sel): Delete mode attribute.
11877 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
11878 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
11879 TARGET_ISEL instead of TARGET_ISEL<sel>.
11881 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
11883 * config/rs6000/rs6000.c
11884 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
11886 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11888 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
11890 (reg_or_cint_operand): Fix comment.
11891 (reg_or_zero_operand): New predicate.
11892 * config/rs6000/rs6000-protos.h (output_isel): Delete.
11893 * config/rs6000/rs6000.c (output_isel): Delete.
11894 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
11895 instead of reg_or_cint_operand. Output instruction directly (not via
11897 (isel_unsigned_<mode>): Ditto.
11898 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
11899 gpc_reg_operand. Add an instruction alternative for this. Output
11900 instruction directly.
11901 (*isel_reversed_unsigned_<mode>): Ditto.
11903 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11905 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
11906 (TARGET_CANONICALIZE_COMPARISON): Define.
11908 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
11911 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11912 Check whether the dest is REG before adding REG_EQUIV note.
11914 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11917 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11919 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11920 (make_hard_regno_born, make_hard_regno_dead): Update
11921 bb_killed_pseudos and bb_gen_pseudos.
11923 2017-10-11 Nathan Sidwell <nathan@acm.org>
11925 * incpath.h (enum incpath_kind): Name enum, prefix values.
11926 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
11927 * incpath.c (heads, tails): Use INC_MAX.
11928 (add_env_var_paths, add_standard_paths): Use incpath_kind.
11929 (merge_include_chains, split_quote_chain,
11930 register_include_chains): Update incpath_kind names.
11931 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
11932 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
11934 (add_framework_path, darwin_register_objc_includes): Likewise.
11935 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
11937 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11939 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
11940 Do not use float_operator operator predicate.
11941 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
11942 * config/i386/predicates.md (float_operator): Remove predicate.
11944 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11946 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
11947 (*jccxf_i387): Ditto.
11948 (*jcc<mode>_i387): Ditto.
11949 (*jccu<mode>_i387): Ditto.
11950 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
11951 (*jcc_*_i387 splitters): Remove.
11952 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
11953 * config/i386/i386.c (ix86_split_fp_branch): Remove.
11954 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
11957 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11959 * profile-count.h (slow_safe_scale_64bit): New function.
11960 (safe_scale_64bit): New inline.
11961 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
11962 * profile-count.c: Include wide-int.h
11963 (slow_safe_scale_64bit): New.
11965 2017-10-11 Nathan Sidwell <nathan@acm.org>
11967 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
11968 HAS_DECL_ASSEMBLER_NAME_P.
11969 * gimple-expr.c (gimple_decl_printable_name: Check
11970 HAS_DECL_ASSEMBLER_NAME_P too.
11971 * ipa-utils.h (type_in_anonymous_namespace_p): Check
11972 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
11973 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
11974 * passes.c (rest_of_decl_compilation): Check
11975 HAS_DECL_ASSEMBLER_NAME_P too.
11976 * recog.c (verify_changes): Likewise.
11977 * tree-pretty-print.c (dump_decl_name): Likewise.
11978 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
11980 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
11981 (SET_DECL_ASSEMBLER_NAME): Use it.
11982 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
11983 (COPY_DECL_ASSEMBLER_NAME): Likewise.
11984 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
11986 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11988 * config.gcc (i386, x86_64): Add extra objects.
11989 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
11990 (ix86_min_insn_size): Declare.
11991 (ix86_issue_rate): Declare.
11992 (ix86_adjust_cost): Declare.
11993 (ia32_multipass_dfa_lookahead): Declare.
11994 (ix86_macro_fusion_p): Declare.
11995 (ix86_macro_fusion_pair_p): Declare.
11996 (ix86_bd_has_dispatch): Declare.
11997 (ix86_bd_do_dispatch): Declare.
11998 (ix86_core2i7_init_hooks): Declare.
11999 (ix86_atom_sched_reorder): Declare.
12000 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12001 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12002 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12003 (rip_relative_addr_p): Rename to ...
12004 (ix86_rip_relative_addr_p): ... this one; export.
12005 (memory_address_length): Update.
12006 (ix86_issue_rate): Move to x86-tune-sched.c.
12007 (ix86_flags_dependent): Move to x86-tune-sched.c.
12008 (ix86_agi_dependent): Move to x86-tune-sched.c.
12009 (exact_dependency_1): Move to x86-tune-sched.c.
12010 (exact_store_load_dependency): Move to x86-tune-sched.c.
12011 (ix86_adjust_cost): Move to x86-tune-sched.c.
12012 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12013 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12014 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12015 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12016 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12017 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12018 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12019 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12020 (min_insn_size): Rename to ...
12021 (ix86_min_insn_size): ... this one; export.
12022 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12023 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12024 (core2i7_first_cycle_multipass_backtrack): Move to
12025 x86-tune-sched-core.c.
12026 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12027 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12028 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12029 (ix86_avoid_jump_mispredicts): Update.
12030 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12031 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12032 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12033 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12034 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12035 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12036 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12037 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12038 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12039 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12040 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12041 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12042 (BIG): Move to ix86-tune-sched-bd.c.
12043 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12044 (enum insn_path): Move to ix86-tune-sched-bd.c.
12045 (get_mem_group): Move to ix86-tune-sched-bd.c.
12046 (is_cmp): Move to ix86-tune-sched-bd.c.
12047 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12048 (is_branch): Move to ix86-tune-sched-bd.c.
12049 (is_prefetch): Move to ix86-tune-sched-bd.c.
12050 (init_window): Move to ix86-tune-sched-bd.c.
12051 (allocate_window): Move to ix86-tune-sched-bd.c.
12052 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12053 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
12054 (process_end_window): Move to ix86-tune-sched-bd.c.
12055 (allocate_next_window): Move to ix86-tune-sched-bd.c.
12056 (find_constant): Move to ix86-tune-sched-bd.c.
12057 (get_num_immediates): Move to ix86-tune-sched-bd.c.
12058 (has_immediate): Move to ix86-tune-sched-bd.c.
12059 (get_insn_path): Move to ix86-tune-sched-bd.c.
12060 (get_insn_group): Move to ix86-tune-sched-bd.c.
12061 (count_num_restricted): Move to ix86-tune-sched-bd.c.
12062 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
12063 (add_insn_window): Move to ix86-tune-sched-bd.c.
12064 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
12065 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
12066 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
12067 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
12068 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
12069 (do_dispatch): Move to ix86-tune-sched-bd.c.
12070 (has_dispatch): Move to ix86-tune-sched-bd.c.
12071 * config/i386/t-i386: Add new object files.
12072 * config/i386/x86-tune-costs.h: New file.
12073 * config/i386/x86-tune-sched-atom.c: New file.
12074 * config/i386/x86-tune-sched-bd.c: New file.
12075 * config/i386/x86-tune-sched-core.c: New file.
12076 * config/i386/x86-tune-sched.c: New file.
12078 2017-10-11 Liu Hao <lh_mouse@126.com>
12080 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
12081 the generic version below instead.
12082 (should_colorize): Recognize Windows consoles as terminals
12084 * pretty-print.c [__MINGW32__] (write_all): New function.
12085 [__MINGW32__] (find_esc_head): Likewise.
12086 [__MINGW32__] (find_esc_terminator): Likewise.
12087 [__MINGW32__] (eat_esc_sequence): Likewise.
12088 [__MINGW32__] (mingw_ansi_fputs): New function that handles
12090 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
12093 2017-10-11 Richard Biener <rguenther@suse.de>
12095 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
12096 Properly call analyze_scalar_evolution with the loop of the stmt.
12098 2017-10-11 Richard Biener <rguenther@suse.de>
12100 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
12101 * tree-core.h (tree_base): Add chrec_var union member.
12102 * tree.h (CHREC_VAR): Remove.
12103 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
12104 * tree-chrec.h (build_polynomial_chrec): Adjust.
12105 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
12106 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
12108 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
12110 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
12111 * match.pd: ... here.
12112 ((T) X == (T) Y): Relax condition.
12114 2017-10-11 Bin Cheng <bin.cheng@arm.com>
12116 PR tree-optimization/82472
12117 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
12119 (break_alias_scc_partitions): Update postorder number.
12121 2017-10-11 Martin Liska <mliska@suse.cz>
12124 * opts.c (parse_no_sanitize_attribute): Do not use error_value
12126 * opts.h (parse_no_sanitize_attribute): Remove last argument.
12128 2017-10-11 Martin Liska <mliska@suse.cz>
12130 * print-rtl.c (print_insn): Move declaration of idbuf
12131 to same scope as name.
12133 2017-10-11 Martin Liska <mliska@suse.cz>
12138 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12139 volatile arguments.
12141 2017-10-11 Martin Liska <mliska@suse.cz>
12144 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12145 volatile arguments.
12147 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
12149 * config.gcc (default_gnu_indirect_function): Default to yes for
12150 arm*-*-linux* with glibc.
12152 2017-10-11 Richard Biener <rguenther@suse.de>
12154 * tree-scalar-evolution.c (get_scalar_evolution): Handle
12155 default-defs and types we do not want to analyze.
12156 (interpret_loop_phi): Replace unreachable code with an assert.
12157 (compute_scalar_evolution_in_loop): Remove and inline ...
12158 (analyze_scalar_evolution_1): ... here, replacing condition with
12159 what makes the intent clearer. Remove handling of cases
12160 get_scalar_evolution now handles.
12162 2017-10-10 Jim Wilson <wilson@tuliptree.org>
12164 PR rtl-optimization/81434
12165 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
12166 comment for main loop. In sched_group_found if, also add checks for
12167 pass and min_cost_group.
12169 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
12171 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
12172 (rs6000_insn_cost): New function.
12173 * config/rs6000/rs6000.md (cost): New attribute.
12175 2017-10-10 Jakub Jelinek <jakub@redhat.com>
12176 H.J. Lu <hongjiu.lu@intel.com>
12180 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
12181 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
12182 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
12183 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
12184 to OPTION_MASK_ISA_AVX512VL - builtins that have both
12185 OPTION_MASK_ISA_MMX and some other bit set require both
12186 mmx and the ISAs without the mmx bit.
12187 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
12188 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
12189 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
12190 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
12191 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
12192 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
12193 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
12194 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
12195 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
12196 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
12197 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
12198 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
12199 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
12200 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
12201 Add OPTION_MASK_ISA_MMX.
12203 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
12205 * config.gcc (armv7*-*-freebsd*): New target.
12206 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
12208 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
12210 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
12211 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
12214 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
12216 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
12217 a template parameter.
12218 (WIDE_INT_REF_FOR): Update accordingly.
12219 * tree.h (wi::int_traits <const_tree>): Delete.
12220 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
12221 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
12222 (wi::tree_to_wide_ref): New typedef.
12223 (wi::to_wide): New function.
12224 * calls.c (get_size_range): Use wi::to_wide when operating on
12225 trees as wide_ints.
12226 * cgraph.c (cgraph_node::create_thunk): Likewise.
12227 * config/i386/i386.c (ix86_data_alignment): Likewise.
12228 (ix86_local_alignment): Likewise.
12229 * dbxout.c (stabstr_O): Likewise.
12230 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
12231 * expr.c (const_vector_from_tree): Likewise.
12232 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
12233 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
12234 (fold_negate_expr_1, int_const_binop_1, const_binop)
12235 (fold_convert_const_int_from_real, optimize_bit_field_compare)
12236 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
12237 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
12238 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
12239 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
12240 (fold_not_const, round_up_loc): Likewise.
12241 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
12242 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
12243 (alloca_call_type): Likewise.
12244 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
12245 * godump.c (go_output_typedef): Likewise.
12246 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
12247 * internal-fn.c (get_min_precision): Likewise.
12248 * ipa-cp.c (ipcp_store_vr_results): Likewise.
12249 * ipa-polymorphic-call.c
12250 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
12251 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
12252 (ipa_modify_call_arguments): Likewise.
12253 * match.pd: Likewise.
12254 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
12255 * print-tree.c (print_node_brief, print_node): Likewise.
12256 * stmt.c (expand_case): Likewise.
12257 * stor-layout.c (layout_type): Likewise.
12258 * tree-affine.c (tree_to_aff_combination): Likewise.
12259 * tree-cfg.c (group_case_labels_stmt): Likewise.
12260 * tree-data-ref.c (dr_analyze_indices): Likewise.
12261 (prune_runtime_alias_test_list): Likewise.
12262 * tree-dump.c (dequeue_and_dump): Likewise.
12263 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
12264 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
12265 * tree-pretty-print.c (dump_generic_node): Likewise.
12266 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
12267 (simple_iv_with_niters): Likewise.
12268 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
12269 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
12270 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
12271 * tree-ssa-loop-niter.c (split_to_var_and_offset)
12272 (refine_value_range_using_guard, number_of_iterations_ne_max)
12273 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
12274 (get_cst_init_from_scev, record_nonwrapping_iv)
12275 (scev_var_range_cant_overflow): Likewise.
12276 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
12277 * tree-ssa-pre.c (compute_avail): Likewise.
12278 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
12279 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
12280 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
12281 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
12282 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
12283 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
12284 (dump_case_nodes, try_switch_expansion): Likewise.
12285 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
12286 (vect_do_peeling): Likewise.
12287 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
12288 * tree-vect-stmts.c (vectorizable_load): Likewise.
12289 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
12290 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
12291 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
12292 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
12293 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
12294 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
12295 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
12296 (evrp_dom_walker::before_dom_children): Likewise.
12297 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
12298 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
12299 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
12300 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
12301 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
12302 (get_range_pos_neg): Likewise.
12303 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
12304 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
12305 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
12306 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
12307 * config/avr/avr.c (avr_fold_builtin): Likewise.
12308 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
12309 * config/msp430/msp430.c (msp430_attr): Likewise.
12310 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
12311 * config/powerpcspe/powerpcspe-c.c
12312 (altivec_resolve_overloaded_builtin): Likewise.
12313 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
12314 (rs6000_expand_ternop_builtin): Likewise.
12315 * config/rs6000/rs6000-c.c
12316 (altivec_resolve_overloaded_builtin): Likewise.
12317 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
12318 (rs6000_expand_ternop_builtin): Likewise.
12319 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
12321 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12323 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
12324 when copying loop nest with only one inner loop.
12326 2017-10-10 Richard Biener <rguenther@suse.de>
12328 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
12329 blocks if SCEV is active.
12330 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
12332 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
12333 (scev_initialize): Assert we are not yet initialized.
12335 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12337 * tree-loop-distribution.c (generate_loops_for_partition): Remove
12338 inner loop's exit stmt by making it always exit the loop, otherwise
12339 we would generate an infinite empty loop.
12341 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12343 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
12344 renaming variables in new preheader if it's deleted.
12346 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12348 * tree-loop-distribution.c (struct partition): Remove unused field
12349 loops of the structure.
12350 (partition_alloc, partition_free): Ditto.
12351 (build_rdg_partition_for_vertex): Ditto.
12353 2017-10-09 Jeff Law <law@redhat.com>
12355 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
12356 return type to match prototype and documentation.
12358 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12360 * config/rs6000/rs6000.c (processor_costs): Move to ...
12361 * config/rs6000/rs6000.h: ... here.
12362 (rs6000_cost): Declare.
12364 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
12366 * except.c (setjmp_fn): New global variable.
12367 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
12368 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
12369 if DONT_USE_BUILTIN_SETJMP is defined.
12371 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12373 * target.def (insn_cost): New hook.
12374 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
12375 * doc/tm.texi: Regenerate.
12376 * rtlanal.c (insn_cost): Use the new hook.
12378 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12380 * combine.c (combine_validate_cost): Compute the new insn_cost,
12381 not just pattern_cost.
12382 (try_combine): Adjust comment.
12384 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12386 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
12388 * combine.c (uid_insn_cost): Adjust comment.
12389 (combine_validate_cost): Adjust comment. Use pattern_cost instead
12391 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
12392 * dse.c (find_shift_sequence): Ditto.
12393 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
12394 (bb_valid_for_noce_process_p): Use pattern_cost.
12395 * rtl.h (insn_rtx_cost): Delete.
12396 (pattern_cost): New prototype.
12397 (insn_cost): New prototype.
12398 * rtlanal.c (insn_rtx_cost): Rename to...
12399 (pattern_cost): ... this.
12402 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
12404 * config/i386/i386.md (*jcc_2): Remove insn pattern.
12405 (*jcc<mode>_0_r_i387): Ditto.
12406 (*jccxf_r_i387): Ditto.
12407 (*jcc<mode>_r_i387): Ditto.
12408 (*jccu<mode>_r_i387): Ditto.
12409 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
12410 (*jcc): Rename from *jcc_1.
12412 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12414 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
12415 deferred rescans after the lvx/stvx recombination pre-pass.
12417 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
12419 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
12420 memory operation instruction support.
12421 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
12422 (rs6000-ibm-aix[789]*): Likewise.
12423 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
12424 Document new functions.
12426 2017-10-09 Richard Biener <rguenther@suse.de>
12428 PR tree-optimization/82397
12429 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
12430 equality only for semantically equal trees.
12432 2017-10-09 Richard Biener <rguenther@suse.de>
12434 PR tree-optimization/82449
12435 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
12436 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
12437 allow constant addresses.
12438 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
12441 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12443 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
12446 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12449 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
12452 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12455 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
12457 2017-10-09 Jakub Jelinek <jakub@redhat.com>
12460 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
12461 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
12463 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
12465 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
12466 (WI_BINARY_PREDICATE_RESULT): Likewise.
12467 (wi::binary_traits::operator_result): New type.
12468 (wi::binary_traits::predicate_result): Likewise.
12469 (generic_wide_int::operator~, unary generic_wide_int::operator-)
12470 (generic_wide_int::operator==, generic_wide_int::operator!=)
12471 (generic_wide_int::operator&, generic_wide_int::and_not)
12472 (generic_wide_int::operator|, generic_wide_int::or_not)
12473 (generic_wide_int::operator^, generic_wide_int::operator+
12474 (binary generic_wide_int::operator-, generic_wide_int::operator*):
12476 (operator~, unary operator-, operator==, operator!=, operator&)
12477 (operator|, operator^, operator+, binary operator-, operator*): New
12479 * expr.c (get_inner_reference): Use wi::bit_and_not.
12480 * fold-const.c (fold_binary_loc): Likewise.
12481 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
12482 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
12483 (bit_value_binop): Likewise.
12484 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
12485 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
12486 (extract_range_from_binary_expr_1): Likewise.
12487 (masked_increment): Likewise.
12488 (simplify_bit_ops_using_ranges): Likewise.
12490 2017-10-09 Martin Jambor <mjambor@suse.cz>
12493 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
12494 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
12495 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
12497 (hsa_fixup_mov_insn_type): New function.
12498 (hsa_op_with_type::get_in_type): Use it.
12499 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
12500 immediates in an assert.
12501 (hsa_op_with_type::extend_int_to_32bit): New method.
12502 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
12503 types. Convert to dest type if necessary.
12504 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
12505 (reg_for_gimple_ssa): Pass false as min32int to
12506 hsa_type_for_scalar_tree_type.
12507 (gen_hsa_addr): Fixup type when creating addresable temporary.
12508 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
12509 (gen_hsa_unary_operation): Extend operands and convert to dest type if
12510 necessary. Call hsa_fixup_mov_insn_type.
12511 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
12512 extend operands and convert to dest type if necessary.
12513 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
12514 to dest type if necessary.
12515 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
12516 if conversion nt necessary and size matches.
12517 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
12518 to dest type if necessary.
12519 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
12520 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
12522 (gen_hsa_clrsb): Likewise.
12523 (gen_hsa_ffs): Likewise.
12524 (gen_hsa_divmod): Extend operands and convert to dest type if
12526 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
12528 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12530 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
12531 Remove empty default arguments. Use a brace block as output
12533 (conditional return): Ditto.
12535 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
12536 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
12538 (group_ending_nop): Ditto.
12539 (doloop_end): Ditto.
12540 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
12541 (splitters for those): Ditto.
12543 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12545 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
12546 a conditional jump (and the compare for it) so that pc_rtx is the
12548 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
12549 for the deleted and renamed ctr<mode>_internal[234] patterns.
12550 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
12551 Delete second conditional return pattern.
12552 (ctr<mode>_internal2): Delete this second bdnz pattern.
12553 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
12554 (ctr<mode>_internal4): Delete this second bdz pattern.
12556 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12558 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
12559 (always_initialized_rtx_for_ssa_name_p): New predicate.
12560 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
12561 (finish_out_of_ssa): Free new field of SA.
12562 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
12563 * tree-ssa-coalesce.c: Include tree-ssa.h.
12564 (get_parm_default_def_partitions): Remove extern keyword.
12565 (get_undefined_value_partitions): New function.
12566 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12567 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12568 uninitialized bits.
12569 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12571 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12573 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12575 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12577 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12578 for targets that preffer 128bit.
12580 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12582 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12584 2017-10-08 Olivier Hainque <hainque@adacore.com>
12586 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12587 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12589 (thumb_set_return_address): Likewise.
12591 2017-10-08 Olivier Hainque <hainque@adacore.com>
12593 * common/config/arm/arm-common.c (arm_except_unwind_info):
12594 Handle DWARF2_UNWIND_INFO.
12596 2017-10-07 Michael Collison <michael.collison@arm.com>
12598 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12601 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12603 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12604 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12605 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12606 defined, force the creation of a new block for a dispatch label.
12608 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12610 * invoke.texi (Wsuggest-attribute=cold): Document.
12611 * common.opt (Wsuggest-attribute=cold): New
12612 * ipa-pure-const.c (warn_function_cold): New function.
12613 * predict.c (compute_function_frequency): Use it.
12614 * predict.h (warn_function_cold): Declare.
12616 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12618 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12621 2017-10-06 Martin Liska <mliska@suse.cz>
12623 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12624 keyword for member functions.
12625 (struct sanopt_tree_couple): New struct.
12626 (struct sanopt_tree_couple_hash): New function.
12627 (struct sanopt_ctx): Add new hash_map.
12628 (has_dominating_ubsan_ptr_check): New function.
12629 (record_ubsan_ptr_check_stmt): Likewise.
12630 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12631 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12632 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12634 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12637 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12638 aarch64_simd_valid_immediate on CONST_VECTORs.
12639 (aarch64_reg_or_bic_imm): Likewise.
12641 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12643 PR rtl-optimization/82396
12644 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12646 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12648 * graphite-dependences.c (scop_get_reads): Move code to...
12649 (scop_get_must_writes): Move code to...
12650 (scop_get_may_writes): Move code to...
12651 (scop_get_reads_and_writes): ... here.
12652 (scop_get_dependences): Call scop_get_reads_and_writes.
12654 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12656 PR tree-optimization/82434
12657 * fold-const.h (can_native_encode_type_p,
12658 can_native_encode_string_p): Remove.
12659 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12660 don't encode anything, just return what would be otherwise returned.
12661 (native_encode_fixed, native_encode_complex, native_encode_vector):
12663 (native_encode_string): Likewise. Inline by hand
12664 can_native_encode_string_p.
12665 (can_native_encode_type_p): Remove.
12666 (can_native_encode_string_p): Remove.
12667 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12668 STRING_CSTs using can_native_encode_string_p, test all
12669 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12670 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12671 argument from native_encode_expr.
12672 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12673 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12676 2017-10-06 Richard Biener <rguenther@suse.de>
12678 PR tree-optimization/82397
12679 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12680 operand_equal_p but rely on data_ref_compare_tree for detecting
12682 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12683 to match up with dr_group_sort_cmp.
12685 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12688 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12690 * config/s390/s390-builtin-types.def: Regenerate.
12692 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12695 * config/s390/s390-builtin-types.def: Regenerate.
12696 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12697 Change flag from B_VXE to B_VX.
12698 (s390_vec_min_dbl): Remove B_VXE flag.
12700 2017-10-06 Richard Biener <rguenther@suse.de>
12702 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12703 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12704 translate_isl_ast_to_gimple::is_valid_rename,
12705 translate_isl_ast_to_gimple::get_rename,
12706 translate_isl_ast_to_gimple::get_def_bb_for_const,
12707 translate_isl_ast_to_gimple::get_new_name,
12708 translate_isl_ast_to_gimple::collect_all_ssa_names,
12709 translate_isl_ast_to_gimple::copy_loop_phi_args,
12710 translate_isl_ast_to_gimple::collect_all_ssa_names,
12711 translate_isl_ast_to_gimple::copy_loop_phi_args,
12712 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12713 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12714 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12715 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12716 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12717 translate_isl_ast_to_gimple::copy_cond_phi_args,
12718 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12719 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12720 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12721 translate_isl_ast_to_gimple::rename_uses,
12722 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12723 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12724 (set_rename_for_each_def): Likewise.
12725 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12726 here. Handle rewriting SCEV analyzable uses here.
12727 (copy_bb_and_scalar_dependences): Generate code for PHI
12729 (graphite_regenerate_ast_isl): Adjust.
12730 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12731 (add_write, add_read): New functions.
12732 (build_cross_bb_scalars_def): Use it and simplify.
12733 (build_cross_bb_scalars_use): Likewise.
12734 (graphite_find_cross_bb_scalar_vars): Inline into...
12735 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12736 simulating out-of-SSA. Compute liveout and add dependencies.
12737 (build_scops): Force an empty entry block.
12738 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12740 (sese_build_liveouts): Declare.
12741 (sese_trivially_empty_bb_p): Likewise.
12742 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12743 compute liveout and debug_liveout.
12744 (sese_bad_liveouts_use): Remove.
12745 (sese_reset_debug_liveouts_bb): Likewise.
12746 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12747 (sese_build_liveouts): Build liveout and debug_liveout and store
12749 (new_sese_info): Adjust.
12750 (free_sese_info): Likewise.
12751 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12752 do not build liveout here.
12753 (move_sese_in_condition): Adjust region entry.
12754 (scev_analyzable_p): Match up with chrec_apply requirements.
12755 (sese_trivially_empty_bb_p): New.
12756 * tree-into-ssa.c (get_reaching_def): Properly support generating
12757 default-defs for incremental rewrite of anonymous names.
12759 2017-10-06 Richard Biener <rguenther@suse.de>
12761 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12762 precision do not perform modulo reduction.
12764 2017-10-06 Richard Biener <rguenther@suse.de>
12766 PR tree-optimization/82436
12767 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12768 conservatively choose the vectorization factor when checking
12769 whether we can perform the required load permutation.
12770 (vect_transform_slp_perm_load): Assert when we may not fail.
12772 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12774 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12775 message for incompatible -msdata=* and -mcall-* options.
12777 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12779 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12780 rate for post-reload scheduling.
12782 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12784 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12786 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12788 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12789 to improve monte carlo in scimark.
12791 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12793 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12794 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12795 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12796 pentium4_cost, nocona_cost): Set reassociation width to 1.
12797 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12798 width to 2 for fp operations and 1 otherwise.
12799 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12801 (atom_cost): Set reassociation width to 2.
12802 (slm_cost, generic_cost): Set fp reassociation width
12803 to 2 and 1 otherwise.
12804 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12805 (core_cost): Set fp reassociation width to 4 and vector to 2.
12806 (ix86_reassociation_width): Rewrite using cost table; special case
12807 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12808 and TARGET_AVX128_OPTIMAL.
12809 * config/i386/i386.h (processor_costs): Add
12810 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
12811 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
12812 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
12813 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
12814 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
12815 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
12817 2017-10-05 Nathan Sidwell <nathan@acm.org>
12819 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
12821 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12823 * config/arm/arm.c (arm_test_fpu_data): New.
12824 (arm_run_selftests): Call arm_test_fpu_data.
12826 2017-10-04 Nathan Sidwell <nathan@acm.org>
12828 * toplev.c (toplev::main): Remove excess parens on pretty_printer
12830 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
12832 2017-10-04 Sudakshina Das <sudi.das@arm.com>
12834 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
12835 check type for aarch64_simd_valid_immediate.
12836 (aarch64_output_simd_mov_immediate): Update prototype.
12837 (aarch64_simd_valid_immediate): Update prototype.
12838 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
12839 support for ORR-immediate.
12840 (and<mode>3): modified pattern to add support for BIC-immediate.
12841 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
12842 now checks for valid immediate for BIC and ORR based on new enum
12844 (aarch64_output_simd_mov_immediate): Function now used to output
12845 BIC/ORR imm as well based on new enum argument.
12846 * config/aarch64/constraints.md (Do): New vector immediate constraint.
12848 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
12849 (aarch64_reg_or_bic_imm): Likewise.
12851 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12853 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
12854 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
12856 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12860 PR rtl-optimization/82396
12861 * haifa-sched.c (autopref_multipass_init): Simplify
12863 (autopref_rank_data): Simplify sort order.
12864 * sched-int.h (autopref_multipass_data_): Remove
12865 multi_mem_insn_p, min_offset and max_offset.
12867 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12869 * doc/sourcebuild.texi: Document vect_peeling_profitable.
12871 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12873 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
12874 vect_doubleint_cvt.
12876 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12878 * doc/sourcebuild.texi: Document vect_long_mult.
12880 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
12882 PR tree-optimization/82413
12883 * fold-const.c (build_range_check): Use widest_int when comparing
12884 the maximum ETYPE value with HIGH.
12886 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12888 PR rtl-optimization/82396
12889 * haifa-sched.c (autopref_multipass_init): Simplify
12891 (autopref_rank_data): Simplify sort order.
12892 * sched-int.h (autopref_multipass_data_): Remove
12893 multi_mem_insn_p, min_offset and max_offset.
12895 2017-10-04 Jakub Jelinek <jakub@redhat.com>
12897 PR tree-optimization/82381
12898 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
12899 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
12902 PR tree-optimization/82374
12903 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
12904 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
12905 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
12906 current_function_decl to the new decl.
12908 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
12910 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
12911 helper macro for IEEE float128 hardware built-in functions.
12912 (SQRTF128_ODD): Add built-in functions with the round-to-odd
12914 (TRUNCF128_ODD): Likewise.
12915 (ADDF128_ODD): Likewise.
12916 (SUBF128_ODD): Likewise.
12917 (MULF128_ODD): Likewise.
12918 (DIVF128_ODD): Likewise.
12919 (FMAF128_ODD): Likewise.
12920 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
12921 UNSPEC_TRUNC_ROUND_TO_ODD.
12922 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
12923 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
12924 floating point round to odd instructions.
12925 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
12926 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
12927 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
12928 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
12929 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
12930 (trunc<mode>sf2_hw): Change the truncate with round to odd
12931 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
12932 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
12933 to odd hardware instructions.
12934 (sub<mode>3_odd): Likewise.
12935 (mul<mode>3_odd): Likewise.
12936 (div<mode>3_odd): Likewise.
12937 (sqrt<mode>2_odd): Likewise.
12938 (fma<mode>4_odd): Likewise.
12939 (fms<mode>4_odd): Likewise.
12940 (nfma<mode>4_odd): Likewise.
12941 (nfms<mode>4_odd): Likewise.
12942 (trunc<mode>df2_odd): Change the truncate with round to odd
12943 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
12945 * doc/extend.texi (PowerPC built-in functions): Update documentation
12946 for existing IEEE float128-bit built-in functions. Add built-in
12947 functions that generate the IEEE 128-bit floating point round to
12950 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
12952 PR rtl-optimization/77729
12953 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
12954 to (X&(C1&~C2))|C2 transformations.
12956 2017-10-03 Martin Jambor <mjambor@suse.cz>
12958 PR tree-optimization/82363
12959 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
12960 mismatch, mark lacc written regardless of racc.
12962 2017-10-03 Jakub Jelinek <jakub@redhat.com>
12964 PR tree-optimization/82381
12965 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
12966 stmt_to_insert nor wheather SSA_NAMEs are default defs.
12967 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
12968 fallthrough into reassoc_stmt_dominates_stmt_p.
12971 * combine.c (combine_instructions): Don't combine in unreachable
12974 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
12977 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
12978 function to not use the have_cpu variable. Do not set cpu_index,
12979 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
12980 or the default cpu.
12981 (rs6000_valid_attribute_p): Remove duplicate initializations of
12982 old_optimize and func_optimize.
12983 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
12984 (rs6000_activate_target_options): Make global.
12985 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
12988 2017-10-02 Jakub Jelinek <jakub@redhat.com>
12990 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
12991 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
12992 Set *poffset to 0 and *psize and *pmax_size to -1 if
12993 *poffset + *psize overflows in HOST_WIDE_INT.
12995 PR tree-optimization/82387
12996 PR tree-optimization/82388
12997 PR tree-optimization/82389
12998 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
12999 instead of live_bytes non-NULL.
13001 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13004 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13007 2017-10-02 Richard Biener <rguenther@suse.de>
13009 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13010 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13011 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13013 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13015 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13016 requested precision matches the type's.
13017 * calls.c (alloc_max_size): Calculate the new candidate size as
13018 a widest_int and use wi::to_widest when comparing it with the
13019 current candidate size.
13020 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13021 zero rather than integer_zero_node.
13022 * match.pd: Check for a no-op conversion before using wi::add
13023 rather than after. Use tree_to_uhwi when summing small shift
13024 counts into an unsigned int.
13026 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13027 Alan Hayward <alan.hayward@arm.com>
13028 David Sherwood <david.sherwood@arm.com>
13031 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13032 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13033 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13034 POINTER_AND_FP_REGS.
13036 2017-10-02 Richard Biener <rguenther@suse.de>
13038 PR tree-optimization/82355
13039 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13040 a mapping for the enclosing loop but avoid generating one for
13041 the loop tree root.
13042 (copy_bb_and_scalar_dependences): Remove premature codegen
13043 error on PHIs in blocks duplicated into multiple places.
13044 * graphite-scop-detection.c
13045 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13046 in the region use it as loop and nest to analyze the DR in.
13047 (try_generate_gimple_bb): Likewise.
13048 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13049 (add_loop_constraints): For blocks in a loop not in the region
13050 create a dimension with a single iteration.
13051 * sese.h (gbb_loop_at_index): Remove assert.
13053 2017-10-01 Kevin Buettner <kevinb@redhat.com>
13055 * omp-expand.c (adjust_context_scope): New function.
13056 (expand_parallel_call): Call adjust_context_scope.
13058 2017-10-01 Jeff Law <law@redhat.com>
13060 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
13061 dom_opt_dom_walker class with direct access to private members.
13062 Add comments. Call test_for_singularity.
13063 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
13064 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
13065 m_dummy_cond anymore.
13066 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
13068 (pass_dominator:execute): Build the dummy_cond here and pass it
13069 to the dom_opt_dom_walker ctor.
13070 (test_for_singularity): New function.
13072 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
13073 Maya Rashish <coypu@sdf.org>
13075 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
13076 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
13077 (alpha*-*-netbsd*) Use nbsd_tm_file.
13078 (arm*-*-netbsdelf*) Likewise.
13079 (i[34567]86-*-netbsdelf*) Likewise.
13080 (x86_64-*-netbsd*) Likewise.
13081 (mips*-*-netbsd*) Likewise.
13082 (powerpc-*-netbsd*) Likewise.
13083 (sh*-*-netbsd*) Likewise.
13084 (sparc-*-netbsdelf*) Likewise.
13085 (sparc64-*-netbsd*) Likewise.
13086 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
13088 (vax-*-netbsdelf*) Likewise.
13089 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
13090 (UINT_FAST8_TYPE) Likewise.
13091 (INT_FAST16_TYPE) Check CHAR_FAST16.
13092 (UINT_FAST16_TYPE) Likewise.
13094 2017-09-30 Jakub Jelinek <jakub@redhat.com>
13097 * config/i386/i386.md
13098 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
13099 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
13100 *divmodsi4_zext_2): New define_insn_and_split.
13101 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
13102 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
13103 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
13104 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
13105 New define_insn_and_split.
13106 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
13107 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
13108 operands[1] having DImode when mode is SImode.
13110 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
13111 always SImode for DIV and MOD in REG_EQUAL notes.
13113 2017-09-29 Yury Gribov <tetra2005@gmail.com>
13115 PR middle-end/82319
13116 * match.pd: Fix handling of NaNs in pattern.
13118 2017-09-29 Jeff Law <law@redhat.com>
13120 * sbitmap.c (bitmap_bit_in_range_p): New function.
13121 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
13122 * tree-ssa-dse.c (live_bytes_read): New function.
13123 (dse_classify_store): Ignore reads of dead bytes.
13125 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
13126 typos and whitespace errors.
13127 * config/i386/predicates.md (address_no_seg_operand): Likewise.
13128 * config/s390/s390.c (s390_emit_prologue): Likewise.
13130 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13133 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
13134 with a symbol for LRA.
13136 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13138 PR rtl-optimization/82338
13139 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
13141 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
13143 * genmodes.c (calc_wider_mode): Suppress qsort macro.
13144 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
13145 (qsort_chk): Declare.
13146 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
13147 (qsort_chk): New function.
13149 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13151 PR tree-optimization/82337
13152 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
13153 phi definition if the PHI result appears in an abnormal PHI.
13154 (find_basis_for_base_expr): Don't record a basis if the LHS of the
13155 basis appears in an abnormal PHI.
13157 2017-09-29 Richard Biener <rguenther@suse.de>
13159 * graphite-isl-ast-to-gimple.c
13160 (translate_isl_ast_to_gimple::set_codegen_error): New function.
13161 (binary_op_to_tree): Use it.
13162 (get_rename_from_scev): Likewise.
13163 (copy_loop_phi_nodes): Likewise.
13164 (copy_bb_and_scalar_dependences): Likewise.
13165 (translate_pending_phi_nodes): Likewise.
13167 2017-09-29 Jakub Jelinek <jakub@redhat.com>
13170 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
13171 for movabsq $(i32 << shift), r64.
13173 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
13175 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
13176 index when encoding %esp as %rsp to avoid 0x67 prefix.
13178 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
13180 * config/i386/i386.md (*movsf_internal, *movdf_internal):
13181 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
13183 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
13185 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
13186 Extensions with more than 16 double VFP registers.
13187 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
13188 to_clear_mask and all code related to it. Replace the remaining
13189 entry by a sbitmap and adapt code accordingly.
13191 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
13193 * brig-builtins.def: Change pure attributes to const.
13195 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13197 * config.gcc (default_gnu_indirect_function): Default to yes for
13198 sparc*-*-linux* with glibc.
13200 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13202 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
13203 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
13204 when creating .init_array and .fini_array sections with priority
13207 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
13210 * config/aarch64/aarch64.c
13211 (aarch64_builtin_support_vector_misalignment): Always return false
13212 when misalignment is unknown.
13214 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
13216 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
13217 this function to return false if the definition used by the swap
13218 instruction is artificial, or if the memory address from which the
13219 constant value is loaded is not represented by a base address held
13220 in a register or if the base address register is a frame or stack
13221 pointer. Additionally, return false if the base address of the
13222 loaded constant is a SYMBOL_REF but is not considered to be a
13224 (replace_swapped_load_constant): New function.
13225 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
13226 loaded constant vector with a load of a swapped constant vector.
13228 2017-09-27 Carl Love <cel@us.ibm.com>
13230 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
13231 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
13232 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
13235 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
13237 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
13238 first, always call autopref_rank_data otherwise.
13240 2017-09-27 Richard Biener <rguenther@suse.de>
13242 * graphite-scop-detection.c (find_scop_parameters): Move
13243 loop bound handling ...
13244 (gather_bbs::before_dom_children): ... here, avoiding the need
13245 to build scop_info->loop_nest.
13246 (record_loop_in_sese): Remove.
13247 * sese.h (sese_info_t::loop_nest): Remove.
13248 * sese.c (new_sese_info): Do not allocate loop_nest.
13249 (free_sese_info): Do not free loop_nest.
13251 2017-09-27 Jakub Jelinek <jakub@redhat.com>
13254 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
13255 lhs from calls if the lhs has addressable type.
13257 2017-09-27 Richard Biener <rguenther@suse.de>
13259 * graphite.h (scop::max_alias_set): New member.
13260 * graphite-scop-detection.c: Remove references to non-existing
13261 --param in comments.
13262 (build_alias_sets): Record the maximum alias set used for drs.
13263 (build_scops): Support zero as unlimited for
13264 --param graphite-max-arrays-per-scop.
13265 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
13266 and inline into ...
13267 (build_poly_sr_1): ... here. Compute alias set based on the
13268 maximum alias set used for drs rather than
13269 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
13271 2017-09-27 Richard Biener <rguenther@suse.de>
13273 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
13274 --param loop-block-tile-size=0 to disable tiling.
13276 2017-09-27 Richard Biener <rguenther@suse.de>
13278 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
13279 (graphite-max-nb-scop-params): Document special value zero.
13280 * domwalk.h (dom_walker::STOP): New symbolical constant.
13281 (dom_walker::dom_walker): Add optional parameter for bb to
13283 (dom_walker::~dom_walker): Declare.
13284 (dom_walker::before_dom_children): Document STOP return value.
13285 (dom_walker::m_user_bb_to_rpo): New member.
13286 (dom_walker::m_bb_to_rpo): Likewise.
13287 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
13288 mapping here if not provided by the user.
13289 (dom_walker::~dom_walker): Free bb to RPO mapping if not
13290 provided by the user.
13291 (dom_walker::STOP): Define.
13292 (dom_walker::walk): Do not compute bb to RPO mapping here.
13293 Support STOP return value from before_dom_children to stop
13295 * graphite-optimize-isl.c (optimize_isl): If the schedule
13296 is the same still generate code if -fgraphite-identity
13297 or -floop-parallelize-all are given.
13298 * graphite-scop-detection.c: Include cfganal.h.
13299 (gather_bbs::gather_bbs): Get and pass through bb to RPO
13301 (gather_bbs::before_dom_children): Return STOP for BBs
13303 (build_scops): Compute bb to RPO mapping and pass it to
13304 the domwalk. Treat --param graphite-max-nb-scop-params=0
13305 as not limiting the number of params.
13306 * graphite.c (graphite_initialize): Remove limit on the
13307 number of basic-blocks in a function.
13308 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
13309 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
13310 default value of 10.
13312 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13314 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
13315 Adjust code to eliminate needing to do the shift right 32-bits
13316 operation after XSCVDPSPN.
13318 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13320 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
13321 ((X / Y) != 0 -> X >= Y): Likewise.
13323 2017-09-26 Carl Love <cel@us.ibm.com>
13325 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
13326 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
13327 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
13328 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
13329 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
13330 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
13331 definitions and overloading.
13332 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
13333 statement for P9V_BUILTIN_XST_LEN_R.
13334 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
13335 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
13336 define_expand and define_insn for the instructions and builtins.
13337 * doc/extend.texi: Update the built-in documentation file for the new
13338 built-in functions.
13339 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
13340 define_insn for the instructions
13342 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
13345 * gcc/config/netbsd-protos.h: New file.
13346 * gcc/config/netbsd.c: New file.
13347 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
13348 * gcc/config/t-netbsd: New file.
13349 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
13350 (tmake_file) Add t-netbsd.
13351 (extra_objs) Add netbsd.o.
13353 2017-09-26 Janus Weil <janus@gcc.gnu.org>
13357 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
13359 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13361 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
13362 sign extension from a vector register to a GPR by doing a 32-bit
13363 direct move and then an EXTSW.
13364 (extendsi<mode>2 splitter): Likewise.
13365 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
13366 right or vector extract after doing XSCVDPSPN. Use
13367 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
13369 (movdi_from_sf_zero_ext): Likewise.
13370 (reload_gpr_from_vsxsf): Likewise.
13371 (p8_mfvsrd_4_disf): Delete, no longer used.
13372 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
13373 then needing to move the SFmode to a GPR to use the XSCVDPSP
13374 instruction instead of FRSP and XSCVDPSPN.
13375 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
13376 it is adjacent to the other XSCVSPDP insns.
13377 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
13378 SFmode to be in traditional Altivec registers.
13379 (vsx_xscvdpspn): Eliminate useless alternative constraint.
13380 (vsx_xscvspdpn): Likewise.
13381 (vsx_xscvspdpn_scalar): Likewise.
13383 2017-09-26 Martin Jambor <mjambor@suse.cz>
13385 * tree-sra.c (compare_access_positions): Put integral types first,
13386 stabilize sorting of integral types, remove conditions putting
13387 non-full-precision integers last.
13388 (sort_and_splice_var_accesses): Disable scalarization if a
13389 non-integert would be represented by a non-full-precision integer.
13391 2017-09-26 Joseph Myers <joseph@codesourcery.com>
13393 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
13394 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
13395 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
13396 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
13397 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
13398 conditionals inside the function instead of around it. Call
13399 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
13400 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
13402 2017-09-26 Richard Biener <rguenther@suse.de>
13404 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
13406 (scop_detection::build_scop_breadth): ... this. Removed.
13407 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
13408 (scop_detection::harmful_stmt_in_bb): Likewise.
13409 (scop_detection::graphite_can_represent_stmt): Likewise.
13410 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
13411 (scop_detection::can_represent_loop): Remove recursion, fold in ...
13412 (scop_detection::can_represent_loop_1): ... this. Removed.
13413 (scop_detection::harmful_loop_in_region): Simplify after inlining
13414 the above and remove more quadraticness.
13415 (build_scops): Adjust.
13416 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
13419 2017-09-26 Jakub Jelinek <jakub@redhat.com>
13422 * config/i386/i386.c (ix86_print_operand_address_as): Only test
13423 REGNO (base) == SP_REG if base is a REG.
13425 PR middle-end/35691
13426 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
13427 if it is different SSA_NAME.
13428 (optimize_range_tests_cmp_bitwise): New function.
13429 (optimize_range_tests): Call it.
13431 2017-09-26 Richard Biener <rguenther@suse.de>
13433 PR tree-optimization/82321
13434 * graphite.c (canonicalize_loop_closed_ssa): Properly check
13435 for the def being inside the loop.
13437 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13439 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
13441 * config/s390/s390-builtins.def: Fix constraint on op4.
13443 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13445 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
13446 independent expanders.
13447 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
13448 ("vec_ordered", "vec_unordered"): New expanders.
13450 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13452 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
13455 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13457 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
13458 vec_unpacks_lo_v16qi.
13459 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
13461 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13463 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
13464 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
13465 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
13467 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13469 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
13471 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
13473 ("*vec_slb<mode>"): New insn pattern.
13474 ("vec_shr_<mode>"): New expander.
13475 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
13476 and force the shift count operand to V16QImode.
13477 ("vec_srb<mode>"): Set shift count mode to V16QI.
13479 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13481 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
13482 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
13483 ("vec_widen_smult_hi_<mode>"): New expander definitions.
13485 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
13488 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
13490 2017-09-26 Richard Biener <rguenther@suse.de>
13492 PR tree-optimization/82320
13493 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
13496 2017-09-25 Jeff Law <law@redhat.com>
13498 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
13499 prototype for new argument.
13500 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
13501 mostly extracted from rs6000_emit_allocate_stack.
13502 (rs6000_emit_probe_stack_range_stack_clash): New function.
13503 (rs6000_emit_allocate_stack): Call
13504 rs6000_emit_probe_stack_range_stack_clash as needed.
13505 (rs6000_emit_probe_stack_range): Add additional argument
13506 to call to gen_probe_stack_range{si,di}.
13507 (output_probe_stack_range): New.
13508 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
13509 (output_probe_stack_range_stack_clash): New.
13510 (rs6000_emit_prologue): Emit notes into dump file as requested.
13511 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
13512 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
13513 Add additional operand and pass it to output_probe_stack_range.
13515 2017-09-25 Bin Cheng <bin.cheng@arm.com>
13517 PR tree-optimization/82163
13518 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
13519 (checking_verify_loop_closed_ssa): New parameter.
13520 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
13521 (check_loop_closed_ssa_stmt): Delete.
13522 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
13523 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
13524 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
13527 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
13529 * brig-builtins.def: Treat HSAIL barrier builtins as
13530 setjmp/longjump style functions.
13532 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13534 * target.def (constant_alignment): New hook.
13535 * defaults.h (CONSTANT_ALIGNMENT): Delete.
13536 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
13537 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
13538 * doc/tm.texi: Regenerate.
13539 * targhooks.h (default_constant_alignment): Declare.
13540 (constant_alignment_word_strings): Likewise.
13541 * targhooks.c (default_constant_alignment): New function.
13542 (constant_alignment_word_strings): Likewise.
13543 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
13544 instead of CONSTANT_ALIGNMENT.
13545 * varasm.c (align_variable, get_variable_align, build_constant_desc)
13546 (force_const_mem): Likewise.
13547 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
13548 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
13549 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
13550 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13551 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
13553 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
13554 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13555 constant_alignment_word_strings.
13556 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
13557 (CONSTANT_ALIGNMENT): Likewise.
13558 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13559 (arm_constant_alignment): New function.
13560 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
13561 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13562 constant_alignment_word_strings.
13563 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
13564 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13565 constant_alignment_word_strings.
13566 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13567 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13568 (cris_constant_alignment): New function.
13569 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13570 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13571 (epiphany_constant_alignment): New function.
13572 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13573 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13574 constant_alignment_word_strings.
13575 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13576 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13577 constant_alignment_word_strings.
13578 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13579 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13580 constant_alignment_word_strings.
13581 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13582 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13583 * config/i386/i386.c (ix86_constant_alignment): Make static.
13584 Use the same interface as the target hook.
13585 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13586 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13587 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13588 constant_alignment_word_strings.
13589 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13590 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13591 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13592 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13593 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13594 constant_alignment_word_strings.
13595 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13596 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13597 constant_alignment_word_strings.
13598 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13599 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13600 constant_alignment_word_strings.
13601 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13602 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13604 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13605 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13606 * config/mips/mips.c (mips_constant_alignment): New function.
13607 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13608 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13609 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13610 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13611 (mmix_constant_alignment): Make static. Use the same interface
13612 as the target hook.
13613 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13614 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13615 constant_alignment_word_strings.
13616 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13617 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13618 constant_alignment_word_strings.
13619 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13620 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13621 constant_alignment_word_strings.
13622 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13623 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13624 (rs6000_constant_alignment): New function.
13625 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13626 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13627 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13628 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13629 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13630 (rs6000_constant_alignment): New function.
13631 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13632 * config/s390/s390.c (s390_constant_alignment): New function.
13633 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13634 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13635 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13636 constant_alignment_word_strings.
13637 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13638 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13639 (sparc_constant_alignment): New function.
13640 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13641 * config/spu/spu.c (spu_constant_alignment): New function.
13642 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13643 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13644 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13645 constant_alignment_word_strings.
13646 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13647 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13648 constant_alignment_word_strings.
13649 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13650 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13651 constant_alignment_word_strings.
13652 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13653 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13654 (visium_constant_alignment): New function.
13655 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13656 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13657 (xtensa_constant_alignment): New function.
13658 * system.h (CONSTANT_ALIGNMENT): Poison.
13660 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13662 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13663 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13664 (rs6000_builtin_valid_without_lhs): New helper function.
13665 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13666 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13668 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13670 * target.h (vec_perm_indices): Use unsigned short rather than
13672 (auto_vec_perm_indices): Likewise.
13673 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13674 Use unsigned int rather than unsigned char.
13675 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13677 2017-09-25 Richard Biener <rguenther@suse.de>
13679 * cfgloop.h (sort_sibling_loops): Declare.
13680 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13681 (sort_sibling_loops): New function sorting the sibling loop list
13683 * graphite.c (graphite_transform_loops): Sort sibling loops.
13685 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13687 * target.def (vec_perm_const_ok): Change sel parameter to
13689 * optabs-query.c (can_vec_perm_p): Update accordingly.
13690 * doc/tm.texi: Regenerate.
13691 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13692 auto_vec_perm_indices and remove separate nelt field.
13693 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13694 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13695 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13696 (aarch64_expand_vec_perm_const): Update accordingly.
13697 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13698 to vec_perm_indices.
13699 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13700 auto_vec_perm_indices and remove separate nelt field.
13701 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13702 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13703 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13705 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13706 to vec_perm_indices.
13707 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13708 sel to vec_perm_indices.
13709 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13710 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13711 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13713 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13716 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13719 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13720 on the FUNCTION_DECL function context if it has a DIE that is a
13723 2017-09-25 Richard Biener <rguenther@suse.de>
13725 PR tree-optimization/82285
13726 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13729 2017-09-25 Tom de Vries <tom@codesourcery.com>
13733 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13736 2017-09-25 Richard Biener <rguenther@suse.de>
13738 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13739 ISL errors other than isl_error_quota happen. Dump if the
13740 schedule is the same.
13741 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13742 errors instead of aborting inside ISL.
13744 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13747 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13748 of libgcc_eh for m64.
13749 * config/i386/darwin64.h: Likewise.
13751 2017-09-25 Richard Biener <rguenther@suse.de>
13753 PR middle-end/82144
13754 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13755 attribute for incomplete types nor twice for complete ones.
13757 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13760 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13761 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13764 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13767 * config/i386/i386.opt (mprefer-avx256): Use
13768 ix86_target_flags variable.
13769 * config/i386/i386.c (ix86_target_string): Move
13770 -mprefer-avx256 to flag2_opts.
13772 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13774 PR middle-end/35691
13775 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13776 and x != -1 | y != -1 into (x & y) != -1.
13778 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13780 * config.gcc: Add new case statement to set
13781 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13782 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13783 s390x-*-linux* case statements. Added aarch64 to the list of
13784 supported architectures.
13786 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13788 PR tree-optimization/82289
13789 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13790 STMT_VINFO_RELEVANT_P.
13792 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13793 Alan Hayward <alan.hayward@arm.com>
13794 David Sherwood <david.sherwood@arm.com>
13796 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
13797 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13798 (extract_range_from_binary_expr_1): Don't call
13799 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
13801 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13802 Alan Hayward <alan.hayward@arm.com>
13803 David Sherwood <david.sherwood@arm.com>
13805 * target.def (preferred_vector_alignment): New hook.
13806 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13808 * doc/tm.texi: Regenerate.
13809 * targhooks.h (default_preferred_vector_alignment): Declare.
13810 * targhooks.c (default_preferred_vector_alignment): New function.
13811 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
13813 (DR_TARGET_ALIGNMENT): New macro.
13814 (aligned_access_p): Update commentary.
13815 (vect_known_alignment_in_bytes): New function.
13816 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
13818 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
13819 Calculate the misalignment based on the target alignment rather than
13821 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
13822 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
13823 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
13824 the target alignment, rather than masking the element misalignment
13825 with the number of elements in a vector. Also use the target
13826 alignment when calculating the maximum number of peels.
13827 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
13828 instead of TYPE_ALIGN_UNIT.
13829 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
13830 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
13831 (vect_create_addr_base_for_vector_ref): Update call accordingly.
13832 (vect_create_data_ref_ptr): Likewise.
13833 (vect_setup_realignment): Realign by ANDing with
13834 -DR_TARGET_MISALIGNMENT.
13835 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
13836 the number of peels based on DR_TARGET_ALIGNMENT.
13837 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
13838 with the guaranteed alignment boundary when deciding whether
13840 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
13841 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
13842 (ensure_base_align): Remove stmt_info parameter. Get the
13843 target base alignment from DR_TARGET_ALIGNMENT.
13844 (vectorizable_store): Update call accordingly. Interpret
13845 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
13847 (vectorizable_load): Likewise.
13849 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13850 Alan Hayward <alan.hayward@arm.com>
13851 David Sherwood <david.sherwood@arm.com>
13853 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
13854 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
13855 (vect_enhance_data_refs_alignment): Likewise.
13857 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13859 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13860 error. Only quit immediately if parsing is complete.
13861 (BEGIN): Initialize fatal_err and parse_done.
13862 (begin fpu, end fpu): Check number of arguments.
13863 (begin arch, end arch): Likewise.
13864 (begin cpu, end cpu): Likewise.
13865 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13866 (optalias): Likewise.
13868 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13870 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13871 * config/arm/arm-isa.h: Delete. Move definitions to ...
13872 * arm-cpus.in: ... here. Use new feature and fgroup values.
13873 * config/arm/arm.c (arm_option_override): Use lower case for feature
13875 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13876 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13877 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13879 (print_isa_bits_for): New function.
13880 (gen_isa): New function.
13881 (gen_comm_data): Use print_isa_bits_for.
13882 (define feature): New keyword.
13883 (define fgroup): New keyword.
13884 * config/arm/t-arm (TM_H): Remove.
13885 (GTM_H): Add arm-isa.h.
13886 (arm-isa.h): Add rule to generate file.
13887 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13888 case for feature bit names.
13890 2017-09-22 Richard Biener <rguenther@suse.de>
13892 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
13894 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
13895 print of no dependency loops ...
13896 * graphite.c (graphite_transform_loops): ... here.
13897 (canonicalize_loop_closed_ssa_form): Work from inner to outer
13899 (same_close_phi_node, remove_duplicate_close_phi,
13900 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
13901 (canonicalize_loop_closed_ssa): ... here and simplify.
13902 * graphite-optimize-isl.c: Include tree-vectorizer.h.
13903 (optimize_isl): Use dump_printf_loc to tell when we stopped
13904 optimizing because of an ISL timeout.
13906 2017-09-22 Richard Biener <rguenther@suse.de>
13908 PR tree-optimization/82291
13909 * tree-if-conv.c (predicate_mem_writes): Make sure to
13910 remove writes in blocks predicated with false.
13912 2017-09-22 Richard Biener <rguenther@suse.de>
13914 * sese.c: Include cfganal.h.
13915 (if_region_set_false_region): Remove.
13916 (create_if_region_on_edge): Likewise.
13917 (move_sese_in_condition): Re-implement without destroying
13920 2017-09-22 Richard Biener <rguenther@suse.de>
13922 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
13923 Verify both BBs contain loop PHI nodes before dispatching to
13924 copy_loop_phi_args.
13925 (graphite_regenerate_ast_isl): Do not recompute dominators,
13926 do not verify three times. Restructure for clarity.
13927 * graphite-scop-detection.c (same_close_phi_node,
13928 remove_duplicate_close_phi, make_close_phi_nodes_unique,
13929 defined_in_loop_p, canonicalize_loop_closed_ssa,
13930 canonicalize_loop_closed_ssa_form): Simplify, remove excess
13931 checking and SSA rewrite, move to ...
13932 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
13933 (graphite_initialize): Do not pass in ctx, do not reset the
13934 SCEV cache, compute only dominators.
13935 (graphite_transform_loops): Allocate ISL ctx after
13936 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
13937 Maintain post-dominators only around build_scops.
13938 * sese.c (if_region_set_false_region): Make static. Free
13939 and recompute dominators.
13940 (move_sese_in_condition): Assert we don't get called with
13941 post-dominators computed.
13942 * sese.h (if_region_set_false_region): Remove.
13944 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
13946 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
13947 mode attribute for TARGET_AVX512VL.
13949 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
13951 * config/i386/i386.opt (mprefer-avx256): New option.
13952 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
13954 (ix86_preferred_simd_mode): Return 256-bit AVX modes
13955 for TARGET_PREFER_AVX256.
13956 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
13958 2017-09-21 Jeff Law <law@redhat.com>
13960 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
13961 Fix dump output if the only stack space is for pushed registers.
13963 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13965 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
13968 2017-09-21 Martin Sebor <msebor@redhat.com>
13971 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
13972 code (in C++) or code that triggers warnings.
13974 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
13976 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
13978 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
13980 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
13981 * sched-rgn.c: Ditto.
13982 * sel-sched-ir.c: Ditto.
13984 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
13986 * toplev.h (set_random_seed): Adjust return type.
13987 * toplev.c (init_local_tick): Move eager initialization of random_seed
13988 to get_random_seed. Adjust comment.
13989 (init_random_seed): Inline to get_random_seed, delete.
13990 (get_random_seed): Initialize random_seed lazily.
13991 (set_random_seed): Do not return previous value.
13992 (print_switch_value): Do not call get_random_seed.
13994 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
13996 * cgraph.c (delete_function_version): New, broken out from...
13997 (cgraph_node::delete_function_version): ...here. Rename to
13998 cgraph_node::delete_function_version_by_decl. Update all uses.
13999 (cgraph_node::remove): Call delete_function_version.
14001 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14004 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14005 VAR_DECLs to which addressable non-volatile parameters are mapped
14006 and for id->retvar after the return value assignment. Clear
14007 id->retval and id->retbnd after inlining.
14009 2017-09-21 Richard Biener <rguenther@suse.de>
14011 PR tree-optimization/82276
14012 PR tree-optimization/82244
14013 * tree-vrp.c (build_assert_expr_for): Set
14014 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14016 (remove_range_assertions): Revert earlier change.
14018 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14021 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14023 2017-09-21 Richard Biener <rguenther@suse.de>
14025 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14026 Restore valid IL after code generation errors.
14027 * graphite.c (graphite_transform_loops): Diagnose code
14028 generation issues as MSG_MISSED_OPTIMIZATION and continue
14029 with processing SCOPs.
14031 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14032 Alan Hayward <alan.hayward@arm.com>
14033 David Sherwood <david.sherwood@arm.com>
14035 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14036 rather than choosing between plus_constant and gen_rtx_<CODE>.
14037 * expr.c (emit_push_insn): Likewise.
14038 (expand_expr_real_2): Likewise.
14040 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14041 Alan Hayward <alan.hayward@arm.com>
14042 David Sherwood <david.sherwood@arm.com>
14044 * loop-unroll.c (split_iv): Call copy_rtx on the step.
14046 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14047 Alan Hayward <alan.hayward@arm.com>
14048 David Sherwood <david.sherwood@arm.com>
14050 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14051 calling tree_to_uhwi.
14053 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14054 Alan Hayward <alan.hayward@arm.com>
14055 David Sherwood <david.sherwood@arm.com>
14057 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
14058 INTEGER_CST rather than a negative test for ADDR_EXPR.
14060 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14061 Alan Hayward <alan.hayward@arm.com>
14062 David Sherwood <david.sherwood@arm.com>
14064 * tree-vrp.c (extract_range_from_binary_expr_1): Check
14065 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
14067 2017-09-21 Richard Biener <rguenther@suse.de>
14069 PR tree-optimization/71351
14070 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
14071 graphite_create_new_loop_guard): Remove, fold remaining parts
14073 (translate_isl_ast_node_for): ... here and simplify.
14075 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14078 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
14079 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
14080 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
14081 alternative always use QI mode, for -Os imov (=R,R) alternative
14082 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
14085 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14086 Jeff Law <law@redhat.com>
14088 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
14089 (allocate_stack_space): New function, partially extracted from
14090 s390_emit_prologue.
14091 (s390_emit_prologue): Track offset to most recent stack probe.
14092 Code to allocate space moved into allocate_stack_space.
14093 Dump actions when no stack is allocated.
14094 (s390_prologue_plus_offset): New function.
14095 (s390_emit_stack_probe): Likewise.
14097 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
14099 * common.opt (Wa, Wl, Wp, g, gz=): Add
14101 (gno-column-info): Remove.
14102 (gcolumn-info): Drop RejectNegative.
14103 (gno-): New prefix.
14104 (gno-record-gcc-switches): Remove.
14105 (grecord-gcc-switches): Drop RejectNegative.
14106 (gno-split-dwarf): Remove.
14107 (gsplit-dwarf): Drop RejectNegative.
14108 (gno-strict-dwarf): Remove.
14109 (gstrict-dwarf): Drop RejectNegative.
14110 * config/darwin.opt (gfull, gused): Add RejectNegative.
14111 * dwarf2out.c (gen_producer_string): Drop
14112 gno-record-gcc-switches handler.
14113 * optc-gen.awk: Add g to prefixes with negative forms.
14114 * opts-common.c (remapping_prefix_p): New.
14115 (find_opt): Check it.
14116 (generate_canonical_option): Test g prefix.
14117 (option_map): Add -gno- mapping.
14118 (add_misspelling_candidates): Check remapping_prefix_p.
14120 2017-09-20 Jeff Law <law@redhat.com>
14122 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
14123 thinko in stack clash protection support.
14125 * explow.c (compute_stack_clash_protection_loop_data): Use
14126 CONST_INT_P instead of explicit test. Verify object is a
14127 CONST_INT_P before looking at INTVAL.
14128 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
14129 instead of explicit test.
14131 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
14134 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
14135 address instead of to r1 and r11.
14137 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
14139 * config.gcc: Support "knm".
14140 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
14141 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14143 * config/i386/i386.c (m_KNM): Define.
14144 (processor_target_table): Add "knm".
14146 (ix86_option_override_internal): Add "knm".
14147 (ix86_issue_rate): Add PROCESSOR_KNM.
14148 (ix86_adjust_cost): Ditto.
14149 (ia32_multipass_dfa_lookahead): Ditto.
14150 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
14151 (fold_builtin_cpu): Add M_INTEL_KNM.
14152 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
14153 (processor_type): Add PROCESSOR_KNM.
14154 * config/i386/x86-tune.def: Add m_KNM.
14155 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
14157 2017-09-20 Richard Biener <rguenther@suse.de>
14159 PR tree-optimization/80213
14160 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
14161 are allowed in empty BBs as well.
14162 (canonicalize_loop_closed_ssa): Also look for other complex
14164 (scop_detection::get_sese): Include the loop-closed PHI block
14166 (scop_detection::merge_sese): Remove code adding extra blocks.
14167 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
14168 (build_scops): Assert the final returned scop is invalid.
14170 2017-09-20 Richard Biener <rguenther@suse.de>
14172 PR tree-optimization/82264
14173 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
14175 (vn_phi_lookup): Likewise.
14176 (vn_phi_insert): Likewise.
14178 2017-09-20 Jakub Jelinek <jakub@redhat.com>
14180 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
14181 that fits into uhwi or shwi, add DW_AT_const_value regardless
14182 of early_dwarf without going through RTL, using add_AT_unsigned
14185 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
14186 (DEBUG_LTO_ABBREV_SECTION): Likewise.
14187 (DEBUG_LTO_MACINFO_SECTION): Likewise.
14188 (DEBUG_MACRO_SECTION): Likewise.
14189 (DEBUG_LTO_MACRO_SECTION): Likewise.
14190 (DEBUG_STR_DWO_SECTION): Likewise.
14191 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
14192 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
14193 (DEBUG_LTO_DWO_LINE_SECTION): Define.
14194 (DEBUG_LTO_LINE_STR_SECTION): Define.
14195 (init_sections_and_labels): Initialize debug_line_str_section
14196 variable. Initialize debug_loc_section for -gdwarf-5 to
14197 DEBUG_LOCLISTS_SECTION. Formatting fixes.
14199 2017-09-20 Richard Biener <rguenther@suse.de>
14201 * graphite-sese-to-poly.c (extract_affine): Properly handle
14202 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
14204 2017-09-20 Richard Biener <rguenther@suse.de>
14206 PR tree-optimization/81373
14207 * graphite-scop-detection.c (build_cross_bb_scalars_def):
14208 Force SESE live-out defs to be handled even if they are
14211 2017-09-19 Jeff Law <law@redhat.com>
14213 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
14214 nothing for stack adjustments with REG_STACK_CHECK.
14215 * sched-deps.c (parse_add_or_inc): Reject insns with
14216 REG_STACK_CHECK from dependency breaking.
14217 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
14218 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
14219 * reg-notes.def (STACK_CHECK): New note.
14221 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
14222 (ix86_expand_prologue): Dump stack clash info as needed.
14223 Call ix86_adjust_stack_and_probe_stack_clash as needed.
14225 * function.c (dump_stack_clash_frame_info): New function.
14226 * function.h (dump_stack_clash_frame_info): Prototype.
14227 (enum stack_clash_probes): New enum.
14229 * config/alpha/alpha.c (alpha_expand_prologue): Also check
14230 flag_stack_clash_protection.
14231 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
14232 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
14233 (arm_frame_pointer_required): Likewise.
14234 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
14235 (ia64_expand_prologue): Likewise.
14236 * config/mips/mips.c (mips_expand_prologue): Likewise.
14237 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
14238 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14239 (sparc_flat_expand_prologue): Likewise.
14240 * config/spu/spu.c (spu_expand_prologue): Likewise.
14242 * explow.c: Include "params.h".
14243 (anti_adjust_stack_and_probe_stack_clash): New function.
14244 (get_stack_check_protect): Likewise.
14245 (compute_stack_clash_protection_loop_data): Likewise.
14246 (emit_stack_clash_protection_loop_start): Likewise.
14247 (emit_stack_clash_protection_loop_end): Likewise.
14248 (allocate_dynamic_stack_space): Use get_stack_check_protect.
14249 Use anti_adjust_stack_and_probe_stack_clash.
14250 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
14251 (emit_stack_clash_protection_loop_start): Likewise.
14252 (emit_stack_clash_protection_loop_end): Likewise.
14253 * rtl.h (get_stack_check_protect): Prototype.
14254 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
14255 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
14256 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
14258 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
14260 * doc/tm.texi: Rebuilt.
14261 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
14262 get_stack_check_protect.
14263 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
14264 * config/arm/arm.c (arm_expand_prologue): Likewise.
14265 (arm_frame_pointer_required): Likewise.
14266 * config/i386/i386.c (ix86_expand_prologue): Likewise.
14267 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
14268 * config/mips/mips.c (mips_expand_prologue): Likewise.
14269 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
14270 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
14271 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14272 (sparc_flat_expand_prologue): Likewise.
14274 * common.opt (-fstack-clash-protection): New option.
14275 * flag-types.h (enum stack_check_type): Note difference between
14276 -fstack-check= and -fstack-clash-protection.
14277 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
14278 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
14279 * toplev.c (process_options): Issue warnings/errors for cases
14280 not handled with -fstack-clash-protection.
14281 * doc/invoke.texi (-fstack-clash-protection): Document new option.
14282 (-fstack-check): Note additional problem with -fstack-check=generic.
14283 Note that -fstack-check is primarily for Ada and refer users
14284 to -fstack-clash-protection for stack-clash-protection.
14285 Document new params for stack clash protection.
14287 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
14289 * config/i386/i386.c (ix86_split_long_move): Do not handle
14290 address used for LEA in a special way.
14292 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
14294 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
14296 2017-09-19 Martin Sebor <msebor@redhat.com>
14299 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
14300 of incompatible types.
14302 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
14304 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14305 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
14306 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14307 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
14309 2017-09-19 Richard Biener <rguenther@suse.de>
14311 PR tree-optimization/82244
14312 * tree-vrp.c (remove_range_assertions): Do not propagate
14313 a constant to abnormals but replace the assert with a copy.
14315 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
14317 PR rtl-optimization/57878
14318 PR rtl-optimization/68988
14319 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
14320 avoidance test involving non_reload_pseudos. Move frequency test
14321 below the general fragmentation avoidance test.
14323 2017-09-19 Richard Biener <rguenther@suse.de>
14325 PR tree-optimization/69728
14326 * graphite-sese-to-poly.c (schedule_error): New global.
14327 (add_loop_schedule): Handle empty domain by failing the
14329 (build_original_schedule): Handle schedule_error.
14331 2017-09-19 Richard Biener <rguenther@suse.de>
14333 * graphite-scop-detection.c (scop_detection::can_represent_loop):
14334 Do not iterate to sibling loops but only to siblings of inner
14337 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
14340 * config/m68k/m68k.md (moveq feeding equality comparison): Check
14341 that the registers are different.
14343 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
14345 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
14346 to processor_model and "amdfam17h" to arch_names_table.
14347 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
14349 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14352 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
14354 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14355 Alan Hayward <alan.hayward@arm.com>
14356 David Sherwood <david.sherwood@arm.com>
14358 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
14360 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
14362 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
14363 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
14364 vect_schedule_slp_instance.
14365 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
14366 Update call to vect_slp_analyze_node_operations. Simplify return
14368 (vect_slp_analyze_bb_1): Update call accordingly.
14369 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
14370 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
14371 (vect_schedule_slp): Update call accordingly.
14373 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14374 Alan Hayward <alan.hayward@arm.com>
14375 David Sherwood <david.sherwood@arm.com>
14377 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
14378 with types that aren't in fact scalar.
14380 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14382 * tree-vect-slp.c (vect_record_max_nunits): New function,
14384 (vect_build_slp_tree_1): ...here.
14385 (vect_build_slp_tree_2): Call it for phis too.
14387 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14389 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
14390 to vect_get_vec_def_for_operand when getting the mask operand.
14392 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14393 Alan Hayward <alan.hayward@arm.com>
14394 David Sherwood <david.sherwood@arm.com>
14396 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
14399 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14400 Alan Hayward <alan.hayward@arm.com>
14401 David Sherwood <david.sherwood@arm.com>
14403 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
14404 calculation for vector booleans.
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-vect-stmts.c (can_vectorize_live_stmts): New function,
14412 (vect_transform_stmt): ...here.
14413 (vect_analyze_stmt): Use it instead of calling
14414 vectorizable_live_operation directly.
14416 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
14418 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
14419 non-SIMT targets in acc vector loops.
14421 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
14423 * configure.ac: Add arc and check if assembler supports gdwarf2.
14424 * configure: Regenerate.
14426 2017-09-18 Richard Biener <rguenther@suse.de>
14428 PR tree-optimization/82220
14429 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
14430 epilogue niters from the min_profitable_iters compute.
14432 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14435 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
14436 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
14437 (ix86_init_pic_reg): Revert 2017-09-01 changes.
14439 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
14442 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
14443 switching to a new text section.
14445 2017-09-18 Richard Biener <rguenther@suse.de>
14447 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
14449 (build_alias_set): Reject aliases with no access function.
14451 2017-09-18 Richard Biener <rguenther@suse.de>
14453 PR tree-optimization/79622
14454 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
14456 (build_cross_bb_scalars_use): Likewise.
14458 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
14460 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
14462 2017-09-18 Alan Modra <amodra@gmail.com>
14465 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
14466 stack_pointer_rtx for count 0. Update comments. Break up
14467 large rtl expression.
14469 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
14471 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
14472 Increase to 20 bytes.
14473 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
14474 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
14475 or avx version of the stub.
14477 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
14480 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
14481 compute the minimum stack alignment. Also update preferred stack
14482 boundary for leaf functions.
14484 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
14486 PR tree-optimization/82228
14487 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
14490 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14492 * common/config/nds32/nds32-common.c
14493 (nds32_option_optimization_table): Refine formatting.
14494 (nds32_option_optimization_table): Use -fsched-pressure and
14495 -fomit-frame-pointer for specific optimization level.
14497 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14499 * config/nds32/nds32.c: Refine formatting and comments.
14500 * config/nds32/nds32.h: Likewise.
14501 * config/nds32/nds32.md: Likewise.
14502 * config/nds32/nds32-cost.c: Likewise.
14503 * config/nds32/nds32-isr.c: Likewise.
14504 * config/nds32/nds32-md-auxiliary.c: Likewise.
14505 * config/nds32/nds32-multiple.md: Likewise.
14506 * config/nds32/nds32-predicates.c: Likewise.
14508 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
14509 Jakub Jelinek <jakub@redhat.com>
14511 Add support for -std=c++2a.
14512 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
14514 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
14516 2017-09-15 Steve Ellcey <sellcey@cavium.com>
14519 * doc/extend.texi (Common Function Attributes): Add
14520 references to ARM, AArch64, and S/390 specific attributes.
14521 (Function Specific Option Pragmas): Add AArch64 and S/390
14522 to list of back ends that support the target pragma.
14524 2017-09-15 Nathan Sidwell <nathan@acm.org>
14526 * doc/standards.texi: Fix C++17 description. Update URLs for
14529 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14531 * common.opt (Wcast-align=strict): New warning option.
14532 * doc/invoke.texi: Document -Wcast-align=strict.
14534 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
14536 * cgraph.h (cgraph_thunk_info): Add comments.
14537 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
14538 assert for VIRTUAL_* arguments stricter.
14540 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
14542 PR tree-optimization/71026
14543 * match.pd: Move RDIV patterns from fold-const.c
14544 * fold-const.c (distribute_real_division): Removed.
14545 (fold_binary_loc): Remove calls to distribute_real_divison.
14547 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14549 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
14550 c++1z and gnu++1z as deprecated. Change other references to
14551 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
14552 Change -Wc++1z-compat to -Wc++17-compat.
14553 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
14554 * dwarf2out.c (highest_c_language): Handle C++17.
14555 (gen_compile_unit_die): Likewise.
14557 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14559 PR rtl-optimization/82192
14560 * combine.c (make_extraction): Don't look through non-paradoxical
14561 SUBREGs or TRUNCATE if pos + len is or might be bigger than
14564 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14565 Alan Hayward <alan.hayward@arm.com>
14566 David Sherwood <david.sherwood@arm.com>
14568 * target.def (function_arg_offset): New hook.
14569 * targhooks.h (default_function_arg_offset): Declare.
14570 * targhooks.c (default_function_arg_offset): New function.
14571 * function.c (locate_and_pad_parm): Use
14572 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14573 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14574 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14575 * doc/tm.texi: Regenerate.
14576 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14577 * config/spu/spu.c (spu_function_arg_offset): New function.
14578 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14579 * system.h (FUNCTION_ARG_OFFSET): Poison.
14581 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14582 Alan Hayard <alan.hayward@arm.com>
14583 David Sherwood <david.sherwood@arm.com>
14585 * target.def (truly_noop_truncation): New hook.
14586 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14587 than TRULY_NOOP_TRUNCATION.
14588 * hooks.h (hook_bool_uint_uint_true): Declare.
14589 * hooks.c (hook_bool_uint_uint_true): New function.
14590 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14591 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14592 * doc/tm.texi: Regenerate.
14593 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14594 rather than TRULY_NOOP_TRUNCATION in comments.
14595 (simplify_comparison): Likewise.
14596 (record_truncated_value): Likewise.
14597 * expmed.c (extract_bit_field_1): Likewise.
14598 (extract_split_bit_field): Likewise.
14599 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14600 instead of TRULY_NOOP_TRUNCATION.
14601 * function.c (assign_parm_setup_block): Likewise.
14602 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14603 * rtlhooks.c: Include target.h.
14604 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14605 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14606 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14607 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14608 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14609 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14610 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14611 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14612 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14613 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14614 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14615 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14616 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14617 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14618 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14619 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14620 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14621 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14622 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14623 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14624 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14625 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14626 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14627 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14628 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14629 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14630 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14631 rather than TRULY_NOOP_TRUNCATION in comments.
14632 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14633 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14634 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14635 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14636 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14637 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14638 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14639 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14640 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14641 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14642 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14643 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14644 rather than TRULY_NOOP_TRUNCATION in comments.
14645 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14646 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14647 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14648 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14649 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14650 TRULY_NOOP_TRUNCATION condition.
14651 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14652 (TRULY_NOOP_TRUNCATION): Delete.
14653 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14654 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14655 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14656 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14657 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14658 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14659 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14660 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14661 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14662 rather than TRULY_NOOP_TRUNCATION in comments.
14663 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14664 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14665 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14666 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14667 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14668 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14670 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14673 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14674 (*cmp_ior): Likewise.
14675 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14676 (*ior_scc_scc_cmp): Likewise.
14677 (*and_scc_scc): Likewise.
14678 (*and_scc_scc_cmp): Likewise.
14680 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14681 Alan Hayard <alan.hayward@arm.com>
14682 David Sherwood <david.sherwood@arm.com>
14684 * target.def (can_change_mode_class): New hook.
14685 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14686 (hard_regno_nregs): Likewise.
14687 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14688 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14689 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14690 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14691 (LOAD_EXTEND_OP): Update accordingly.
14692 * doc/tm.texi: Regenerate.
14693 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14694 CANNOT_CHANGE_MODE_CLASS.
14695 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14696 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14697 * combine.c (simplify_set): Update accordingly.
14698 * emit-rtl.c (validate_subreg): Likewise.
14699 * recog.c (general_operand): Likewise.
14700 * regcprop.c (mode_change_ok): Likewise.
14701 * reload1.c (choose_reload_regs): Likewise.
14702 (inherit_piecemeal_p): Likewise.
14703 * rtlanal.c (simplify_subreg_regno): Likewise.
14704 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14705 instead of CANNOT_CHANGE_MODE_CLASS.
14706 (reload_cse_simplify_operands): Likewise.
14707 * reload.c (push_reload): Use targetm.can_change_mode_class
14708 instead of CANNOT_CHANGE_MODE_CLASS.
14709 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14710 REG_CANNOT_CHANGE_MODE_P.
14711 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14712 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14713 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14714 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14715 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14716 (arm_can_change_mode_class): New function.
14717 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14718 than CANNOT_CHANGE_MODE_CLASS in comments.
14719 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14720 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14721 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14722 (ix86_can_change_mode_class): ...this new function, inverting the
14723 sense of the return value.
14724 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14725 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14726 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14727 (ia64_can_change_mode_class): New function.
14728 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14729 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14730 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14731 (m32c_can_change_mode_class): ...this new function, inverting the
14732 sense of the return value.
14733 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14734 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14735 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14736 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14737 (mips_can_change_mode_class): ...this new function, inverting the
14738 sense of the return value.
14739 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14740 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14741 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14742 (msp430_can_change_mode_class): New function.
14743 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14744 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14745 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14746 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14747 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14748 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14749 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14750 (pa_cannot_change_mode_class): Replace with...
14751 (pa_can_change_mode_class): ...this new function, inverting the
14752 sense of the return value.
14753 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14754 than CANNOT_CHANGE_MODE_CLASS in comments.
14755 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14756 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14757 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14758 (pdp11_cannot_change_mode_class): Replace with...
14759 (pdp11_can_change_mode_class): ...this new function, inverting the
14760 sense of the return value.
14761 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14762 * config/powerpcspe/powerpcspe-protos.h
14763 (rs6000_cannot_change_mode_class_ptr): Delete.
14764 * config/powerpcspe/powerpcspe.c
14765 (rs6000_cannot_change_mode_class_ptr): Delete.
14766 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14767 (rs6000_option_override_internal): Assign to
14768 targetm.can_change_mode_class instead of
14769 rs6000_cannot_change_mode_class_ptr.
14770 (rs6000_cannot_change_mode_class): Replace with...
14771 (rs6000_can_change_mode_class): ...this new function, inverting the
14772 sense of the return value.
14773 (rs6000_debug_cannot_change_mode_class): Replace with...
14774 (rs6000_debug_can_change_mode_class): ...this new function.
14775 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14776 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14777 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14778 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14779 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14781 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14782 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14783 (rs6000_option_override_internal): Assign to
14784 targetm.can_change_mode_class instead of
14785 rs6000_cannot_change_mode_class_ptr.
14786 (rs6000_cannot_change_mode_class): Replace with...
14787 (rs6000_can_change_mode_class): ...this new function, inverting the
14788 sense of the return value.
14789 (rs6000_debug_cannot_change_mode_class): Replace with...
14790 (rs6000_debug_can_change_mode_class): ...this new function.
14791 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14792 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14793 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14794 (s390_can_change_mode_class): ...this new function, inverting the
14795 sense of the return value.
14796 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14797 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14798 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14799 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14800 (sh_cannot_change_mode_class): Replace with...
14801 (sh_can_change_mode_class): ...this new function, inverting the
14802 sense of the return value.
14803 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14804 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14805 (sparc_can_change_mode_class): New function.
14806 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14807 * config/spu/spu.c (spu_can_change_mode_class): New function.
14808 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14809 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14810 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14811 (visium_can_change_mode_class): New function.
14812 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
14814 2017-09-15 Richard Biener <rguenther@suse.de>
14816 PR tree-optimization/82217
14817 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
14818 but not undefined case.
14820 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14823 * postreload.c (reload_cse_simplify_operands): Skip
14824 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
14826 2017-09-15 Richard Biener <rguenther@suse.de>
14828 PR tree-optimization/68823
14829 * graphite-scop-detection.c (build_alias_set): If we have a
14830 possible dependence check whether we can handle them by just
14831 looking at the DRs DR_ACCESS_FNs.
14832 (build_scops): If build_alias_set fails, fail the SCOP.
14834 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
14836 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
14837 to support float128 built-in functions that require the ISA 3.0
14839 (BU_FLOAT128_3_HW): Likewise.
14840 (SQRTF128): Add support for the IEEE 128-bit square root and fma
14841 built-in functions.
14842 (FMAF128): Likewise.
14844 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14845 support for built-in functions that need the ISA 3.0 IEEE 128-bit
14846 floating point instructions.
14847 (rs6000_invalid_builtin): Likewise.
14848 (rs6000_builtin_mask_names): Likewise.
14849 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
14850 (RS6000_BTM_FLOAT128_HW): Likewise.
14851 (RS6000_BTM_COMMON): Likewise.
14852 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
14854 * doc/extend.texi (RS/6000 built-in functions): Document the
14855 IEEE 128-bit floating point square root and fused multiply-add
14856 built-in functions.
14858 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
14860 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
14861 reg (r2) isn't in the set of registers defined in the prologue.
14863 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14864 Alan Hayward <alan.hayward@arm.com>
14865 David Sherwood <david.sherwood@arm.com>
14867 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
14868 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
14869 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
14870 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
14871 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
14873 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
14874 max_vectorization_factor.
14875 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
14877 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14878 Alan Hayward <alan.hayward@arm.com>
14879 David Sherwood <david.sherwood@arm.com>
14881 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
14882 (vect_worthwhile_without_simd_p): Declare.
14883 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
14884 (vectorizable_reduction): Use it.
14885 * tree-vect-stmts.c (vectorizable_shift): Likewise.
14886 (vectorizable_operation): Likewise.
14888 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14889 Alan Hayward <alan.hayward@arm.com>
14890 David Sherwood <david.sherwood@arm.com>
14892 * tree-vectorizer.h (vect_get_num_copies): New function.
14893 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
14894 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14895 (vectorizable_induction): Likewise.
14896 (vectorizable_live_operation): Likewise.
14897 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14898 (vectorizable_bswap): Likewise.
14899 (vectorizable_call): Likewise.
14900 (vectorizable_conversion): Likewise.
14901 (vectorizable_assignment): Likewise.
14902 (vectorizable_shift): Likewise.
14903 (vectorizable_operation): Likewise.
14904 (vectorizable_store): Likewise.
14905 (vectorizable_load): Likewise.
14906 (vectorizable_condition): Likewise.
14907 (vectorizable_comparison): Likewise.
14908 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
14910 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14911 Alan Hayward <alan.hayward@arm.com>
14912 David Sherwood <david.sherwood@arm.com>
14914 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
14915 of vect_init_vector.
14917 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14918 Alan Hayward <alan.hayward@arm.com>
14919 David Sherwood <david.sherwood@arm.com>
14921 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
14922 an inline wrapper that provides a location.
14923 (gimple_build_vector): Likewise.
14924 * gimple-fold.c (gimple_build_vector_from_val): New function.
14925 (gimple_build_vector): Likewise.
14926 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
14927 functions to build the initial value. Always return a gimple value.
14928 (get_initial_defs_for_reduction): Likewise. Only compute
14930 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
14931 vect_init_vector on the results from get_initial_def(s)_for_reduction.
14932 (vectorizable_induction): Use gimple_build_vector rather than
14935 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14936 Alan Hayward <alan.hayward@arm.com>
14937 David Sherwood <david.sherwood@arm.com>
14939 * target.h (vec_perm_indices): New typedef.
14940 (auto_vec_perm_indices): Likewise.
14941 * optabs-query.h: Include target.h
14942 (can_vec_perm_p): Take a vec_perm_indices *.
14943 * optabs-query.c (can_vec_perm_p): Likewise.
14944 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
14945 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14946 * tree-vect-generic.c (lower_vec_perm): Likewise.
14947 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
14948 (vect_grouped_load_supported): Likewise.
14949 (vect_shift_permute_load_chain): Likewise.
14950 (vect_permute_store_chain): Use auto_vec_perm_indices.
14951 (vect_permute_load_chain): Likewise.
14952 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
14953 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
14954 Update uses of can_vec_perm_p.
14955 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
14956 mode with a number of elements. Take a vec_perm_indices *.
14957 (vect_create_epilog_for_reduction): Update accordingly.
14958 Use auto_vec_perm_indices.
14959 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
14960 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
14961 (vect_transform_slp_perm_load): Likewise.
14962 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
14963 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
14964 (vect_gen_perm_mask_checked): Likewise.
14965 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
14966 (vect_gen_perm_mask_checked): Likewise.
14967 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
14968 (vectorizable_store): Likewise.
14969 (vectorizable_load): Likewise.
14970 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
14971 (vectorizable_bswap): Likewise.
14973 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14974 Alan Hayward <alan.hayward@arm.com>
14975 David Sherwood <david.sherwood@arm.com>
14977 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
14978 * tree.c (build_vector): Likewise.
14979 (build_vector_from_ctor): Update accordingly.
14980 (build_vector_from_val): Likewise.
14981 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14982 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14983 * tree-vect-generic.c (add_rshift): Likewise.
14984 (expand_vector_divmod): Likewise.
14985 (optimize_vector_constructor): Likewise.
14986 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
14987 (vect_transform_slp_perm_load): Likewise.
14988 (vect_schedule_slp_instance): Likewise.
14989 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
14990 (vectorizable_call): Likewise.
14991 (vect_gen_perm_mask_any): Likewise. Add elements in order.
14992 * expmed.c (make_tree): Likewise.
14993 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
14994 a vector passed to build_vector.
14995 (fold_convert_const): Likewise.
14996 (exact_inverse): Likewise.
14997 (fold_ternary_loc): Likewise.
14998 (fold_relational_const): Likewise.
14999 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15000 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15001 (const_unop): Likewise. Store the reduction accumulator in a
15002 variable rather than an array.
15003 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15004 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15005 the new vector, rather than constructing it after the input arrays.
15006 (native_interpret_vector): Use auto_vec<tree> when building
15007 a vector passed to build_vector. Add elements in order.
15008 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15009 auto_vec<tree> when building a vector passed to build_vector.
15010 (vect_create_epilog_for_reduction): Likewise.
15011 (vectorizable_induction): Likewise.
15012 (get_initial_def_for_reduction): Likewise. Fix indentation of
15014 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15016 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15017 passed to build_vector.
15019 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15020 Alan Hayward <alan.hayward@arm.com>
15021 David Sherwood <david.sherwood@arm.com>
15023 * tree-core.h (tree_base::u): Add an "nelts" field.
15024 (tree_vector): Use VECTOR_CST_NELTS as the length.
15025 * tree.c (tree_size): Likewise.
15026 (make_vector): Initialize VECTOR_CST_NELTS.
15027 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15028 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15029 TYPE_VECTOR_SUBPARTS.
15030 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15031 as the number of units, setting it from VECTOR_CST_NELTS.
15032 (const_vector_from_tree): Likewise.
15033 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15034 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15035 (fold_negate_expr_1): Likewise.
15036 (fold_convert_const): Likewise.
15037 (const_binop): Likewise. Differentiate the number of output and
15039 (const_unop): Likewise.
15040 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15041 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15042 in cases that did the opposite.
15044 2017-09-14 Richard Biener <rguenther@suse.de>
15046 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15049 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15051 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15053 2017-09-14 Jakub Jelinek <jakub@redhat.com>
15056 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
15057 if and where to split a bb, except for splitting before debug insn
15058 sequences followed by non-label real insn. Delete debug insns
15059 in between basic blocks.
15061 * combine.c (make_compound_operation_int): Formatting fixes.
15063 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
15064 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15065 * config/netbsd.h (LINK_EH_SPEC): Likewise.
15066 * config/sol2.h (LINK_EH_SPEC): Likewise.
15067 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15068 * config/s390/linux.h (LINK_SPEC): Likewise.
15069 * config/freebsd.h (LINK_EH_SPEC): Likewise.
15070 * config/openbsd.h (LINK_EH_SPEC): Likewise.
15071 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15072 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
15073 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
15074 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15075 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
15076 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
15078 2017-09-13 Jakub Jelinek <jakub@redhat.com>
15080 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
15082 (ENDFILE_LINUX_SPEC): Likewise.
15083 (LINK_EH_SPEC): Likewise.
15084 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
15085 (LINK_OS_LINUX_SPEC32): Likewise.
15086 (LINK_OS_LINUX_SPEC64): Likewise.
15087 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
15088 (LINK_OS_LINUX_SPEC): Likewise.
15090 2017-09-13 Martin Liska <mliska@suse.cz>
15092 PR middle-end/82154
15093 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
15094 CASE_HIGH is NULL_TREE.
15096 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15097 Alan Hayward <alan.hayward@arm.com>
15098 David Sherwood <david.sherwood@arm.com>
15100 * target.def (secondary_memory_needed): New hook.
15101 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
15102 instead of SECONDARY_MEMORY_NEEDED.
15103 (secondary_memory_needed_mode): Likewise.
15104 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
15105 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
15106 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
15107 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
15108 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
15109 * doc/tm.texi: Regenerate.
15110 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
15111 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
15112 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15113 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
15114 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
15115 * config/i386/i386.c (inline_secondary_memory_needed): Put the
15116 mode argument first and change the reg_class arguments to reg_class_t.
15117 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
15118 Make static. Update the call to inline_secondary_memory_needed.
15119 (ix86_register_move_cost): Update the call to
15120 inline_secondary_memory_needed.
15121 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15122 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
15124 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
15125 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15127 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
15128 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
15129 * config/mips/mips.c (mips_secondary_memory_needed): Make static
15130 and match hook interface. Add comment from mips.h.
15131 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15132 * config/mmix/mmix.md (truncdfsf2): Refer to
15133 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15135 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
15136 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
15137 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15138 (pa_secondary_memory_needed): New function.
15139 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
15140 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
15141 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15142 (pdp11_secondary_memory_needed): Make static and match hook interface.
15143 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
15144 * config/powerpcspe/powerpcspe-protos.h
15145 (rs6000_secondary_memory_needed_ptr): Delete.
15146 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
15148 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15149 (rs6000_option_override_internal): Assign to
15150 targetm.secondary_memory_needed rather than
15151 rs6000_secondary_memory_needed_ptr.
15152 (rs6000_secondary_memory_needed): Match hook interface.
15153 (rs6000_debug_secondary_memory_needed): Likewise.
15154 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
15155 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
15156 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
15157 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15158 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
15159 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
15161 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
15162 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15163 (rs6000_option_override_internal): Assign to
15164 targetm.secondary_memory_needed rather than
15165 rs6000_secondary_memory_needed_ptr.
15166 (rs6000_secondary_memory_needed): Match hook interface.
15167 (rs6000_debug_secondary_memory_needed): Likewise.
15168 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
15169 * config/s390/s390.c (s390_secondary_memory_needed): New function.
15170 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15171 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
15172 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15173 (sparc_secondary_memory_needed): New function.
15174 * lra-constraints.c (check_and_process_move): Refer to
15175 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15177 (curr_insn_transform): Likewise.
15178 (process_alt_operands): Use targetm.secondary_memory_needed
15179 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15180 (check_secondary_memory_needed_p): Likewise.
15181 (choose_split_class): Likewise.
15182 * reload.c: Unconditionally include code that was previously
15183 conditional on SECONDARY_MEMORY_NEEDED.
15184 (push_secondary_reload): Use targetm.secondary_memory_needed
15185 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15186 (push_reload): Likewise.
15187 * reload1.c: Unconditionally include code that was previously
15188 conditional on SECONDARY_MEMORY_NEEDED.
15189 (choose_reload_regs): Use targetm.secondary_memory_needed
15190 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15191 (gen_reload): Likewise.
15192 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
15194 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15195 Alan Hayward <alan.hayward@arm.com>
15196 David Sherwood <david.sherwood@arm.com>
15198 * target.def (secondary_memory_needed_mode): New hook:
15199 * targhooks.c (default_secondary_memory_needed_mode): Declare.
15200 * targhooks.h (default_secondary_memory_needed_mode): New function.
15201 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
15202 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
15203 * doc/tm.texi: Regenerate.
15204 * lra-constraints.c (check_and_process_move): Use
15205 targetm.secondary_memory_needed_mode instead of
15206 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
15207 (curr_insn_transform): Likewise.
15208 * reload.c (get_secondary_mem): Likewise.
15209 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15210 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
15212 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15213 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15214 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
15215 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15216 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
15218 * config/powerpcspe/powerpcspe-protos.h
15219 (rs6000_secondary_memory_needed_mode): Delete.
15220 * config/powerpcspe/powerpcspe.c
15221 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15222 (rs6000_secondary_memory_needed_mode): Make static.
15223 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15224 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
15226 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15228 (rs6000_secondary_memory_needed_mode): Make static.
15229 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15230 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
15231 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15232 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15233 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15235 (sparc_secondary_memory_needed_mode): New function.
15236 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
15238 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
15240 * config/aarch64/constraints.md (Umq): New constraint.
15241 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15243 (mov<mode>): Update condition.
15245 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15247 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
15248 when bitposition is the same.
15250 2017-09-13 Richard Biener <rguenther@suse.de>
15252 * dwarf2out.c (output_die_symbol): Remove.
15253 (output_die): Do not output a DIEs symbol.
15255 2017-09-13 Richard Biener <rguenther@suse.de>
15257 PR middle-end/82128
15258 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
15259 default-def to avoid breaking iterator update with the weird
15260 interaction with cgraph_update_edges_for_call_stmt_node.
15262 2017-09-13 Richard Biener <rguenther@suse.de>
15264 * tree-cfg.c (verify_gimple_assign_binary): Add verification
15265 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
15266 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
15267 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
15269 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
15271 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
15272 Disable pc relative literal load irrespective of
15273 TARGET_FIX_ERR_A53_84341 for default.
15275 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
15277 * config/sparc/sparc.c (output_return): Output the source location of
15278 the insn in the delay slot, if any.
15279 (output_sibcall): Likewise.
15281 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
15284 * common.opt (-static-pie): New alias.
15285 (shared): Negate static-pie.
15286 (-no-pie): Update help text.
15288 (static-pie): New option.
15289 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
15290 -static-pie support.
15291 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
15292 (LINK_EH_SPEC): Likewise.
15293 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15294 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15295 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15296 * gcc.c (LINK_COMMAND_SPEC): Likewise.
15297 (init_gcc_specs): Likewise.
15298 (init_spec): Likewise.
15299 (display_help): Update help message for -pie.
15300 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
15303 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
15305 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
15306 (movdi_aarch64): Likewise.
15307 (movti_aarch64): Likewise.
15309 2017-09-12 Simon Wright <simon@pushface.org>
15312 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
15313 calculation of the minor version, always output as 0.
15315 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15318 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15319 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
15320 on it early, rather than manual conversion late. For
15321 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
15322 instead of performing manual conversion.
15324 2017-09-12 Carl Love <cel@us.ibm.com>
15326 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
15327 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
15329 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15330 VMULOSW): Add definitions.
15331 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15332 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15333 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15334 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15335 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15337 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15339 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
15341 (movti_aarch64): Likewise.
15342 (movdf_aarch64): Likewise.
15343 (movtf_aarch64): Likewise.
15344 (load_pairdi): Likewise.
15345 (store_pairdi): Likewise.
15346 (load_pairdf): Likewise.
15347 (store_pairdf): Likewise.
15348 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
15349 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
15350 (ldr_got_small_<mode>): Likewise.
15351 (ldr_got_small_28k_<mode>): Likewise.
15352 (ldr_got_tiny): Likewise.
15353 * config/aarch64/iterators.md (ldst_sz): New.
15354 (ldpstp_sz): Likewise.
15355 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
15357 (thunderx_load): Split load_8 to load_16.
15358 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
15360 (thunderx2t99_storepair_basic): Split store_8 to store_16.
15361 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
15362 (xgene1_store_pair): Split store_8 to store_16.
15363 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
15364 (falkor_st_0_st_sd): Split store_8 to store_16.
15366 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15368 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
15369 and store1/2/3/4 to store_4/8/12/16.
15370 * config/aarch64/aarch64.md: Update for rename.
15371 * config/arm/arm.md: Likewise.: Likewise.
15372 * config/arm/arm.c: Likewise.
15373 * config/arm/thumb1.md: Likewise.
15374 * config/arm/thumb2.md: Likewise.
15375 * config/arm/vfp.md: Likewise.
15376 * config/arm/arm-generic.md: Likewise.
15377 * config/arm/arm1020e.md: Likewise.
15378 * config/arm/arm1026ejs.md: Likewise.
15379 * config/arm/arm1136jfs.md: Likewise.
15380 * config/arm/arm926ejs.md: Likewise.
15381 * config/arm/cortex-a15.md: Likewise.
15382 * config/arm/cortex-a17.md: Likewise.
15383 * config/arm/cortex-a5.md: Likewise.
15384 * config/arm/cortex-a53.md: Likewise.
15385 * config/arm/cortex-a57.md: Likewise.
15386 * config/arm/cortex-a7.md: Likewise.
15387 * config/arm/cortex-a8.md: Likewise.
15388 * config/arm/cortex-a9.md: Likewise.
15389 * config/arm/cortex-m4.md: Likewise.
15390 * config/arm/cortex-m7.md: Likewise.
15391 * config/arm/cortex-r4.md: Likewise.
15392 * config/arm/exynos-m1.md: Likewise.
15393 * config/arm/fa526.md: Likewise.
15394 * config/arm/fa606te.md: Likewise.
15395 * config/arm/fa626te.md: Likewise.
15396 * config/arm/fa726te.md: Likewise.
15397 * config/arm/fmp626.md: Likewise.
15398 * config/arm/iwmmxt.md: Likewise.
15399 * config/arm/ldmstm.md: Likewise.
15400 * config/arm/marvell-pj4.md: Likewise.
15401 * config/arm/xgene1.md: Likewise.
15402 * config/aarch64/thunderx.md: Likewise.
15403 * config/aarch64/thunderx2t99.md: Likewise.
15404 * config/aarch64/falkor.md: Likewise.
15406 2017-09-12 Martin Liska <mliska@suse.cz>
15408 * attribs.c (private_lookup_attribute): New function.
15409 * attribs.h (private_lookup_attribute): Declared here.
15410 (lookup_attribute): Called from this place.
15412 2017-09-12 Richard Biener <rguenther@suse.de>
15414 PR tree-optimization/82157
15415 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
15416 stmts with side-effects.
15418 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15419 Alan Hayward <alan.hayward@arm.com>
15420 David Sherwood <david.sherwood@arm.com>
15422 * target.def (hard_regno_nregs): New hook.
15423 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
15424 * targhooks.h (default_hard_regno_nregs): Declare.
15425 * targhooks.c (default_hard_regno_nregs): New function.
15426 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
15427 (TARGET_HARD_REGNO_NREGS): ...this hook.
15428 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
15429 (CLASS_MAX_NREGS): Likewise.
15430 * doc/tm.texi: Regenerate.
15431 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
15432 instead of HARD_REGNO_NREGS.
15433 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
15434 HARD_REGNO_NREGS in the comment.
15435 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
15436 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
15437 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
15438 Return an unsigned int.
15439 (TARGET_HARD_REGNO_NREGS): Redefine.
15440 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
15441 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
15442 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15443 (arc_hard_regno_nregs): New function.
15444 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
15445 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
15446 (arm_hard_regno_nregs): New function.
15447 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
15448 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
15449 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
15450 (TARGET_HARD_REGNO_NREGS): Redefine.
15451 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
15452 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
15453 (HARD_REGNO_NREGS): Delete.
15454 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
15455 (cr16_hard_regno_nregs): New function.
15456 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
15457 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
15458 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
15459 (cris_hard_regno_nregs): New function.
15460 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
15461 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
15462 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15463 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
15464 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
15465 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
15466 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
15467 (frv_hard_regno_nregs): Make static. Take and return an
15469 (frv_class_max_nregs): Remove outdated copy of documentation.
15470 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
15471 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
15472 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
15473 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
15474 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
15475 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
15476 (TARGET_HARD_REGNO_NREGS): Redefine.
15477 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
15478 (CLASS_MAX_NREGS): Update comment.
15479 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
15480 (ia64_hard_regno_nregs): New function.
15481 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
15482 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
15483 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
15484 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
15485 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
15487 (m32c_hard_regno_nregs): Likewise. Make static.
15488 (TARGET_HARD_REGNO_NREGS): Redefine.
15489 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
15490 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
15491 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
15492 (m68k_hard_regno_nregs): New function.
15493 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
15494 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
15495 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
15496 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
15497 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
15498 Take and return an unsigned int.
15499 (TARGET_HARD_REGNO_NREGS): Redefine.
15500 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
15501 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15502 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
15503 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
15504 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
15505 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
15506 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
15507 (msp430_hard_regno_nregs): Make static. Take and return an
15509 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
15510 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
15511 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
15512 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
15513 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
15514 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
15515 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
15516 (TARGET_HARD_REGNO_NREGS): Redefine.
15517 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
15518 (PA_HARD_REGNO_NREGS): ...this.
15519 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
15520 (PA_HARD_REGNO_NREGS): ...this.
15521 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15522 (pa_hard_regno_nregs): New function.
15523 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
15524 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
15525 (pdp11_hard_regno_nregs): New function.
15526 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
15527 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
15528 (rs6000_hard_regno_nregs_hook): New function.
15529 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
15530 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
15531 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
15532 Take and return an unsigned int. Move earlier in file.
15533 (TARGET_HARD_REGNO_NREGS): Redefine.
15534 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
15535 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
15536 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
15537 (rl78_hard_regno_nregs): Make static. Take and return an
15539 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
15540 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
15541 (rs6000_hard_regno_nregs_hook): New function.
15542 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
15543 * config/rx/rx.c (rx_hard_regno_nregs): New function.
15544 (TARGET_HARD_REGNO_NREGS): Redefine.
15545 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
15546 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
15547 instead of HARD_REGNO_NREGS.
15548 (s390_hard_regno_nregs): New function.
15549 (s390_hard_regno_mode_ok): Add comment from s390.h.
15550 (TARGET_HARD_REGNO_NREGS): Redefine.
15551 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
15552 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
15553 (sh_hard_regno_nregs): New function.
15554 (sh_pass_in_reg_p): Use it.
15555 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
15556 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15557 (sparc_hard_regno_nregs): New function.
15558 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
15559 * config/spu/spu.c (spu_hard_regno_nregs): New function.
15560 (spu_function_arg_advance): Use it, supplying a valid register number.
15561 (TARGET_HARD_REGNO_NREGS): Redefine.
15562 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
15563 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
15564 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
15565 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15566 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15567 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15568 (CLASS_MAX_NREGS): Remove copy of old documentation.
15569 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15570 (visium_hard_regno_nregs): New function.
15571 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15572 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15573 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15574 xtensa_hard_regno_nregs): New function.
15575 * system.h (HARD_REGNO_NREGS): Poison.
15577 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15579 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15580 hard_regno_nregs instead of HARD_REGNO_NREGS.
15581 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15582 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15583 (c6x_expand_epilogue): Likewise.
15584 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15585 (frv_read_iacc_argument): Likewise.
15586 * config/sh/sh.c: Include regs.h.
15587 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15588 (regs_used): Likewise.
15589 (output_stack_adjust): Likewise.
15590 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15591 * expmed.c: Include regs.h.
15592 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15593 * ree.c: Include regs.h.
15594 (combine_reaching_defs): Use hard_regno_nregs instead of
15596 (add_removable_extension): Likewise.
15598 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15600 * regs.h (hard_regno_nregs): Turn into a function.
15601 (end_hard_regno): Update accordingly.
15602 * caller-save.c (setup_save_areas): Likewise.
15603 (save_call_clobbered_regs): Likewise.
15604 (replace_reg_with_saved_mem): Likewise.
15605 (insert_restore): Likewise.
15606 (insert_save): Likewise.
15607 * combine.c (can_change_dest_mode): Likewise.
15608 (move_deaths): Likewise.
15609 (distribute_notes): Likewise.
15610 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15611 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15612 (rs6000_split_multireg_move): Likewise.
15613 (rs6000_register_move_cost): Likewise.
15614 (rs6000_memory_move_cost): Likewise.
15615 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15616 (rs6000_split_multireg_move): Likewise.
15617 (rs6000_register_move_cost): Likewise.
15618 (rs6000_memory_move_cost): Likewise.
15619 * cselib.c (cselib_reset_table): Likewise.
15620 (cselib_lookup_1): Likewise.
15621 * emit-rtl.c (set_mode_and_regno): Likewise.
15622 * function.c (aggregate_value_p): Likewise.
15623 * ira-color.c (setup_profitable_hard_regs): Likewise.
15624 (check_hard_reg_p): Likewise.
15625 (calculate_saved_nregs): Likewise.
15626 (assign_hard_reg): Likewise.
15627 (improve_allocation): Likewise.
15628 (calculate_spill_cost): Likewise.
15629 * ira-emit.c (modify_move_list): Likewise.
15630 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15631 (ira_hard_reg_in_set_p): Likewise.
15632 * ira.c (setup_reg_mode_hard_regset): Likewise.
15633 (clarify_prohibited_class_mode_regs): Likewise.
15634 (check_allocation): Likewise.
15635 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15636 (lra_setup_reg_renumber): Likewise.
15637 (setup_try_hard_regno_pseudos): Likewise.
15638 (spill_for): Likewise.
15639 (assign_hard_regno): Likewise.
15640 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15641 * lra-constraints.c (in_class_p): Likewise.
15642 (lra_constraint_offset): Likewise.
15643 (simplify_operand_subreg): Likewise.
15644 (lra_constraints): Likewise.
15645 (split_reg): Likewise.
15646 (split_if_necessary): Likewise.
15647 (invariant_p): Likewise.
15648 (inherit_in_ebb): Likewise.
15649 * lra-lives.c (process_bb_lives): Likewise.
15650 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15651 (get_hard_regs): Likewise.
15652 (do_remat): Likewise.
15653 * lra-spills.c (assign_spill_hard_regs): Likewise.
15654 * mode-switching.c (create_pre_exit): Likewise.
15655 * postreload.c (reload_combine_recognize_pattern): Likewise.
15656 * recog.c (peep2_find_free_register): Likewise.
15657 * regcprop.c (kill_value_regno): Likewise.
15658 (set_value_regno): Likewise.
15659 (copy_value): Likewise.
15660 (maybe_mode_change): Likewise.
15661 (find_oldest_value_reg): Likewise.
15662 (copyprop_hardreg_forward_1): Likewise.
15663 * regrename.c (check_new_reg_p): Likewise.
15664 (regrename_do_replace): Likewise.
15665 * reload.c (push_reload): Likewise.
15666 (combine_reloads): Likewise.
15667 (find_dummy_reload): Likewise.
15668 (operands_match_p): Likewise.
15669 (find_reloads): Likewise.
15670 (find_equiv_reg): Likewise.
15671 (reload_adjust_reg_for_mode): Likewise.
15672 * reload1.c (count_pseudo): Likewise.
15673 (count_spilled_pseudo): Likewise.
15674 (find_reg): Likewise.
15675 (clear_reload_reg_in_use): Likewise.
15676 (free_for_value_p): Likewise.
15677 (allocate_reload_reg): Likewise.
15678 (choose_reload_regs): Likewise.
15679 (reload_adjust_reg_for_temp): Likewise.
15680 (emit_reload_insns): Likewise.
15681 (delete_output_reload): Likewise.
15682 * rtlanal.c (subreg_get_info): Likewise.
15683 * sched-deps.c (sched_analyze_reg): Likewise.
15684 * sel-sched.c (init_regs_for_mode): Likewise.
15685 (mark_unavailable_hard_regs): Likewise.
15686 (choose_best_reg_1): Likewise.
15687 (verify_target_availability): Likewise.
15688 * valtrack.c (dead_debug_insert_temp): Likewise.
15689 * var-tracking.c (track_loc_p): Likewise.
15690 (emit_note_insn_var_location): Likewise.
15691 * varasm.c (make_decl_rtl): Likewise.
15692 * reginfo.c (choose_hard_reg_mode): Likewise.
15693 (init_reg_modes_target): Refer directly to
15694 this_target_regs->x_hard_regno_nregs.
15696 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15698 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15699 instead of hard_regno_nregs.
15701 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15703 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15704 end_hard_regno instead of hard_regno_nregs.
15705 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15706 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15707 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15708 * ira-color.c (improve_allocation): Likewise.
15709 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15710 * lra-lives.c (mark_regno_live): Likewise.
15711 (mark_regno_dead): Likewise.
15712 * lra-remat.c (operand_to_remat): Likewise.
15713 * lra.c (collect_non_operand_hard_regs): Likewise.
15714 * postreload.c (reload_combine_note_store): Likewise.
15715 (move2add_valid_value_p): Likewise.
15716 * reload.c (regno_clobbered_p): Likewise.
15718 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15720 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15722 * config/v850/v850.c (v850_reorg): Likewise.
15723 * reload.c (refers_to_regno_for_reload_p): Likewise.
15724 (find_equiv_reg): Likewise.
15725 * reload1.c (reload_reg_reaches_end_p): Likewise.
15727 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15729 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15731 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15732 * config/arm/arm.c (output_move_neon): Likewise.
15733 (arm_attr_length_move_neon): Likewise.
15734 (neon_split_vcombine): Likewise.
15735 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15736 (c6x_mark_reg_written): Likewise.
15737 (c6x_dwarf_register_span): Likewise.
15738 * config/i386/i386.c (ix86_save_reg): Likewise.
15739 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15740 (rws_access_reg): Likewise.
15741 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15742 * mode-switching.c (create_pre_exit): Likewise.
15743 * ree.c (combine_reaching_defs): Likewise.
15744 (add_removable_extension): Likewise.
15745 * regcprop.c (find_oldest_value_reg): Likewise.
15746 (copyprop_hardreg_forward_1): Likewise.
15747 * reload.c (reload_inner_reg_of_subreg): Likewise.
15748 (push_reload): Likewise.
15749 (combine_reloads): Likewise.
15750 (find_dummy_reload): Likewise.
15751 (reload_adjust_reg_for_mode): Likewise.
15752 * reload1.c (find_reload_regs): Likewise.
15753 (forget_old_reloads_1): Likewise.
15754 (reload_reg_free_for_value_p): Likewise.
15755 (reload_adjust_reg_for_temp): Likewise.
15756 (emit_reload_insns): Likewise.
15757 (delete_output_reload): Likewise.
15758 * sel-sched.c (choose_best_reg_1): Likewise.
15759 (choose_best_pseudo_reg): Likewise.
15761 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15762 Alan Hayward <alan.hayward@arm.com>
15763 David Sherwood <david.sherwood@arm.com>
15765 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15766 * target.def (slow_unaligned_access): New hook.
15767 * targhooks.h (default_slow_unaligned_access): Declare.
15768 * targhooks.c (default_slow_unaligned_access): New function.
15769 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15770 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15771 * doc/tm.texi: Regenerate.
15772 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15773 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15774 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15776 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15777 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15779 (rs6000_slow_unaligned_access): New function.
15780 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15781 (expand_block_compare): Likewise.
15782 (expand_strn_compare): Likewise.
15783 (rs6000_rtx_costs): Likewise.
15784 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15785 (riscv_slow_unaligned_access): Likewise.
15786 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15787 (riscv_slow_unaligned_access_p): ...this and make static.
15788 (riscv_option_override): Update accordingly.
15789 (riscv_slow_unaligned_access): New function.
15790 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15791 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15792 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15793 (rs6000_slow_unaligned_access): New function.
15794 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15795 (rs6000_rtx_costs): Likewise.
15796 * config/rs6000/rs6000-string.c (expand_block_compare)
15797 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15798 of SLOW_UNALIGNED_ACCESS.
15799 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15800 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15801 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15802 of SLOW_UNALIGNED_ACCESS.
15803 * expmed.c (simple_mem_bitfield_p): Likewise.
15804 * expr.c (alignment_for_piecewise_move): Likewise.
15805 (emit_group_load_1): Likewise.
15806 (emit_group_store): Likewise.
15807 (copy_blkmode_from_reg): Likewise.
15808 (emit_push_insn): Likewise.
15809 (expand_assignment): Likewise.
15810 (store_field): Likewise.
15811 (expand_expr_real_1): Likewise.
15812 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15813 * lra-constraints.c (simplify_operand_subreg): Likewise.
15814 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15815 * gimple-ssa-store-merging.c: Likewise in block comment at start
15817 * tree-ssa-strlen.c: Include target.h.
15818 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
15819 of SLOW_UNALIGNED_ACCESS.
15820 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
15822 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15824 PR rtl-optimization/82185
15825 * expmed.c (emit_store_flag_int): Only test tem if it has been
15828 2017-09-12 Richard Biener <rguenther@suse.de>
15830 PR middle-end/82149
15831 * match.pd ((FTYPE) N CMP CST): Fix typo.
15833 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
15835 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
15837 (mips_near_type_p): Add 'short_call' attribute as a synonym
15839 * doc/extend.texi (short_call): Document new function attribute.
15841 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15844 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
15845 assertion check that in the condition.
15846 (get_atomic_generic_size): Likewise. Before testing if parameter
15847 has pointer type, if it has array type, call for C++
15848 default_conversion to perform array-to-pointer conversion.
15850 2017-09-12 Richard Biener <rguenther@suse.de>
15852 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
15853 for operations we cannot scalarize.
15855 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
15857 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
15858 vectors heap vectors. Clean up comments.
15859 Make visited_bbs a reference.
15860 (profitable_jump_thread_path): Make GC
15861 vectors heap vectors. Clean up comments.
15863 (convert_and_register_jump_thread_path): Make GC vectors heap
15865 (check_subpath_and_update_thread_path): Same. Clean up comments.
15866 Make visited_bbs a reference.
15867 (handle_phi): Abstract common code to to
15868 register_jump_thread_path_if_profitable.
15869 Rename VAR_BB to DEF_BB.
15871 Make GC vectors heap vectors.
15872 Make visited_bbs a reference.
15873 (handle_assignment): Same.
15874 (register_jump_thread_path_if_profitable): New.
15875 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
15877 Make GC vectors heap vectors. Clean up comments.
15878 Make visited_bbs a reference.
15879 (find_jump_threads_backwards): Make visited_bbs live in the stack.
15880 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
15883 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
15886 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
15887 words of E_DImode object are reachable by xtensa_uimm8x4 access.
15889 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
15891 Revert r251800 and r251799.
15893 2017-09-11 Martin Jambor <mjambor@suse.cz>
15896 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
15897 arguments in advance.
15898 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
15899 use it to find predecessor edges.
15900 (naive_outof_ssa): Collect vector of predecessors.
15902 2017-09-08 Jason Merrill <jason@redhat.com>
15904 PR c++/70029 - ICE with ref-qualifier and -flto
15905 * langhooks.h (struct lang_hooks_for_types): Add
15906 copy_lang_qualifiers.
15907 * attribs.c (build_type_attribute_qual_variant): Use it.
15908 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
15910 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
15911 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
15913 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
15916 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
15917 (*mulsi3_sp64): New instruction.
15918 (mulsi3): New expander.
15920 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
15922 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
15924 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15926 * sancov.c: Include memmodel.h.
15928 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
15931 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
15934 2017-09-07 Carl Love <cel@us.ibm.com>
15936 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
15937 the sldi instruction.
15939 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
15941 * sancov.c: Include tm_p.h.
15943 2017-09-07 Jakub Jelinek <jakub@redhat.com>
15946 * output.h (switch_to_other_text_partition): New declaration.
15947 * varasm.c (switch_to_other_text_partition): New function.
15948 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
15949 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
15950 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
15951 to the other text partition before emitting LCL label and switch back
15952 after emitting the word after it.
15954 2017-09-07 Richard Biener <rguenther@suse.de>
15956 * passes.def (pass_split_crit_edges): Remove instance before PRE.
15957 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
15958 critical edges here, after loop init.
15959 (pass_data_pre): Remove PROP_no_crit_edges flags.
15960 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
15961 for valueization of call args to avoid leaking VN_TOP.
15962 (visit_use): Assert we do not visit default defs.
15963 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
15964 Use error_mark_node to more easily detect leaking VN_TOP.
15965 All default-defs are varying, not VN_TOP. Mark them visited.
15966 (run_scc_vn): Make code match comment.
15968 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15970 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
15971 OPTION_MASK_FLOAT128_KEYWORD.
15972 (POWERPC_MASKS): Likewise.
15973 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
15974 support for the -mfloat128-type option, and make -mfloat128
15975 default on PowerPC Linux systems. Define or undefine
15976 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
15977 Define __float128 to be __ieee128 if IEEE 128-bit support is
15978 enabled, or undefine it.
15979 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
15980 Delete defining __FLOAT128_TYPE__.
15981 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
15982 -mfloat128-type option and make -mfloat128 default on PowerPC
15984 (TARGET_FLOAT128_TYPE): Likewise.
15985 (-mfloat128-type): Likewise.
15986 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15987 Delete the -mfloat128-type option and make -mfloat128 default on
15988 PowerPC Linux systems. Always use __ieee128 to be the keyword for
15989 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
15990 128-bit floating point is enabled. Change tests from using
15991 -mfloat128-type to -mfloat128.
15992 (rs6000_mangle_type): Use the correct mangling for the __float128
15993 type even if normal long double is restricted to 64-bits.
15994 (floatn_mode): Enable the _Float128 type by default on VSX Linux
15996 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
15997 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
15998 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16000 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16001 documentation for -mfloat128.
16003 2017-09-06 Olivier Hainque <hainque@adacore.com>
16005 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16007 2017-09-06 Wish Wu <wishwu007@gmail.com>
16008 Jakub Jelinek <jakub@redhat.com>
16010 * asan.c (initialize_sanitizer_builtins): Add
16011 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16012 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16013 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16014 BT_FN_VOID_UINT64_PTR variables.
16015 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16016 (BT_FN_VOID_UINT16_UINT16): Likewise.
16017 (BT_FN_VOID_UINT32_UINT32): Likewise.
16018 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16019 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16020 (BT_FN_VOID_UINT64_PTR): Likewise.
16021 * common.opt (flag_sanitize_coverage): New variable.
16022 (fsanitize-coverage=trace-pc): Remove.
16023 (fsanitize-coverage=): Add.
16024 * flag-types.h (enum sanitize_coverage_code): New enum.
16025 * fold-const.c (fold_range_test): Disable non-short-circuit
16026 optimization if flag_sanitize_coverage.
16027 (fold_truth_andor): Likewise.
16028 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16029 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16030 (coverage_sanitizer_opts): New array.
16031 (get_closest_sanitizer_option): Add OPTS argument, handle also
16032 OPT_fsanitize_coverage_.
16033 (parse_sanitizer_options): Adjusted to also handle
16034 OPT_fsanitize_coverage_.
16035 (common_handle_option): Add OPT_fsanitize_coverage_.
16036 * sancov.c (instrument_comparison, instrument_switch): New function.
16037 (sancov_pass): Add trace-cmp support.
16038 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16039 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16040 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16041 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16042 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16043 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16044 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16045 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16046 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16047 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16049 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16051 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16052 error. Only quit immediately if parsing is complete.
16053 (BEGIN): Initialize fatal_err and parse_done.
16054 (begin fpu, end fpu): Check number of arguments.
16055 (begin arch, end arch): Likewise.
16056 (begin cpu, end cpu): Likewise.
16057 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
16058 (optalias): Likewise.
16060 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16062 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
16063 * config/arm/arm-isa.h: Delete. Move definitions to ...
16064 * arm-cpus.in: ... here. Use new feature and fgroup values.
16065 * config/arm/arm.c (arm_option_override): Use lower case for feature
16067 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
16068 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
16069 * config/arm/parsecpu.awk (END): Add new command 'isa'.
16071 (print_isa_bits_for): New function.
16072 (gen_isa): New function.
16073 (gen_comm_data): Use print_isa_bits_for.
16074 (define feature): New keyword.
16075 (define fgroup): New keyword.
16076 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
16077 (arm-isa.h): Add rule to generate file.
16078 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
16079 case for feature bit names.
16081 2017-09-06 Richard Biener <rguenther@suse.de>
16083 * tree-ssa-pre.c (NECESSARY): Remove.
16084 (create_expression_by_pieces): Do not touch pass-local flags.
16085 (insert_into_preds_of_block): Likewise.
16086 (do_pre_regular_insertion): Likewise.
16087 (eliminate_insert): Likewise.
16088 (eliminate_dom_walker::before_dom_children): Likewise.
16089 (fini_eliminate): Do not look at inserted_exprs.
16090 (mark_operand_necessary): Remove.
16091 (remove_dead_inserted_code): Replace with simple work-list
16092 algorithm based on inserted_exprs and SSA uses.
16093 (pass_pre::execute): Re-order fini_eliminate and
16094 remove_dead_inserted_code.
16096 2017-09-06 Olivier Hainque <hainque@adacore.com>
16098 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
16099 for VxWorks 7. Adjust surrounding comments.
16101 2017-09-06 Richard Biener <rguenther@suse.de>
16103 * gimple-ssa-strength-reduction.c
16104 (find_candidates_dom_walker::before_dom_children): Also allow
16107 2017-09-06 Richard Biener <rguenther@suse.de>
16109 PR tree-optimization/82108
16110 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
16111 for gap in the non-permutation SLP case.
16113 2017-09-06 Martin Jambor <mjambor@suse.cz>
16115 PR tree-optimization/82078
16116 * tree-sra.c (sort_and_splice_var_accesses): Move call to
16117 add_access_to_work_queue...
16118 (build_accesses_from_assign): ...here.
16119 (propagate_all_subaccesses): Make sure racc is the group
16120 representative, if there is one.
16122 2017-09-06 Jakub Jelinek <jakub@redhat.com>
16124 PR middle-end/82095
16125 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
16128 2017-09-06 Richard Biener <rguenther@suse.de>
16130 * gimple-ssa-strength-reduction.c
16131 (find_candidates_dom_walker::before_doom_children): Use a
16132 type and not a mode check.
16134 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
16137 * config/arm/predicates.md (arm_general_adddi_operand): Create new
16139 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
16141 2017-09-05 Jeff Law <law@redhat.com>
16143 PR tree-optimization/64910
16144 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
16145 cases where we have 3 or more operands.
16147 2017-09-05 Jakub Jelinek <jakub@redhat.com>
16149 PR middle-end/81768
16150 * omp-low.c (lower_omp_for): Recompute tree invariant if
16151 gimple_omp_for_initial/final is ADDR_EXPR.
16153 PR middle-end/81768
16154 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
16155 into gimple val before gimplification fo the COND_EXPR.
16157 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
16159 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
16160 REGION_COPY argument.
16161 (thread_through_all_blocks): Remove unused argument to
16162 duplicate_thread_path.
16164 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16165 Alan Hayward <alan.hayward@arm.com>
16166 David Sherwood <david.sherwood@arm.com>
16168 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
16169 Take a scalar_mode rather than a machine_mode.
16170 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16171 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
16172 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16173 (aarch64_gen_adjusted_ldpstp): Likewise.
16174 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
16176 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16177 Alan Hayward <alan.hayward@arm.com>
16178 David Sherwood <david.sherwood@arm.com>
16180 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
16181 Take a scalar_int_mode instead of a machine_mode.
16182 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16183 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16184 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16185 (aarch64_simd_attr_length_rglist): Delete.
16186 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
16187 a scalar_int_mode instead of a machine_mode.
16188 (aarch64_add_offset): Likewise.
16189 (aarch64_internal_mov_immediate): Likewise
16190 (aarch64_add_constant_internal): Likewise.
16191 (aarch64_add_constant): Likewise.
16192 (aarch64_movw_imm): Likewise.
16193 (aarch64_rtx_arith_op_extract_p): Likewise.
16194 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16195 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16196 Remove assert that the mode isn't a vector.
16197 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16198 (aarch64_expand_mov_immediate): Update calls after above changes.
16199 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
16200 (aarch64_and_bitmask_imm): Check for scalar integer modes.
16201 (aarch64_move_imm): Likewise.
16202 (aarch64_can_const_movi_rtx_p): Likewise.
16203 (aarch64_strip_extend): Likewise.
16204 (aarch64_extr_rtx_p): Likewise.
16205 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
16206 a CONST_INT when the mode parameter is VOIDmode.
16207 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
16209 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16211 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
16212 * stor-layout.c (bitwise_mode_for_mode): Likewise.
16213 (bitwise_type_for_mode): Update accordingly.
16215 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16217 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
16218 * stor-layout.c (mode_for_size_tree): Likewise.
16219 (mode_for_array): Update accordingly.
16220 (layout_decl): Likewise.
16221 (compute_record_mode): Likewise. Only set the mode once.
16223 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16225 * target.def (get_mask_mode): Change return type to opt_mode.
16227 * doc/tm.texi: Regenerate.
16228 * targhooks.h (default_get_mask_mode): Return an opt_mode.
16229 * targhooks.c (default_get_mask_mode): Likewise.
16230 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16231 * optabs-query.c (can_vec_mask_load_store_p): Update use of
16232 targetm.get_mask_mode.
16233 * tree.c (build_truth_vector_type): Likewise.
16235 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16237 * machmode.h (mode_for_vector): Return an opt_mode.
16238 * stor-layout.c (mode_for_vector): Likewise.
16239 (mode_for_int_vector): Update accordingly.
16240 (layout_type): Likewise.
16241 * config/i386/i386.c (emit_memmov): Likewise.
16242 (ix86_expand_set_or_movmem): Likewise.
16243 (ix86_expand_vector_init): Likewise.
16244 (ix86_get_mask_mode): Likewise.
16245 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
16247 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
16248 * expmed.c (extract_bit_field_1): Likewise.
16249 * expr.c (expand_expr_real_2): Likewise.
16250 * optabs-query.c (can_vec_perm_p): Likewise.
16251 (can_vec_mask_load_store_p): Likewise.
16252 * optabs.c (expand_vec_perm): Likewise.
16253 * targhooks.c (default_get_mask_mode): Likewise.
16254 * tree-vect-stmts.c (vectorizable_store): Likewise.
16255 (vectorizable_load): Likewise.
16256 (get_vectype_for_scalar_type_and_size): Likewise.
16258 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16260 * machmode.h (mode_for_int_vector): New function.
16261 * stor-layout.c (mode_for_int_vector): Likewise.
16262 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
16263 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
16264 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16265 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16266 (s390_expand_vcond): Likewise.
16268 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16270 * machmode.h (opt_machine_mode): New type.
16271 (opt_mode<T>): Allow construction from anything that can be
16273 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
16274 (mode_for_size): Return an opt_machine_mode.
16275 * stor-layout.c (mode_for_size): Likewise.
16276 (mode_for_size_tree): Update call accordingly.
16277 (bitwise_mode_for_mode): Likewise.
16278 (make_fract_type): Likewise.
16279 (make_accum_type): Likewise.
16280 * caller-save.c (replace_reg_with_saved_mem): Update call
16282 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16283 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16284 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16285 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16286 * expmed.c (extract_bit_field_1): Likewise.
16287 * reload.c (get_secondary_mem): Likewise.
16288 * varasm.c (assemble_integer): Likewise.
16289 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
16292 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16294 * machmode.h (decimal_float_mode_for_size): New function.
16295 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
16296 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
16297 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
16298 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
16299 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
16300 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
16302 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16304 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
16305 (get_builtin_sync_mode): Likewise.
16306 (expand_ifn_atomic_compare_exchange): Likewise.
16307 (expand_builtin_atomic_clear): Likewise.
16308 (expand_builtin_atomic_test_and_set): Likewise.
16309 (fold_builtin_atomic_always_lock_free): Likewise.
16310 * calls.c (compute_argument_addresses): Likewise.
16311 (emit_library_call_value_1): Likewise.
16312 (store_one_arg): Likewise.
16313 * combine.c (combine_instructions): Likewise.
16314 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
16315 * config/arm/arm.c (arm_function_value): Likewise.
16316 (aapcs_allocate_return_reg): Likewise.
16317 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
16318 * config/i386/i386.c (construct_container): Likewise.
16319 (ix86_gimplify_va_arg): Likewise.
16320 (ix86_expand_sse_cmp): Likewise.
16321 (emit_memmov): Likewise.
16322 (emit_memset): Likewise.
16323 (expand_small_movmem_or_setmem): Likewise.
16324 (ix86_expand_pextr): Likewise.
16325 (ix86_expand_pinsr): Likewise.
16326 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
16327 * config/microblaze/microblaze.c (microblaze_block_move_straight):
16329 * config/mips/mips.c (mips_function_value_1) Likewise.
16330 (mips_block_move_straight): Likewise.
16331 (mips_expand_ins_as_unaligned_store): Likewise.
16332 * config/powerpcspe/powerpcspe.c
16333 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16334 (rs6000_darwin64_record_arg_flush): Likewise.
16335 * config/rs6000/rs6000.c
16336 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16337 (rs6000_darwin64_record_arg_flush): Likewise.
16338 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
16339 (sparc_function_value_1): Likewise.
16340 * config/spu/spu.c (adjust_operand): Likewise.
16341 (spu_emit_branch_or_set): Likewise.
16342 (arith_immediate_p): Likewise.
16343 * emit-rtl.c (gen_lowpart_common): Likewise.
16344 * expr.c (expand_expr_real_1): Likewise.
16345 * function.c (assign_parm_setup_block): Likewise.
16346 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
16347 * reload1.c (alter_reg): Likewise.
16348 * stor-layout.c (mode_for_vector): Likewise.
16349 (layout_type): Likewise.
16351 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16353 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
16354 (spu_convert_move): Likewise.
16355 * lower-subreg.c (resolve_simple_move): Likewise.
16357 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16360 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
16361 define_insn to a define_expand.
16362 (altivec_vsum2sws_direct): New define_insn.
16363 (altivec_vsumsws): Convert from a define_insn to a define_expand.
16365 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16367 * config/arm/arm.c (arm_option_params_internal): Improve setting of
16370 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
16375 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
16376 realign stack if stack alignment needed is less than incoming
16379 2017-09-05 Marek Polacek <polacek@redhat.com>
16382 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
16385 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16387 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
16389 2017-09-05 Richard Biener <rguenther@suse.de>
16391 PR tree-optimization/82084
16392 * fold-const.c (can_native_encode_string_p): Handle wide characters.
16394 2017-09-05 Richard Biener <rguenther@suse.de>
16396 PR tree-optimization/82102
16397 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
16399 2017-09-05 Martin Liska <mliska@suse.cz>
16401 PR tree-optimization/82032
16402 * tree-cfg.c (generate_range_test): New function.
16403 * tree-cfg.h (generate_range_test): Declared here.
16404 * tree-cfgcleanup.c (convert_single_case_switch): New function.
16405 (cleanup_control_expr_graph): Use it.
16406 * tree-switch-conversion.c (try_switch_expansion): Remove
16408 (emit_case_nodes): Use generate_range_test.
16410 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16413 * config/i386/i386.md (*<btsc><mode>_mask): Add
16414 TARGET_USE_BT to insn constraint.
16415 (*btr<mode>_mask): Ditto.
16417 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
16419 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
16420 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
16422 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
16425 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
16426 TARGET_NEON and TARGET_IWMMXT.
16427 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
16428 TARGET_NEON and TARGET_IWMMXT.
16429 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
16431 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16433 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
16434 (ix86_rewrite_tls_address): Ditto.
16435 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
16436 (ix86_rewrite_tls_address_1): Ditto.
16437 (ix86_rewrite_tls_address): Ditto.
16438 * config/i386/predicates.md (tls_address_pattern): New predicate.
16439 * config/i386/i386.md (TLS address splitter): New splitter.
16441 2017-09-04 Richard Biener <rguenther@suse.de>
16443 PR tree-optimization/82084
16444 * fold-const.h (can_native_encode_string_p): Declare.
16445 * fold-const.c (can_native_encode_string_p): Factor out from ...
16446 (native_encode_string): ... here.
16447 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
16448 vectorizing stores from constants we later cannot handle.
16450 2017-09-04 Marek Polacek <polacek@redhat.com>
16453 * doc/invoke.texi: Update -Wtautological-compare documentation.
16455 2017-09-04 Jeff Law <law@redhat.com>
16457 PR tree-optimization/64910
16458 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
16459 swap the first and last operand if the last is a constant.
16461 2017-09-04 Marek Polacek <polacek@redhat.com>
16464 * convert.c (do_narrow): When sanitizing signed integer overflows,
16465 bail out for signed types.
16466 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
16468 2017-09-04 Richard Biener <rguenther@suse.de>
16470 PR tree-optimization/82060
16471 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16472 Move devirtualization after stmt folding and before EH/AB/noreturn
16473 cleanup to get the stmt refs canonicalized. Use a bool instead
16474 of gimple_modified_p since that doesn't work for NOPs. Schedule
16475 NOPs generated by folding for removal.
16477 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16478 Alan Hayward <alan.hayward@arm.com>
16479 David Sherwood <david.sherwood@arm.com>
16481 * coretypes.h (pad_direction): New enum.
16482 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
16483 (FUNCTION_ARG_PADDING): Likewise.
16484 * target.def (function_arg_padding): New hook.
16485 * targhooks.h (default_function_arg_padding): Declare.
16486 * targhooks.c (default_function_arg_padding): New function.
16487 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
16488 (TARGET_FUNCTION_ARG_PADDING): ...this.
16489 * doc/tm.texi: Regenerate.
16490 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16491 instead of direction.
16492 (compute_argument_addresses): Likewise.
16493 (load_register_parameters): Likewise.
16494 (emit_library_call_value_1): Likewise.
16495 (store_one_arg): Use targetm.calls.function_arg_padding instead
16496 of FUNCTION_ARG_PADDING.
16497 (must_pass_in_stack_var_size_or_pad): Likewise.
16498 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
16499 (emit_group_store): Likewise.
16500 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
16501 instead of FUNCTION_ARG_PADDING.
16502 (emit_push_insn): Likewise, and propagate enum change throughout
16504 * function.h (direction): Delete.
16505 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
16507 * function.c (assign_parm_find_stack_rtl): Likewise.
16508 (assign_parm_setup_block_p): Likewise.
16509 (assign_parm_setup_block): Likewise.
16510 (gimplify_parameters): Likewise.
16511 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
16512 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
16514 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
16515 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16516 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
16517 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
16518 (aarch64_function_arg_padding): ...this new function.
16519 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
16520 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16521 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
16522 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16523 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
16524 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16525 (arm_pad_arg_upward): Replace with...
16526 (arm_function_arg_padding): ...this new function.
16527 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
16529 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
16530 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
16531 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16532 (ia64_hpux_function_arg_padding): Replace with...
16533 (ia64_function_arg_padding): ...this new function. Use pad_direction
16534 instead of direction. Check for TARGET_HPUX.
16535 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
16536 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16537 (iq2000_function_arg_padding): New function.
16538 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
16539 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
16540 (mips_function_arg_padding): ...this new function.
16541 (mips_pad_reg_upward): Update accordingly.
16542 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16543 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
16544 targetm.calls.function_arg_padding.
16545 (FUNCTION_ARG_PADDING): Delete.
16546 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16547 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
16548 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16549 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
16550 (nios2_block_reg_padding): Return pad_direction instead of direction.
16551 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
16552 instead of direction.
16553 (nios2_function_arg_padding): Likewise. Make static.
16554 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16555 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
16556 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
16557 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
16558 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16559 (pa_function_arg_padding): Make static. Return pad_direction instead
16561 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
16562 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16563 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
16564 instead of direction. Use targetm.calls.function_arg_padding.
16565 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16566 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16567 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16568 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16569 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16571 (function_arg_padding): Rename to...
16572 (rs6000_function_arg_padding): ...this. Make static. Return
16573 pad_direction instead of direction.
16574 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16575 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16576 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16577 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16578 instead of direction. Use targetm.calls.function_arg_padding.
16579 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16580 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16581 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16582 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16583 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16584 (function_arg_padding): Rename to...
16585 (rs6000_function_arg_padding): ...this. Make static. Return
16586 pad_direction instead of direction.
16587 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16588 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16589 * config/s390/s390.c (s390_function_arg_padding): New function.
16590 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16591 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16592 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16593 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16594 (function_arg_padding): Rename to...
16595 (sparc_function_arg_padding): ...this. Make static. Return
16596 pad_direction instead of direction.
16597 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16598 * config/spu/spu.c (spu_function_arg_padding): New function.
16599 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16600 * system.h (FUNCTION_ARG_PADDING): Poison.
16602 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16603 Alan Hayward <alan.hayward@arm.com>
16604 David Sherwood <david.sherwood@arm.com>
16606 * target.def (modes_tieable_p): New hook.
16607 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16608 (TARGET_MODES_TIEABLE_P): ...this.
16609 * doc/tm.texi.in: Regenerate.
16610 * hooks.h (hook_bool_mode_mode_true): Declare.
16611 * hooks.c (hook_bool_mode_mode_true): New function.
16612 * combine.c (subst): Use targetm.modes_tieable_p instead of
16614 * dse.c (find_shift_sequence): Likewise.
16615 * expmed.c (extract_low_bits): Likewise.
16616 * lower-subreg.c: Include target.h.
16617 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16619 * rtlanal.c (rtx_cost): Likewise.
16620 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16621 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16622 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16623 (TARGET_MODES_TIEABLE_P): Redefine.
16624 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16625 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16626 (TARGET_MODES_TIEABLE_P): Redefine.
16627 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16628 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16629 (arc_modes_tieable_p): New function.
16630 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16631 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16632 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16633 (arm_modes_tieable_p): Make static.
16634 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16635 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16636 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16637 (TARGET_MODES_TIEABLE_P): Redefine.
16638 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16639 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16640 (TARGET_MODES_TIEABLE_P): Redefine.
16641 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16642 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16643 (cr16_modes_tieable_p): New function.
16644 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16645 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16646 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16647 (TRULY_NOOP_TRUNCATION): Update comment.
16648 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16649 (TRULY_NOOP_TRUNCATION): Update comment.
16650 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16651 (frv_modes_tieable_p): New function.
16652 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16653 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16654 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16655 (TARGET_MODES_TIEABLE_P): Redefine.
16656 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16657 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16658 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16659 (TARGET_MODES_TIEABLE_P): Redefine.
16660 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16661 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16662 (ia64_modes_tieable_p): New function.
16663 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16664 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16665 (iq2000_modes_tieable_p): New function.
16666 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16667 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16668 (lm32_modes_tieable_p): New function.
16669 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16670 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16671 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16672 (TARGET_MODES_TIEABLE_P): Redefine.
16673 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16674 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16675 (m32r_modes_tieable_p): New function.
16676 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16677 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16678 (m68k_modes_tieable_p): New function.
16679 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16680 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16681 (mcore_modes_tieable_p): New function.
16682 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16683 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16685 (TARGET_MODES_TIEABLE_P): Redefine.
16686 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16687 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16688 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16689 (TARGET_MODES_TIEABLE_P): Redefine.
16690 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16691 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16692 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16693 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16694 (mn10300_modes_tieable_p): ...this and make static.
16695 (TARGET_MODES_TIEABLE_P): Redefine.
16696 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16697 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16698 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16699 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16700 (msp430_modes_tieable_p): Make static.
16701 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16702 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16703 (TARGET_MODES_TIEABLE_P): Redefine.
16704 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16705 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16706 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16707 (TARGET_MODES_TIEABLE_P): Redefine.
16708 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16709 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16710 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16711 (TARGET_MODES_TIEABLE_P): Redefine.
16712 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16713 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16714 (pdp11_modes_tieable_p): New function.
16715 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16716 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16717 (rs6000_modes_tieable_p): New function.
16718 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16719 * config/powerpcspe/powerpcspe.md: Update comment.
16720 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16721 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16722 (TARGET_MODES_TIEABLE_P): Redefine.
16723 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16724 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16725 (rl78_modes_tieable_p): New function.
16726 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16727 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16728 (rs6000_modes_tieable_p): New function.
16729 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16730 * config/rs6000/rs6000.md: Update comment.
16731 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16732 * config/rx/rx.c (rx_modes_tieable_p): New function.
16733 (TARGET_MODES_TIEABLE_P): Redefine.
16734 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16735 * config/s390/s390.c (s390_modes_tieable_p): New function.
16736 (TARGET_MODES_TIEABLE_P): Redefine.
16737 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16738 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16739 (sh_modes_tieable_p): New function.
16740 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16741 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16742 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16743 (sparc_modes_tieable_p): Make static.
16744 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16745 * config/spu/spu.c (spu_modes_tieable_p): New function.
16746 (TARGET_MODES_TIEABLE_P): Redefine.
16747 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16748 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16749 (TARGET_MODES_TIEABLE_P): Redefine.
16750 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16751 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16752 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16753 * config/v850/v850.c (v850_modes_tieable_p): New function.
16754 (TARGET_MODES_TIEABLE_P): Redefine.
16755 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16756 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16757 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16758 (visium_modes_tieable_p): New function.
16759 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16760 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16761 (xtensa_modes_tieable_p): New function.
16762 * system.h (MODES_TIEABLE_P): Poison.
16764 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16765 Alan Hayward <alan.hayward@arm.com>
16766 David Sherwood <david.sherwood@arm.com>
16768 * target.def (hard_regno_mode_ok): New hook.
16769 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16770 (TARGET_HARD_REGNO_MODE_OK): ...this.
16771 * doc/tm.texi.in: Regenerate.
16772 * hooks.h (hook_bool_uint_mode_true): Declare.
16773 * hooks.c (hook_bool_uint_mode_true): New function.
16774 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16775 HARD_REGNO_MODE_OK.
16776 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16777 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16778 instead of HARD_REGNO_MODE_OK.
16779 * caller-save.c: Include target.h.
16780 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16781 HARD_REGNO_MODE_OK.
16782 * combine.c (can_combine_p): Likewise.
16783 (combinable_i3pat): Likewise.
16784 (can_change_dest_mode): Likewise.
16785 * expr.c (init_expr_target): Likewise.
16786 (convert_move): Likewise.
16787 (convert_modes): Likewise.
16788 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16789 (setup_prohibited_mode_move_regs): Likewise.
16790 * ira.h (target_ira): Likewise.
16791 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16792 * lra-constraints.c (process_alt_operands): Likewise.
16793 (split_reg): Likewise.
16794 * recog.c (peep2_find_free_register): Likewise.
16795 * ree.c (combine_reaching_defs): Likewise.
16796 * regcprop.c (maybe_mode_change): Likewise.
16797 * reginfo.c (init_reg_sets_1): Likewise.
16798 (choose_hard_reg_mode): Likewise.
16799 (simplifiable_subregs): Likewise.
16800 * regrename.c (check_new_reg_p): Likewise.
16801 * reload.c (find_valid_class): Likewise.
16802 (find_valid_class_1): Likewise.
16803 (reload_inner_reg_of_subreg): Likewise.
16804 (push_reload): Likewise.
16805 (combine_reloads): Likewise.
16806 (find_dummy_reload): Likewise.
16807 (find_reloads): Likewise.
16808 * reload1.c (find_reg): Likewise.
16809 (set_reload_reg): Likewise.
16810 (allocate_reload_reg): Likewise.
16811 (choose_reload_regs): Likewise.
16812 (reload_adjust_reg_for_temp): Likewise.
16813 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
16814 (simplify_subreg_regno): Likewise.
16815 * sel-sched.c (init_regs_for_mode): Likewise.
16816 * varasm.c (make_decl_rtl): Likewise.
16817 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
16818 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
16819 HARD_REGNO_MODE_OK.
16820 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
16821 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
16822 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16823 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
16824 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
16825 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16826 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
16827 (arc_mode_class): Delete.
16828 (HARD_REGNO_MODE_OK): Delete.
16829 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16830 (arc_hard_regno_mode_ok): Rename old array to...
16831 (arc_hard_regno_mode_ok_modes): ...this.
16832 (arc_conditional_register_usage): Update accordingly.
16833 (arc_mode_class): Make static.
16834 (arc_hard_regno_mode_ok): New function.
16835 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
16836 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
16837 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16838 (arm_hard_regno_mode_ok): Make static.
16839 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
16840 HARD_REGNO_MODE_OK.
16841 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
16842 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
16843 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
16845 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16846 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
16847 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
16848 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
16849 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
16850 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16851 * config/bfin/predicates.md (valid_reg_operand): Use
16852 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
16853 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
16854 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
16855 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16856 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
16857 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
16858 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16859 (cr16_hard_regno_mode_ok): Make static and return a bool.
16860 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
16861 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16862 (cris_hard_regno_mode_ok): New function.
16863 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
16864 (epiphany_mode_class): Delete.
16865 (HARD_REGNO_MODE_OK): Delete.
16866 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
16867 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16868 (hard_regno_mode_ok): Rename to...
16869 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
16870 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
16871 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
16872 HARD_REGNO_MODE_OK.
16873 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
16874 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
16875 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16876 (frv_hard_regno_mode_ok): Make static and return a bool.
16877 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
16878 HARD_REGNO_MODE_OK.
16879 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
16880 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
16881 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
16882 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
16884 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16885 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
16886 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
16887 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
16889 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16890 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
16891 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16892 (ia64_hard_regno_mode_ok): New function.
16893 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
16894 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16895 (iq2000_hard_regno_mode_ok): New function.
16896 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
16897 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16898 (lm32_hard_regno_mode_ok): New function.
16899 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
16900 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
16901 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
16902 instead of HARD_REGNO_MODE_OK.
16903 (m32c_hard_regno_ok): Rename to...
16904 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
16905 (m32c_cannot_change_mode_class): Update accordingly.
16906 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16907 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
16908 (m32r_mode_class): Delete.
16909 (HARD_REGNO_MODE_OK): Delete.
16910 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16911 (m32r_hard_regno_mode_ok): Rename to...
16912 (m32r_hard_regno_modes): ...this.
16913 (m32r_mode_class): Make static.
16914 (m32r_hard_regno_mode_ok): New function.
16915 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
16916 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
16917 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16918 (m68k_hard_regno_mode_ok): Make static.
16919 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
16920 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16921 (mcore_hard_regno_mode_ok): New function.
16922 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
16923 (HARD_REGNO_MODE_OK): Delete.
16924 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
16926 (microblaze_hard_regno_mode_ok_p): ...this and make static.
16927 (microblaze_hard_regno_mode_ok): New function.
16928 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16929 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
16930 (mips_hard_regno_mode_ok): Delete.
16931 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
16932 (mips_hard_regno_mode_ok_p): ...this and make static.
16933 (mips_hard_regno_mode_ok_p): Rename to...
16934 (mips_hard_regno_mode_ok_uncached): ...this.
16935 (mips_hard_regno_mode_ok): New function.
16936 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
16937 of HARD_REGNO_MODE_OK.
16938 (mips_option_override): Update after above name changes.
16939 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16940 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
16941 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
16942 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
16943 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
16944 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16945 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
16946 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
16947 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
16948 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16949 (msp430_hard_regno_mode_ok): Make static and return a bool.
16950 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
16951 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
16952 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
16954 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16955 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
16956 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
16957 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
16958 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
16959 (PA_HARD_REGNO_MODE_OK): ...this
16960 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
16961 (PA_HARD_REGNO_MODE_OK): ...this.
16962 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16963 (pa_hard_regno_mode_ok): New function.
16964 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
16965 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16966 (pdp11_hard_regno_mode_ok): New function.
16967 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
16968 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
16970 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
16972 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16973 (rs6000_hard_regno_mode_ok): Rename to...
16974 (rs6000_hard_regno_mode_ok_uncached): ...this.
16975 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16976 (rs6000_hard_regno_mode_ok): New function.
16977 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
16978 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
16979 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
16980 (riscv_hard_regno_mode_ok): ...this and make static.
16981 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16982 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
16983 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
16984 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16985 (rl78_hard_regno_mode_ok): Make static and return bool.
16986 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
16987 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
16989 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
16990 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16991 (rs6000_hard_regno_mode_ok): Rename to...
16992 (rs6000_hard_regno_mode_ok_uncached): ...this.
16993 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16994 (rs6000_hard_regno_mode_ok): New function.
16995 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
16996 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
16997 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16998 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
16999 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17000 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17001 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17002 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17003 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17004 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17005 (sh_hard_regno_mode_ok): Make static.
17006 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17007 instead of HARD_REGNO_MODE_OK.
17008 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17009 (sparc_mode_class): Delete.
17010 (HARD_REGNO_MODE_OK): Delete.
17011 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17012 (hard_regno_mode_classes): Make static.
17013 (sparc_mode_class): Likewise.
17014 (sparc_hard_regno_mode_ok): New function.
17015 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17016 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17017 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17019 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17020 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17021 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17022 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17023 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17024 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17025 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17026 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17027 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17028 (visium_hard_regno_mode_ok): New function.
17029 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17030 instead of HARD_REGNO_MODE_OK.
17031 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17032 (HARD_REGNO_MODE_OK): Delete.
17033 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17034 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17035 (xtensa_option_override): Update accordingly.
17036 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17037 (xtensa_hard_regno_mode_ok): New function.
17038 * system.h (HARD_REGNO_MODE_OK): Poison.
17040 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17041 Alan Hayward <alan.hayward@arm.com>
17042 David Sherwood <david.sherwood@arm.com>
17044 * target.def (hard_regno_call_part_clobbered): New hook.
17045 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17046 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17047 * doc/tm.texi: Regenerate.
17048 * hooks.h (hook_bool_uint_mode_false): Declare.
17049 * hooks.c (hook_bool_uint_mode_false): New function.
17050 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17051 * cselib.c (cselib_process_insn): Use
17052 targetm.hard_regno_call_part_clobbered instead of
17053 HARD_REGNO_CALL_PART_CLOBBERED.
17054 * ira-conflicts.c (ira_build_conflicts): Likewise.
17055 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17056 * lra-constraints.c (need_for_call_save_p): Likewise.
17057 * lra-lives.c: Include target.h.
17058 (check_pseudos_live_through_calls): Use
17059 targetm.hard_regno_call_part_clobbered instead of
17060 HARD_REGNO_CALL_PART_CLOBBERED.
17061 * regcprop.c: Include target.h.
17062 (copyprop_hardreg_forward_1): Use
17063 targetm.hard_regno_call_part_clobbered instead of
17064 HARD_REGNO_CALL_PART_CLOBBERED.
17065 * reginfo.c (choose_hard_reg_mode): Likewise.
17066 * regrename.c (check_new_reg_p): Likewise.
17067 * reload.c (find_equiv_reg): Likewise.
17068 * reload1.c (emit_reload_insns): Likewise.
17069 * sched-deps.c (deps_analyze_insn): Likewise.
17070 * sel-sched.c (init_regs_for_mode): Likewise.
17071 (mark_unavailable_hard_regs): Likewise.
17072 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
17073 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17074 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
17076 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17077 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17078 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
17080 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
17082 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17083 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17084 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
17086 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17087 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17088 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
17090 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17091 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
17093 * config/powerpcspe/powerpcspe.c
17094 (rs6000_hard_regno_call_part_clobbered): New function.
17095 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17096 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17097 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
17099 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17100 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17101 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
17103 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17104 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17105 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
17107 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17108 Alan Hayward <alan.hayward@arm.com>
17109 David Sherwood <david.sherwood@arm.com>
17111 * rtl.h (subreg_memory_offset): Declare.
17112 * emit-rtl.c (subreg_memory_offset): New function.
17113 * expmed.c (store_bit_field_1): Use it.
17114 * expr.c (undefined_operand_subword_p): Likewise.
17115 * simplify-rtx.c (simplify_subreg): Likewise.
17117 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
17119 PR rtl-optimization/57448
17122 * optabs.c (expand_atomic_load): Place compiler memory barriers if
17123 using atomic_load pattern.
17124 (expand_atomic_store): Likewise.
17126 2017-09-04 Jakub Jelinek <jakub@redhat.com>
17129 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
17130 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
17131 handling. Use replace_call_with_value with NULL instead of
17132 gsi_replace, unlink_stmt_vdef and release_defs.
17134 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
17137 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
17139 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17142 * rtl.h (emit_library_call_value_1): Declare.
17143 (emit_library_call): Replace declaration with a series of overloads.
17144 Remove the parameter count argument.
17145 (emit_library_call_value): Likewise.
17146 * calls.c (emit_library_call_value_1): Make global. Replace varargs
17147 with an "rtx_mode_t *".
17148 (emit_library_call_value): Delete.
17149 (emit_library_call): Likewise.
17150 * asan.c (asan_emit_stack_protection): Update calls accordingly.
17151 (asan_emit_allocas_unpoison): Likewise.
17152 * builtins.c (expand_builtin_powi): Likewise.
17153 (expand_asan_emit_allocas_unpoison): Likewise.
17154 * cfgexpand.c (expand_main_function): Likewise.
17155 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
17156 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
17157 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
17158 * config/arm/arm.c (arm_trampoline_init): Likewise.
17159 (arm_call_tls_get_addr): Likewise.
17160 (arm_expand_divmod_libfunc): Likewise.
17161 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
17162 (smulsi3_highpart): Likewise.
17163 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
17164 (c6x_expand_compare): Likewise.
17165 (c6x_expand_movmem): Likewise.
17166 * config/frv/frv.c (frv_trampoline_init): Likewise.
17167 * config/i386/i386.c (ix86_trampoline_init): Likewise.
17168 (ix86_expand_divmod_libfunc): Likewise.
17169 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
17170 (ia64_expand_compare): Likewise.
17171 (ia64_profile_hook): Likewise.
17172 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
17173 (nonlocal_goto): Likewise.
17174 (restore_stack_nonlocal): Likewise.
17175 * config/m32r/m32r.c (block_move_call): Likewise.
17176 (m32r_trampoline_init): Likewise.
17177 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
17178 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
17179 (m68k_call_m68k_read_tp): Likewise.
17180 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
17181 (microblaze_expand_divide): Likewise.
17182 * config/mips/mips.h (mips_args): Likewise.
17183 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
17184 (MIPS_ICACHE_SYNC): Likewise.
17185 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
17186 (nios2_trampoline_init): Likewise.
17187 * config/pa/pa.c (hppa_tls_call): Likewise.
17188 (pa_trampoline_init): Likewise.
17189 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
17190 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
17191 (expand_strn_compare): Likewise.
17192 (rs6000_generate_compare): Likewise.
17193 (rs6000_expand_float128_convert): Likewise.
17194 (output_profile_hook): Likewise.
17195 (rs6000_trampoline_init): Likewise.
17196 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
17197 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
17198 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
17199 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
17200 (rs6000_generate_compare): Likewise.
17201 (rs6000_expand_float128_convert): Likewise.
17202 (output_profile_hook): Likewise.
17203 (rs6000_trampoline_init): Likewise.
17204 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
17205 * config/sh/sh.c (sh_trampoline_init): Likewise.
17206 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
17207 (sparc_emit_float_lib_cmp): Likewise.
17208 (sparc32_initialize_trampoline): Likewise.
17209 (sparc64_initialize_trampoline): Likewise.
17210 (sparc_profile_hook): Likewise.
17211 * config/spu/spu.c (ea_load_store): Likewise.
17212 * config/spu/spu.md (floatunssidf2): Likewise.
17213 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
17214 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
17215 * config/visium/visium.c (expand_block_move_4): Likewise.
17216 (expand_block_move_2): Likewise.
17217 (expand_block_move_1): Likewise.
17218 (expand_block_set_4): Likewise.
17219 (expand_block_set_2): Likewise.
17220 (expand_block_set_1): Likewise.
17221 (visium_trampoline_init): Likewise.
17222 (visium_profile_hook): Likewise.
17223 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
17224 (xtensa_setup_frame_addresses): Likewise.
17225 (xtensa_trampoline_init): Likewise.
17226 * except.c (sjlj_emit_function_enter): Likewise.
17227 (sjlj_emit_function_exit): Likewise.
17228 * explow.c (allocate_dynamic_stack_space): Likewise.
17229 (probe_stack_range): Likewise.
17230 * expr.c (convert_mode_scalar): Likewise.
17231 * optabs.c (expand_binop): Likewise.
17232 (expand_twoval_binop_libfunc): Likewise.
17233 (expand_unop): Likewise.
17234 (prepare_cmp_insn): Likewise.
17235 (prepare_float_lib_cmp): Likewise.
17236 (expand_float): Likewise.
17237 (expand_fix): Likewise.
17238 (expand_fixed_convert): Likewise.
17239 (maybe_emit_sync_lock_test_and_set): Likewise.
17240 (expand_atomic_compare_and_swap): Likewise.
17241 (expand_mem_thread_fence): Likewise.
17242 (expand_atomic_fetch_op): Likewise.
17244 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
17246 * doc/generic.texi (OpenACC): Adjust URL.
17247 * doc/invoke.texi (C Dialect Options): Ditto.
17249 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
17251 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
17252 predicate for operand 1. Add (m,<S>) constraint.
17253 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
17254 Prevent memory operand 1 with register operand 2.
17256 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
17258 PR rtl-optimization/82024
17259 * combine.c (try_combine): If the combination result is a PARALLEL,
17260 and we only need to retain the SET in there that would be placed
17261 at I2, check that we can place that at I3 instead, before doing so.
17263 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17266 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
17268 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
17269 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
17272 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
17273 Jeff Law <law@redhat.com>
17275 * varasm.c (bss_initializer_p): Do not put constants into .bss
17276 (categorize_decl_for_section): Handle bss_initializer_p returning
17277 false when DECL_INITIAL is NULL.
17279 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17282 * config/s390/s390.c (s390_can_inline_p): New function.
17284 2017-09-01 Jeff Law <law@redhat.com>
17286 PR tree-optimization/82052
17287 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17288 Always initialize the returned slot after a hash table miss
17289 when INSERT is true.
17291 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
17293 * config/s390/s390.md (mem_signal_fence): Remove.
17294 * doc/md.texi (mem_signal_fence): Remove.
17295 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
17297 * target-insns.def (mem_signal_fence): Remove.
17299 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17302 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
17305 * asan.c (create_odr_indicator): Strip name encoding from assembler
17306 name before appending it after __odr_asan_.
17308 2017-09-01 Martin Liska <mliska@suse.cz>
17310 PR tree-optimization/82059
17311 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
17312 frequency only when an edge is redirected.
17314 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17316 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
17317 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
17318 (arc_conditional_register_usage): Remove ARC600 lp_count
17320 (arc_file_start): Emit Tag_ARC_CPU_variation.
17321 (arc_can_use_doloop_p): New conditions to use ZOLs.
17322 (hwloop_fail): New function.
17323 (hwloop_optimize): Likewise.
17324 (hwloop_pattern_reg): Likewise.
17325 (arc_doloop_hooks): New struct, to be used with reorg_loops.
17326 (arc_reorg_loops): New function, calls reorg_loops.
17327 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
17328 (arc600_corereg_hazard): Remove ZOL checking, case handled by
17330 (arc_loop_hazard): Remove function, functionality moved into
17332 (arc_hazard): Remove arc_loop_hazard call.
17333 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
17334 into hwloop_optimize.
17335 (arc_label_align): Remove ZOL handling.
17336 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
17337 * config/arc/arc.md (doloop_begin): Remove pattern.
17338 (doloop_begin_i): Likewise.
17339 (doloop_end_i): Likewise.
17340 (doloop_fallback): Likewise.
17341 (doloop_fallback_m): Likewise.
17342 (doloop_end): Reimplement expand.
17343 (arc_lp): New pattern for LP instruction.
17344 (loop_end): New pattern.
17345 (loop_fail): Likewise.
17346 (decrement_and_branch_until_zero): Likewise.
17347 * config/arc/arc.opt (mlpc-width): New option.
17348 * doc/invoke.texi (mlpc-width): Document option.
17350 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17352 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
17353 (arc_ccfsm_advance): Fix checking for delay slots.
17354 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
17356 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17358 * config/arc/arc.md (movqi_insn): Add stores to save constant long
17360 (movhi_insn): Update store instruction constraint which are saving
17361 6-bit short immediates.
17362 (movsi_insn): Consider also short scaled load operations.
17363 (zero_extendhisi2_i): Use Usd constraint instead of T.
17364 (extendhisi2_i): Add q constraint.
17365 (arc_clzsi2): Add type and length attributes.
17366 (arc_ctzsi2): Likewise.
17367 * config/arc/constraints.md (Usc): Update constraint, the
17368 assembler can parse two relocations for a single instruction.
17370 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17372 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
17373 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
17375 2017-08-31 Olivier Hainque <hainque@adacore.com>
17377 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
17378 match as powerpc-wrs-vxworks*.
17380 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
17382 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
17383 register constraint for by-element operand.
17384 (aarch64_mls_elt_merge<mode>): Likewise.
17386 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17388 * config/arc/arc.c (arc_can_follow_jump): Check for short
17391 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17393 * config.gcc: Use g.opt for arc.
17394 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
17395 functionality moved to ...
17396 (legitimate_scaled_address_p): New function, ...here.
17397 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
17398 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
17399 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
17401 (arc_override_options): Handle G option.
17402 (arc_output_pic_addr_const): Correct function definition.
17403 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
17404 (arc_decl_anon_ns_mem_p): Delete.
17405 (arc_in_small_data_p): Overhaul this function to take into
17406 consideration the value given via G option.
17407 (arc_rewrite_small_data_1): Renamed and corrected old
17408 arc_rewrite_small_data function.
17409 (arc_rewrite_small_data): New function.
17410 (small_data_pattern): Don't use pic_offset_table_rtx.
17411 * config/arc/arc.h (CC1_SPEC): Recognize G option.
17412 * config/arc/simdext.md (movmisalignv2hi): Use
17413 prepare_move_operands function.
17415 (movmisalign*): Likewise.
17416 * doc/invoke.texi (ARC options): Document -G option.
17418 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17420 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
17422 * config/arc/arc.c (arc_print_operand): Output scalled address for
17423 sdata whenever is possible.
17424 (arc_in_small_data_p): Allow sdata for 64bit datum when double
17425 load/stores are available.
17426 (compact_sda_memory_operand): Check for the alignment required by
17427 code density instructions.
17428 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
17430 * config/arc/constraints.md (Usd): Update constraint.
17431 (Us0): New constraint.
17432 (Usc): Update constraint.
17434 2017-08-31 Richard Biener <rguenther@suse.de>
17436 PR middle-end/82054
17437 * dwarf2out.c (dwarf2out_early_global_decl): Process each
17438 function only once.
17440 2017-08-31 Tamar Christina <tamar.christina@arm.com>
17442 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
17443 Resize type_signature.
17445 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17446 Alan Hayward <alan.hayward@arm.com>
17447 David Sherwood <david.sherwood@arm.com>
17449 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
17450 subregs whose inner modes can be stored in GPRs.
17451 (aarch64_classify_index): Likewise.
17453 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17454 Alan Hayward <alan.hayward@arm.com>
17455 David Sherwood <david.sherwood@arm.com>
17457 * config/aarch64/iterators.md (V_cmp_result): Rename to...
17458 (V_INT_EQUIV): ...this.
17459 (v_cmp_result): Rename to...
17460 (v_int_equiv): ...this.
17461 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
17462 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
17463 (copysign<mode>3): Likewise.
17464 (aarch64_simd_bsl<mode>_internal): Likewise.
17465 (aarch64_simd_bsl<mode>): Likewise.
17466 (vec_cmp<mode><mode>): Likewise.
17467 (vcond<mode><mode>): Likewise.
17468 (vcond<v_cmp_mixed><mode>): Likewise.
17469 (vcondu<mode><v_cmp_mixed>): Likewise.
17470 (aarch64_cm<optab><mode>): Likewise.
17471 (aarch64_cmtst<mode>): Likewise.
17472 (aarch64_fac<optab><mode>): Likewise.
17473 (vec_perm_const<mode>): Likewise.
17474 (vcond_mask_<mode><v_cmp_result>): Rename to...
17475 (vcond_mask_<mode><v_int_equiv>): ...this.
17476 (vec_cmp<mode><v_cmp_result>): Rename to...
17477 (vec_cmp<mode><v_int_equiv>): ...this.
17479 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17480 Alan Hayward <alan.hayward@arm.com>
17481 David Sherwood <david.sherwood@arm.com>
17483 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
17485 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
17486 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
17487 (UNSPEC_LD4_DREG): New unspecs.
17488 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
17489 (aarch64_ld2<mode>_dreg_be): Replace with...
17490 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
17492 (aarch64_ld3<mode>_dreg_le)
17493 (aarch64_ld3<mode>_dreg_be): Replace with...
17494 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
17496 (aarch64_ld4<mode>_dreg_le)
17497 (aarch64_ld4<mode>_dreg_be): Replace with...
17498 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
17501 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17503 PR tree-optimization/81987
17504 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
17505 insert an initializer in a location not dominated by the stride
17508 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17510 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
17511 on the entire header of the finally block in the fallthru case.
17513 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17515 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
17517 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
17519 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
17520 rs6000_emit_move_from_cr and call renamed function.
17521 (rs6000_emit_prologue): Call renamed functions.
17522 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
17523 movesi_from_cr, remove volatile CRs.
17525 2017-08-30 Jon Beniston <jon@beniston.com>
17526 Richard Biener <rguenther@suse.de>
17528 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
17529 of VECTOR_MODE_P check.
17530 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17531 element vector types.
17533 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17535 * df.h (df_read_modify_subreg_p): Remove in favor of...
17536 * rtl.h (read_modify_subreg_p): ...this new function. Take a
17537 const_rtx instead of an rtx.
17538 * cprop.c (local_cprop_find_used_regs): Update accordingly.
17539 * df-problems.c (df_word_lr_mark_ref): Likewise.
17540 * ira-lives.c (mark_pseudo_reg_live): Likewise.
17541 (mark_pseudo_reg_dead): Likewise.
17542 (mark_ref_dead): Likewise.
17543 * reginfo.c (init_subregs_of_mode): Likewise.
17544 * sched-deps.c (sched_analyze_1): Likewise.
17545 * df-scan.c (df_def_record_1): Likewise.
17546 (df_uses_record): Likewise.
17547 (df_read_modify_subreg_p): Remove in favor of...
17548 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
17549 const_rtx instead of an rtx.
17551 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17552 Alan Hayward <alan.hayward@arm.com>
17553 David Sherwood <david.sherwood@arm.com>
17555 * rtl.h (partial_subreg_p): New function.
17556 * caller-save.c (save_call_clobbered_regs): Use it.
17557 * calls.c (expand_call): Likewise.
17558 * combine.c (combinable_i3pat): Likewise.
17559 (simplify_set): Likewise.
17560 (make_extraction): Likewise.
17561 (make_compound_operation_int): Likewise.
17562 (gen_lowpart_or_truncate): Likewise.
17563 (force_to_mode): Likewise.
17564 (make_field_assignment): Likewise.
17565 (reg_truncated_to_mode): Likewise.
17566 (record_truncated_value): Likewise.
17567 (move_deaths): Likewise.
17568 * cse.c (record_jump_cond): Likewise.
17569 (cse_insn): Likewise.
17570 * cselib.c (cselib_lookup_1): Likewise.
17571 * expmed.c (extract_bit_field_using_extv): Likewise.
17572 * function.c (assign_parm_setup_reg): Likewise.
17573 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17574 * ira-build.c (create_insn_allocnos): Likewise.
17575 * lra-coalesce.c (merge_pseudos): Likewise.
17576 * lra-constraints.c (match_reload): Likewise.
17577 (simplify_operand_subreg): Likewise.
17578 (curr_insn_transform): Likewise.
17579 * lra-lives.c (process_bb_lives): Likewise.
17580 * lra.c (new_insn_reg): Likewise.
17581 (lra_substitute_pseudo): Likewise.
17582 * regcprop.c (mode_change_ok): Likewise.
17583 (maybe_mode_change): Likewise.
17584 (copyprop_hardreg_forward_1): Likewise.
17585 * reload.c (push_reload): Likewise.
17586 (find_reloads): Likewise.
17587 (find_reloads_subreg_address): Likewise.
17588 * reload1.c (alter_reg): Likewise.
17589 (eliminate_regs_1): Likewise.
17590 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17592 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17594 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17595 back to if statements, including unpack.
17597 2017-08-30 Martin Liska <mliska@suse.cz>
17599 PR inline-asm/82001
17600 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17602 (func_checker::compare_asm_inputs_outputs): ... this function.
17603 (func_checker::compare_gimple_asm): Use the function to compare
17604 also ASM constrains.
17605 * ipa-icf-gimple.h: Rename the function.
17607 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17608 Alan Hayward <alan.hayward@arm.com>
17609 David Sherwood <david.sherwood@arm.com>
17611 * coretypes.h (complex_mode): New type.
17612 * gdbhooks.py (build_pretty_printer): Handle it.
17613 * machmode.h (complex_mode): New class.
17614 (complex_mode::includes_p): New function.
17615 (is_complex_int_mode): Likewise.
17616 (is_complex_float_mode): Likewise.
17617 * genmodes.c (get_mode_class): Handle complex mode classes.
17618 * function.c (expand_function_end): Use is_complex_int_mode.
17620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17621 Alan Hayward <alan.hayward@arm.com>
17622 David Sherwood <david.sherwood@arm.com>
17624 * coretypes.h (scalar_mode_pod): New typedef.
17625 * gdbhooks.py (build_pretty_printer): Handle it.
17626 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17627 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17628 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17629 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17630 as_a <scalar_mode>.
17632 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17633 Alan Hayward <alan.hayward@arm.com>
17634 David Sherwood <david.sherwood@arm.com>
17636 * machmode.h (mode_for_vector): Take a scalar_mode instead
17638 * stor-layout.c (mode_for_vector): Likewise.
17639 * explow.c (promote_mode): Use as_a <scalar_mode>.
17640 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
17642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17643 Alan Hayward <alan.hayward@arm.com>
17644 David Sherwood <david.sherwood@arm.com>
17646 * target.def (preferred_simd_mode): Take a scalar_mode
17647 instead of a machine_mode.
17648 * targhooks.h (default_preferred_simd_mode): Likewise.
17649 * targhooks.c (default_preferred_simd_mode): Likewise.
17650 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17651 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17652 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17653 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17654 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17655 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17656 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17657 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17659 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17660 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17661 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17662 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17663 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17664 * doc/tm.texi: Regenerate.
17665 * optabs-query.c (can_vec_mask_load_store_p): Return false for
17668 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17669 Alan Hayward <alan.hayward@arm.com>
17670 David Sherwood <david.sherwood@arm.com>
17672 * target.def (scalar_mode_supported_p): Take a scalar_mode
17673 instead of a machine_mode.
17674 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17675 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17676 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17677 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17678 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17679 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17680 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17681 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17682 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17683 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17684 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17685 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17686 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17687 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17689 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17690 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17691 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17692 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17693 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17695 * doc/tm.texi: Regenerate.
17697 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17698 Alan Hayward <alan.hayward@arm.com>
17699 David Sherwood <david.sherwood@arm.com>
17701 * coretypes.h (opt_scalar_mode): New typedef.
17702 * gdbhooks.py (build_pretty_printers): Handle it.
17703 * machmode.h (mode_iterator::get_2xwider): Add overload for
17705 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17707 * expr.c (convert_mode_scalar): Likewise.
17708 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17709 * optabs.c (expand_float): Likewise.
17710 (expand_fix): Likewise.
17711 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17713 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17714 Alan Hayward <alan.hayward@arm.com>
17715 David Sherwood <david.sherwood@arm.com>
17717 * optabs.c (expand_float): Explicitly check for scalars before
17718 using a branching expansion.
17719 (expand_fix): Likewise.
17721 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17722 Alan Hayward <alan.hayward@arm.com>
17723 David Sherwood <david.sherwood@arm.com>
17725 * expr.c (convert_mode): Split scalar handling out into...
17726 (convert_mode_scalar): ...this new function. Treat the modes
17729 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17730 Alan Hayward <alan.hayward@arm.com>
17731 David Sherwood <david.sherwood@arm.com>
17733 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17735 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17737 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17738 Alan Hayward <alan.hayward@arm.com>
17739 David Sherwood <david.sherwood@arm.com>
17741 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17742 rather than a machine_mode.
17743 (fixed_from_string): Likewise.
17744 (fixed_convert): Likewise.
17745 (fixed_convert_from_int): Likewise.
17746 (fixed_convert_from_real): Likewise.
17747 (real_convert_from_fixed): Likewise.
17748 * fixed-value.c (fixed_from_double_int): Likewise.
17749 (fixed_from_string): Likewise.
17750 (fixed_convert): Likewise.
17751 (fixed_convert_from_int): Likewise.
17752 (fixed_convert_from_real): Likewise.
17753 (real_convert_from_fixed): Likewise.
17754 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17757 Alan Hayward <alan.hayward@arm.com>
17758 David Sherwood <david.sherwood@arm.com>
17760 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17761 of separate mode class checks. Do not allow vector modes here.
17762 (immed_wide_int_const): Use as_a <scalar_mode>.
17763 * explow.c (trunc_int_for_mode): Likewise.
17764 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17765 (wi::shwi): Likewise.
17766 (wi::min_value): Likewise.
17767 (wi::max_value): Likewise.
17768 * dwarf2out.c (loc_descriptor): Likewise.
17769 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17770 for CONST_WIDE_INT.
17772 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17773 Alan Hayward <alan.hayward@arm.com>
17774 David Sherwood <david.sherwood@arm.com>
17776 * tree.h (SCALAR_TYPE_MODE): New macro.
17777 * expr.c (expand_expr_addr_expr_1): Use it.
17778 (expand_expr_real_2): Likewise.
17779 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17780 (fold_convert_const_fixed_from_int): Likewise.
17781 (fold_convert_const_fixed_from_real): Likewise.
17782 (native_encode_fixed): Likewise
17783 (native_encode_complex): Likewise
17784 (native_encode_vector): Likewise.
17785 (native_interpret_fixed): Likewise.
17786 (native_interpret_real): Likewise.
17787 (native_interpret_complex): Likewise.
17788 (native_interpret_vector): Likewise.
17789 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17790 (simd_clone_adjust_argument_types): Likewise.
17791 (simd_clone_init_simd_arrays): Likewise.
17792 (simd_clone_adjust): Likewise.
17793 * stor-layout.c (layout_type): Likewise.
17794 * tree.c (build_minus_one_cst): Likewise.
17795 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17796 * tree-inline.c (estimate_move_cost): Likewise.
17797 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17798 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17799 (vectorizable_reduction): Likewise.
17800 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17801 (vect_recog_mixed_size_cond_pattern): Likewise.
17802 (check_bool_pattern): Likewise.
17803 (adjust_bool_pattern): Likewise.
17804 (search_type_for_mask_1): Likewise.
17805 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17806 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17807 (vectorizable_load): Likewise.
17808 (vectorizable_store): Likewise.
17809 * ubsan.c (ubsan_encode_value): Likewise.
17810 * varasm.c (output_constant): Likewise.
17812 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17813 Alan Hayward <alan.hayward@arm.com>
17814 David Sherwood <david.sherwood@arm.com>
17816 * coretypes.h (scalar_mode): New class.
17817 * machmode.h (scalar_mode): Likewise.
17818 (scalar_mode::includes_p): New function.
17819 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
17820 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
17821 * genmodes.c (get_mode_class): Handle remaining scalar modes.
17822 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
17823 * expmed.c (store_bit_field_1): Likewise.
17824 (extract_bit_field_1): Likewise.
17825 * expr.c (write_complex_part): Likewise.
17826 (read_complex_part): Likewise.
17827 (emit_move_complex_push): Likewise.
17828 (expand_expr_real_2): Likewise.
17829 * function.c (assign_parm_setup_reg): Likewise.
17830 (assign_parms_unsplit_complex): Likewise.
17831 * optabs.c (expand_binop): Likewise.
17832 * rtlanal.c (subreg_get_info): Likewise.
17833 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17834 * varasm.c (output_constant_pool_2): Likewise.
17836 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17837 Alan Hayward <alan.hayward@arm.com>
17838 David Sherwood <david.sherwood@arm.com>
17840 * expmed.c (extract_high_half): Use scalar_int_mode and remove
17842 (expmed_mult_highpart_optab): Likewise.
17843 (expmed_mult_highpart): Likewise.
17845 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17846 Alan Hayward <alan.hayward@arm.com>
17847 David Sherwood <david.sherwood@arm.com>
17849 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
17850 instead of a machine_mode.
17851 (builtin_memset_read_str): Likewise.
17852 * builtins.c (c_readstr): Likewise.
17853 (builtin_memcpy_read_str): Likewise.
17854 (builtin_strncpy_read_str): Likewise.
17855 (builtin_memset_read_str): Likewise.
17856 (builtin_memset_gen_str): Likewise.
17857 (expand_builtin_signbit): Use scalar_int_mode for local variables.
17858 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
17859 instead of a machine_mode.
17860 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
17862 (make_extraction): Likewise.
17863 (try_widen_shift_mode): Take and return scalar_int_modes instead
17865 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
17866 a scalar_int_mode instead of a machine_mode.
17867 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
17868 (avr_addr_space_pointer_mode): Likewise.
17869 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
17870 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
17871 (msp430_unwind_word_mode): Likewise.
17872 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
17873 (spu_addr_space_pointer_mode): Likewise.
17874 (spu_addr_space_address_mode): Likewise.
17875 (spu_libgcc_cmp_return_mode): Likewise.
17876 (spu_libgcc_shift_count_mode): Likewise.
17877 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
17878 (rl78_addr_space_pointer_mode): Likewise.
17879 (fl78_unwind_word_mode): Likewise.
17880 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
17882 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
17883 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
17884 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
17885 (mips_valid_pointer_mode): Likewise.
17886 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
17887 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
17888 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
17890 (ft32_addr_space_address_mode): Likewise.
17891 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
17892 scalar_int_mode instead of a machine_mode.
17893 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
17895 (m32c_addr_space_address_mode): Likewise.
17896 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
17897 (rs6000_eh_return_filter_mode): Likewise.
17898 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
17899 (rs6000_eh_return_filter_mode): Likewise.
17900 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
17901 (s390_libgcc_shift_count_mode): Likewise.
17902 (s390_unwind_word_mode): Likewise.
17903 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
17905 * target.def (mode_rep_extended): Likewise.
17906 (valid_pointer_mode): Likewise.
17907 (addr_space.valid_pointer_mode): Likewise.
17908 (eh_return_filter_mode): Return a scalar_int_mode rather than
17910 (libgcc_cmp_return_mode): Likewise.
17911 (libgcc_shift_count_mode): Likewise.
17912 (unwind_word_mode): Likewise.
17913 (addr_space.pointer_mode): Likewise.
17914 (addr_space.address_mode): Likewise.
17915 * doc/tm.texi: Regenerate.
17916 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
17918 (do_jump): Use scalar_int_mode for local variables.
17919 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
17920 rather than a machine_mode.
17921 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
17922 (scompare_loc_descriptor_wide): Likewise.
17923 (scompare_loc_descriptor_narrow): Likewise.
17924 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
17926 * except.c (sjlj_emit_dispatch_table): Likewise.
17927 (expand_builtin_eh_copy_values): Likewise.
17928 * explow.c (convert_memory_address_addr_space_1): Likewise.
17929 Take a scalar_int_mode rather than a machine_mode.
17930 (convert_memory_address_addr_space): Take a scalar_int_mode rather
17931 than a machine_mode.
17932 (memory_address_addr_space): Use scalar_int_mode for local variables.
17933 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
17934 rather than a machine_mode.
17935 * expmed.c (mask_rtx): Likewise.
17936 (init_expmed_one_conv): Likewise.
17937 (expand_mult_highpart_adjust): Likewise.
17938 (extract_high_half): Likewise.
17939 (expmed_mult_highpart_optab): Likewise.
17940 (expmed_mult_highpart): Likewise.
17941 (expand_smod_pow2): Likewise.
17942 (expand_sdiv_pow2): Likewise.
17943 (emit_store_flag_int): Likewise.
17944 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
17946 (extract_low_bits): Likewise.
17947 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
17949 * expr.c (pieces_addr::adjust): Likewise.
17950 (can_store_by_pieces): Likewise.
17951 (store_by_pieces): Likewise.
17952 (clear_by_pieces_1): Likewise.
17953 (expand_expr_addr_expr_1): Likewise.
17954 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
17955 (expand_expr_real_1): Likewise.
17956 (try_casesi): Likewise.
17957 * final.c (shorten_branches): Likewise.
17958 * fold-const.c (fold_convert_const_int_from_fixed): Change the
17959 type of "mode" to machine_mode.
17960 * internal-fn.c (expand_arith_overflow_result_store): Take a
17961 scalar_int_mode rather than a machine_mode.
17962 (expand_mul_overflow): Use scalar_int_mode for local variables.
17963 * loop-doloop.c (doloop_modify): Likewise.
17964 (doloop_optimize): Likewise.
17965 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
17966 than a machine_mode.
17967 (expand_doubleword_shift_condmove): Likewise.
17968 (expand_doubleword_shift): Likewise.
17969 (expand_doubleword_clz): Likewise.
17970 (expand_doubleword_popcount): Likewise.
17971 (expand_doubleword_parity): Likewise.
17972 (expand_absneg_bit): Use scalar_int_mode for local variables.
17973 (prepare_float_lib_cmp): Likewise.
17974 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
17975 rather than a machine_mode.
17976 (convert_memory_address_addr_space): Likewise.
17977 (get_mode_bounds): Likewise.
17978 (get_address_mode): Return a scalar_int_mode rather than a
17980 * rtlanal.c (get_address_mode): Likewise.
17981 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
17982 than a machine_mode.
17983 * targhooks.c (default_mode_rep_extended): Likewise.
17984 (default_valid_pointer_mode): Likewise.
17985 (default_addr_space_valid_pointer_mode): Likewise.
17986 (default_eh_return_filter_mode): Return a scalar_int_mode rather
17987 than a machine_mode.
17988 (default_libgcc_cmp_return_mode): Likewise.
17989 (default_libgcc_shift_count_mode): Likewise.
17990 (default_unwind_word_mode): Likewise.
17991 (default_addr_space_pointer_mode): Likewise.
17992 (default_addr_space_address_mode): Likewise.
17993 * targhooks.h (default_eh_return_filter_mode): Likewise.
17994 (default_libgcc_cmp_return_mode): Likewise.
17995 (default_libgcc_shift_count_mode): Likewise.
17996 (default_unwind_word_mode): Likewise.
17997 (default_addr_space_pointer_mode): Likewise.
17998 (default_addr_space_address_mode): Likewise.
17999 (default_mode_rep_extended): Take a scalar_int_mode rather than
18001 (default_valid_pointer_mode): Likewise.
18002 (default_addr_space_valid_pointer_mode): Likewise.
18003 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18005 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18006 rather than a machine_mode.
18007 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18008 for local variables.
18009 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18010 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18011 than a machine_mode.
18013 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18014 Alan Hayward <alan.hayward@arm.com>
18015 David Sherwood <david.sherwood@arm.com>
18017 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18018 the mode argument to scalar_int_mode.
18019 (do_jump_by_parts_zero_rtx): Likewise.
18020 (do_jump_by_parts_equality_rtx): Likewise.
18021 (do_jump_by_parts_greater): Take a mode argument.
18022 (do_jump_by_parts_equality): Likewise.
18023 (do_jump_1): Update calls accordingly.
18025 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18026 Alan Hayward <alan.hayward@arm.com>
18027 David Sherwood <david.sherwood@arm.com>
18029 * is-a.h (safe_dyn_cast): New function.
18030 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18031 (jump_table_for_label): Likewise.
18032 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18033 instead of an rtx_insn *.
18034 (shorten_branches): Use dyn_cast instead of LABEL_P and
18035 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18036 rtx_jump_table_data::get_data_mode.
18037 (final_scan_insn): Likewise.
18039 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18040 Alan Hayward <alan.hayward@arm.com>
18041 David Sherwood <david.sherwood@arm.com>
18043 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18044 trying to combine a full-register integer set with a subreg
18047 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18048 Alan Hayward <alan.hayward@arm.com>
18049 David Sherwood <david.sherwood@arm.com>
18051 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18052 that is always either address_mode or pointer_mode.
18054 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18055 Alan Hayward <alan.hayward@arm.com>
18056 David Sherwood <david.sherwood@arm.com>
18058 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
18059 when the two are known to be equal.
18061 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18062 Alan Hayward <alan.hayward@arm.com>
18063 David Sherwood <david.sherwood@arm.com>
18065 * simplify-rtx.c (simplify_const_unary_operation): Use
18066 is_a <scalar_int_mode> instead of checking for a nonzero
18067 precision. Forcibly convert op_mode to a scalar_int_mode
18068 in that case. More clearly differentiate the operand and
18069 result modes and use the former when deciding what the value
18070 of a count-bits operation should be. Use is_int_mode instead
18071 of checking for a MODE_INT. Remove redundant check for whether
18072 this mode has a zero precision.
18074 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18075 Alan Hayward <alan.hayward@arm.com>
18076 David Sherwood <david.sherwood@arm.com>
18078 * optabs.c (widen_leading): Change the type of the mode argument
18079 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
18080 (widen_bswap): Likewise.
18081 (expand_parity): Likewise.
18082 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
18083 (expand_ffs): Likewise.
18084 (epand_unop): Check for scalar integer modes before calling the
18087 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18088 Alan Hayward <alan.hayward@arm.com>
18089 David Sherwood <david.sherwood@arm.com>
18091 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
18093 (expand_expr_real_1): Update call accordingly.
18095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18096 Alan Hayward <alan.hayward@arm.com>
18097 David Sherwood <david.sherwood@arm.com>
18099 * expmed.c (store_bit_field_using_insv): Add op0_mode and
18100 value_mode arguments. Use scalar_int_mode internally.
18101 (store_bit_field_1): Rename the new integer mode from imode
18102 to op0_mode and use it instead of GET_MODE (op0). Update calls
18103 to store_split_bit_field, store_bit_field_using_insv and
18104 store_fixed_bit_field.
18105 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
18106 Use scalar_int_mode internally. Use a bit count rather than a mode
18107 when calculating the largest bit size for get_best_mode.
18108 Update calls to store_split_bit_field and store_fixed_bit_field_1.
18109 (store_fixed_bit_field_1): Add mode and value_mode arguments.
18110 Remove assertion that OP0 has a scalar integer mode.
18111 (store_split_bit_field): Add op0_mode and value_mode arguments.
18112 Update calls to extract_fixed_bit_field.
18113 (extract_bit_field_using_extv): Add an op0_mode argument.
18114 Use scalar_int_mode internally.
18115 (extract_bit_field_1): Rename the new integer mode from imode to
18116 op0_mode and use it instead of GET_MODE (op0). Update calls to
18117 extract_split_bit_field, extract_bit_field_using_extv and
18118 extract_fixed_bit_field.
18119 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
18120 to extract_split_bit_field and extract_fixed_bit_field_1.
18121 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
18122 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
18123 on the target mode.
18124 (extract_split_bit_field): Add an op0_mode argument. Update call
18125 to extract_fixed_bit_field.
18127 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18128 Alan Hayward <alan.hayward@arm.com>
18129 David Sherwood <david.sherwood@arm.com>
18131 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
18132 * explow.c (hard_function_value): Likewise.
18133 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
18134 convert_to_mode call outside the loop.
18135 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
18136 for the mode iterator. Require the mode specified by max_pieces
18138 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
18140 (copy_blkmode_to_reg): Likewise.
18141 (set_storage_via_setmem): Likewise.
18142 * optabs.c (prepare_cmp_insn): Likewise.
18143 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18144 * stor-layout.c (finish_bitfield_representative): Likewise.
18146 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18147 Alan Hayward <alan.hayward@arm.com>
18148 David Sherwood <david.sherwood@arm.com>
18150 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
18151 * expr.c (convert_move): Use them.
18152 (convert_modes): Likewise.
18153 (store_expr_with_bounds): Likewise.
18155 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18156 Alan Hayward <alan.hayward@arm.com>
18157 David Sherwood <david.sherwood@arm.com>
18159 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
18160 parameter for the mode of "x". Remove the "known_x", "known_mode"
18161 and "known_ret" arguments. Change the type of the mode argument
18162 to scalar_int_mode.
18163 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
18164 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
18165 (reg_num_sign_bit_copies_for_combine): Likewise.
18166 * rtlanal.c (nonzero_bits1): Likewise.
18167 (num_sign_bit_copies1): Likewise.
18168 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
18169 (reg_num_sign_bit_copies_general): Likewise.
18170 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
18171 (reg_nonzero_bits_general): Likewise.
18173 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18174 Alan Hayward <alan.hayward@arm.com>
18175 David Sherwood <david.sherwood@arm.com>
18177 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
18178 than in subroutines. Return 1 for non-integer modes.
18179 (cached_num_sign_bit_copies): Change the type of the mode parameter
18180 to scalar_int_mode.
18181 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
18182 classes. Handle CONST_INT_P first and then check whether X also
18183 has a scalar integer mode. Check the same thing for inner registers
18184 of a SUBREG and for values that are being extended or truncated.
18186 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18187 Alan Hayward <alan.hayward@arm.com>
18188 David Sherwood <david.sherwood@arm.com>
18190 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
18191 in subroutines. Return the mode mask for non-integer modes.
18192 (cached_nonzero_bits): Change the type of the mode parameter
18193 to scalar_int_mode.
18194 (nonzero_bits1): Likewise. Remove early exit for other mode
18195 classes. Handle CONST_INT_P first and then check whether X
18196 also has a scalar integer mode.
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 * expr.c (widest_int_mode_for_size): Make the comment match the code.
18203 Return a scalar_int_mode and assert that the size is greater than
18205 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
18206 (op_by_pieces_d::op_by_pieces_d): Likewise.
18207 (op_by_pieces_d::run): Likewise.
18208 (can_store_by_pieces): Likewise.
18210 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18211 Alan Hayward <alan.hayward@arm.com>
18212 David Sherwood <david.sherwood@arm.com>
18214 * combine.c (extract_left_shift): Add a mode argument and update
18216 (make_compound_operation_int): Change the type of the mode parameter
18217 to scalar_int_mode and update the call to extract_left_shift.
18219 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18220 Alan Hayward <alan.hayward@arm.com>
18221 David Sherwood <david.sherwood@arm.com>
18223 * combine.c (simplify_and_const_int): Change the type of the mode
18224 parameter to scalar_int_mode.
18225 (simplify_and_const_int_1): Likewise. Update recursive call.
18227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18228 Alan Hayward <alan.hayward@arm.com>
18229 David Sherwood <david.sherwood@arm.com>
18231 * combine.c (simplify_compare_const): Check that the mode is a
18232 scalar_int_mode (rather than VOIDmode) before testing its
18234 (simplify_comparison): Move COMPARISON_P handling out of the
18235 loop and restrict the latter part of the loop to scalar_int_modes.
18236 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
18237 and when considering SUBREG_REGs. Use is_int_mode instead of
18238 checking GET_MODE_CLASS against MODE_INT.
18240 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18241 Alan Hayward <alan.hayward@arm.com>
18242 David Sherwood <david.sherwood@arm.com>
18244 * combine.c (try_widen_shift_mode): Move check for equal modes to...
18245 (simplify_shift_const_1): ...here. Use scalar_int_mode for
18246 shift_unit_mode and for modes involved in scalar shifts.
18248 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18249 Alan Hayward <alan.hayward@arm.com>
18250 David Sherwood <david.sherwood@arm.com>
18252 * combine.c (force_int_to_mode): New function, split out from...
18253 (force_to_mode): ...here. Keep xmode up-to-date and use it
18254 instead of GET_MODE (x).
18256 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18257 Alan Hayward <alan.hayward@arm.com>
18258 David Sherwood <david.sherwood@arm.com>
18260 * optabs-query.h (extraction_insn::struct_mode): Change type to
18261 opt_scalar_int_mode and update comment.
18262 (extraction_insn::field_mode): Change type to scalar_int_mode.
18263 (extraction_insn::pos_mode): Likewise.
18264 * combine.c (make_extraction): Update accordingly.
18265 * optabs-query.c (get_traditional_extraction_insn): Likewise.
18266 (get_optab_extraction_insn): Likewise.
18267 * recog.c (simplify_while_replacing): Likewise.
18268 * expmed.c (narrow_bit_field_mem): Change the type of the mode
18269 parameter to opt_scalar_int_mode.
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 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
18276 to a scalar_int_mode instead of a machine_mode.
18277 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
18278 (get_best_mode): Return a boolean and use a pointer argument to store
18279 the selected mode. Replace the limit mode parameter with a bit limit.
18280 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
18281 for the values returned by bit_field_mode_iterator::next_mode.
18282 (store_bit_field): Update call to get_best_mode.
18283 (store_fixed_bit_field): Likewise.
18284 (extract_fixed_bit_field): Likewise.
18285 * expr.c (optimize_bitfield_assignment_op): Likewise.
18286 * fold-const.c (optimize_bit_field_compare): Likewise.
18287 (fold_truth_andor_1): Likewise.
18288 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
18289 Update for new type of m_mode.
18290 (get_best_mode): As above.
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 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
18297 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
18298 (store_bit_field): Check is_a <scalar_int_mode> before calling
18299 strict_volatile_bitfield_p.
18300 (extract_bit_field): Likewise.
18302 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18303 Alan Hayward <alan.hayward@arm.com>
18304 David Sherwood <david.sherwood@arm.com>
18306 * target.def (cstore_mode): Return a scalar_int_mode.
18307 * doc/tm.texi: Regenerate.
18308 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
18309 * targhooks.h (default_cstore_mode): Likewise.
18310 * targhooks.c (default_cstore_mode): Likewise, using a forced
18312 * expmed.c (emit_cstore): Expect the target of the cstore to be
18315 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18316 Alan Hayward <alan.hayward@arm.com>
18317 David Sherwood <david.sherwood@arm.com>
18319 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
18321 (niter_desc): Likewise mode.
18322 (iv_analyze): Add a mode parameter.
18324 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
18325 and change its type to scalar_int_mode.
18326 * loop-iv.c: Update commentary at head of file.
18327 (iv_constant): Pass the mode paraeter before the rtx it describes
18328 and change its type to scalar_int_mode. Remove VOIDmode handling.
18329 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
18330 (iv_extend): Likewise.
18331 (shorten_into_mode): Likewise.
18332 (iv_add): Use scalar_int_mode.
18333 (iv_mult): Likewise.
18334 (iv_shift): Likewise.
18335 (canonicalize_iv_subregs): Likewise.
18336 (get_biv_step_1): Pass the outer_mode parameter before the rtx
18337 it describes and change its mode to scalar_int_mode. Also change
18338 the type of the returned inner_mode to scalar_int_mode.
18339 (get_biv_step): Likewise, turning outer_mode from a pointer
18340 into a direct parameter. Update call to get_biv_step_1.
18341 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
18342 iv_constant and get_biv_step.
18343 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
18344 and change its type to scalar_int_mode. Don't initialise iv->mode
18345 to VOIDmode and remove later checks for its still being VOIDmode.
18346 Update calls to iv_analyze_op and iv_analyze_expr. Check
18347 is_a <scalar_int_mode> when changing the mode under consideration.
18348 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
18349 Update call to iv_analyze_expr.
18350 (iv_analyze_op): Add a mode parameter. Reject subregs whose
18351 inner register is not also a scalar_int_mode. Update call to
18353 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
18354 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
18355 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
18356 separate mode class checks. Update calls to iv_analyze. Remove
18357 fix-up of VOIDmodes after iv_analyze_biv.
18358 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
18359 don't have a scalar_int_mode. Update call to biv_p.
18361 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18362 Alan Hayward <alan.hayward@arm.com>
18363 David Sherwood <david.sherwood@arm.com>
18365 * cfgexpand.c (convert_debug_memory_address): Use
18366 as_a <scalar_int_mode>.
18367 * combine.c (expand_compound_operation): Likewise.
18368 (make_extraction): Likewise.
18369 (change_zero_ext): Likewise.
18370 (simplify_comparison): Likewise.
18371 * cse.c (cse_insn): Likewise.
18372 * dwarf2out.c (minmax_loc_descriptor): Likewise.
18373 (mem_loc_descriptor): Likewise.
18374 (loc_descriptor): Likewise.
18375 * expmed.c (init_expmed_one_mode): Likewise.
18376 (synth_mult): Likewise.
18377 (emit_store_flag_1): Likewise.
18378 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
18379 of a comparison with size.
18380 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
18381 (reduce_to_bit_field_precision): Likewise.
18382 * function.c (expand_function_end): Likewise.
18383 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
18384 * loop-doloop.c (doloop_modify): Likewise.
18385 * optabs.c (expand_binop): Likewise.
18386 (expand_unop): Likewise.
18387 (expand_copysign_absneg): Likewise.
18388 (prepare_cmp_insn): Likewise.
18389 (maybe_legitimize_operand): Likewise.
18390 * recog.c (const_scalar_int_operand): Likewise.
18391 * rtlanal.c (get_address_mode): Likewise.
18392 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18393 (simplify_cond_clz_ctz): Likewise.
18394 * tree-nested.c (get_nl_goto_field): Likewise.
18395 * tree.c (build_vector_type_for_mode): Likewise.
18396 * var-tracking.c (use_narrower_mode): Likewise.
18398 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18399 Alan Hayward <alan.hayward@arm.com>
18400 David Sherwood <david.sherwood@arm.com>
18402 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
18403 * builtins.c (expand_builtin_signbit): Use it.
18404 * cfgexpand.c (expand_debug_expr): Likewise.
18405 * dojump.c (do_jump): Likewise.
18406 (do_compare_and_jump): Likewise.
18407 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
18408 * expmed.c (make_tree): Likewise.
18409 * expr.c (expand_expr_real_2): Likewise.
18410 (expand_expr_real_1): Likewise.
18411 (try_casesi): Likewise.
18412 * fold-const-call.c (fold_const_call_ss): Likewise.
18413 * fold-const.c (unextend): Likewise.
18414 (extract_muldiv_1): Likewise.
18415 (fold_single_bit_test): Likewise.
18416 (native_encode_int): Likewise.
18417 (native_encode_string): Likewise.
18418 (native_interpret_int): Likewise.
18419 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
18420 * internal-fn.c (expand_addsub_overflow): Likewise.
18421 (expand_neg_overflow): Likewise.
18422 (expand_mul_overflow): Likewise.
18423 (expand_arith_overflow): Likewise.
18424 * match.pd: Likewise.
18425 * stor-layout.c (layout_type): Likewise.
18426 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18427 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
18428 * tree-ssanames.c (get_range_info): Likewise.
18429 * tree-switch-conversion.c (array_value_type) Likewise.
18430 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
18431 (vect_recog_divmod_pattern): Likewise.
18432 (vect_recog_mixed_size_cond_pattern): Likewise.
18433 * tree-vrp.c (extract_range_basic): Likewise.
18434 (simplify_float_conversion_using_ranges): Likewise.
18435 * tree.c (int_fits_type_p): Likewise.
18436 * ubsan.c (instrument_bool_enum_load): Likewise.
18437 * varasm.c (mergeable_string_section): Likewise.
18438 (narrowing_initializer_constant_valid_p): Likewise.
18439 (output_constant): Likewise.
18441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18442 Alan Hayward <alan.hayward@arm.com>
18443 David Sherwood <david.sherwood@arm.com>
18445 * machmode.h (NARROWEST_INT_MODE): New macro.
18446 * expr.c (alignment_for_piecewise_move): Use it instead of
18447 GET_CLASS_NARROWEST_MODE (MODE_INT).
18448 (push_block): Likewise.
18449 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
18451 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18453 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18454 Alan Hayward <alan.hayward@arm.com>
18455 David Sherwood <david.sherwood@arm.com>
18457 * postreload.c (move2add_valid_value_p): Change the type of the
18458 mode parameter to scalar_int_mode.
18459 (move2add_use_add2_insn): Add a mode parameter and use it instead
18461 (move2add_use_add3_insn): Likewise.
18462 (reload_cse_move2add): Update accordingly.
18464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18465 Alan Hayward <alan.hayward@arm.com>
18466 David Sherwood <david.sherwood@arm.com>
18468 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
18470 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
18471 * optabs.c (expand_unop): Likewise.
18473 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18474 Alan Hayward <alan.hayward@arm.com>
18475 David Sherwood <david.sherwood@arm.com>
18477 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
18478 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
18479 (popcount_loc_descriptor): Likewise.
18480 (bswap_loc_descriptor): Likewise.
18481 (rotate_loc_descriptor): Likewise.
18482 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
18483 calling the functions above.
18485 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18486 Alan Hayward <alan.hayward@arm.com>
18487 David Sherwood <david.sherwood@arm.com>
18489 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
18491 (try_combine): Likewise.
18492 (simplify_if_then_else): Likewise.
18493 * cse.c (cse_insn): Likewise.
18494 * dwarf2out.c (mem_loc_descriptor): Likewise.
18495 * emit-rtl.c (gen_lowpart_common): Likewise.
18496 * simplify-rtx.c (simplify_truncation): Likewise.
18497 (simplify_binary_operation_1): Likewise.
18498 (simplify_const_relational_operation): Likewise.
18499 (simplify_ternary_operation): Likewise.
18500 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
18502 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18503 Alan Hayward <alan.hayward@arm.com>
18504 David Sherwood <david.sherwood@arm.com>
18506 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
18507 * reload.c (push_reload): Likewise.
18508 (find_reloads): Likewise.
18510 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18511 Alan Hayward <alan.hayward@arm.com>
18512 David Sherwood <david.sherwood@arm.com>
18514 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
18515 (make_compound_operation_int): Likewise.
18516 (change_zero_ext): Likewise.
18517 * expr.c (convert_move): Likewise.
18518 (convert_modes): Likewise.
18519 * fwprop.c (forward_propagate_subreg): Likewise.
18520 * loop-iv.c (get_biv_step_1): Likewise.
18521 * optabs.c (widen_operand): Likewise.
18522 * postreload.c (move2add_valid_value_p): Likewise.
18523 * recog.c (simplify_while_replacing): Likewise.
18524 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18525 (simplify_binary_operation_1): Likewise. Remove redundant
18526 mode equality check.
18528 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18529 Alan Hayward <alan.hayward@arm.com>
18530 David Sherwood <david.sherwood@arm.com>
18532 * combine.c (combine_simplify_rtx): Add checks for
18533 is_a <scalar_int_mode>.
18534 (simplify_if_then_else): Likewise.
18535 (make_field_assignment): Likewise.
18536 (simplify_comparison): Likewise.
18537 * ifcvt.c (noce_try_bitop): Likewise.
18538 * loop-invariant.c (canonicalize_address_mult): Likewise.
18539 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18541 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18542 Alan Hayward <alan.hayward@arm.com>
18543 David Sherwood <david.sherwood@arm.com>
18545 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
18546 is_a <scalar_int_mode> instead of != BLKmode.
18548 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18549 Alan Hayward <alan.hayward@arm.com>
18550 David Sherwood <david.sherwood@arm.com>
18552 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
18553 instead of != VOIDmode.
18554 * combine.c (if_then_else_cond): Likewise.
18555 (change_zero_ext): Likewise.
18556 * dwarf2out.c (mem_loc_descriptor): Likewise.
18557 (loc_descriptor): Likewise.
18558 * rtlanal.c (canonicalize_condition): Likewise.
18559 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
18561 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18562 Alan Hayward <alan.hayward@arm.com>
18563 David Sherwood <david.sherwood@arm.com>
18565 * simplify-rtx.c (simplify_binary_operation_1): Use
18566 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18568 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18569 Alan Hayward <alan.hayward@arm.com>
18570 David Sherwood <david.sherwood@arm.com>
18572 * wide-int.h (int_traits<unsigned char>) New class.
18573 (int_traits<unsigned short>) Likewise.
18574 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18575 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18577 * combine.c (set_nonzero_bits_and_sign_copies): Use
18578 is_a <scalar_int_mode>.
18579 (find_split_point): Likewise.
18580 (combine_simplify_rtx): Likewise.
18581 (simplify_logical): Likewise.
18582 (expand_compound_operation): Likewise.
18583 (expand_field_assignment): Likewise.
18584 (make_compound_operation): Likewise.
18585 (extended_count): Likewise.
18586 (change_zero_ext): Likewise.
18587 (simplify_comparison): Likewise.
18588 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18589 (ucompare_loc_descriptor): Likewise.
18590 (minmax_loc_descriptor): Likewise.
18591 (mem_loc_descriptor): Likewise.
18592 (loc_descriptor): Likewise.
18593 * expmed.c (init_expmed_one_mode): Likewise.
18594 * lra-constraints.c (lra_constraint_offset): Likewise.
18595 * optabs.c (prepare_libcall_arg): Likewise.
18596 * postreload.c (move2add_note_store): Likewise.
18597 * reload.c (operands_match_p): Likewise.
18598 * rtl.h (load_extend_op): Likewise.
18599 * rtlhooks.c (gen_lowpart_general): Likewise.
18600 * simplify-rtx.c (simplify_truncation): Likewise.
18601 (simplify_unary_operation_1): Likewise.
18602 (simplify_binary_operation_1): Likewise.
18603 (simplify_const_binary_operation): Likewise.
18604 (simplify_const_relational_operation): Likewise.
18605 (simplify_subreg): Likewise.
18606 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18607 * var-tracking.c (adjust_mems): Likewise.
18608 (prepare_call_arguments): Likewise.
18610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18611 Alan Hayward <alan.hayward@arm.com>
18612 David Sherwood <david.sherwood@arm.com>
18614 * machmode.h (is_int_mode): New fuction.
18615 * combine.c (find_split_point): Use it.
18616 (combine_simplify_rtx): Likewise.
18617 (simplify_if_then_else): Likewise.
18618 (simplify_set): Likewise.
18619 (simplify_shift_const_1): Likewise.
18620 (simplify_comparison): Likewise.
18621 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18622 * cse.c (notreg_cost): Likewise.
18623 (cse_insn): Likewise.
18624 * cselib.c (cselib_lookup_1): Likewise.
18625 * dojump.c (do_jump_1): Likewise.
18626 (do_compare_rtx_and_jump): Likewise.
18627 * dse.c (get_call_args): Likewise.
18628 * dwarf2out.c (rtl_for_decl_init): Likewise.
18629 (native_encode_initializer): Likewise.
18630 * expmed.c (emit_store_flag_1): Likewise.
18631 (emit_store_flag): Likewise.
18632 * expr.c (convert_modes): Likewise.
18633 (store_field): Likewise.
18634 (expand_expr_real_1): Likewise.
18635 * fold-const.c (fold_read_from_constant_string): Likewise.
18636 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18637 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18638 * optabs.c (expand_binop): Likewise.
18639 (expand_unop): Likewise.
18640 (expand_abs_nojump): Likewise.
18641 (expand_one_cmpl_abs_nojump): Likewise.
18642 * simplify-rtx.c (mode_signbit_p): Likewise.
18643 (val_signbit_p): Likewise.
18644 (val_signbit_known_set_p): Likewise.
18645 (val_signbit_known_clear_p): Likewise.
18646 (simplify_relational_operation_1): Likewise.
18647 * tree.c (vector_type_mode): Likewise.
18649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18650 Alan Hayward <alan.hayward@arm.com>
18651 David Sherwood <david.sherwood@arm.com>
18653 * machmode.h (smallest_mode_for_size): Fix formatting.
18654 (smallest_int_mode_for_size): New function.
18655 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18656 instead of smallest_mode_for_size.
18657 * combine.c (make_extraction): Likewise.
18658 * config/arc/arc.c (arc_expand_movmem): Likewise.
18659 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18660 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18661 * config/s390/s390.c (s390_expand_insv): Likewise.
18662 * config/sparc/sparc.c (assign_int_registers): Likewise.
18663 * config/spu/spu.c (spu_function_value): Likewise.
18664 (spu_function_arg): Likewise.
18665 * coverage.c (get_gcov_type): Likewise.
18666 (get_gcov_unsigned_t): Likewise.
18667 * dse.c (find_shift_sequence): Likewise.
18668 * expmed.c (store_bit_field_1): Likewise.
18669 * expr.c (convert_move): Likewise.
18670 (store_field): Likewise.
18671 * internal-fn.c (expand_arith_overflow): Likewise.
18672 * optabs-query.c (get_best_extraction_insn): Likewise.
18673 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18674 * stor-layout.c (layout_type): Likewise.
18675 (initialize_sizetypes): Likewise.
18676 * targhooks.c (default_get_mask_mode): Likewise.
18677 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18680 Alan Hayward <alan.hayward@arm.com>
18681 David Sherwood <david.sherwood@arm.com>
18683 * machmode.h (opt_mode::else_blk): New function.
18684 (int_mode_for_mode): Declare.
18685 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18686 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18688 * cfgexpand.c (expand_debug_expr): Likewise.
18689 * combine.c (gen_lowpart_or_truncate): Likewise.
18690 (gen_lowpart_for_combine): Likewise.
18691 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18692 * config/avr/avr.c (avr_to_int_mode): Likewise.
18693 (avr_out_plus_1): Likewise.
18694 (avr_out_plus): Likewise.
18695 (avr_out_round): Likewise.
18696 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18697 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18698 (s390_expand_vcond): Likewise.
18699 * config/spu/spu.c (spu_split_immediate): Likewise.
18700 (spu_expand_mov): Likewise.
18701 * dse.c (get_stored_val): Likewise.
18702 * expmed.c (store_bit_field_1): Likewise.
18703 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18705 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18706 (extract_low_bits): Likewise.
18707 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18708 handling rather than repeating the check.
18709 (emit_group_store): Likewise.
18710 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18711 * optabs.c (expand_absneg_bit): Likewise.
18712 (expand_copysign_absneg): Likewise.
18713 (expand_copysign_bit): Likewise.
18714 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18715 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18716 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18717 * var-tracking.c (prepare_call_arguments): Likewise.
18718 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18719 int_mode_for_mode instead of mode_for_size.
18720 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18722 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18723 Alan Hayward <alan.hayward@arm.com>
18724 David Sherwood <david.sherwood@arm.com>
18726 * machmode.h (int_mode_for_size): New function.
18727 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18728 instead of mode_for_size.
18729 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18731 * combine.c (expand_field_assignment): Use int_mode_for_size
18732 instead of mode_for_size.
18733 (make_extraction): Likewise.
18734 (simplify_shift_const_1): Likewise.
18735 (simplify_comparison): Likewise.
18736 * dojump.c (do_jump): Likewise.
18737 * dwarf2out.c (mem_loc_descriptor): Likewise.
18738 * emit-rtl.c (init_derived_machine_modes): Likewise.
18739 * expmed.c (flip_storage_order): Likewise.
18740 (convert_extracted_bit_field): Likewise.
18741 * expr.c (copy_blkmode_from_reg): Likewise.
18742 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18743 * internal-fn.c (expand_mul_overflow): Likewise.
18744 * lower-subreg.c (simple_move): Likewise.
18745 * optabs-libfuncs.c (init_optabs): Likewise.
18746 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18747 * tree.c (vector_type_mode): Likewise.
18748 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18749 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18750 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18751 * tree-vect-stmts.c (vectorizable_load): Likewise.
18752 (vectorizable_store): Likewise.
18754 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18755 Alan Hayward <alan.hayward@arm.com>
18756 David Sherwood <david.sherwood@arm.com>
18758 * coretypes.h (pod_mode): New type.
18759 (scalar_int_mode_pod): New typedef.
18760 * machmode.h (pod_mode): New class.
18761 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18762 * genmodes.c (emit_mode_int_n): Update accordingly.
18763 * lower-subreg.h (target_lower_subreg): Change type to
18764 scalar_int_mode_pod.
18765 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18766 scalar_int_mode_pod.
18768 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18769 Alan Hayward <alan.hayward@arm.com>
18770 David Sherwood <david.sherwood@arm.com>
18772 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18773 machine_mode to scalar_int_mode.
18774 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18775 (rs6000_option_override_internal): Remove cast to int.
18776 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18777 machine_mode to scalar_int_mode.
18778 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18779 (rs6000_option_override_internal): Remove cast to int.
18780 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18781 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18783 * config/s390/s390.c (s390_expand_builtin): Likewise.
18784 * coretypes.h (scalar_int_mode): New type.
18785 (opt_scalar_int_mode): New typedef.
18786 * machmode.h (scalar_int_mode): New class.
18787 (scalar_int_mode::includes_p): New function.
18788 (byte_mode): Change type to scalar_int_mode.
18789 (word_mode): Likewise.
18790 (ptr_mode): Likewise.
18791 * emit-rtl.c (byte_mode): Likewise.
18792 (word_mode): Likewise.
18793 (ptr_mode): Likewise.
18794 (init_derived_machine_modes): Update accordingly.
18795 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18796 and MODE_PARTIAL_INT.
18797 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18798 opt_scalar_int_mode.
18800 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18801 Alan Hayward <alan.hayward@arm.com>
18802 David Sherwood <david.sherwood@arm.com>
18804 * target.def (libgcc_floating_mode_supported_p): Take a
18806 * doc/tm.texi: Regenerate.
18807 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18809 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18810 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
18813 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18814 Alan Hayward <alan.hayward@arm.com>
18815 David Sherwood <david.sherwood@arm.com>
18817 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
18818 * doc/tm.texi: Regenerate.
18819 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
18820 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
18821 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
18822 * targhooks.h (default_floatn_mode): Likewise.
18823 * targhooks.c (default_floatn_mode): Likewise.
18824 * tree.c (build_common_tree_nodes): Update accordingly.
18826 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18827 Alan Hayward <alan.hayward@arm.com>
18828 David Sherwood <david.sherwood@arm.com>
18830 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
18831 (mode_iterator::iterate_p): Likewise.
18832 (mode_iterator::get_wider): Likewise.
18833 * expr.c (init_expr_target): Use opt_scalar_float_mode.
18835 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18836 Alan Hayward <alan.hayward@arm.com>
18837 David Sherwood <david.sherwood@arm.com>
18839 * coretypes.h (opt_scalar_float_mode): New typedef.
18840 * machmode.h (float_mode_for_size): New function.
18841 * emit-rtl.c (double_mode): Delete.
18842 (init_emit_once): Use float_mode_for_size.
18843 * stor-layout.c (layout_type): Likewise.
18844 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
18846 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18847 Alan Hayward <alan.hayward@arm.com>
18848 David Sherwood <david.sherwood@arm.com>
18850 * output.h (assemble_real): Take a scalar_float_mode.
18851 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
18852 * config/arm/arm.md (consttable_4): Likewise.
18853 (consttable_8): Likewise.
18854 (consttable_16): Likewise.
18855 * config/mips/mips.md (consttable_float): Likewise.
18856 * config/s390/s390.c (s390_output_pool_entry): Likewise.
18857 * varasm.c (assemble_real): Take a scalar_float_mode.
18858 (output_constant_pool_2): Update accordingly.
18859 (output_constant): Likewise.
18861 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18862 Alan Hayward <alan.hayward@arm.com>
18863 David Sherwood <david.sherwood@arm.com>
18865 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
18866 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
18867 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
18868 (native_encode_real): Likewise.
18869 (native_interpret_real): Likewise.
18870 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
18871 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18873 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18874 Alan Hayward <alan.hayward@arm.com>
18875 David Sherwood <david.sherwood@arm.com>
18877 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
18878 <scalar_float_mode>. Simplify.
18879 (gen_extend_conv_libfunc): Likewise.
18881 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18882 Alan Hayward <alan.hayward@arm.com>
18883 David Sherwood <david.sherwood@arm.com>
18885 * coretypes.h (scalar_float_mode): New type.
18886 * machmode.h (mode_traits::from_int): Use machine_mode if
18887 USE_ENUM_MODES is defined.
18888 (is_a): New function.
18890 (dyn_cast): Likewise.
18891 (scalar_float_mode): New class.
18892 (scalar_float_mode::includes_p): New function.
18893 (is_float_mode): Likewise.
18894 * gdbhooks.py (MachineModePrinter): New class.
18895 (build_pretty_printer): Use it for scalar_float_mode.
18896 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
18897 (format_helper::format_helper): Turn into a template.
18898 * genmodes.c (get_mode_class): New function.
18899 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
18900 or machine_mode if none.
18901 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
18902 as_a <scalar_float_mode>.
18903 * dwarf2out.c (mem_loc_descriptor): Likewise.
18904 (insert_float): Likewise.
18905 (add_const_value_attribute): Likewise.
18906 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18907 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
18908 (expand_unop): Update accordingly.
18909 (expand_abs_nojump): Likewise.
18910 (expand_copysign_absneg): Take a scalar_float_mode.
18911 (expand_copysign_bit): Likewise.
18912 (expand_copysign): Update accordingly.
18914 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18915 Alan Hayward <alan.hayward@arm.com>
18916 David Sherwood <david.sherwood@arm.com>
18918 * coretypes.h (opt_mode): New class.
18919 * machmode.h (opt_mode): Likewise.
18920 (opt_mode::else_void): New function.
18921 (opt_mode::require): Likewise.
18922 (opt_mode::exists): Likewise.
18923 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
18924 (GET_MODE_2XWIDER_MODE): Likewise.
18925 (mode_iterator::get_wider): Update accordingly.
18926 (mode_iterator::get_2xwider): Likewise.
18927 (mode_iterator::get_known_wider): Likewise, turning into a template.
18928 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
18929 forcing a wider mode to exist.
18930 * config/cr16/cr16.h (LONG_REG_P): Likewise.
18931 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18932 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
18933 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
18934 * lower-subreg.c (init_lower_subreg): Likewise.
18935 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
18936 on the final iteration.
18937 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
18938 a wider mode exists before asking for a move pattern.
18939 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
18940 forcing a wider mode to exist.
18941 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
18942 returning false if no such mode exists.
18943 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
18944 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
18945 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
18946 Avoid checking for a MODE_INT if we already know the mode is not a
18948 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
18949 forcing a wider mode to exist.
18950 (expmed_mult_highpart_optab): Likewise.
18951 (expmed_mult_highpart): Likewise.
18952 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
18954 * lto-streamer-in.c (lto_input_mode_table): Likewise.
18955 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
18956 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
18957 * internal-fn.c (expand_mul_overflow): Update use of
18958 GET_MODE_2XWIDER_MODE.
18959 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18960 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
18961 GET_MODE_WIDER_MODE.
18962 (convert_plusminus_to_widen): Likewise.
18963 * tree-switch-conversion.c (array_value_type): Likewise.
18964 * var-tracking.c (emit_note_insn_var_location): Likewise.
18965 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18966 Return false inside rather than outside the loop if no wider mode
18968 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
18969 and GET_MODE_2XWIDER_MODE
18970 (can_compare_p): Use else_void.
18971 * gdbhooks.py (OptMachineModePrinter): New class.
18972 (build_pretty_printer): Use it for opt_mode.
18974 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18975 Alan Hayward <alan.hayward@arm.com>
18976 David Sherwood <david.sherwood@arm.com>
18978 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
18979 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
18981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18982 Alan Hayward <alan.hayward@arm.com>
18983 David Sherwood <david.sherwood@arm.com>
18985 * machmode.h (mode_traits): New structure.
18986 (get_narrowest_mode): New function.
18987 (mode_iterator::start): Likewise.
18988 (mode_iterator::iterate_p): Likewise.
18989 (mode_iterator::get_wider): Likewise.
18990 (mode_iterator::get_known_wider): Likewise.
18991 (mode_iterator::get_2xwider): Likewise.
18992 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
18993 (FOR_EACH_MODE): Likewise.
18994 (FOR_EACH_MODE_FROM): Likewise.
18995 (FOR_EACH_MODE_UNTIL): Likewise.
18996 (FOR_EACH_WIDER_MODE): Likewise.
18997 (FOR_EACH_2XWIDER_MODE): Likewise.
18998 * builtins.c (expand_builtin_strlen): Use new mode iterators.
18999 * combine.c (simplify_comparison): Likewise
19000 * config/i386/i386.c (type_natural_mode): Likewise.
19001 * cse.c (cse_insn): Likewise.
19002 * dse.c (find_shift_sequence): Likewise.
19003 * emit-rtl.c (init_derived_machine_modes): Likewise.
19004 (init_emit_once): Likewise.
19005 * explow.c (hard_function_value): Likewise.
19006 * expmed.c (extract_fixed_bit_field_1): Likewise.
19007 (extract_bit_field_1): Likewise.
19008 (expand_divmod): Likewise.
19009 (emit_store_flag_1): Likewise.
19010 * expr.c (init_expr_target): Likewise.
19011 (convert_move): Likewise.
19012 (alignment_for_piecewise_move): Likewise.
19013 (widest_int_mode_for_size): Likewise.
19014 (emit_block_move_via_movmem): Likewise.
19015 (copy_blkmode_to_reg): Likewise.
19016 (set_storage_via_setmem): Likewise.
19017 (compress_float_constant): Likewise.
19018 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19019 * optabs-query.c (get_best_extraction_insn): Likewise.
19020 * optabs.c (expand_binop): Likewise.
19021 (expand_twoval_unop): Likewise.
19022 (expand_twoval_binop): Likewise.
19023 (widen_leading): Likewise.
19024 (widen_bswap): Likewise.
19025 (expand_parity): Likewise.
19026 (expand_unop): Likewise.
19027 (prepare_cmp_insn): Likewise.
19028 (prepare_float_lib_cmp): Likewise.
19029 (expand_float): Likewise.
19030 (expand_fix): Likewise.
19031 (expand_sfix_optab): Likewise.
19032 * postreload.c (move2add_use_add2_insn): Likewise.
19033 * reg-stack.c (reg_to_stack): Likewise.
19034 * reginfo.c (choose_hard_reg_mode): Likewise.
19035 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19036 * stor-layout.c (mode_for_size): Likewise.
19037 (smallest_mode_for_size): Likewise.
19038 (mode_for_vector): Likewise.
19039 (finish_bitfield_representative): Likewise.
19040 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19041 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19042 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19043 * var-tracking.c (prepare_call_arguments): Likewise.
19045 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19046 Alan Hayward <alan.hayward@arm.com>
19047 David Sherwood <david.sherwood@arm.com>
19049 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19050 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19051 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19052 * machmode.h (mode_size): Move earlier in file.
19053 (mode_precision): Likewise.
19054 (mode_inner): Likewise.
19055 (mode_nunits): Likewise.
19056 (mode_unit_size): Likewise.
19057 (unit_unit_precision): Likewise.
19058 (mode_wider): Likewise.
19059 (mode_2xwider): Likewise.
19060 (machine_mode): New class.
19061 (mode_to_bytes): New function.
19062 (mode_to_bits): Likewise.
19063 (mode_to_precision): Likewise.
19064 (mode_to_inner): Likewise.
19065 (mode_to_unit_size): Likewise.
19066 (mode_to_unit_precision): Likewise.
19067 (mode_to_nunits): Likewise.
19068 (GET_MODE_SIZE): Use mode_to_bytes.
19069 (GET_MODE_BITSIZE): Use mode_to_bits.
19070 (GET_MODE_PRECISION): Use mode_to_precision.
19071 (GET_MODE_INNER): Use mode_to_inner.
19072 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
19073 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
19074 (GET_MODE_NUNITS): Use mode_to_nunits.
19075 * system.h (ALWAYS_INLINE): New macro.
19076 * config/powerpcspe/powerpcspe-c.c
19077 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
19078 int for arg1_mode and arg2_mode.
19080 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19081 Alan Hayward <alan.hayward@arm.com>
19082 David Sherwood <david.sherwood@arm.com>
19084 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19085 Prefix mode names with E_ in case statements.
19086 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19087 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
19088 (aarch64_split_simd_move): Likewise.
19089 (aarch64_gen_storewb_pair): Likewise.
19090 (aarch64_gen_loadwb_pair): Likewise.
19091 (aarch64_gen_store_pair): Likewise.
19092 (aarch64_gen_load_pair): Likewise.
19093 (aarch64_get_condition_code_1): Likewise.
19094 (aarch64_constant_pool_reload_icode): Likewise.
19095 (get_rsqrte_type): Likewise.
19096 (get_rsqrts_type): Likewise.
19097 (get_recpe_type): Likewise.
19098 (get_recps_type): Likewise.
19099 (aarch64_gimplify_va_arg_expr): Likewise.
19100 (aarch64_simd_container_mode): Likewise.
19101 (aarch64_emit_load_exclusive): Likewise.
19102 (aarch64_emit_store_exclusive): Likewise.
19103 (aarch64_expand_compare_and_swap): Likewise.
19104 (aarch64_gen_atomic_cas): Likewise.
19105 (aarch64_emit_bic): Likewise.
19106 (aarch64_emit_atomic_swap): Likewise.
19107 (aarch64_emit_atomic_load_op): Likewise.
19108 (aarch64_evpc_trn): Likewise.
19109 (aarch64_evpc_uzp): Likewise.
19110 (aarch64_evpc_zip): Likewise.
19111 (aarch64_evpc_ext): Likewise.
19112 (aarch64_evpc_rev): Likewise.
19113 (aarch64_evpc_dup): Likewise.
19114 (aarch64_gen_ccmp_first): Likewise.
19115 (aarch64_gen_ccmp_next): Likewise.
19116 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
19117 (alpha_emit_xfloating_libcall): Likewise.
19118 (emit_insxl): Likewise.
19119 (alpha_arg_type): Likewise.
19120 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
19121 (arc_preferred_simd_mode): Likewise.
19122 (arc_secondary_reload): Likewise.
19123 (get_arc_condition_code): Likewise.
19124 (arc_print_operand): Likewise.
19125 (arc_legitimate_constant_p): Likewise.
19126 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19127 * config/arc/arc.md (casesi_load): Likewise.
19128 (casesi_compact_jump): Likewise.
19129 * config/arc/predicates.md (proper_comparison_operator): Likewise.
19130 (cc_use_register): Likewise.
19131 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19132 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19133 (arm_init_iwmmxt_builtins): Likewise.
19134 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
19135 (neon_expand_vector_init): Likewise.
19136 (arm_attr_length_move_neon): Likewise.
19137 (maybe_get_arm_condition_code): Likewise.
19138 (arm_emit_vector_const): Likewise.
19139 (arm_preferred_simd_mode): Likewise.
19140 (arm_output_iwmmxt_tinsr): Likewise.
19141 (thumb1_output_casesi): Likewise.
19142 (thumb2_output_casesi): Likewise.
19143 (arm_emit_load_exclusive): Likewise.
19144 (arm_emit_store_exclusive): Likewise.
19145 (arm_expand_compare_and_swap): Likewise.
19146 (arm_evpc_neon_vuzp): Likewise.
19147 (arm_evpc_neon_vzip): Likewise.
19148 (arm_evpc_neon_vrev): Likewise.
19149 (arm_evpc_neon_vtrn): Likewise.
19150 (arm_evpc_neon_vext): Likewise.
19151 (arm_validize_comparison): Likewise.
19152 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
19153 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
19154 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
19155 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
19156 (c6x_preferred_simd_mode): Likewise.
19157 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
19158 (epiphany_rtx_costs): Likewise.
19159 * config/epiphany/predicates.md (proper_comparison_operator):
19161 * config/frv/frv.c (condexec_memory_operand): Likewise.
19162 (frv_emit_move): Likewise.
19163 (output_move_single): Likewise.
19164 (output_condmove_single): Likewise.
19165 (frv_hard_regno_mode_ok): Likewise.
19166 (frv_matching_accg_mode): Likewise.
19167 * config/h8300/h8300.c (split_adds_subs): Likewise.
19168 (h8300_rtx_costs): Likewise.
19169 (h8300_print_operand): Likewise.
19170 (compute_mov_length): Likewise.
19171 (output_logical_op): Likewise.
19172 (compute_logical_op_length): Likewise.
19173 (compute_logical_op_cc): Likewise.
19174 (h8300_shift_needs_scratch_p): Likewise.
19175 (output_a_shift): Likewise.
19176 (compute_a_shift_length): Likewise.
19177 (compute_a_shift_cc): Likewise.
19178 (expand_a_rotate): Likewise.
19179 (output_a_rotate): Likewise.
19180 * config/i386/i386.c (classify_argument): Likewise.
19181 (function_arg_advance_32): Likewise.
19182 (function_arg_32): Likewise.
19183 (function_arg_64): Likewise.
19184 (function_value_64): Likewise.
19185 (ix86_gimplify_va_arg): Likewise.
19186 (ix86_legitimate_constant_p): Likewise.
19187 (put_condition_code): Likewise.
19188 (split_double_mode): Likewise.
19189 (ix86_avx256_split_vector_move_misalign): Likewise.
19190 (ix86_expand_vector_logical_operator): Likewise.
19191 (ix86_split_idivmod): Likewise.
19192 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
19193 (ix86_build_const_vector): Likewise.
19194 (ix86_build_signbit_mask): Likewise.
19195 (ix86_match_ccmode): Likewise.
19196 (ix86_cc_modes_compatible): Likewise.
19197 (ix86_expand_branch): Likewise.
19198 (ix86_expand_sse_cmp): Likewise.
19199 (ix86_expand_sse_movcc): Likewise.
19200 (ix86_expand_int_sse_cmp): Likewise.
19201 (ix86_expand_vec_perm_vpermi2): Likewise.
19202 (ix86_expand_vec_perm): Likewise.
19203 (ix86_expand_sse_unpack): Likewise.
19204 (ix86_expand_int_addcc): Likewise.
19205 (ix86_split_to_parts): Likewise.
19206 (ix86_vectorize_builtin_gather): Likewise.
19207 (ix86_vectorize_builtin_scatter): Likewise.
19208 (avx_vpermilp_parallel): Likewise.
19209 (inline_memory_move_cost): Likewise.
19210 (ix86_tieable_integer_mode_p): Likewise.
19211 (x86_maybe_negate_const_int): Likewise.
19212 (ix86_expand_vector_init_duplicate): Likewise.
19213 (ix86_expand_vector_init_one_nonzero): Likewise.
19214 (ix86_expand_vector_init_one_var): Likewise.
19215 (ix86_expand_vector_init_concat): Likewise.
19216 (ix86_expand_vector_init_interleave): Likewise.
19217 (ix86_expand_vector_init_general): Likewise.
19218 (ix86_expand_vector_set): Likewise.
19219 (ix86_expand_vector_extract): Likewise.
19220 (emit_reduc_half): Likewise.
19221 (ix86_emit_i387_round): Likewise.
19222 (ix86_mangle_type): Likewise.
19223 (ix86_expand_round_sse4): Likewise.
19224 (expand_vec_perm_blend): Likewise.
19225 (canonicalize_vector_int_perm): Likewise.
19226 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
19227 (expand_vec_perm_1): Likewise.
19228 (expand_vec_perm_interleave3): Likewise.
19229 (expand_vec_perm_even_odd_pack): Likewise.
19230 (expand_vec_perm_even_odd_1): Likewise.
19231 (expand_vec_perm_broadcast_1): Likewise.
19232 (ix86_vectorize_vec_perm_const_ok): Likewise.
19233 (ix86_expand_vecop_qihi): Likewise.
19234 (ix86_expand_mul_widen_hilo): Likewise.
19235 (ix86_expand_sse2_abs): Likewise.
19236 (ix86_expand_pextr): Likewise.
19237 (ix86_expand_pinsr): Likewise.
19238 (ix86_preferred_simd_mode): Likewise.
19239 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
19240 * config/i386/sse.md (*andnot<mode>3): Likewise.
19241 (<mask_codefor><code><mode>3<mask_name>): Likewise.
19242 (*<code><mode>3): Likewise.
19243 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
19244 (ia64_expand_atomic_op): Likewise.
19245 (ia64_arg_type): Likewise.
19246 (ia64_mode_to_int): Likewise.
19247 (ia64_scalar_mode_supported_p): Likewise.
19248 (ia64_vector_mode_supported_p): Likewise.
19249 (expand_vec_perm_broadcast): Likewise.
19250 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
19251 (iq2000_function_arg_advance): Likewise.
19252 (iq2000_function_arg): Likewise.
19253 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
19254 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
19255 (m68k_libcall_value): Likewise.
19256 (m68k_function_value): Likewise.
19257 (sched_attr_op_type): Likewise.
19258 * config/mcore/mcore.c (mcore_output_move): Likewise.
19259 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
19261 (microblaze_function_arg): Likewise.
19262 * config/mips/mips.c (mips16_build_call_stub): Likewise.
19263 (mips_print_operand): Likewise.
19264 (mips_mode_ok_for_mov_fmt_p): Likewise.
19265 (mips_vector_mode_supported_p): Likewise.
19266 (mips_preferred_simd_mode): Likewise.
19267 (mips_expand_vpc_loongson_even_odd): Likewise.
19268 (mips_expand_vec_unpack): Likewise.
19269 (mips_expand_vi_broadcast): Likewise.
19270 (mips_expand_vector_init): Likewise.
19271 (mips_expand_vec_reduc): Likewise.
19272 (mips_expand_msa_cmp): Likewise.
19273 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
19274 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
19275 (cc_flags_for_mode): Likewise.
19276 * config/msp430/msp430.c (msp430_print_operand): Likewise.
19277 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
19278 (nds32_output_casesi_pc_relative): Likewise.
19279 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19280 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
19281 (nvptx_gen_unpack): Likewise.
19282 (nvptx_gen_pack): Likewise.
19283 (nvptx_gen_shuffle): Likewise.
19284 (nvptx_gen_wcast): Likewise.
19285 (nvptx_preferred_simd_mode): Likewise.
19286 * config/pa/pa.c (pa_secondary_reload): Likewise.
19287 * config/pa/predicates.md (base14_operand): Likewise.
19288 * config/powerpcspe/powerpcspe-c.c
19289 (altivec_resolve_overloaded_builtin): Likewise.
19290 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
19292 (rs6000_preferred_simd_mode): Likewise.
19293 (output_vec_const_move): Likewise.
19294 (rs6000_expand_vector_extract): Likewise.
19295 (rs6000_split_vec_extract_var): Likewise.
19296 (reg_offset_addressing_ok_p): Likewise.
19297 (rs6000_legitimate_offset_address_p): Likewise.
19298 (rs6000_legitimize_address): Likewise.
19299 (rs6000_emit_set_const): Likewise.
19300 (rs6000_const_vec): Likewise.
19301 (rs6000_emit_move): Likewise.
19302 (spe_build_register_parallel): Likewise.
19303 (rs6000_darwin64_record_arg_recurse): Likewise.
19304 (swap_selector_for_mode): Likewise.
19305 (spe_init_builtins): Likewise.
19306 (paired_init_builtins): Likewise.
19307 (altivec_init_builtins): Likewise.
19308 (do_load_for_compare): Likewise.
19309 (rs6000_generate_compare): Likewise.
19310 (rs6000_expand_float128_convert): Likewise.
19311 (emit_load_locked): Likewise.
19312 (emit_store_conditional): Likewise.
19313 (rs6000_output_function_epilogue): Likewise.
19314 (rs6000_handle_altivec_attribute): Likewise.
19315 (rs6000_function_value): Likewise.
19316 (emit_fusion_gpr_load): Likewise.
19317 (emit_fusion_p9_load): Likewise.
19318 (emit_fusion_p9_store): Likewise.
19319 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
19320 (fusion_gpr_mem_load): Likewise.
19321 (fusion_addis_mem_combo_load): Likewise.
19322 (fusion_addis_mem_combo_store): Likewise.
19323 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
19324 (fusion_gpr_mem_load): Likewise.
19325 (fusion_addis_mem_combo_load): Likewise.
19326 (fusion_addis_mem_combo_store): Likewise.
19327 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19329 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
19330 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19331 (rs6000_preferred_simd_mode): Likewise.
19332 (output_vec_const_move): Likewise.
19333 (rs6000_expand_vector_extract): Likewise.
19334 (rs6000_split_vec_extract_var): Likewise.
19335 (reg_offset_addressing_ok_p): Likewise.
19336 (rs6000_legitimate_offset_address_p): Likewise.
19337 (rs6000_legitimize_address): Likewise.
19338 (rs6000_emit_set_const): Likewise.
19339 (rs6000_const_vec): Likewise.
19340 (rs6000_emit_move): Likewise.
19341 (rs6000_darwin64_record_arg_recurse): Likewise.
19342 (swap_selector_for_mode): Likewise.
19343 (paired_init_builtins): Likewise.
19344 (altivec_init_builtins): Likewise.
19345 (rs6000_expand_float128_convert): Likewise.
19346 (emit_load_locked): Likewise.
19347 (emit_store_conditional): Likewise.
19348 (rs6000_output_function_epilogue): Likewise.
19349 (rs6000_handle_altivec_attribute): Likewise.
19350 (rs6000_function_value): Likewise.
19351 (emit_fusion_gpr_load): Likewise.
19352 (emit_fusion_p9_load): Likewise.
19353 (emit_fusion_p9_store): Likewise.
19354 * config/rx/rx.c (rx_gen_move_template): Likewise.
19355 (flags_from_mode): Likewise.
19356 * config/s390/predicates.md (s390_alc_comparison): Likewise.
19357 (s390_slb_comparison): Likewise.
19358 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
19359 (s390_vector_mode_supported_p): Likewise.
19360 (s390_cc_modes_compatible): Likewise.
19361 (s390_match_ccmode_set): Likewise.
19362 (s390_canonicalize_comparison): Likewise.
19363 (s390_emit_compare_and_swap): Likewise.
19364 (s390_branch_condition_mask): Likewise.
19365 (s390_rtx_costs): Likewise.
19366 (s390_secondary_reload): Likewise.
19367 (__SECONDARY_RELOAD_CASE): Likewise.
19368 (s390_expand_cs): Likewise.
19369 (s390_preferred_simd_mode): Likewise.
19370 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
19371 * config/sh/sh.c (sh_print_operand): Likewise.
19372 (dump_table): Likewise.
19373 (sh_secondary_reload): Likewise.
19374 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19375 * config/sh/sh.md (casesi_worker_1): Likewise.
19376 (casesi_worker_2): Likewise.
19377 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
19378 (fcc_comparison_operator): Likewise.
19379 * config/sparc/sparc.c (sparc_expand_move): Likewise.
19380 (emit_soft_tfmode_cvt): Likewise.
19381 (sparc_preferred_simd_mode): Likewise.
19382 (output_cbranch): Likewise.
19383 (sparc_print_operand): Likewise.
19384 (sparc_expand_vec_perm_bmask): Likewise.
19385 (vector_init_bshuffle): Likewise.
19386 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
19387 (spu_vector_mode_supported_p): Likewise.
19388 (spu_expand_insv): Likewise.
19389 (spu_emit_branch_or_set): Likewise.
19390 (spu_handle_vector_attribute): Likewise.
19391 (spu_builtin_splats): Likewise.
19392 (spu_builtin_extract): Likewise.
19393 (spu_builtin_promote): Likewise.
19394 (spu_expand_sign_extend): Likewise.
19395 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
19396 (tilegx_simd_int): Likewise.
19397 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
19398 (tilepro_simd_int): Likewise.
19399 * config/v850/v850.c (const_double_split): Likewise.
19400 (v850_print_operand): Likewise.
19401 (ep_memory_offset): Likewise.
19402 * config/vax/vax.c (vax_rtx_costs): Likewise.
19403 (vax_output_int_move): Likewise.
19404 (vax_output_int_add): Likewise.
19405 (vax_output_int_subtract): Likewise.
19406 * config/visium/predicates.md (visium_branch_operator): Likewise.
19407 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
19408 (visium_print_operand_address): Likewise.
19409 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19410 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
19411 (xtensa_expand_conditional_branch): Likewise.
19412 (xtensa_copy_incoming_a7): Likewise.
19413 (xtensa_output_literal): Likewise.
19414 * dfp.c (decimal_real_maxval): Likewise.
19415 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19417 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19418 Alan Hayward <alan.hayward@arm.com>
19419 David Sherwood <david.sherwood@arm.com>
19421 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
19422 (mode_nunits_inline): Likewise.
19423 (mode_inner_inline): Likewise.
19424 (mode_unit_size_inline): Likewise.
19425 (mode_unit_precision_inline): Likewise.
19426 (emit_insn_modes_h): Likewise. Also emit a #define of the
19428 (emit_mode_wider): Add an E_ prefix to mode names.
19429 (emit_mode_complex): Likewise.
19430 (emit_mode_inner): Likewise.
19431 (emit_mode_adjustments): Likewise.
19432 (emit_mode_int_n): Likewise.
19433 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
19434 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
19435 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
19436 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
19437 (CRC32_BUILTIN, ENTRY): Likewise.
19438 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
19439 (aarch64_pop_regs): Likewise.
19440 (aarch64_process_components): Likewise.
19441 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
19442 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
19443 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
19444 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
19445 * config/arm/arm.c (arm_init_libfuncs): Likewise.
19446 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
19448 * config/i386/i386-builtin.def (pcmpestr): Likewise.
19449 (pcmpistr): Likewise.
19450 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
19451 * config/mmix/mmix.c (mmix_output_condition): Likewise.
19452 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19454 * config/rl78/rl78.c (mduc_regs): Likewise.
19455 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19456 (htm_expand_builtin): Likewise.
19457 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
19458 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
19459 * config/xtensa/xtensa.c (print_operand): Likewise.
19460 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
19461 (NUM_MODE_VECTOR_INT): Likewise.
19462 * genoutput.c (null_operand): Likewise.
19463 (output_operand_data): Likewise.
19464 * genrecog.c (print_parameter_value): Likewise.
19465 * lra.c (debug_operand_data): Likewise.
19467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19469 * dwarf2out.c (scompare_loc_descriptor_wide)
19470 (scompare_loc_descriptor_narrow): New functions, split out from...
19471 (scompare_loc_descriptor): ...here.
19472 * expmed.c (emit_store_flag_int): New function, split out from...
19473 (emit_store_flag): ...here.
19475 2017-08-30 Richard Biener <rguenther@suse.de>
19477 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
19478 (dwarf2out_early_finish): Move setting of AT_pubnames from
19479 early debug output to early finish.
19481 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19483 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
19484 and -mdata-region to the assembler.
19486 2017-08-30 Richard Biener <rguenther@suse.de>
19488 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
19490 (gen_subprogram_die): Add DW_AT_object_pointer only early.
19491 (dwarf2out_early_global_decl): Only generate a DIE for the
19492 abstract origin if it doesn't already exist or is a declaration DIE.
19493 (resolve_addr): Do not add the linkage name twice when
19494 generating a stub DIE for the DW_TAG_GNU_call_site target.
19496 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19498 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19499 Use machine_mode rather than int for arg1_mode.
19501 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19504 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
19505 that the second argument of the built-in functions to unpack
19506 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
19507 switch statement instead a lot of if statements.
19508 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
19509 Allow 64-bit values to be in Altivec registers as well as
19510 traditional floating point registers.
19511 (pack<mode>, FMOVE128_VSX iterator): Likewise.
19513 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
19515 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
19516 MAX_REGS_PER_ADDRESS == 1.
19518 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
19520 * config/i386/i386.opt (flag_fentry): Do not init to -1.
19521 * config/i386/i386.c (ix86_option_override_internal): Simplify
19522 setting of opts->x_flag_entry.
19524 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19525 Jakub Jelinek <jakub@redhat.com>
19526 Richard Biener <rguenther@suse.de>
19528 PR tree-optimization/81503
19529 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
19530 folded constant fits in the target type; reorder tests for clarity.
19532 2017-08-29 Martin Liska <mliska@suse.cz>
19534 * passes.def: Include pass_lower_switch.
19535 * stmt.c (dump_case_nodes): Remove and move to
19536 tree-switch-conversion.
19537 (case_values_threshold): Likewise.
19538 (expand_switch_as_decision_tree_p): Likewise.
19539 (emit_case_decision_tree): Likewise.
19540 (expand_case): Likewise.
19541 (balance_case_nodes): Likewise.
19542 (node_has_low_bound): Likewise.
19543 (node_has_high_bound): Likewise.
19544 (node_is_bounded): Likewise.
19545 (emit_case_nodes): Likewise.
19546 (struct simple_case_node): New struct.
19547 (add_case_node): Remove.
19548 (emit_case_dispatch_table): Use vector instead of case_list.
19549 (reset_out_edges_aux): Remove.
19550 (compute_cases_per_edge): Likewise.
19551 (expand_case): Build list of simple_case_node.
19552 (expand_sjlj_dispatch_table): Use it.
19553 * tree-switch-conversion.c (struct case_node): Moved from
19554 stmt.c and adjusted.
19555 (emit_case_nodes): Likewise.
19556 (node_has_low_bound): Likewise.
19557 (node_has_high_bound): Likewise.
19558 (node_is_bounded): Likewise.
19559 (case_values_threshold): Likewise.
19560 (reset_out_edges_aux): Likewise.
19561 (compute_cases_per_edge): Likewise.
19562 (add_case_node): Likewise.
19563 (dump_case_nodes): Likewise.
19564 (balance_case_nodes): Likewise.
19565 (expand_switch_as_decision_tree_p): Likewise.
19566 (emit_jump): Likewise.
19567 (emit_case_decision_tree): Likewise.
19568 (try_switch_expansion): Likewise.
19569 (do_jump_if_equal): Likewise.
19570 (emit_cmp_and_jump_insns): Likewise.
19571 (fix_phi_operands_for_edge): New function.
19572 (record_phi_operand_mapping): Likewise.
19573 (class pass_lower_switch): New pass.
19574 (pass_lower_switch::execute): New function.
19575 (make_pass_lower_switch): Likewise.
19576 (conditional_probability):
19577 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19578 * tree-pass.h: Add make_pass_lower_switch.
19580 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19583 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19586 2017-08-29 Richard Biener <rguenther@suse.de>
19588 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19589 we do not add a DW_AT_inline attribute twice.
19590 (gen_subprogram_die): Remove code setting DW_AT_inline on
19591 DECL_ABSTRACT_P nodes.
19593 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19595 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19596 calls to internal functions.
19597 (gimplify_modify_expr): Likewise.
19598 * tree-call-cdce.c (use_internal_fn): Likewise.
19599 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19600 (convert_to_divmod): Set the nothrow flag.
19601 * tree-if-conv.c (predicate_mem_writes): Likewise.
19602 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19603 (vectorizable_call): Likewise.
19604 (vectorizable_store): Likewise.
19605 (vectorizable_load): Likewise.
19606 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19607 (vect_recog_mask_conversion_pattern): Likewise.
19609 2017-08-29 Martin Liska <mliska@suse.cz>
19612 * gcc.c (driver_handle_option): Add new argument.
19613 * opts-common.c (handle_option): Pass
19614 target_option_override_hook.
19615 * opts-global.c (lang_handle_option): Add new option.
19616 (set_default_handlers): Add new argument.
19617 (decode_options): Likewise.
19618 * opts.c (target_handle_option): Likewise.
19619 (common_handle_option): Call target_option_override_hook.
19620 * opts.h (struct cl_option_handler_func): Add hook for
19621 target option override.
19622 (struct cl_option_handlers): Likewise.
19623 (set_default_handlers): Add new argument.
19624 (decode_options): Likewise.
19625 (common_handle_option): Likewise.
19626 (target_handle_option): Likewise.
19627 * toplev.c (toplev::main): Pass targetm.target_option.override
19630 2017-08-29 Richard Biener <rguenther@suse.de>
19631 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19633 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19634 life to the active subtree.
19636 2017-08-28 Jeff Law <law@redhat.com>
19638 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19639 derive_equivalences.
19640 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19641 Code moved into....
19642 (edge_info::derive_equivalences): New private member function
19644 * tree-ssa-dom.c (class edge_info): Changed from a struct
19645 to a class. Add ctor/dtor, methods and data members.
19646 (edge_info::edge_info): Renamed from allocate_edge_info.
19647 Initialize additional members.
19648 (edge_info::~edge_info): New.
19649 (free_dom_edge_info): Delete the edge info.
19650 (record_edge_info): Use new class & associated member functions.
19651 Tighten forms for testing for edge equivalences.
19652 (record_temporary_equivalences): Iterate over the simple
19653 equivalences rather than assuming there's only one per edge.
19654 (cprop_into_successor_phis): Iterate over the simple
19655 equivalences rather than assuming there's only one per edge.
19656 (optimize_stmt): Use operand_equal_p rather than pointer
19657 equality for mini-DSE code.
19659 2017-08-28 Nathan Sidwell <nathan@acm.org>
19661 * gcc.c (execute): Fold SIGPIPE handling into switch
19662 statement. Adjust internal error message.
19664 2017-08-28 Richard Biener <rguenther@suse.de>
19667 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19668 Do nothing for removed DIEs.
19670 2017-08-28 Richard Biener <rguenther@suse.de>
19672 PR tree-optimization/81977
19673 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19676 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19679 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19680 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19681 using targetm.gen_mem_thread_fence.
19683 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19686 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19687 predicate to register_operand. Reorder operands.
19688 (*btr<mode>): Ditto.
19689 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19690 (*btr<mode>_mask): Ditto.
19692 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19694 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19695 * config/rs6000/xmmintrin.h: New file.
19696 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19698 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19701 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19702 parameter and_insn and return it.
19703 (recombine_lvx_pattern): Insert a copy to ensure availability of
19704 the base register of the copied masking operation at the point of
19705 the instruction replacement.
19706 (recombine_stvx_pattern): Likewise.
19708 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19710 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19711 undocumented switches.
19712 (-mpower9-dform-vector): Likewise.
19713 (-mpower9-dform): Likewise.
19714 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19715 comments to delete references to -mpower9-dform* switches.
19716 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19717 Delete reference to -mpower9-dform* switches, test for
19718 -mpower9-vector instead.
19719 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19720 (OTHER_P9_VECTOR_MASKS): Likewise.
19721 (POWERPC_MASKS): Likewise.
19722 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19723 tests against -mpower9-dform* to -mpower9-vector. Delete code
19724 that checked for -mpower9-dform* consistancy with other options.
19725 Add test for -mpower9-misc to enable other power9 switches.
19726 (rs6000_init_hard_regno_mode_ok): Likewise.
19727 (rs6000_option_override_internal): Likewise.
19728 (rs6000_emit_prologue): Likewise.
19729 (rs6000_emit_epilogue): Likewise.
19730 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19731 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19732 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19734 (emit_fusion_p9_store): Likewise.
19735 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19736 resetting these macros if the assembler does not support ISA 3.0
19738 (TARGET_P9_DFORM_VECTOR): Likewise.
19739 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19740 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19742 2017-08-25 Alan Modra <amodra@gmail.com>
19744 PR rtl-optimization/81747
19745 * cse.c (cse_extended_basic_block): Don't attempt to record
19746 equivalences for degenerate conditional branches that branch
19747 to their fall-through.
19749 2017-08-24 Martin Sebor <msebor@redhat.com>
19751 PR middle-end/81908
19752 * gimple-fold.c (size_must_be_zero_p): New function.
19753 (gimple_fold_builtin_memory_op): Call it.
19755 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19757 * config/rs6000/mm_malloc.h: New file.
19759 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19761 PR tree-optimization/81913
19762 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19763 analysis when either IVs in condition can wrap.
19765 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19767 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19768 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19770 2017-08-24 Richard Biener <rguenther@suse.de>
19773 * targhooks.c (default_target_can_inline_p): Properly
19774 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19775 is present and always compare.
19776 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19777 infer -mfpmath=sse from TARGET_SSE_P.
19778 (ix86_can_inline_p): Properly use target_option_default_node when
19779 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19781 2017-08-24 Richard Biener <rguenther@suse.de>
19784 * dwarf2out.c (output_die): Handle flag_generate_offload like
19786 (output_comp_unit): Likewise.
19787 (gen_array_type_die): Likewise.
19788 (dwarf2out_early_finish): Likewise.
19789 (note_variable_value_in_expr): Likewise.
19790 (dwarf2out_finish): Likewise. Adjust assert.
19791 * cgraphunit.c (symbol_table::compile): Move setting of
19792 flag_generate_offload earlier ...
19793 (symbol_table::finalize_compilation_unit): ... here, before
19794 early debug finalization.
19796 2017-08-24 Richard Biener <rguenther@suse.de>
19798 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19799 and ipa-fnsummary.h.
19800 (ix86_can_inline_p): When ix86_fpmath flags do not match
19801 check whether the callee uses FP math at all.
19803 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19805 PR middle-end/81931
19806 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19807 instead of TYPE_PRECISION.
19809 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19810 Alan Hayward <alan.hayward@arm.com>
19811 David Sherwood <david.sherwood@arm.com>
19813 * combine.c (make_extraction): Use subreg_offset_from_lsb.
19815 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
19817 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
19819 (ix86_frame::stack_realign_allocate): New field.
19820 (struct machine_frame_state): Modify comments.
19821 (machine_frame_state::sp_realigned_fp_end): New field.
19822 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
19823 layout calculation.
19824 (sp_valid_at): Add assertion to assure no attempt to access invalid
19825 offset of a realigned stack.
19826 (fp_valid_at): Likewise.
19827 (choose_baseaddr): Modify comments.
19828 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
19829 ix86_expand_prologue.
19830 (ix86_expand_prologue): Modify stack realignment and allocation.
19831 (ix86_expand_epilogue): Modify comments.
19832 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
19833 avx2_runtime, avx512f, and avx512f_runtime.
19835 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
19837 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
19838 (mstackrealign): Do not init to -1.
19839 * config/i386/i386.c (ix86_option_override_internal):
19840 Check opts_set, not opts when setting default value of
19841 opts->x_ix86_force_align_arg_pointer.
19843 2017-08-23 Richard Biener <rguenther@suse.de>
19845 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
19846 lang_hooks.decl_printable_name.
19847 * print-rtl-function.c (print_rtx_function): Likewise.
19848 * tree-pretty-print.c (dump_function_header): Likewise.
19850 2017-08-23 Richard Biener <rguenther@suse.de>
19853 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
19854 -g0 at compile-time.
19856 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19858 PR middle-end/19706
19859 * doc/sourcebuild.texi (Other hardware attributes):
19862 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19864 PR middle-end/19706
19865 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
19866 Fix single-use check.
19868 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
19870 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
19872 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19874 * doc/install.texi: Modify to add more details on running selected
19877 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19879 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
19880 is combined with -mabi=ms.
19881 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
19884 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19886 PR tree-optimization/81488
19887 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
19888 and cached_basis fields.
19889 (MAX_SPREAD): New constant.
19890 (alloc_cand_and_find_basis): Initialize new fields.
19891 (clear_visited): New function.
19892 (create_phi_basis_1): Rename from create_phi_basis, set visited
19893 and cached_basis fields.
19894 (create_phi_basis): New wrapper function.
19895 (phi_add_costs_1): Rename from phi_add_costs, add spread
19896 parameter, set visited field, short-circuit when limits reached.
19897 (phi_add_costs): New wrapper function.
19898 (record_phi_increments_1): Rename from record_phi_increments, set
19900 (record_phi_increments): New wrapper function.
19901 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
19902 (phi_incr_cost): New wrapper function.
19903 (all_phi_incrs_profitable_1): Rename from
19904 all_phi_incrs_profitable, set visited field.
19905 (all_phi_incrs_profitable): New wrapper function.
19907 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19908 Alan Hayward <alan.hayward@arm.com>
19909 David Sherwood <david.sherwood@arm.com>
19911 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
19912 that takes the outer and inner modes.
19913 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
19914 comparison as the canonical test for a paradoxical subreg.
19915 * combine.c (simplify_set): Use paradoxical_subreg_p.
19916 (make_extraction): Likewise.
19917 (force_to_mode): Likewise.
19918 (rtx_equal_for_field_assignment_p): Likewise.
19919 (gen_lowpart_for_combine): Likewise.
19920 (simplify_comparison): Likewise.
19921 * cse.c (equiv_constant): Likewise.
19922 * expmed.c (store_bit_field_1): Likewise.
19923 * final.c (alter_subreg): Likewise.
19924 * fwprop.c (propagate_rtx): Likewise.
19925 (forward_propagate_subreg): Likewise.
19926 * ira-conflicts.c (ira_build_conflicts): Likewise.
19927 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
19928 * lra-constraints.c (curr_insn_transform): Likewise.
19929 (split_reg): Likewise.
19930 * lra-eliminations.c (move_plus_up): Likewise.
19931 (lra_eliminate_regs_1): Likewise.
19932 * recog.c (general_operand): Likewise.
19933 * ree.c (combine_reaching_defs): Likewise.
19934 * reload.c (push_reload): Likewise.
19935 (find_reloads): Likewise.
19936 * reload1.c (elimination_effects): Likewise.
19937 (compute_reload_subreg_offset): Likewise.
19938 (choose_reload_regs): Likewise.
19939 * rtlanal.c (subreg_lsb_1): Likewise.
19940 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19941 (simplify_subreg): Likewise.
19942 * var-tracking.c (track_loc_p): Likewise.
19943 * emit-rtl.c (byte_lowpart_offset): Likewise.
19944 (paradoxical_subreg_p): Delete out-of-line definition.
19946 2017-08-22 Jeff Law <law@redhat.com>
19948 PR tree-optimization/81741
19949 PR tree-optimization/71947
19950 * tree-ssa-dom.c: Include tree-inline.h.
19951 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
19952 equivalences if one is more expensive to compute than the other.
19953 * tree-ssa-scopedtables.h (class const_or_copies): Make
19954 record_const_or_copy_raw method private.
19955 (class avail_exprs_stack): New method simplify_binary_operation.
19956 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
19957 avail_exprs_stack::simplify_binary_operation as needed.
19958 (avail_exprs_stack::simplify_binary_operation): New function.
19960 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
19962 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
19963 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
19964 (DOT_SYMBOLS): Likewise.
19965 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
19966 (RELOCATABLE_NEEDS_FIXUP): Likewise.
19967 (RS6000_ABI_NAME): Likewise.
19968 (TARGET_CMODEL): Likewise.
19969 (TOC_SECTION_ASM_OP): Likewise.
19970 (SET_CMODEL): New macro.
19971 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
19973 2017-08-22 Richard Biener <rguenther@suse.de>
19975 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
19976 to allow for free-lang-data replacements similar to verify_type_variant.
19978 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19979 Alan Hayward <alan.hayward@arm.com>
19980 David Sherwood <david.sherwood@arm.com>
19982 * config/aarch64/aarch64.md (casesi): Use DImode rather than
19983 VOIDmode for the LABEL_REF.
19985 2017-08-22 Richard Biener <rguenther@suse.de>
19987 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
19989 2017-08-22 Richard Biener <rguenther@suse.de>
19991 * common.opt (feliminate-dwarf2-dups): Ignore.
19992 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
19993 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
19994 same_die_p_wrap, compute_section_prefix,
19995 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
19996 (comdat_symbol_id, comdat_symbol_number): Likewise.
19997 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
19999 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20000 (output_die): Mark unreachable path unreachable.
20001 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20002 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20003 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20004 (dwarf2out_early_finish): Likewise.
20006 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20008 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20010 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20013 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20014 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20015 (avr_attribute_table) <io, io_low, address>: Initialize
20016 .decl_required with true.
20018 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20020 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20021 undocumented debugging options.
20022 (-mvsx-scalar-double): Likewise.
20023 (-mallow-df-permute): Likewise.
20024 (-mvectorize-builtins): Likewise.
20025 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20026 (rs6000_builtin_vectorized_function): Likewise.
20027 (rs6000_builtin_md_vectorized_function): Likewise.
20028 (rs6000_opt_vars): Likewise.
20030 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20033 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20034 (*btrq_imm): Rename from *btrq.
20035 (*btcq_imm): Rename from *btcq.
20036 (btsc): New code attribute.
20037 (*<btsc><mode>): New insn pattern.
20038 (*btr<mode>): Ditto.
20039 (*<btsc><mode>_mask): New insn_and_split pattern.
20040 (*btr<mode>_mask): Ditto.
20042 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20043 Alan Hayward <alan.hayward@arm.com>
20044 David Sherwood <david.sherwood@arm.com>
20046 * function.c (pad_below): Simplify padding calculation.
20048 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20049 Alan Hayward <alan.hayward@arm.com>
20050 David Sherwood <david.sherwood@arm.com>
20052 * target.def (function_prologue): Remove frame size argument.
20053 (function_epilogue): Likewise.
20054 * doc/tm.texi: Regenerate.
20055 * final.c (final_start_function): Update call to function_prologue.
20056 (final_end_function): Update call to function_epilogue.
20057 (default_function_pro_epilogue): Remove frame size argument.
20058 * output.h (default_function_pro_epilogue): Likewise.
20059 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
20060 (arm_output_function_prologue): Likewise.
20061 * config/frv/frv.c (frv_function_prologue): Likewise.
20062 (frv_function_epilogue): Likewise.
20063 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
20064 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
20065 (ia64_output_function_epilogue): Likewise.
20066 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
20067 (m32r_output_function_epilogue): Likewise.
20068 * config/microblaze/microblaze.c (microblaze_function_prologue)
20069 (microblaze_function_epilogue): Likewise.
20070 * config/mips/mips.c (mips_output_function_prologue): Likewise.
20071 (mips_output_function_epilogue): Likewise.
20072 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
20073 (mmix_target_asm_function_epilogue): Likewise.
20074 * config/msp430/msp430.c (msp430_start_function): Likewise.
20075 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
20076 (nds32_asm_function_epilogue): Likewise.
20077 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
20078 * config/pa/pa.c (pa_output_function_prologue): Likewise.
20079 (pa_output_function_epilogue): Likewise.
20080 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
20081 (rs6000_output_function_epilogue): Likewise.
20082 * config/rl78/rl78.c (rl78_start_function): Likewise.
20083 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
20084 (rs6000_output_function_epilogue): Likewise.
20085 * config/rx/rx.c (rx_output_function_prologue): Likewise.
20086 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
20087 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
20088 (sparc_asm_function_epilogue): Likewise.
20090 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20092 * tree.h (type_has_mode_precision_p): New function.
20093 * convert.c (convert_to_integer_1): Use it.
20094 * expr.c (expand_expr_real_2): Likewise.
20095 (expand_expr_real_1): Likewise.
20096 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
20097 * match.pd: Likewise.
20098 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
20099 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
20100 * tree-tailcall.c (process_assignment): Likewise.
20101 * tree-vect-loop.c (vectorizable_reduction): Likewise.
20102 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
20103 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
20104 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
20105 (vectorizable_assignment): Likewise.
20106 (vectorizable_shift): Likewise.
20107 (vectorizable_operation): Likewise.
20108 * tree-vrp.c (register_edge_assert_for_2): Likewise.
20110 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
20112 * match.pd: Add pow (C, x) simplification.
20114 2017-08-21 Richard Biener <rguenther@suse.de>
20116 PR tree-optimization/81900
20117 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
20118 for blocks with abnormal predecessors.
20119 (compute_antic): Do not set visited flag prematurely.
20121 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
20124 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
20126 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20128 * stor-layout.h (vector_type_mode): Move to...
20129 * tree.h (vector_type_mode): ...here.
20130 * stor-layout.c (vector_type_mode): Move to...
20131 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
20133 2017-08-21 Richard Biener <rguenther@suse.de>
20135 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
20136 register_external_die hooks.
20137 (debug_false_tree_charstarstar_uhwistar): Declare.
20138 (debug_nothing_tree_charstar_uhwi): Likewise.
20139 * debug.c (do_nothing_debug_hooks): Adjust.
20140 (debug_false_tree_charstarstar_uhwistar): New do nothing.
20141 (debug_nothing_tree_charstar_uhwi): Likewise.
20142 * dbxout.c (dbx_debug_hooks): Adjust.
20143 (xcoff_debug_hooks): Likewise.
20144 * sdbout.c (sdb_debug_hooks): Likewise.
20145 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20146 * dwarf2out.c (macinfo_label_base): New global.
20147 (dwarf2out_register_external_die): New function for the
20148 register_external_die hook.
20149 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
20150 (dwarf2_debug_hooks): Use them.
20151 (dwarf2_lineno_debug_hooks): Adjust.
20152 (struct die_struct): Add with_offset flag.
20153 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
20154 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
20155 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
20156 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
20157 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
20158 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
20159 defining section names for the early LTO debug variants.
20160 (reset_indirect_string): New helper.
20161 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
20162 (print_dw_val): Add support for offsetted symbol references.
20163 (get_ultimate_context): Split out from is_cxx.
20164 (is_cxx): Use get_ultimate_context.
20165 (is_fortran): Add decl overload.
20166 (compute_comp_unit_symbol): Split out worker from
20167 compute_section_prefix.
20168 (compute_section_prefix): Call compute_comp_unit_symbol and
20169 set comdat_type_p here.
20170 (output_die): Skip DIE symbol output for the LTO added one.
20171 Handle DIE symbol references with offset.
20172 (output_comp_unit): Guard section name mangling properly.
20173 For LTO debug sections emit a symbol at the section beginning
20174 which we use to refer to its DIEs.
20175 (add_abstract_origin_attribute): For DIEs registered via
20176 dwarf2out_register_external_die directly refer to the early
20177 DIE rather than indirectly through the shadow one we created.
20178 Remove obsolete call to dwarf2out_abstract_function for
20179 non-function/block origins.
20180 (gen_array_type_die): When generating early LTO debug do
20181 not emit DW_AT_string_length.
20182 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
20183 late when in LTO. As suggested place a gcc_unreachable for
20184 the DECL_ABSTRACT_P case.
20185 (gen_subprogram_die): Avoid another specification DIE
20186 for early built declarations/definitions for the late LTO case.
20187 (gen_variable_die): Add type references for late duplicated VLA dies
20189 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
20190 we have the abstract instance already.
20191 (process_scope_var): Adjust decl DIE contexts in LTO which
20192 first puts them in limbo.
20193 (gen_decl_die): Do not generate type DIEs late apart from
20194 types for VLAs or for decls we do not yet have a DIE. Do not
20195 call dwarf2out_abstract_function late.
20196 (dwarf2out_early_global_decl): Make sure to create DIEs
20197 for abstract instances of a decl first.
20198 (dwarf2out_late_global_decl): Adjust comment.
20199 (output_macinfo_op): With multiple macro sections use
20200 macinfo_label_base to distinguish labels.
20201 (output_macinfo): Likewise. Update macinfo_label_base.
20202 Pass in the line info label.
20203 (note_variable_value_in_expr): When generating LTO resolve
20204 all variable values here by generating DIEs as needed.
20205 (init_sections_and_labels): Add early LTO debug flag parameter
20206 and generate different sections and names if set. Add generation
20207 counter for the labels so we can have multiple of them.
20208 (reset_dies): Helper to allow DIEs to be output multiple times.
20209 (dwarf2out_finish): When outputting DIEs to the fat part of an
20210 LTO object first reset DIEs.
20211 (dwarf2out_early_finish): Output early DIEs when generating LTO.
20212 (modified_type_die): Check for decl_ultimate_origin being self
20214 (gen_type_die_with_usage): Likewise.
20215 (gen_typedef_die): Allow decl_ultimate_origin being self.
20216 (set_decl_abstract_flags): Remove.
20217 (set_block_abstract_flags): Likewise.
20218 (dwarf2out_abstract_function): Treat the early generated DIEs
20219 as the abstract copy and only add DW_AT_inline and
20220 DW_AT_artificial here and call set_decl_origin_self.
20221 If the DIE has an abstract origin don't do anything.
20222 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
20223 if we have none yet (Go fails to build one, PR78628).
20224 (variably_modified_type_p): Prevent endless recursion for Ada
20225 cyclic pointer types.
20226 * lto-streamer-in.c: Include debug.h.
20227 (dref_queue): New global.
20228 (lto_read_tree_1): Stream in DIE references.
20229 (lto_input_tree): Register DIE references.
20230 (input_function): Stream DECL_DEBUG_ARGS.
20231 * lto-streamer-out.c: Include debug.h.
20232 (lto_write_tree_1): Output DIE references.
20233 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
20234 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
20235 (output_function): Stream DECL_DEBUG_ARGS.
20236 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
20237 Stream DECL_ABSTRACT_ORIGIN.
20238 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
20239 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
20240 DECL_CONTEXT for file-scope decls.
20241 * lto-streamer.h (struct dref_entry): Declare.
20242 (dref_queue): Likewise.
20243 * cfgexpand.c (pass_expand::execute): Do not call the
20244 outlining_inline_function hook here.
20245 * lto-wrapper.c (debug_obj): New global.
20246 (tool_cleanup): Unlink it if required.
20247 (debug_objcopy): New function.
20248 (run_gcc): Handle early debug sections in the IL files by
20249 extracting them to separate files, partially linkin them and
20250 feeding the result back as result to the linker.
20251 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
20252 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
20253 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
20254 sections into a separate segment.
20255 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
20257 (darwin_asm_dwarf_section): Likewise.
20258 (darwin_asm_output_dwarf_offset): Likewise.
20259 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
20261 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20262 Alan Hayward <alan.hayward@arm.com>
20263 David Sherwood <david.sherwood@arm.com>
20265 * read-md.h (md_reader::record_potential_iterator_use): Replace
20266 pointer argument with an rtx and an index.
20267 * read-rtl.c (iterator_group::apply_iterator): Likewise.
20268 (apply_mode_iterator): Likewise.
20269 (apply_code_iterator): Likewise.
20270 (apply_int_iterator): Likewise.
20271 (apply_subst_iterator): Likewise.
20272 (record_iterator_use): Likewise.
20273 (record_attribute_use): Likewise.
20274 (md_reader::record_potential_iterator_use): Likewise. Update calls
20275 to record_iterator_use and apply_iterator.
20276 (iterator_use): Replace ptr with x and index.
20277 (attribute_use): Likewise.
20278 (apply_attribute_uses): Update calls to apply_iterator.
20279 (apply_iterators): Likewise. Update initialization of iterator_use.
20280 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
20281 and record_potential_iterator_use.
20282 (rtx_reader::read_rtx_operand): Likewise.
20284 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20285 Alan Hayward <alan.hayward@arm.com>
20286 David Sherwood <david.sherwood@arm.com>
20288 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
20291 2017-08-21 Richard Biener <rguenther@suse.de>
20293 PR middle-end/81884
20294 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
20295 at struct end conservatively when comparing common bases.
20297 2017-08-21 Richard Biener <rguenther@suse.de>
20299 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
20300 (mem_ref_in_stmt): Remove.
20301 (determine_max_movement): Use ref index to get at the reference.
20302 (invariantness_dom_walker::before_dom_children): Deal with
20303 lim data already initialized.
20304 (gather_mem_refs_stmt): Initialize lim data and record ref index.
20306 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20308 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
20309 (TARGET_ISA_ROUND): Ditto.
20310 (TARGET_ROUND): Ditto.
20311 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
20312 * config/i386/i386.md: Ditto.
20313 * config/i386/sse.md: Ditto.
20314 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
20315 with OPTION_MASK_ISA_SSE4_1.
20317 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20320 * doc/extend.texi (x86 Built-in Functions): Correct the name of
20321 __builtin_ia32_lzcnt_u16.
20323 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
20326 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
20327 (rs6000_set_current_function): Rewrite function to use it.
20329 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
20332 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
20333 and TYPE_WARN_IF_NOT_ALIGN.
20334 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
20335 (handle_warn_if_not_align): New.
20336 (place_union_field): Call handle_warn_if_not_align.
20337 (place_field): Call handle_warn_if_not_align.
20338 Copy TYPE_WARN_IF_NOT_ALIGN.
20339 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
20340 (layout_type): Likewise.
20341 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
20343 (tree_decl_common): Add warn_if_not_align.
20344 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
20345 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
20346 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
20347 (DECL_WARN_IF_NOT_ALIGN): Likewise.
20348 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
20349 * doc/extend.texi: Document warn_if_not_aligned attribute.
20350 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
20352 2017-08-17 Martin Liska <mliska@suse.cz>
20355 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
20356 (get_data_dependence): Use it as pointer type.
20357 (distribute_loop): Likewise.
20359 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20361 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
20362 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
20363 (p8_vmrgow_<mode>_direct): New define_insn.
20364 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
20365 handle endianness for vmrgew and vmrgow permute patterns.
20367 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20369 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
20370 * config/rs6000/rs6000-cpus.def: Remove comment.
20371 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
20372 (POWERPC_MASKS): Likewise.
20373 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
20374 use of TARGET_VSX_TIMODE.
20375 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
20376 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
20377 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
20378 (rs6000_option_override_internal): Remove dead code.
20379 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
20380 (rs6000_legitimize_reload_address): Likewise.
20381 (rs6000_legitimate_address_p): Likewise.
20382 (rs6000_opt_masks): Delete "vsx-timode".
20383 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
20384 from function comment.
20385 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
20386 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
20387 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
20389 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
20390 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
20391 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
20392 (VSX_TI): Likewise.
20394 (define_peephole2): Likewise.
20396 2017-08-17 Martin Sebor <msebor@redhat.com>
20399 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
20400 past the end of an array.
20401 (test_pp_format): Add test cases.
20403 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
20405 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
20406 missing ECF_NOTHROW flags.
20408 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20411 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
20412 operands residing in integer registers.
20413 (*vsx_le_perm_load_<mode>): Likewise.
20414 (*vsx_le_perm_store_<mode>): Likewise.
20415 (define_peephole2): Add peepholes to optimize the above.
20417 2017-08-17 Marek Polacek <polacek@redhat.com>
20419 PR middle-end/81814
20420 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
20421 to mimic what shorten_compare did. Change the return type to bool.
20422 (fold_cond_expr_with_comparison): Update call to
20423 operand_equal_for_comparison_p.
20424 (fold_ternary_loc): Likewise.
20426 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
20428 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
20430 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
20432 2017-08-17 Richard Biener <rguenther@suse.de>
20434 * tree-ssa-structalias.c (solve_graph): When propagating
20435 to successors update the graphs succ edges and avoid duplicate work.
20437 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
20440 * config/i386/i386.c (ix86_option_override_internal): Save target
20441 specific options after ix86_stack_protector_guard_reg was changed.
20443 2017-08-17 Richard Biener <rguenther@suse.de>
20445 PR tree-optimization/81827
20446 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
20447 (new_var_info): Initialize it conservatively.
20448 (get_call_vi): Mark register vars.
20449 (new_scalar_tmp_constraint_exp): Likewise.
20450 (handle_rhs_call): Likewise.
20451 (handle_const_call): Likewise.
20452 (create_function_info_for): Likewise.
20453 (solve_constraints): Sort varinfos to separate register from
20454 non-register vars to pack points-to solution bitmaps during
20457 2017-08-17 Marek Polacek <polacek@redhat.com>
20459 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
20461 2017-08-17 Richard Biener <rguenther@suse.de>
20463 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
20464 to true when overflow is undefined and we saturated the result.
20466 2017-08-17 Alan Modra <amodra@gmail.com>
20469 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
20470 Don't use store multiple if only one reg needs saving.
20471 (interesting_frame_related_regno): New function.
20472 (rs6000_frame_related): Don't emit frame info for regs that
20474 (rs6000_emit_epilogue): Likewise.
20476 2017-08-16 Nathan Sidwell <nathan@acm.org>
20478 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
20479 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
20480 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
20481 * tree.c (free_lang_data_in_type): Use else-if chain. Always
20482 clear TYPE_LANG_1. Remove obsolete member-function stripping.
20483 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
20484 (verify_type): Adjust for TYPE_BINFO move.
20485 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
20486 process TYPE_BINFO directly.
20487 (hash_tree): Likewise.
20488 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20490 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20493 2017-08-16 David Malcolm <dmalcolm@redhat.com>
20495 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
20497 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
20500 * config/i386/i386.md (*anddi_1_btr): Change predicates of
20501 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
20502 Add ix86_binary_operator_ok to insn constraint.
20503 (*iordi_1_bts): Ditto.
20504 (*xordi_1_btc): Ditto.
20505 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
20506 Update corresponding peephole2 pattern.
20510 2017-08-16 Bin Cheng <bin.cheng@arm.com>
20512 PR tree-optimization/81832
20513 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
20514 copy loop header which has IFN_LOOP_DIST_ALIAS call.
20516 2017-08-16 Marek Polacek <polacek@redhat.com>
20519 * fold-const.c (fold_indirect_ref_1): Restore original behavior
20520 regarding size_zero_node.
20522 2017-08-16 Martin Liska <mliska@suse.cz>
20525 * config.gcc: Respect previously set extra_objs in case
20528 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
20530 PR tree-optimization/81835
20531 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
20532 the phi SSA_NAME. Check that the condition in a COND_EXPR does
20533 not depend on the phi.
20535 2017-08-16 Alan Modra <amodra@gmail.com>
20537 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
20540 2017-08-16 Alan Modra <amodra@gmail.com>
20542 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
20543 (save_reg_p): ..into this. Update all callers.
20544 (first_reg_to_save): Simplify.
20546 2017-08-16 Alan Modra <amodra@gmail.com>
20548 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
20551 2017-08-15 Joseph Myers <joseph@codesourcery.com>
20555 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
20556 constant count if that count is less than 32.
20558 2017-08-15 Nathan Sidwell <nathan@acm.org>
20560 * gcc.c (execute): Emit friendlier message if inferior is killed
20561 by an external cause.
20563 2017-08-15 Richard Biener <rguenther@suse.de>
20565 PR tree-optimization/81790
20566 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20567 CONSTRUCTORs from simplifying and VN.
20569 2017-08-14 Martin Sebor <msebor@redhat.com>
20571 * builtin-attrs.def: Add comments.
20573 2017-08-14 Martin Sebor <msebor@redhat.com>
20576 * doc/extend.texi (attribute nonstring): Document new attribute.
20578 2017-08-14 Martin Sebor <msebor@redhat.com>
20581 * tree-diagnostic.c (default_tree_printer): Handle %G.
20582 * gimple-pretty-print.h (percent_G_format): Declare new function.
20583 * gimple-pretty-print.c (percent_G_format): Define.
20584 * tree-pretty-print.c (percent_K_format): Add argument.
20586 2017-08-14 Martin Sebor <msebor@redhat.com>
20588 PR translation/79998
20589 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20590 Remove a stray space.
20592 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20595 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20596 (*iordi_1_bts): Ditto.
20597 (*xordi_1_btc): Ditto.
20599 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20602 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20603 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20605 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20606 quoted strings, and make more translator-friendly.
20607 (darwin_rs6000_override_options): Likewise.
20608 (rs6000_option_override_internal): Likewise.
20609 (rs6000_return_in_memory): Fix overlong line.
20610 (init_cmulative_args): Use quoted strings, and make more
20611 translator-friendly.
20612 (rs6000_pass_by_reference): Fix overlong line.
20613 (def_builtin): Use quoted strings.
20614 (altivec_expand_predicate_builtin): Use quoted strings, and make
20615 more translator-friendly.
20616 (htm_expand_builtin): Use quoted strings.
20617 (cpu_expand_builtin): Use quoted strings, and make more
20618 translator-friendly.
20619 (altivec_expand_builtin): Likewise.
20620 (paired_expand_predicate_builtin): Likewise.
20621 (rs6000_invalid_builtin): Likewise.
20622 (builtin_function_type): Use quoted strings.
20623 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20624 more translator-friendly.
20625 (rs6000_trampoline_init): Likewise.
20626 (rs6000_handle_altivec_attribute): Likewise.
20627 (rs6000_inner_target_options): Use quoted strings.
20628 (rs6000_disable_incompatible_switches): Likewise.
20629 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20630 strings, and make more translator-friendly.
20631 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20633 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20635 PR tree-optimization/81799
20636 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20637 cond_expr to simple gimple operand.
20639 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20641 PR middle-end/46932
20642 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20644 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20648 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20649 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20650 TARGET_GASISR_PROLOGUES.
20651 * config/avr/avr.c (avr_option_override): Same.
20652 (avr_pass_pre_proep::execute): Same.
20654 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20657 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20658 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20660 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20662 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20663 define_insn_and_split. Split to a memory load from 0 in
20664 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20665 using PTR mode iterator.
20666 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20667 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20668 (*add_tp_<mode>): Redefine as define_insn_and_split.
20669 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20670 address space. Merge with *add_tp_x32 using PTR mode iterator.
20671 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20672 Split to an add with a memory load from 0 in
20673 DEFAULT_TLS_SEG_REG address space.
20675 2017-08-12 Andrew Pinski <apinski@cavium.com>
20677 * config/aarch64/aarch64-option-extensions.def (rdma):
20678 Fix feature string to what Linux prints out in /proc/cpuinfo.
20680 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20683 * dwarf2out.c (modified_type_die): For C typedef types that have
20684 an ultimate origin, process the ultimate origin instead of the
20686 (gen_typedef_die): Assert that input DECLs have no ultimate
20688 (gen_type_die_with_usage): For typedef variants that have an
20689 ultimate origin, just call gen_decl_die on the original DECL.
20690 (process_scope_var): Avoid creating DIEs for local typedefs and
20691 concrete static variables.
20693 2017-08-12 Alan Modra <amodra@gmail.com>
20697 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20698 match gnu-user.h startfile.
20699 (ENDFILE_LINUX_SPEC): Similarly.
20701 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20704 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20706 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20708 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20709 * config/aarch64/aarch64.md (mov<mode>): Change.
20710 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20711 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20712 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20714 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20716 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20717 for storage order barriers.
20719 2017-08-11 Martin Liska <mliska@suse.cz>
20722 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20723 variables of void type.
20725 2017-08-11 Martin Liska <mliska@suse.cz>
20727 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20728 TARGET_SUPPORTS_ALIASES.
20729 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20730 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20731 (optimize_weakref): Likewise.
20732 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20733 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20734 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20736 2017-08-11 Martin Liska <mliska@suse.cz>
20739 * config/i386/i386.c (make_resolver_func): Do complete
20740 refactoring of the function.
20742 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20745 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20746 * config/i386/i386.c (ix86_stack_protect_guard): Use
20747 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20748 * doc/invoke.texi (x86 Options): Document
20749 -mstack-protector-guard-symbol= option.
20751 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20753 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20754 * config/i386/i386.c (ix86_split_stack_guard): New function.
20755 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20756 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20757 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20758 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20759 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20760 (split_stack_space_check): Call ix86_split_stack_guard.
20762 2017-08-10 Martin Sebor <msebor@redhat.com>
20764 * print-tree.c (print_node): Print location using the established
20766 Replace spaces with colons.
20767 (debug_raw, debug): Ditto.
20769 2017-08-10 Martin Sebor <msebor@redhat.com>
20772 * pretty-print.c (pp_format): Correct the handling of %s precision.
20774 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20777 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20779 (ix86_finalize_stack_frame_flags): This. Also clear
20780 frame_pointer_needed if -fno-omit-frame-pointer is used without
20782 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20783 with ix86_finalize_stack_frame_flags.
20784 (ix86_expand_epilogue): Likewise.
20785 (ix86_expand_split_stack_prologue): Likewise.
20786 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20788 2017-08-10 Martin Liska <mliska@suse.cz>
20791 * c-attribs.c (handle_target_attribute):
20792 Report warning for an empty string argument of target attribute.
20794 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20797 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20799 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20800 or DECL_NONLOCAL labels.
20801 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20802 or DECL_NONLOCAL labels here.
20804 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20806 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20807 to indicate when early gimple folding has been disabled.
20808 (rs6000_gimple_fold_builtin): Add debug content.
20809 (rs6000_invalid_builtin): Fix whitespace.
20810 (rs6000_expand_builtin): Fix whitespace.
20811 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
20813 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
20816 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
20817 SAVE_MULTIPLE if not all the registers that saves, should be saved.
20819 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
20821 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
20822 (qdf24xx): Likewise.
20823 * config/aarch64/aarch64.md: Include falkor.md.
20824 * config/aarch64/falkor.md: New.
20826 2017-08-09 Marek Polacek <polacek@redhat.com>
20829 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
20830 * diagnostic.c (emit_diagnostic): Add a comment.
20831 (emit_diagnostic_valist): New function.
20833 2017-08-09 Marek Polacek <polacek@redhat.com>
20836 * input.c (make_location): New overload.
20837 * input.h (make_location): Declare.
20839 2017-08-08 Alan Modra <amodra@gmail.com>
20840 H.J. Lu <hongjiu.lu@intel.com>
20843 * gcc.c (NO_PIE_SPEC): Delete.
20844 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
20846 (LINK_PIE_SPEC): ..to here.
20847 (LINK_COMMAND_SPEC): Support -no-pie.
20848 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
20849 chain of crtbegin*.o selection, update for PIE_SPEC changes and
20851 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
20852 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
20853 (ENDFILE_CRTEND_SPEC): Similarly.
20855 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
20858 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
20859 (mstack-protector-guard-offset=): Ditto.
20860 * config/i386/i386.c (ix86_option_override): Handle
20861 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
20863 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
20864 ix86_stack_protect_guard_offset variables.
20865 (TARGET_STACK_PROTECT_GUARD): Always define.
20866 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
20867 and -mstack-protector-guard-offset= options.
20869 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20871 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
20872 boundary case for the last candidate.
20874 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20876 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
20877 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
20879 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20881 PR middle-end/19706
20882 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
20883 * config/aarch64/aarch64-builtins.c
20884 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
20885 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
20886 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
20888 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20889 Andrew Pinski <pinskia@gmail.com>
20891 PR middle-end/19706
20892 * internal-fn.def (XORSIGN): New.
20893 * optabs.def (xorsign_optab): New.
20894 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
20895 (convert_expand_mult_copysign): New.
20896 (pass_optimize_widening_mul::execute): Call
20897 convert_expand_mult_copysign.
20899 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20901 PR tree-optimization/81354
20902 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
20903 Insert on edges rather than explicitly creating landing pads.
20904 (analyze_candidates_and_replace): Commit edge inserts.
20906 2017-08-08 Richard Biener <rguenther@suse.de>
20908 PR middle-end/81719
20909 * tree-ssa-loop-niter.c: Include tree-dfa.h.
20910 (expand_simple_operations): Also look through ADDR_EXPRs with
20911 MEM_REF bases treating them as POINTER_PLUS_EXPR.
20913 2017-08-08 Richard Biener <rguenther@suse.de>
20915 PR tree-optimization/81723
20916 * tree-vect-slp.c (struct bst_traits): New hash traits.
20917 (bst_fail): New global.
20918 (vect_build_slp_tree_2): New worker, split out from ...
20919 (vect_build_slp_tree): ... this now wrapping it with using
20920 bst_fail set to cache SLP tree build fails. Properly handle
20922 (vect_analyze_slp_instance): Allocate and free bst_fail.
20924 2017-08-08 Martin Liska <mliska@suse.cz>
20927 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
20928 LABEL_DECLs that can be from a different function.
20930 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20932 PR tree-optimization/81744
20933 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
20934 loop's number of iterations.
20936 2017-08-08 Martin Liska <mliska@suse.cz>
20938 * asan.c: Include header files.
20939 * attribs.c (build_decl_attribute_variant): New function moved
20941 (build_type_attribute_qual_variant): Likewise.
20942 (cmp_attrib_identifiers): Likewise.
20943 (simple_cst_list_equal): Likewise.
20944 (omp_declare_simd_clauses_equal): Likewise.
20945 (attribute_value_equal): Likewise.
20946 (comp_type_attributes): Likewise.
20947 (build_type_attribute_variant): Likewise.
20948 (lookup_ident_attribute): Likewise.
20949 (remove_attribute): Likewise.
20950 (merge_attributes): Likewise.
20951 (merge_type_attributes): Likewise.
20952 (merge_decl_attributes): Likewise.
20953 (merge_dllimport_decl_attributes): Likewise.
20954 (handle_dll_attribute): Likewise.
20955 (attribute_list_equal): Likewise.
20956 (attribute_list_contained): Likewise.
20957 * attribs.h (lookup_attribute): New function moved from tree.[ch].
20958 (lookup_attribute_by_prefix): Likewise.
20959 * bb-reorder.c: Include header files.
20960 * builtins.c: Likewise.
20961 * calls.c: Likewise.
20962 * cfgexpand.c: Likewise.
20963 * cgraph.c: Likewise.
20964 * cgraphunit.c: Likewise.
20965 * convert.c: Likewise.
20966 * dwarf2out.c: Likewise.
20967 * final.c: Likewise.
20968 * fold-const.c: Likewise.
20969 * function.c: Likewise.
20970 * gimple-expr.c: Likewise.
20971 * gimple-fold.c: Likewise.
20972 * gimple-pretty-print.c: Likewise.
20973 * gimple.c: Likewise.
20974 * gimplify.c: Likewise.
20975 * hsa-common.c: Likewise.
20976 * hsa-gen.c: Likewise.
20977 * internal-fn.c: Likewise.
20978 * ipa-chkp.c: Likewise.
20979 * ipa-cp.c: Likewise.
20980 * ipa-devirt.c: Likewise.
20981 * ipa-fnsummary.c: Likewise.
20982 * ipa-inline.c: Likewise.
20983 * ipa-visibility.c: Likewise.
20985 * lto-cgraph.c: Likewise.
20986 * omp-expand.c: Likewise.
20987 * omp-general.c: Likewise.
20988 * omp-low.c: Likewise.
20989 * omp-offload.c: Likewise.
20990 * omp-simd-clone.c: Likewise.
20991 * opts-global.c: Likewise.
20992 * passes.c: Likewise.
20993 * predict.c: Likewise.
20994 * sancov.c: Likewise.
20995 * sanopt.c: Likewise.
20996 * symtab.c: Likewise.
20997 * toplev.c: Likewise.
20998 * trans-mem.c: Likewise.
20999 * tree-chkp.c: Likewise.
21000 * tree-eh.c: Likewise.
21001 * tree-into-ssa.c: Likewise.
21002 * tree-object-size.c: Likewise.
21003 * tree-parloops.c: Likewise.
21004 * tree-profile.c: Likewise.
21005 * tree-ssa-ccp.c: Likewise.
21006 * tree-ssa-live.c: Likewise.
21007 * tree-ssa-loop.c: Likewise.
21008 * tree-ssa-sccvn.c: Likewise.
21009 * tree-ssa-structalias.c: Likewise.
21010 * tree-ssa.c: Likewise.
21011 * tree-streamer-in.c: Likewise.
21012 * tree-vectorizer.c: Likewise.
21013 * tree-vrp.c: Likewise.
21014 * tsan.c: Likewise.
21015 * ubsan.c: Likewise.
21016 * varasm.c: Likewise.
21017 * varpool.c: Likewise.
21018 * tree.c: Remove functions moved to attribs.[ch].
21019 * tree.h: Likewise.
21020 * config/aarch64/aarch64.c: Add attrs.h header file.
21021 * config/alpha/alpha.c: Likewise.
21022 * config/arc/arc.c: Likewise.
21023 * config/arm/arm.c: Likewise.
21024 * config/avr/avr.c: Likewise.
21025 * config/bfin/bfin.c: Likewise.
21026 * config/c6x/c6x.c: Likewise.
21027 * config/cr16/cr16.c: Likewise.
21028 * config/cris/cris.c: Likewise.
21029 * config/darwin.c: Likewise.
21030 * config/epiphany/epiphany.c: Likewise.
21031 * config/fr30/fr30.c: Likewise.
21032 * config/frv/frv.c: Likewise.
21033 * config/ft32/ft32.c: Likewise.
21034 * config/h8300/h8300.c: Likewise.
21035 * config/i386/winnt.c: Likewise.
21036 * config/ia64/ia64.c: Likewise.
21037 * config/iq2000/iq2000.c: Likewise.
21038 * config/lm32/lm32.c: Likewise.
21039 * config/m32c/m32c.c: Likewise.
21040 * config/m32r/m32r.c: Likewise.
21041 * config/m68k/m68k.c: Likewise.
21042 * config/mcore/mcore.c: Likewise.
21043 * config/microblaze/microblaze.c: Likewise.
21044 * config/mips/mips.c: Likewise.
21045 * config/mmix/mmix.c: Likewise.
21046 * config/mn10300/mn10300.c: Likewise.
21047 * config/moxie/moxie.c: Likewise.
21048 * config/msp430/msp430.c: Likewise.
21049 * config/nds32/nds32-isr.c: Likewise.
21050 * config/nds32/nds32.c: Likewise.
21051 * config/nios2/nios2.c: Likewise.
21052 * config/nvptx/nvptx.c: Likewise.
21053 * config/pa/pa.c: Likewise.
21054 * config/pdp11/pdp11.c: Likewise.
21055 * config/powerpcspe/powerpcspe.c: Likewise.
21056 * config/riscv/riscv.c: Likewise.
21057 * config/rl78/rl78.c: Likewise.
21058 * config/rx/rx.c: Likewise.
21059 * config/s390/s390.c: Likewise.
21060 * config/sh/sh.c: Likewise.
21061 * config/sol2.c: Likewise.
21062 * config/sparc/sparc.c: Likewise.
21063 * config/spu/spu.c: Likewise.
21064 * config/stormy16/stormy16.c: Likewise.
21065 * config/tilegx/tilegx.c: Likewise.
21066 * config/tilepro/tilepro.c: Likewise.
21067 * config/v850/v850.c: Likewise.
21068 * config/vax/vax.c: Likewise.
21069 * config/visium/visium.c: Likewise.
21070 * config/xtensa/xtensa.c: Likewise.
21072 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21075 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
21076 constraints since the -mupper-regs-* switches have been
21078 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
21079 into a vector from a double word element that was extracted from
21080 another vector, and eliminate extra XXPERMDI instructions.
21081 (vsx_concat_<mode>_2): Likewise.
21082 (vsx_concat_<mode>_3): Likewise.
21083 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
21084 concat to allow optimizing inserts from previous extracts.
21086 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
21088 * config/i386/i386.c (ix86_stack_protect_guard): Generate
21089 memory reference to a SSP offset in TLS address space.
21090 (ix86_print_operand) <case '@'>: Remove.
21091 (ix86_print_operand_punct_valid_p): Remove '@' code.
21092 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
21093 UNSPEC_SP_TLS_TEST.
21094 (stack_tls_protect_set_<mode>): Remove.
21095 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
21096 (stack_tls_protect_test_<mode>): Remove.
21097 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
21099 2017-08-07 Olivier Hainque <hainque@adacore.com>
21102 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
21104 2017-08-07 Douglas Rupp <rupp@adacore.com>
21106 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
21107 variable was referenced as multidir in command.
21109 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21112 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
21115 2017-08-07 Martin Liska <mliska@suse.cz>
21117 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
21118 * config/rl78/rl78.c: Add include of attribs.h.
21119 * config/sh/sh.c: Likewise.
21120 * config/v850/v850.c: Likewise.
21122 2017-08-07 Tom de Vries <tom@codesourcery.com>
21124 PR middle-end/78266
21125 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
21127 2017-08-07 Martin Liska <mliska@suse.cz>
21129 * config/mips/mips.c: Include attribs.h.
21131 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
21134 * doc/invoke.texi: Document change in behvaior for -Ofast for
21137 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
21139 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
21141 (aarch64_pop_regs): Likewise.
21142 (aarch64_gen_load_pair): Likewise.
21143 (aarch64_save_callee_saves): Likewise.
21144 (aarch64_restore_callee_saves): Likewise.
21146 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21148 * config/i386/i386.c: Revert the last change.
21150 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21153 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21155 (ix86_finalize_stack_frame_flags): This. Also clear
21156 frame_pointer_needed if -fno-omit-frame-pointer is used without
21158 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21159 with ix86_finalize_stack_frame_flags.
21160 (ix86_expand_epilogue): Likewise.
21161 (ix86_expand_split_stack_prologue): Likewise.
21163 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21166 * config/i386/i386.c (get_builtin_code_for_version): Set priority
21167 to P_AES for Westmere.
21169 2017-08-07 Jonathan Yong <10walls@gmail.com>
21171 * config/i386/mingw.opt (fset-stack-executable): Removed.
21172 * config/i386/cygming.opt (fset-stack-executable): Moved
21174 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
21176 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
21178 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
21180 2017-08-07 Marek Polacek <polacek@redhat.com>
21182 PR middle-end/81737
21183 * fold-const.c (fold_indirect_ref_1): Check type_domain.
21185 2017-08-07 Martin Liska <mliska@suse.cz>
21187 * attribs.h (canonicalize_attr_name): New function.
21188 (cmp_attribs): Move from c-format.c and adjusted.
21189 (is_attribute_p): Moved from tree.h.
21190 * tree-inline.c: Add new includes.
21191 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
21192 (private_is_attribute_p): Remove.
21193 (private_lookup_attribute): Likewise.
21194 (private_lookup_attribute_by_prefix): Simplify.
21195 (remove_attribute): Use is_attribute_p.
21196 * tree.h: Remove removed declarations.
21198 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21200 PR middle-end/81698
21201 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
21202 instead of computing it in the function. Formatting fix.
21203 (expand_case): Don't rely on default_edge being the first edge,
21204 clear it if removing it, pass default_edge to
21205 emit_case_dispatch_table.
21206 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
21209 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
21211 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
21212 insn in the function, emit NOP after the insn.
21214 2017-08-06 Tom de Vries <tom@codesourcery.com>
21216 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
21219 2017-08-06 Tom de Vries <tom@codesourcery.com>
21221 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
21224 2017-08-04 Yury Gribov <tetra2005@gmail.com>
21226 PR tree-optimization/57371
21227 * match.pd: New pattern.
21229 2017-08-04 Marek Polacek <polacek@redhat.com>
21231 PR middle-end/81695
21232 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
21233 perform the computation in offset_int.
21235 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21237 PR tree-optimization/81136
21238 * tree-vectorizer.h: Include tree-hash-traits.h.
21239 (vec_base_alignments): New typedef.
21240 (vec_info): Add a base_alignments field.
21241 (vect_record_base_alignments): Declare.
21242 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
21244 (DR_IS_CONDITIONAL_IN_STMT): New macro.
21245 (create_data_ref): Add an is_conditional_in_stmt argument.
21246 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
21247 the is_conditional_in_stmt field.
21248 (data_ref_loc): Add an is_conditional_in_stmt field.
21249 (get_references_in_stmt): Set the is_conditional_in_stmt field.
21250 (find_data_references_in_stmt): Update call to create_data_ref.
21251 (graphite_find_data_references_in_stmt): Likewise.
21252 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
21253 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
21254 (vect_record_base_alignment): New function.
21255 (vect_record_base_alignments): Likewise.
21256 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
21257 for nested statements even if we fail to compute a misalignment.
21258 Use pooled base alignments for unconditional references.
21259 (vect_find_same_alignment_drs): Compare base addresses instead
21261 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
21262 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
21264 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21266 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
21267 Add an explicit name for the enum. Use auto_vec for slp_instances
21268 and grouped_stores.
21269 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
21271 (_bb_vec_info): Add a constructor and destructor.
21272 (vinfo_for_stmt): Return NULL for uids of -1 as well.
21273 (destroy_loop_vec_info): Delete.
21274 (vect_destroy_datarefs): Likewise.
21275 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
21276 (vec_info::vec_info): New function.
21277 (vec_info::~vec_info): Likewise.
21278 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
21279 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
21280 destroy_loop_vec_info.
21281 * tree-vect-loop.c (new_loop_vec_info): Replace with...
21282 (_loop_vec_info::_loop_vec_info): ...this.
21283 (destroy_loop_vec_info): Replace with...
21284 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
21285 the stmt_vec_infos. Leave handling of vec_info information to its
21286 destructor. Remove explicit vector releases.
21287 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
21288 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
21289 * tree-vect-slp.c (new_bb_vec_info): Replace with...
21290 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
21291 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
21292 (destroy_bb_vec_info): Replace with...
21293 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
21294 information to its destructor.
21295 (vect_slp_analyze_bb_1): Use new and delete instead of
21296 new_bb_vec_info and destroy_bb_vec_info.
21297 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
21300 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21302 * tree-data-ref.h (subscript): Add access_fn field.
21303 (data_dependence_relation): Add could_be_independent_p.
21304 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
21305 (same_access_functions): Move to tree-data-ref.c.
21306 * tree-data-ref.c (ref_contains_union_access_p): New function.
21307 (access_fn_component_p): Likewise.
21308 (access_fn_components_comparable_p): Likewise.
21309 (dr_analyze_indices): Add a reference to access_fn_component_p.
21310 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
21312 (constant_access_functions): Likewise.
21313 (add_other_self_distances): Likewise.
21314 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
21315 (initialize_data_dependence_relation): Use XCNEW and remove
21316 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
21317 of access functions that have the same type. Allow the
21318 subsequence to end with different bases in some circumstances.
21319 Record the chosen access functions in SUB_ACCESS_FN.
21320 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
21321 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
21322 (subscript_dependence_tester_1): Likewise dra and drb.
21323 (build_classic_dist_vector): Update calls accordingly.
21324 (subscript_dependence_tester): Likewise.
21325 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
21326 DDR_COULD_BE_INDEPENDENT_P.
21327 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
21328 comp_alias_ddrs instead of may_alias_ddrs.
21329 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
21331 (vect_analyze_data_ref_dependence): Use it if
21332 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
21333 distance vectors if that fails.
21334 (dependence_distance_ge_vf): New function.
21335 (vect_prune_runtime_alias_test_list): Use it. Don't clear
21336 LOOP_VINFO_MAY_ALIAS_DDRS.
21338 2017-08-04 Richard Biener <rguenther@suse.de>
21340 PR middle-end/81705
21341 * fold-const.c (fold_binary_loc): Properly restrict
21342 minus_var0 && minus_var1 case when associating undefined overflow
21345 2017-08-04 Tom de Vries <tom@codesourcery.com>
21347 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
21349 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21351 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21352 Don't start diagnostic messages with a capital letter.
21353 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21355 (rs6000_invalid_builtin): Likewise.
21356 (rs6000_trampoline_init): Likewise.
21358 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21361 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
21362 after setting changeable df flags.
21364 2017-08-03 Richard Biener <rguenther@suse.de>
21366 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
21367 up if the use is in USE - X.
21369 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
21371 * toplev.c (dumpfile.h): New include.
21372 (internal_error_reentered): New static function. Use it...
21373 (internal_error_function): ...here to handle reentered internal_error.
21375 2017-08-03 Richard Biener <rguenther@suse.de>
21377 PR middle-end/81148
21378 * fold-const.c (split_tree): Add minus_var and minus_con
21379 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
21380 here but always use minus_*.
21381 (associate_trees): Assert we never associate with MINUS_EXPR
21382 and NULL first operand. Do not recurse for PLUS_EXPR operands
21383 when associating as MINUS_EXPR either.
21384 (fold_binary_loc): Track minus_var and minus_con.
21386 2017-08-03 Tom de Vries <tom@codesourcery.com>
21389 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
21390 ACCEL_COMPILER, apply finish_options on
21391 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
21393 2017-08-03 Tom de Vries <tom@codesourcery.com>
21396 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
21397 function_entry_patch_area_size > 0.
21399 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21402 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
21403 instead of 10??LU, perform unit multiplication in wide_int,
21404 don't change alloc_object_size_limit if the limit is larger
21407 PR tree-optimization/81655
21408 PR tree-optimization/81588
21409 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
21410 the case when ranges[i].low and high are 1 for unsigned type with
21413 PR middle-end/81052
21414 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
21415 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
21417 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21419 * tree-vrp.h: Add include guard.
21421 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
21424 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
21425 (ud2): New insn pattern.
21426 * config/i386/i386.c (ix86_expand_epilogue):
21427 For naked functions, generate ud2 instead of trap insn.
21429 2017-08-02 Marek Polacek <polacek@redhat.com>
21432 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
21434 2017-08-02 Tom de Vries <tom@codesourcery.com>
21435 Cesar Philippidis <cesar@codesourcery.com>
21437 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
21438 Add missing edge probabilities.
21440 2017-08-02 Tamar Christina <tamar.christina@arm.com>
21442 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
21443 Correct endianness.
21445 2017-08-02 Jakub Jelinek <jakub@redhat.com>
21447 PR middle-end/79499
21448 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
21449 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
21450 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
21451 prologue_seq sequences - if any.
21453 2017-08-02 Richard Biener <rguenther@suse.de>
21455 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
21456 via vectors if supported, integer extracts via punning if supported
21457 or otherwise vector extracts.
21459 2017-08-02 Richard Biener <rguenther@suse.de>
21461 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
21463 (bitmap_insert_into_set): ... this.
21465 2017-08-02 Richard Biener <rguenther@suse.de>
21467 PR tree-optimization/81633
21469 2015-08-17 Alan Hayward <alan.hayward@arm.com>
21471 PR tree-optimization/71752
21472 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
21474 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
21476 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
21477 (machine_function::call_ms2sysv_pad_out): Remove field.
21478 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
21479 (ix86_compute_frame_layout): Likewise.
21481 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
21484 * config/i386/i386.c (ix86_set_func_type): Disallow naked
21485 attribute with interrupt attribute.
21487 2017-08-01 Andrew Pinski <apinski@cavium.com>
21489 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
21490 BIT_INSERT_EXPR's operand 1
21491 to see if the types precision matches.
21493 2017-08-01 Martin Liska <mliska@suse.cz>
21495 PR middle-end/70140
21496 * builtins.c (expand_builtin_memcpy_args): Remove.
21497 (expand_builtin_memcpy): Call newly added function
21498 expand_builtin_memory_copy_args.
21499 (expand_builtin_memcpy_with_bounds): Likewise.
21500 (expand_builtin_mempcpy): Remove last argument.
21501 (expand_builtin_mempcpy_with_bounds): Likewise.
21502 (expand_builtin_memory_copy_args): New function created from
21503 expand_builtin_mempcpy_args with small modifications.
21504 (expand_builtin_mempcpy_args): Remove.
21505 (expand_builtin_stpcpy): Remove unused argument.
21506 (expand_builtin): Likewise.
21507 (expand_builtin_with_bounds): Likewise.
21509 2017-08-01 Martin Liska <mliska@suse.cz>
21512 Make mempcpy more optimal (PR middle-end/70140).
21514 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21517 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
21518 __builtin_vec_cmpne verify both arguments are compatible vectors
21519 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
21520 verify arg1_type is a pointer or array type. For __builtin_vec_st,
21521 move computation of aligned to after checking the argument types.
21525 * config/rs6000/vsx.md (vextract_fp_from_shorth,
21526 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
21529 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
21530 Doug Rupp <rupp@adacore.com>
21531 Olivier Hainque <hainque@adacore.com>
21533 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
21534 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
21536 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
21537 for TARGET_OS_CPP_BUILTIN.
21538 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
21539 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
21541 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
21542 passing required abi options to the assembler for EABI configurations.
21543 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
21544 of .text.hot and .text.unlikely sections for kernel modules when
21545 using ARM style exceptions.
21546 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
21547 options. Add EXTRA_CC1_SPEC.
21548 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
21550 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
21552 (ARM_TARGET2_DWARF_FORMAT): Define.
21553 * config/arm/t-vxworks: Adjust multilib control to removal of the
21554 Diab command line options.
21556 2017-08-01 Martin Liska <mliska@suse.cz>
21558 PR gcov-profile/81561
21559 * gcov.c (unblock): Make unblocking safe as we need to preserve
21560 index correspondence of blocks and block_lists.
21562 2017-08-01 Richard Biener <rguenther@suse.de>
21564 PR tree-optimization/81181
21565 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21566 (compute_antic): ... end of iteration here.
21568 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21570 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21571 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21572 (ftree-slp-vectorize): Likewise.
21573 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21574 can no longer be set independent of flag_tree_loop_vectorize.
21575 * omp-general.c (emp_max_vf): Likewise.
21576 * opts.c (enable_fdo_optimizations): Remove references to
21577 flag_tree_vectorize, these are now implicit.
21578 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21579 and leave it for the options machinery.
21581 2017-08-01 Martin Liska <mliska@suse.cz>
21583 PR middle-end/70140
21584 * builtins.c (expand_builtin_memcpy_args): Remove.
21585 (expand_builtin_memcpy): Call newly added function
21586 expand_builtin_memory_copy_args.
21587 (expand_builtin_memcpy_with_bounds): Likewise.
21588 (expand_builtin_mempcpy): Remove last argument.
21589 (expand_builtin_mempcpy_with_bounds): Likewise.
21590 (expand_builtin_memory_copy_args): New function created from
21591 expand_builtin_mempcpy_args with small modifications.
21592 (expand_builtin_mempcpy_args): Remove.
21593 (expand_builtin_stpcpy): Remove unused argument.
21594 (expand_builtin): Likewise.
21595 (expand_builtin_with_bounds): Likewise.
21597 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21600 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21601 print "ds:" only for immediates in generic address space.
21603 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21606 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21607 (ix86_function_ok_for_sibcall): Return false for naked functions.
21609 2017-08-01 Richard Biener <rguenther@suse.de>
21611 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21612 (compute_antic): Seed worklist with exit block predecessors.
21613 * cfganal.c (dfs_find_deadend): For a cycle return the source
21614 of the edge closing it.
21616 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21618 * config/aarch64/aarch64.c
21619 (aarch64_can_const_movi_rtx_p): Move 0 check.
21621 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21623 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21624 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21626 * match.pd: Ditto in address comparison pattern.
21628 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21630 PR tree-optimization/81627
21631 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21632 closed ssa form for store-store chain.
21634 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21636 PR tree-optimization/81620
21637 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21638 for store-store chain.
21640 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21642 PR tree-optimization/81588
21643 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21644 ranges[i].in_p, invert comparison code ccode. For >/>=,
21645 swap rhs1 and rhs2 and comparison code unconditionally,
21646 for </<= don't do that. Don't swap rhs1/rhs2 again if
21647 ranges[i].in_p, instead invert comparison code ccode if
21648 opcode or oe->rank is BIT_IOR_EXPR.
21651 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21652 a direct optab to conversion optab.
21653 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21654 with GET_MODE_INNER as last argument instead of optab_handler.
21655 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21656 vector extraction if possible and optab is available.
21657 * expr.c (store_constructor): Use convert_optab_handler instead
21658 of optab_handler. Use vector initialization from smaller
21659 vectors if possible and optab is available.
21660 * tree-vect-stmts.c (vectorizable_load): Likewise.
21661 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21662 now have two modes.
21663 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21664 of vec_init from half-sized vectors with the same element mode.
21665 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21666 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21667 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21668 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21669 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21670 after mode in gen_vec_extract* calls.
21671 (vec_extract<mode>): Renamed to ...
21672 (vec_extract<mode><ssescalarmodelower>): ... this.
21673 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21674 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21675 element mode after mode in gen_vec_init* calls.
21676 (VEC_INIT_HALF_MODE): New mode iterator.
21677 (vec_init<mode>): Renamed to ...
21678 (vec_init<mode><ssescalarmodelower>): ... this.
21679 (vec_init<mode><ssehalfvecmodelower>): New expander.
21680 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21681 (vec_extractv2sfsf): ... this.
21682 (vec_initv2sf): Renamed to ...
21683 (vec_initv2sfsf): ... this.
21684 (vec_extractv2si): Renamed to ...
21685 (vec_extractv2sisi): ... this.
21686 (vec_initv2si): Renamed to ...
21687 (vec_initv2sisi): ... this.
21688 (vec_extractv4hi): Renamed to ...
21689 (vec_extractv4hihi): ... this.
21690 (vec_initv4hi): Renamed to ...
21691 (vec_initv4hihi): ... this.
21692 (vec_extractv8qi): Renamed to ...
21693 (vec_extractv8qiqi): ... this.
21694 (vec_initv8qi): Renamed to ...
21695 (vec_initv8qiqi): ... this.
21696 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21697 (vec_init<mode>): Renamed to ...
21698 (vec_init<mode><VEC_base_l>): ... this.
21699 (vec_extract<mode>): Renamed to ...
21700 (vec_extract<mode><VEC_base_l>): ... this.
21701 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21702 (vec_initv2sfsf): ... this.
21703 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21704 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21705 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21706 element mode after mode in gen_vec_init* calls.
21707 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21708 (vec_init<mode><Vel>): ... this.
21709 (vec_extract<mode>): Renamed to ...
21710 (vec_extract<mode><Vel>): ... this.
21711 * config/aarch64/iterators.md (Vel): New mode attribute.
21712 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21713 Add element mode after mode in gen_vec_extract* calls.
21714 * config/s390/vector.md (non_vec_l): New mode attribute.
21715 (vec_extract<mode>): Renamed to ...
21716 (vec_extract<mode><non_vec_l>): ... this.
21717 (vec_init<mode>): Renamed to ...
21718 (vec_init<mode><non_vec_l>): ... this.
21719 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21720 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21722 * config/arm/iterators.md (V_elem_l): New mode attribute.
21723 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21724 (vec_extract<mode><V_elem_l>): ... this.
21725 (vec_extractv2di): Renamed to ...
21726 (vec_extractv2didi): ... this.
21727 (vec_init<mode>): Renamed to ...
21728 (vec_init<mode><V_elem_l>): ... this.
21729 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21730 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21731 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21732 Add element mode after gen_vec_extract* calls.
21733 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21734 (vec_init<mode><unitmode>): ... this.
21735 (vec_extract<mode>): Renamed to ...
21736 (vec_extract<mode><unitmode>): ... this.
21737 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21738 (vec_init<mode><unitmode>): ... this.
21739 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21740 (vec_initv2sfsf): ... this.
21741 (vec_extractv2sf): Renamed to ...
21742 (vec_extractv2sfsf): ... this.
21743 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21744 Add element mode after gen_vec_extract* calls.
21745 * config/mips/mips.md (unitmode): New mode iterator.
21746 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21747 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21748 * config/spu/spu.md (inner_l): New mode attribute.
21749 (vec_init<mode>): Renamed to ...
21750 (vec_init<mode><inner_l>): ... this.
21751 (vec_extract<mode>): Renamed to ...
21752 (vec_extract<mode><inner_l>): ... this.
21753 * config/sparc/sparc.md (veltmode): New mode iterator.
21754 (vec_init<VMALL:mode>): Renamed to ...
21755 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21756 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21757 (vec_initv2sisi): ... this.
21758 (vec_initv2sf): Renamed to ...
21759 (vec_initv2sfsf): ... this.
21760 (vec_extractv2sf): Renamed to ...
21761 (vec_extractv2sfsf): ... this.
21762 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21763 (vec_init<mode>): Renamed to ...
21764 (vec_init<mode><VEC_base_l>): ... this.
21765 (vec_extract<mode>): Renamed to ...
21766 (vec_extract<mode><VEC_base_l>): ... this.
21767 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21768 (vec_initv2sfsf): ... this.
21769 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21770 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21771 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21772 gen_vec_init* calls.
21774 2017-08-01 Richard Biener <rguenther@suse.de>
21776 PR tree-optimization/81297
21777 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21778 TREE_OVERFLOW from INTEGER_CSTs.
21780 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21782 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21784 2017-07-31 Carl Love <cel@us.ibm.com>
21786 * config/rs6000/rs6000-c: Add support for built-in functions
21787 vector signed char vec_xl_be (signed long long, signed char *);
21788 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21789 vector signed int vec_xl_be (signed long long, signed int *);
21790 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21791 vector signed long long vec_xl_be (signed long long, signed long long *);
21792 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21793 vector signed short vec_xl_be (signed long long, signed short *);
21794 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21795 vector double vec_xl_be (signed long long, double *);
21796 vector float vec_xl_be (signed long long, float *);
21797 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21798 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21799 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21801 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21802 (altivec_expand_builtin): Add switch statement to call
21803 altivec_expand_xl_be for each builtin.
21804 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21805 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21806 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21807 __builtin_vsx_le_be_v16qi.
21808 * doc/extend.texi: Update the built-in documentation file for the
21809 new built-in functions.
21811 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21814 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
21816 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21818 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21820 * config.gcc: Add z14.
21821 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
21822 CPU model numbers for z13s and z14.
21823 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
21825 * config/s390/s390-opts.h (enum processor_type): Rename
21826 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21827 * config/s390/s390.c (processor_table): Add field for CPU name to
21828 be passed to Binutils.
21829 (s390_asm_output_machine_for_arch): Use the new field in
21830 processor_table for Binutils.
21831 (s390_expand_builtin): Replace arch12 with z14.
21832 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21833 (s390_get_sched_attrmask): Likewise.
21834 (s390_get_unit_mask): Likewise.
21835 * config/s390/s390.opt: Add z14 to processor_type enum.
21837 2017-07-31 Martin Jambor <mjambor@suse.cz>
21840 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
21841 regardless of optimization level.
21843 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
21844 Martin Liska <mliska@suse.cz>
21846 * predict.def: Remove old comment and adjust probability.
21847 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
21848 PREDICT statements.
21850 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21853 * config/i386/i386.c (ix86_function_naked): New function.
21854 (ix86_can_use_return_insn_p): Return false for naked functions.
21855 (ix86_expand_prologue): Skip prologue for naked functions.
21856 (ix86_expand_epilogue): Skip epilogue for naked functions
21857 and emit trap instruction.
21858 (ix86_warn_func_return): New function.
21859 (ix86_attribute_table): Add "naked" attribute specification.
21860 (TARGET_WARN_FUNC_RETURN): Define.
21861 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
21863 2017-07-31 Martin Liska <mliska@suse.cz>
21865 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
21866 (dump_gimple_bb_header): Always dump BB info.
21867 (pp_cfg_jump): Do not append info about BB when dumping a jump.
21869 2017-07-31 Martin Liska <mliska@suse.cz>
21872 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
21873 also with current_function_decl non-null equality.
21875 2017-07-31 Jakub Jelinek <jakub@redhat.com>
21878 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
21879 change type to the element type, instead add eltype variable and
21880 use it where we are interested in the element type.
21882 PR tree-optimization/81603
21883 * ipa-polymorphic-call.c
21884 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
21885 offset arithmetic in offset_int, bail out if the resulting bit offset
21886 doesn't fit into shwi.
21888 2017-07-31 Martin Liska <mliska@suse.cz>
21890 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
21891 (gimplify_save_expr): Fix comment.
21893 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
21896 * config/i386/i386.c (ix86_function_arg): Update arguments for
21898 (ix86_compute_frame_layout): Set the initial stack offset to
21899 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
21900 INCOMING_FRAME_SP_OFFSET.
21901 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
21902 stack before exception handler returns.
21903 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
21904 the 'ERROR_CODE' for exception handler.
21906 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
21908 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
21909 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
21910 (ASM_OUTPUT_REG_POP): Ditto.
21911 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
21912 instead of asm_fprintf to output pure string.
21914 2017-07-29 Jakub Jelinek <jakub@redhat.com>
21916 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
21917 to imported_module_or_decl hook.
21918 (debug_nothing_tree_tree_tree_bool): Remove.
21919 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
21920 * debug.c (do_nothing_debug_hooks): Use
21921 debug_nothing_tree_tree_tree_bool_bool instead of
21922 debug_nothing_tree_tree_tree_bool.
21923 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21924 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
21925 * sdbout.c (sdb_debug_hooks): Likewise.
21926 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
21927 (gen_namespace_die): Add DW_AT_export_symbols attribute if
21929 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
21930 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
21931 attribute, don't add anything.
21933 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21935 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
21936 (fold_build2_stat_loc): Likewise.
21937 (fold_build3_stat_loc): Likewise.
21938 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
21939 (fold_build1_loc): Remove macro.
21940 (fold_build2_loc): Likewise.
21941 (fold_build3_loc): Likewise.
21943 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21945 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
21946 (gimple_build_debug_bind_source_stat): Likewise.
21947 * gimple.h (gimple_build_debug_bind): Remove macro.
21948 (gimple_build_debug_bind_source): Likewise.
21950 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21952 * bitmap.c (bitmap_alloc): Adjust.
21953 (bitmap_gc_alloc): Likewise.
21954 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
21956 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21958 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
21959 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
21960 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
21961 (bitmap_gc_alloc_stat): Likewise.
21962 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
21964 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21966 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
21967 * rtl.h (shallow_copy_rtx): Remove macro.
21969 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21971 * emit-rtl.c (gen_raw_REG): Adjust.
21972 * gengenrtl.c (gendef): Likewise.
21973 * rtl.c (rtx_alloc_stat): Remove _stat from name.
21974 * rtl.h (rtx_alloc): Remove macro.
21976 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21978 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
21979 (build_tree_list_stat): Likewise.
21980 * tree.h (build_tree_list): Remove macro.
21981 (build_tree_list_vec): Likewise.
21983 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21985 * tree.c (make_vector_stat): Remove _stat from name.
21986 (build_vector_stat): Likewise.
21987 * tree.h (make_vector_stat): Remove macro.
21988 (build_vector_stat): Likewise.
21990 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21992 * tree.h (build_var_debug_value): Remove prototype.
21994 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21996 * tree.c (tree_cons_stat): Remove _stat from name.
21997 * tree.h (tree_cons): Remove macro.
21999 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22001 * tree.c (build_vl_exp_stat): Remove _stat from name.
22002 * tree.h (build_vl_exp): Remove macro.
22004 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22006 * tree.c (build_decl_stat): Remove _stat from name.
22007 * tree.h (build_decl): Remove macro.
22009 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22011 * gimple.c (gimple_build_with_ops_stat): Adjust.
22012 (gimple_alloc_stat): Remove _stat from name.
22013 * gimple.h (gimple_alloc): Remove macro.
22015 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22017 * tree.c (make_tree_vec_stat): Remove _stat from name.
22018 (grow_tree_vec_stat): Likewise.
22019 * tree.h (make_tree_vec_stat): Adjust prototype.
22020 (grow_tree_vec_stat): Likewise.
22021 (make_tree_vec): Remove macro.
22022 (grow_tree_vec): Likewise.
22024 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22026 * fold-const.c (fold_build1_stat_loc): Adjust.
22027 (fold_build2_stat_loc): Likewise.
22028 (fold_build3_stat_loc): Likewise.
22029 * tree.c (build0_stat): Remove _stat from name.
22030 (build1_stat): Likewise.
22031 (build2_stat): Likewise.
22032 (build3_stat): Likewise.
22033 (build4_stat): Likewise.
22034 (build5_stat): Likewise.
22035 * tree.h (build1_loc): Remove macro, and rename _stat function
22037 (build2_loc): Likewise.
22038 (build3_loc): Likewise.
22039 (build4_loc): Likewise.
22040 (build5_loc): Likewise.
22042 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22044 * tree.c (make_int_cst_stat): Remove _stat from name.
22045 * tree.h (make_int_cst_stat): Adjust prototype.
22046 (make_int_cst): Remove macro.
22048 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22050 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22051 * tree.h (make_tree_binfo_stat): Adjust prototype.
22052 (make_tree_binfo): Remove.
22054 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22056 * tree.c (copy_node_stat): Rename to copy_node.
22057 (build_distinct_type_copy): Adjust.
22058 * tree.h (copy_node_stat): Adjust prototype.
22059 (copy_node): Remove macro.
22061 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22063 * tree.c (make_node_stat): rename to make_node.
22064 (build_tree_list_stat): Adjust.
22065 (build0_stat): Likewise.
22066 (build2_stat): Likewise.
22067 (build3_stat): Likewise.
22068 (build4_stat): Likewise.
22069 (build5_stat): Likewise.
22070 (build_decl_stat): Likewise.
22071 * tree.h (make_node_stat): Adjust prototype.
22072 (make_node): remove macro.
22074 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
22076 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
22077 (PPC_FEATURE2_SCV): Likewise.
22078 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
22080 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22082 * config/aarch64/aarch64.c
22083 (aarch64_internal_mov_immediate): Add new special pattern.
22084 * config/aarch64/aarch64.md (*movdi_aarch64):
22085 Add reg/32bit const mov case.
22087 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22088 Richard Sandiford <richard.sandiford@linaro.org>
22090 * config/aarch64/aarch64.md (mov<mode>): Generalize.
22091 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
22092 Add integer and movi cases.
22093 (movi-split-hf-df-sf split, fp16): New.
22094 (enabled): Added TARGET_FP_F16INST.
22095 * config/aarch64/iterators.md (GPF_HF): New.
22096 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
22098 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22100 * config/aarch64/aarch64.c
22101 (aarch64_simd_container_mode): Add prototype.
22102 (aarch64_expand_mov_immediate): Add HI support.
22103 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
22104 (aarch64_can_const_movi_rtx_p): New.
22105 (aarch64_preferred_reload_class):
22106 Remove restrictions of using FP registers for certain SIMD operations.
22107 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
22108 (aarch64_valid_floating_const): Add integer move validation.
22109 (aarch64_simd_imm_scalar_p): Remove.
22110 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
22111 (aarch64_legitimate_constant_p): Expand list of supported cases.
22112 * config/aarch64/aarch64-protos.h
22113 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
22114 (aarch64_reinterpret_float_as_int): New.
22115 (aarch64_simd_imm_scalar_p): Remove.
22116 * config/aarch64/constraints.md (Uvi): New.
22117 (Dd): Split into Ds and new Dd.
22118 * config/aarch64/aarch64.md (*movsi_aarch64):
22120 (*movdi_aarch64): Add SIMD mov case.
22122 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22124 * tree-predcom.c: (struct chain): Handle store-store chain in which
22125 stores for elimination only store loop invariant values.
22126 (execute_pred_commoning_chain): Ditto.
22127 (prepare_initializers_chain_store_elim): Ditto.
22128 (prepare_finalizers): Ditto.
22129 (is_inv_store_elimination_chain): New function.
22130 (initialize_root_vars_store_elim_1): New function.
22132 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22134 * tree-predcom.c: Revise general description of the pass.
22135 (enum chain_type): New enum type for store elimination.
22136 (struct chain): New field supporting store elimination.
22137 (struct component): Ditto.
22138 (dump_chain): Dump store-stores chain.
22139 (release_chain): Release resources.
22140 (split_data_refs_to_components): Compute and create component
22141 contains only stores for elimination.
22142 (get_chain_last_ref_at): New function.
22143 (make_invariant_chain): Initialization.
22144 (make_rooted_chain): Specify chain type in parameter and record it.
22145 (add_looparound_copies): Skip for store-stores chain.
22146 (determine_roots_comp): Compute type of chain and pass it to
22148 (initialize_root_vars_store_elim_2): New function.
22149 (finalize_eliminated_stores): New function.
22150 (remove_stmt): Handle store for elimination.
22151 (execute_pred_commoning_chain): Execute predictive commoning on
22152 store-store chains.
22153 (determine_unroll_factor): Skip unroll for store-stores chain.
22154 (prepare_initializers_chain_store_elim): New function.
22155 (prepare_initializers_chain): Hanlde store-store chain.
22156 (prepare_finalizers_chain, prepare_finalizers): New function.
22157 (tree_predictive_commoning_loop): Return integer value indicating
22158 if loop is unrolled or lcssa form is corrupted.
22159 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
22161 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22163 * tree-predcom.c (initialize_root): Delete.
22164 (execute_pred_commoning_chain): Initialize root vars and replace
22165 reference of non-combined chain directly, rather than call above
22168 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22170 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
22171 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
22173 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22175 * tree-predcom.c (struct chain): New field init_seq.
22176 (release_chain): Release init_seq.
22177 (prepare_initializers_chain): Record intialization stmts in above
22179 (insert_init_seqs): New function.
22180 (tree_predictive_commoning_loop): Call insert_init_seqs.
22182 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22184 * tree-predcom.c (determine_roots_comp): Skip trivial components.
22186 2017-07-28 Richard Biener <rguenther@suse.de>
22188 * match.pd: Remove superfluous :c.
22189 * genmatch.c (simplify::id): Add member.
22190 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
22192 (current_id): New global.
22193 (dt_node::parent): Move from ...
22194 (dt_operand::parent): ... here. Add for_id member.
22195 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
22196 (decision_tree::find_node): Relax order requirement when
22197 merging DT_TRUE nodes to ones inbetween the current simplify
22198 and the one we try to merge with. Add diagnostic whenever
22199 we need to enforce pattern order by not merging.
22200 (decision_tree::insert): Set current_id.
22201 (decision_tree::print_node): Dump parent node and for_id.
22202 (parser::last_id): Add member.
22203 (parser::push_simplify): Assign unique id.
22204 (parser::parser): Initialize last_id.
22206 2017-07-28 Martin Liska <mliska@suse.cz>
22209 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
22210 gimple_build_debug_bind.
22212 2017-07-28 Richard Biener <rguenther@suse.de>
22214 PR tree-optimization/81502
22215 * match.pd: Add pattern combining BIT_INSERT_EXPR with
22217 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
22219 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
22220 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
22221 for BIT_FIELD_REF args.
22222 * fold-const.c (make_bit_field_ref): Likewise.
22223 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
22225 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22228 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
22229 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
22230 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
22231 Or it into SANITIZER_UNDEFINED.
22232 * ubsan.c: Include gimple-fold.h and varasm.h.
22233 (ubsan_expand_ptr_ifn): New function.
22234 (instrument_pointer_overflow): New function.
22235 (maybe_instrument_pointer_overflow): New function.
22236 (instrument_object_size): Formatting fix.
22237 (pass_ubsan::execute): Call instrument_pointer_overflow
22238 and maybe_instrument_pointer_overflow.
22239 * internal-fn.c (expand_UBSAN_PTR): New function.
22240 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
22241 * sanitizer.def (__ubsan_handle_pointer_overflow,
22242 __ubsan_handle_pointer_overflow_abort): New builtins.
22243 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
22244 * internal-fn.def (UBSAN_PTR): New internal function.
22245 * opts.c (sanitizer_opts): Add pointer-overflow.
22246 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
22247 * fold-const.c (build_range_check): Compute pointer range check in
22248 integral type if pointer arithmetics would be needed. Formatting
22251 2017-07-28 Martin Liska <mliska@suse.cz>
22254 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
22255 parameters that are of a variable-length.
22257 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
22259 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
22261 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
22262 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
22263 (CRT_CALL_STATIC_FUNCTION): Likewise.
22264 (ASM_DEFAULT_SPEC): New define.
22265 (ASM_SPEC32): Likewise.
22266 (ASM_SPEC64): Likewise.
22267 (ASM_SPEC_COMMON): Likewise.
22268 (ASM_SPEC): Likewise.
22269 (INVALID_64BIT): Likewise.
22270 (LINK_OS_DEFAULT_SPEC): Likewise.
22271 (LINK_OS_SPEC32): Likewise.
22272 (LINK_OS_SPEC64): Likewise.
22273 (POWERPC_LINUX): Likewise.
22274 (PTRDIFF_TYPE): Likewise.
22275 (RESTORE_FP_PREFIX): Likewise.
22276 (RESTORE_FP_SUFFIX): Likewise.
22277 (SAVE_FP_PREFIX): Likewise.
22278 (SAVE_FP_SUFFIX): Likewise.
22279 (SIZE_TYPE): Likewise.
22280 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
22281 (TARGET_64BIT): Likewise.
22282 (TARGET_64BIT): Likewise.
22283 (TARGET_AIX): Likewise.
22284 (WCHAR_TYPE_SIZE): Likewise.
22285 (WCHAR_TYPE): Undefine.
22286 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
22287 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
22288 (CPP_OS_RTEMS_SPEC): Delete.
22289 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
22290 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
22292 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
22294 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22296 PR tree-optimization/81578
22297 * tree-parloops.c (build_new_reduction): Bail out if
22298 reduction_code isn't one of the standard OpenMP reductions.
22299 Move the details printing after that decision.
22301 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22303 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
22304 related to reload_in_progress.
22305 (splat_input_operand): Likewise.
22306 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
22308 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
22310 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
22311 (TARGET_INSTANTIATE_DECLS): Likewise.
22312 (legitimate_indexed_address_p): Delete reload_in_progress code.
22313 (rs6000_debug_legitimate_address_p): Likewise.
22314 (rs6000_eliminate_indexed_memrefs): Likewise.
22315 (rs6000_emit_le_vsx_store): Likewise.
22316 (rs6000_emit_move_si_sf_subreg): Likewise.
22317 (rs6000_emit_move): Likewise.
22318 (register_to_reg_type): Likewise.
22319 (rs6000_pre_atomic_barrier): Likewise.
22320 (rs6000_machopic_legitimize_pic_address): Likewise.
22321 (rs6000_allocate_stack_temp): Likewise.
22322 (rs6000_address_for_fpconvert): Likewise.
22323 (rs6000_address_for_altivec): Likewise.
22324 (rs6000_secondary_memory_needed_rtx): Delete function.
22325 (rs6000_check_sdmode): Likewise.
22326 (rs6000_alloc_sdmode_stack_slot): Likewise.
22327 (rs6000_instantiate_decls): Likewise.
22328 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
22329 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
22330 Delete reload_in_progress.
22331 (*vec_reload_and_plus_<mptrsize>): Likewise.
22332 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
22333 (vsx_div_v2di): Likewise.
22334 (vsx_udiv_v2di): Likewise.
22336 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22338 * config/rs6000/rs6000.opt (mlra): Replace with stub.
22339 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
22340 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
22341 (rs6000_debug_reg_global): Delete print of LRA status.
22342 (rs6000_option_override_internal): Delete dead LRA related code.
22343 (rs6000_lra_p): Delete function.
22344 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
22346 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
22348 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
22349 * config/riscv/rtems.h: New file.
22351 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22352 Sudakshina Das <sudi.das@arm.com>
22354 * config/aarch64/aarch64.md
22355 (define_split for and<mode>3nr_compare): Move
22356 non aarch64_logical_operand to a register.
22357 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
22358 register immediate operand to a register.
22359 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
22361 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22363 PR middle-end/81564
22364 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
22366 2017-07-27 Richard Biener <rguenther@suse.de>
22368 PR tree-optimization/81573
22369 PR tree-optimization/81494
22370 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
22371 multi defuse cycle case.
22373 2017-07-27 Richard Biener <rguenther@suse.de>
22375 PR tree-optimization/81571
22376 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
22379 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
22381 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
22382 earlier and only if MASK_FPU is set. Adjust formatting.
22384 2017-07-27 Martin Liska <mliska@suse.cz>
22386 * opt-functions.awk: Add validation of value of Init.
22387 * optc-gen.awk: Pass new argument.
22389 2017-07-27 Martin Liska <mliska@suse.cz>
22391 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
22392 Fix wrong condition.
22394 2017-07-27 Martin Liska <mliska@suse.cz>
22396 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
22397 BBs and edges seen by autoFDO.
22399 2017-07-27 Richard Biener <rguenther@suse.de>
22401 PR tree-optimization/81502
22402 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
22403 with incompatible but same sized type.
22404 (execute_update_addresses_taken): Likewise.
22406 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
22408 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
22409 flag_tree_loop_vectorize rather than flag_tree_vectorize.
22411 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22414 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
22415 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
22416 Change s_operand to memory_operand.
22418 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
22420 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
22421 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
22422 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
22423 Emit instructions rather than returning an expression. Handle TFmode
22424 and KFmode by casting to TImode.
22425 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
22426 (rs6000_emit_le_vsx_store): Likewise.
22427 * config/rs6000/vsx.md (VSX_TI): New iterator.
22428 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
22429 (*vsx_le_undo_permute_<mode>): Likewise.
22430 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
22431 emit the split sequence.
22432 (*vsx_le_perm_store_<mode>): Likewise.
22434 2017-07-27 Jakub Jelinek <jakub@redhat.com>
22436 PR tree-optimization/81555
22437 PR tree-optimization/81556
22438 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
22439 if true, force CHANGED for the recursive invocation.
22440 (reassociate_bb): Remember original length of ops array, pass
22441 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
22443 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
22444 attributes for noipa attribute. For naked attribute use
22445 lookup_attribute first before lookup_attribute_spec.
22446 * final.c (rest_of_handle_final): Disable IPA RA for functions with
22448 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
22449 for functions with noipa attribute.
22450 (cgraph_externally_visible_p): Return true for functions with noipa
22452 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
22453 for functions with noipa attribute.
22454 * doc/extend.texi: Document noipa function attribute.
22455 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
22456 also for functions with noipa attribute.
22457 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
22459 2017-07-26 Andrew Pinski <apinski@cavium.com>
22461 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
22462 vec_unalign_load_cost and vec_unalign_store_cost.
22464 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
22466 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
22467 -mvsx-small-integer option.
22468 (ISA_3_0_MASKS_IEEE): Likewise.
22469 (OTHER_VSX_VECTOR_MASKS): Likewise.
22470 (POWERPC_MASKS): Likewise.
22471 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
22472 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
22473 code, only testing for DImode being allowed in non-VSX floating
22475 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
22476 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
22478 (rs6000_option_override_internal): Delete -mvsx-small-integer.
22479 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
22480 TARGET_P8_VECTOR test.
22481 (rs6000_secondary_reload_simple_move): Likewise.
22482 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
22483 since TARGET_P9_VECTOR was already tested.
22484 (rs6000_opt_masks): Remove -mvsx-small-integer.
22485 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
22486 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22488 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
22489 test for TARGET_VEXTRACTUB was used, and that uses
22491 (p9 extract splitter): Likewise.
22492 (vsx_extract_<mode>_di_p9): Likewise.
22493 (vsx_extract_<mode>_store_p9): Likewise.
22494 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
22495 for TARGET_P9_VECTOR was used. Delete code that is now dead with
22496 the elimination of TARGET_VSX_SMALL_INTEGER.
22497 (vsx_extract_<mode>_p8): Likewise.
22498 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
22499 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
22500 (vsx_set_<mode>_p9): Likewise.
22501 (vsx_set_v4sf_p9): Likewise.
22502 (vsx_set_v4sf_p9_zero): Likewise.
22503 (vsx_insert_extract_v4sf_p9): Likewise.
22504 (vsx_insert_extract_v4sf_p9_2): Likewise.
22505 * config/rs6000/rs6000.md (sign extend splitter): Change
22506 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
22507 (floatsi<mode>2_lfiwax_mem): Likewise.
22508 (floatunssi<mode>2_lfiwzx_mem): Likewise.
22509 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
22510 since a test for TARGET_P9_VECTOR was used.
22511 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22512 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
22513 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22514 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22515 TARGET_P8_VECTOR test.
22516 (fix_trunc<mode>si2_stfiwx): Likewise.
22517 (fix_trunc<mode>si2_internal): Likewise.
22518 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
22519 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22521 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22522 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22523 TARGET_P8_VECTOR test.
22524 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22525 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
22526 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22528 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22529 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
22530 since a test for TARGET_P9_VECTOR was used.
22531 (splitter for loading small constants): Likewise.
22533 2017-07-26 Andrew Pinski <apinski@cavium.com>
22535 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
22538 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
22541 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
22542 (fp_valid_at): Likewise.
22544 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22546 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
22547 (qdf24xx_addrcost_table): Likewise.
22548 (cortexa57_tunings): Update to use generic_branch_cost.
22549 (cortexa72_tunings): Likewise.
22550 (cortexa73_tunings): Likewise.
22551 (qdf24xx_tunings): Likewise.
22553 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22555 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
22556 (thunderx2t99_branch_cost): Likewise.
22557 (cortexa35_tunings): Update to use generic_branch_cost.
22558 (cortexa53_tunings): Likewise.
22559 (cortexa57_tunings): Likewise.
22560 (cortexa72_tunings): Likewise.
22561 (cortexa73_tunings): Likewise.
22562 (thunderx2t99_tunings): Likewise.
22564 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22566 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22567 (sparc_option_override): Honour MASK_FSMULD.
22568 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22569 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22570 * config/sparc/sparc.opt (mfsmuld): New option.
22571 * doc/invoke.texi (mfsmuld): Document option.
22573 2017-07-26 Marek Polacek <polacek@redhat.com>
22575 PR middle-end/70992
22576 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22578 2017-07-26 Richard Biener <rguenther@suse.de>
22580 * gimple-match-head.c (do_valueize): Return OP if valueize
22582 (get_def): New helper to get at the def stmt of a SSA name
22583 if valueize allows.
22584 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22585 do_valueize to get at the def stmt.
22586 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22588 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22590 PR middle-end/46932
22591 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22593 2017-07-26 Martin Liska <mliska@suse.cz>
22596 * function.c (expand_function_start): Make expansion of
22597 nonlocal_goto_save_area after parm_birth_insn.
22599 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22601 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22602 from all CPU target flags enable members.
22604 2017-07-26 Richard Biener <rguenther@suse.de>
22606 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22607 (decision_tree::gen): Make 'type' const.
22608 (write_predicate): Likewise.
22610 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22612 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22613 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22614 (rs6000_option_override_internal): Likewise.
22615 (rs6000_expand_vector_set): Likewise.
22616 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22617 (TARGET_UPPER_REGS_SF): Likewise.
22618 (TARGET_UPPER_REGS_DI): Likewise.
22619 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22620 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22621 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22622 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22623 (Splitters for DI constants in Altivec registers): Likewise.
22624 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22625 (vsx_set_v4sf_p9): Likewise.
22626 (vsx_set_v4sf_p9_zero): Likewise.
22627 (vsx_insert_extract_v4sf_p9): Likewise.
22628 (vsx_insert_extract_v4sf_p9_2): Likewise.
22630 2017-07-25 Carl Love <cel@us.ibm.com>
22632 * doc/extend.texi: Update the built-in documentation file for the
22633 existing built-in functions
22634 vector signed char vec_cnttz (vector signed char);
22635 vector unsigned char vec_cnttz (vector unsigned char);
22636 vector signed short vec_cnttz (vector signed short);
22637 vector unsigned short vec_cnttz (vector unsigned short);
22638 vector signed int vec_cnttz (vector signed int);
22639 vector unsigned int vec_cnttz (vector unsigned int);
22640 vector signed long long vec_cnttz (vector signed long long);
22641 vector unsigned long long vec_cnttz (vector unsigned long long);
22643 2017-07-25 Andrew Pinski <apinski@cavium.com>
22645 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22646 accesses where the use is for the first operand of a BIT_INSERT.
22648 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22651 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22652 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22654 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22656 * config/i386/gstabs.h: Delete.
22657 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22659 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22661 * config/i386/i386.c (ix86_decompose_address): Do not check for
22662 register RTX when looking at index_reg or base_reg.
22663 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22665 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22667 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22668 to update EH info here.
22670 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22672 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22674 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22676 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22678 2017-07-25 Torsten Duwe <duwe@suse.de>
22680 * common.opt: Introduce -fpatchable-function-entry
22681 command line option, and its variables function_entry_patch_area_size
22682 and function_entry_patch_area_start.
22683 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22684 including a two-value parser.
22685 * target.def (print_patchable_function_entry): New target hook.
22686 * targhooks.h (default_print_patchable_function_entry): New function.
22687 * targhooks.c (default_print_patchable_function_entry): Likewise.
22688 * toplev.c (process_options): Switch off IPA-RA if
22689 patchable function entries are being generated.
22690 * varasm.c (assemble_start_function): Look at the
22691 patchable-function-entry command line switch and current
22692 function attributes and maybe generate NOP instructions by
22693 calling the print_patchable_function_entry hook.
22694 * doc/extend.texi: Document patchable_function_entry attribute.
22695 * doc/invoke.texi: Document -fpatchable_function_entry
22696 command line option.
22697 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22699 * doc/tm.texi: Re-generate.
22701 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22704 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22705 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22707 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22709 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22711 2017-07-25 Richard Biener <rguenther@suse.de>
22713 PR tree-optimization/81455
22714 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22715 not walk in cycles when looking for guards.
22717 2017-07-25 Richard Biener <rguenther@suse.de>
22719 PR tree-optimization/81529
22720 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22721 when optimizing backedge uses.
22723 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22725 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22727 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22728 to dl_section_ref. On AIX, append an expression to subtract
22729 the size of the section length to dl_section_ref.
22731 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22733 * configure.ac: If any of the config.* scripts fail, exit 1.
22734 * configure: Regenerate.
22736 2017-07-25 Richard Biener <rguenther@suse.de>
22738 PR middle-end/81546
22739 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22740 of immediate uses, be more verbose on errors.
22742 2017-07-25 Richard Biener <rguenther@suse.de>
22744 PR tree-optimization/81510
22745 * tree-vect-loop.c (vect_is_simple_reduction): When the
22746 reduction stmt is not inside the loop bail out.
22748 2017-07-25 Richard Biener <rguenther@suse.de>
22750 PR tree-optimization/81303
22751 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22752 profitability check against LOOP_VINFO_NITERSM1.
22754 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22756 * domwalk.c (cmp_bb_postorder): Simplify.
22757 (sort_bbs_postorder): New function. Use it...
22758 (dom_walker::walk): ...here to optimize common cases.
22760 2017-07-25 Martin Liska <mliska@suse.cz>
22763 * ipa-visibility.c (function_and_variable_visibility): Make the
22764 redirection just on target that supports aliasing.
22765 Fix GNU coding style.
22767 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22770 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22772 * config/arm/rtems.h: Likewise.
22773 * config/bfin/rtems.h: Likewise.
22774 * config/i386/rtemself.h: Likewise.
22775 * config/lm32/rtems.h: Likewise.
22776 * config/m32c/rtems.h: Likewise.
22777 * config/m68k/rtemself.h: Likewise.
22778 * config/microblaze/rtems.h: Likewise.
22779 * config/mips/rtems.h: Likewise.
22780 * config/moxie/rtems.h: Likewise.
22781 * config/nios2/rtems.h: Likewise.
22782 * config/powerpcspe/rtems.h: Likewise.
22783 * config/rs6000/rtems.h: Likewise.
22784 * config/rtems.h: Likewise.
22785 * config/sh/rtems.h: Likewise.
22786 * config/sh/rtemself.h: Likewise.
22787 * config/sparc/rtemself.h: Likewise.
22789 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22792 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22793 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22794 * tree-ssa-structalias.c (alias_get_name): Same.
22796 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22799 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22800 instructions if no du chain is found.
22802 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22804 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22806 2017-07-25 Richard Biener <rguenther@suse.de>
22808 PR middle-end/81505
22809 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
22812 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22814 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
22815 upper-regs options.
22816 (ISA_2_7_MASKS_SERVER): Likewise.
22817 (ISA_3_0_MASKS_IEEE): Likewise.
22818 (OTHER_P8_VECTOR_MASKS): Likewise.
22819 (OTHER_VSX_VECTOR_MASKS): Likewise.
22820 (POWERPC_MASKS): Likewise.
22821 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
22822 duplicate list of options.
22823 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
22824 explicit -mupper-regs options.
22825 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
22826 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
22827 alias for -mupper-regs-df.
22828 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
22829 (rs6000_init_hard_regno_mode_ok): Likewise.
22830 (rs6000_option_override_internal): Likewise.
22831 (rs6000_opt_masks): Likewise.
22832 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
22833 options in terms of whether -mvsx or -mpower8-vector was used.
22834 (TARGET_UPPER_REGS_DI): Likewise.
22835 (TARGET_UPPER_REGS_SF): Likewise.
22836 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
22837 -mupper-regs-* options.
22839 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22841 * passes.c (emergency_dump_function): Print some empty lines and a
22842 header before the RTL dump.
22844 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22846 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
22848 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
22851 * config/aarch64/aarch64.c (aarch64_classify_symbol):
22852 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
22854 2017-07-24 Carl Love <cel@us.ibm.com>
22856 * config/rs6000/rs6000-c.c: Add support for built-in functions
22857 vector float vec_extract_fp32_from_shorth (vector unsigned short);
22858 vector float vec_extract_fp32_from_shortl (vector unsigned short);
22859 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
22860 vec_extract_fp_from_shortl): Add defines for the two builtins.
22861 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
22862 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
22864 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
22865 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
22866 * doc/extend.texi: Update the built-in documentation file for the
22867 new built-in function.
22869 2017-07-24 Jakub Jelinek <jakub@redhat.com>
22872 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
22874 * doc/generic.texi: Likewise.
22875 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22876 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22878 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
22880 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
22881 (aarch64_mls_elt_merge<mode>): Likewise.
22883 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
22885 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22886 having __cxa_atexit.
22888 2017-07-23 Michael Collison <michael.collison@arm.com>
22890 * config/arm/arm.c (arm_option_override): Deprecate
22891 use of -mstructure-size-boundary.
22892 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
22893 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
22895 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22898 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22899 Reduce cost estimate for direct moves.
22901 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
22904 * config/i386/i386.c (ix86_option_override_internal): Disable
22905 BMI, BMI2 and TBM instructions for -m16.
22907 2017-07-21 Carl Love <cel@us.ibm.com>
22909 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22910 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22911 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22912 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22913 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22914 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22915 VMULOSW): New enum "unspec" values.
22916 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22917 altivec_vmulosw): New patterns.
22918 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22919 VMULOSW): Add definitions.
22921 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
22923 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
22924 (qdf24xx): Likewise.
22925 * config/aarch64/aarch64-options-extensions.def (rdma); New.
22926 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
22927 (AARCH64_FL_V8_1): Renumber.
22928 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
22929 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
22930 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
22931 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
22932 rdma to feature modifiers list.
22934 2017-07-21 Yury Gribov <tetra2005@gmail.com>
22936 PR middle-end/56727
22937 * ipa-visibility (function_and_variable_visibility): Convert
22938 recursive PLT call to direct call if appropriate.
22940 2017-07-21 Andrew Pinski <apinski@cavium.com>
22942 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
22943 operand 1 to see if the types precision matches.
22944 * fold-const.c (operand_equal_p): Likewise.
22946 2017-07-21 Richard Biener <rguenther@suse.de>
22948 PR tree-optimization/81303
22949 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
22950 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
22951 (vect_peeling_hash_get_lowest_cost): Adjust.
22952 (vect_enhance_data_refs_alignment): Likewise. Use
22953 vect_get_peeling_costs_all_drs to compute the penalty for no
22954 peeling to match up costs.
22956 2017-07-21 Richard Biener <rguenther@suse.de>
22958 PR tree-optimization/81500
22959 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
22960 we didn't identify a reduction path.
22962 2017-07-21 Tom de Vries <tom@codesourcery.com>
22963 Cesar Philippidis <cesar@codesourcery.com>
22965 PR gcov-profile/81442
22966 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
22969 2017-07-21 Tom de Vries <tom@codesourcery.com>
22972 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
22974 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
22975 nvptx_override_options_after_change.
22977 2017-07-21 Ulrich Drepper <drepper@redhat.com>
22979 * dwarf2out.c (output_file_names): Avoid double testing for
22980 dwarf_version >= 5.
22982 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
22984 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
22986 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22988 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
22990 (try_crossjump_to_edge): Do not punt on partitioned functions.
22992 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22994 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
22995 Put all BBs reachable only via paths crossing cold region to cold
22997 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
22999 2016-07-21 Richard Biener <rguenther@suse.de>
23001 PR tree-optimization/81303
23002 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23003 into account prologue and epilogue iterations when raising
23004 min_profitable_iters to sth at least covering one vector iteration.
23006 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23008 * config/arm/arm.c (arm_test_cpu_arch_dat):
23011 2017-07-20 Nathan Sidwell <nathan@acm.org>
23013 Remove TYPE_METHODS.
23014 * tree.h (TYPE_METHODS): Delete.
23015 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23016 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23017 (dbxout_type_methods): Scan TYPE_FIELDS.
23018 (dbxout_type): Don't check TYPE_METHODS here.
23019 * function.c (use_register_for_decl): Always ignore register for
23020 class types when not optimizing.
23021 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23022 * tree.c (free_lang_data_in_type): Stitch out member functions and
23023 templates from TYPE_FIELDS.
23024 (build_distinct_type_copy, verify_type_variant,
23025 verify_type): Member fns are on TYPE_FIELDS.
23026 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23027 * tree-pretty-print.c (dump_generic_node): Likewise.
23029 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23032 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23033 V2TImode and V4TImode.
23034 (ix86_expand_vector_extract): Likewise.
23035 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23036 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23037 (ssescalarmode): Handle V4TImode and V2TImode.
23038 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23039 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23040 (VEXTRACTI128_MODE): New mode iterator.
23041 (splitter for *vec_extractv?ti first element): New.
23042 (VEC_INIT_MODE): New mode iterator.
23043 (vec_init<mode>): Consolidate 3 expanders into one using
23044 VEC_INIT_MODE mode iterator.
23046 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23048 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23049 non_spilled_static_chain_regno_p.
23051 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23053 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
23055 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
23057 * bb-reorder.c (connect_traces): Allow copying of blocks within
23060 2017-07-20 Richard Biener <rguenther@suse.de>
23062 * gimple.h (gimple_phi_result): Add gphi * overload.
23063 (gimple_phi_result_ptr): Likewise.
23064 (gimple_phi_arg): Likewise. Adjust index assert to only
23065 allow actual argument accesses rather than all slots available
23067 (gimple_phi_arg_def): Add gphi * overload.
23068 * tree-phinodes.c (make_phi_node): Initialize only actual
23070 (resize_phi_node): Clear memory not covered by old node,
23071 do not initialize excess argument slots.
23072 (reserve_phi_args_for_new_edge): Initialize new argument slot
23075 2017-07-20 Bin Cheng <bin.cheng@arm.com>
23077 PR tree-optimization/81388
23079 2016-07-21 Bin Cheng <bin.cheng@arm.com>
23081 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
23082 by removing computation of may_be_zero.
23084 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23085 Tom de Vries <tom@codesourcery.com>
23087 PR middle-end/81030
23088 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
23089 when gimple level profile disagrees with what RTL expander did.
23091 2017-07-20 Richard Biener <rguenther@suse.de>
23093 PR tree-optimization/61171
23094 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
23095 (vect_analyze_stmt): Add slp instance parameter.
23096 (vectorizable_reduction): Likewise.
23097 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
23098 (vect_is_simple_reduction): Deal with chains not detected
23099 as SLP reduction chain, specifically not properly associated
23100 chains containing a mix of plus/minus.
23101 (get_reduction_op): Remove.
23102 (get_initial_defs_for_reduction): Simplify, pass in whether
23103 this is a reduction chain, pass in the SLP node for the PHIs.
23104 (vect_create_epilog_for_reduction): Get the SLP instance as
23106 (vectorizable_reduction): Get the SLP instance as arg.
23107 During analysis remember the SLP node with the PHIs in the
23108 instance. Simplify getting at the vectorized reduction PHIs.
23109 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
23110 through SLP instance.
23111 (vect_slp_analyze_operations): Likewise.
23112 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
23113 (vect_transform_stmt): Likewise.
23115 2017-07-20 Tom de Vries <tom@codesourcery.com>
23117 PR tree-optimization/81489
23118 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
23119 read of phi arg location to before loop that modifies phi.
23121 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23123 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
23126 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23128 PR middle-end/81331
23129 * except.c (execute): Fix ordering issue.
23131 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23133 PR rtl-optimization/81423
23134 * combine.c (make_compound_operation_int): Don't try to optimize
23135 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
23137 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23139 PR rtl-optimization/81423
23140 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
23141 with a constant that is -1 in the truncated to mode.
23143 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23145 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
23146 (determine_unlikely_bbs): ... here.
23147 * predict.h (propagate_unlikely_bbs_forward): Declare.
23148 * cfgexpand.c (pass_expand::execute): Use it.
23149 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
23151 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
23152 propagate_unlikely_bbs_forward.
23154 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23156 PR middle-end/81331
23157 * except.c (maybe_add_nop_after_section_switch): New function.
23160 2017-07-19 Tom de Vries <tom@codesourcery.com>
23162 * gimple.h (gimple_phi_set_arg): Make assert more strict.
23164 2017-07-19 Tom de Vries <tom@codesourcery.com>
23166 * gimple.h (gimple_phi_arg): Make assert more strict.
23168 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
23170 * config.gcc (powerpc*-*-*): Add mmintrin.h.
23171 * config/rs6000/mmintrin.h: New file.
23172 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
23174 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23176 PR tree-optimization/81346
23177 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
23179 2017-07-19 Tom de Vries <tom@codesourcery.com>
23181 * config/nvptx/nvptx.md (VECIM): Add V2DI.
23183 2017-07-19 Tom de Vries <tom@codesourcery.com>
23185 * config/nvptx/nvptx-modes.def: Add V2DImode.
23186 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
23187 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
23188 (nvptx_output_mov_insn): Handle lack of mov.b128.
23189 (nvptx_print_operand): Handle 'H' and 'L' codes.
23190 (nvptx_vector_mode_supported): Allow V2DImode.
23191 (nvptx_preferred_simd_mode): New function.
23192 (nvptx_data_alignment): New function.
23193 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
23194 nvptx_preferred_simd_mode.
23195 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
23197 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
23199 2017-07-19 Tom de Vries <tom@codesourcery.com>
23201 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
23202 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
23203 (nvptx_vector_mode_supported): New function. Allow V2SImode.
23204 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
23205 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
23206 (mov<VECIM>_insn): New define_insn.
23207 (define_expand "mov<VECIM>): New define_expand.
23209 2017-07-19 Tom de Vries <tom@codesourcery.com>
23211 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
23213 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23215 PR tree-optimization/81346
23216 * fold-const.h (fold_div_compare, range_check_type): Declare.
23217 * fold-const.c (range_check_type): New function.
23218 (build_range_check): Use range_check_type.
23219 (fold_div_compare): No longer static, rewritten into
23220 a match.pd helper function.
23221 (fold_comparison): Don't call fold_div_compare here.
23222 * match.pd (X / C1 op C2): New optimization using fold_div_compare
23223 as helper function.
23225 2017-07-19 Nathan Sidwell <nathan@acm.org>
23227 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
23228 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
23229 * tree.c (find_decls_types_r, verify_type): Use
23230 TYPE_{MIN,MAX}_VALUE_RAW.
23231 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
23232 (hash_tree): Likewise.
23233 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
23235 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
23238 2017-07-18 Tom de Vries <tom@codesourcery.com>
23240 PR middle-end/81464
23241 * omp-expand.c (expand_omp_for_static_chunk): Handle
23242 equal-argument loop exit phi.
23244 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
23247 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
23248 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
23249 operand 2 predicate.
23250 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
23251 operand 2 predicate.
23252 (ror,rol -> rorx splitters): Use const_int_operand as
23253 operand 2 predicate.
23255 2017-06-18 Richard Biener <rguenther@suse.de>
23257 PR tree-optimization/81410
23258 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
23259 the gap in the ! slp_perm SLP case after each group.
23261 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23263 PR middle-end/81463
23264 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
23267 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23269 PR middle-end/81462
23270 * predict.c (set_even_probabilities): Cleanup; do not affect
23271 probabilities that are already known.
23272 (combine_predictions_for_bb): Call even when count is set.
23274 2017-07-18 Nathan Sidwell <nathan@acm.org>
23276 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
23279 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23282 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
23283 optimization for loop niter analysis.
23285 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
23288 * config/avr/avr.c (avr_optimize_casesi): Don't use
23289 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
23291 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
23293 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
23294 body_cost_vec from _vect_peel_extended_info.
23295 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
23296 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
23299 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23301 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
23303 2017-07-18 Richard Biener <rguenther@suse.de>
23305 PR tree-optimization/80620
23306 PR tree-optimization/81403
23307 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
23308 info when re-using a VN table entry.
23310 2017-07-18 Richard Biener <rguenther@suse.de>
23312 PR tree-optimization/81418
23313 * tree-vect-loop.c (vectorizable_reduction): Properly compute
23314 vectype_in. Verify that with lane-reducing reduction operations
23315 we have a single def-use cycle.
23317 2017-07-17 Carl Love <cel@us.ibm.com>
23319 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
23321 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23322 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23323 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23324 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23325 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23326 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23327 VMULOSW): New enum "unspec" values.
23328 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23329 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23330 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23331 altivec_vmulosw): New patterns.
23332 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23333 VMULOSW): Add definitions.
23335 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
23337 * config/alpha/alpha.c: Include predict.h.
23339 2017-07-17 Yury Gribov <tetra2005@gmail.com>
23341 * tree-vrp.c (compare_assert_loc): Fix comparison function
23342 to return predictable results.
23344 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23346 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
23348 (subdi3): Likewise.
23349 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
23350 * doc/invoke.texi (mexpand-adddi): Update text.
23352 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23354 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
23355 that also clobbers the CC register. The old expand code is moved
23357 (*arc_clzsi2): ... here.
23358 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
23359 the CC register. The old expand code is moved to ...
23360 (arc_ctzsi2): ... here.
23362 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23364 * config/arc/arc.opt (mindexed-loads): Use initial value
23365 TARGET_INDEXED_LOADS_DEFAULT.
23366 (mauto-modify-reg): Use initial value
23367 TARGET_AUTO_MODIFY_REG_DEFAULT.
23368 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23369 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23370 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23371 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23373 2017-07-17 Martin Liska <mliska@suse.cz>
23376 * opts.c (finish_options): Do not allow -fgnu-tm
23377 w/ -fsanitize={kernel-,}address. Say sorry.
23379 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23382 * tree-loop-distribution.c (classify_partition): Only assert on
23383 numer of iterations.
23384 (merge_dep_scc_partitions): Delete prameter. Update function call.
23385 (distribute_loop): Remove code handling loop with unknown niters.
23386 (pass_loop_distribution::execute): Skip loop with unknown niters.
23388 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23391 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
23392 function sort_partitions_by_post_order.
23394 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23396 PR tree-optimization/81374
23397 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
23398 the max index of basic blocks, rather than number of basic blocks.
23400 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23402 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
23404 (arc_legitimate_pic_operand_p): Likewise.
23405 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
23407 (arc_needs_pcl_p): Likewise.
23408 (arc_legitimate_pc_offset_p): Likewise.
23409 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
23410 function is also used in constrains.md.
23411 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
23412 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
23413 PLUS. Only return true/false in known cases, otherwise assert.
23414 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
23415 is already called in arc_legitimate_constant_p.
23416 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
23418 (LEGITIMATE_PIC_OPERAND_P): Use
23419 arc_raw_symbolic_reference_mentioned_p function.
23420 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
23425 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23426 Andrew Burgess <andrew.burgess@embecosm.com>
23428 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
23429 (arc_return_address_register): New function.
23430 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
23431 (arc_handle_fndecl_attribute): Add naked attribute.
23432 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
23433 (TARGET_WARN_FUNC_RETURN): Likewise.
23434 (arc_allocate_stack_slots_for_args): New function.
23435 (arc_warn_func_return): Likewise.
23436 (machine_function): Change type fn_type.
23437 (arc_compute_function_type): Consider new naked function type,
23438 change function return type.
23439 (arc_must_save_register): Adapt to handle new
23440 arc_compute_function_type's return type.
23441 (arc_expand_prologue): Likewise.
23442 (arc_expand_epilogue): Likewise.
23443 (arc_return_address_regs): Delete.
23444 (arc_return_address_register): New function.
23445 (arc_epilogue_uses): Use above function.
23446 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
23447 (arc_function_type): Change encoding, add naked type.
23448 (ARC_INTERRUPT_P): Change to handle the new encoding.
23449 (ARC_FAST_INTERRUPT_P): Likewise.
23450 (ARC_NORMAL_P): Define.
23451 (ARC_NAKED_P): Likewise.
23452 (arc_compute_function_type): Delete prototype.
23453 * config/arc/arc.md (in_ret_delay_slot): Use
23454 arc_return_address_register function.
23455 (simple_return): Likewise.
23456 (p_return_i): Likewise.
23458 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23460 PR tree-optimization/81428
23461 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
23462 can't be built for those types.
23464 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23466 Remove stuff dead since r239246.
23468 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
23469 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
23470 (avr_inform_devices): Remove dead stuff.
23472 2017-07-17 Tamar Christina <tamar.christina@arm.com>
23474 * config/arm/arm_neon.h: Fix softp typo.
23476 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23478 PR tree-optimization/81365
23479 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
23480 aggregate moves onto bb predecessor edges, make sure there are no
23481 loads that could alias the lhs in between the start of bb and the
23484 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23487 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
23488 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
23489 [LSHIFTRT, outer_code = TRUNCATE]: Same.
23491 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23493 PR tree-optimization/81396
23494 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
23495 (init_symbolic_number): Initialize it to 1.
23496 (perform_symbolic_merge): Add n_ops from both operands into the new
23498 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
23499 without base_addr as useless if they need more than one operation.
23500 (bswap_replace): Handle !bswap case for NULL base_addr.
23502 2017-07-17 Tom de Vries <tom@codesourcery.com>
23505 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
23508 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
23510 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
23511 conditional builtin define __FIX_LEON3FT_B2BST.
23513 2017-07-17 Daniel Cederman <cederman@gaisler.com>
23515 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
23516 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
23519 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23521 PR rtl-optimization/81424
23522 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
23523 to remove potential trapping from operands if -fnon-call-exceptions.
23525 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23527 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
23528 profile_proability for scalling.
23529 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
23531 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23533 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
23535 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23537 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
23538 fixpoint arithmetics.
23540 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23542 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
23543 fixpoint arithmetics.
23545 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23547 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
23548 fixpoint arithmetics.
23550 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23552 * profile-count.h (profile_probability::from_reg_br_prob_note,
23553 profile_probability::to_reg_br_prob_note): New functions.
23554 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
23555 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
23556 * predict.c (probability_reliable_p): Update.
23557 (edge_probability_reliable_p): Update.
23558 (br_prob_note_reliable_p): Update.
23559 (invert_br_probabilities): Update.
23560 (add_reg_br_prob_note): New function.
23561 (combine_predictions_for_insn): Update.
23562 * asan.c (asan_clear_shadow): Update.
23563 * cfgbuild.c (compute_outgoing_frequencies): Update.
23564 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23565 (update_br_prob_note): Update.
23566 (rtl_verify_edges): Update.
23567 (purge_dead_edges): Update.
23568 (fixup_reorder_chain): Update.
23569 * emit-rtl.c (try_split): Update.
23570 * ifcvt.c (cond_exec_process_insns): Update.
23571 (cond_exec_process_if_block): Update.
23572 (dead_or_predicable): Update.
23573 * internal-fn.c (expand_addsub_overflow): Update.
23574 (expand_neg_overflow): Update.
23575 (expand_mul_overflow): Update.
23576 * loop-doloop.c (doloop_modify): Update.
23577 * loop-unroll.c (compare_and_jump_seq): Update.
23578 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23579 * predict.h: Update.
23580 * reorg.c (mostly_true_jump): Update.
23582 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23583 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23584 * config/arc/arc.c: (emit_unlikely_jump): Update.
23585 * config/arm/arm.c: (emit_unlikely_jump): Update.
23586 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23587 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23588 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23589 (ix86_print_operand): Update.
23590 (ix86_split_fp_branch): Update.
23591 (predict_jump): Update.
23592 * config/ia64/ia64.c (ia64_print_operand): Update.
23593 * config/mmix/mmix.c (mmix_print_operand): Update.
23594 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23595 (rs6000_expand_split_stack_prologue): Update.
23596 * config/rs6000/rs6000.c: Update.
23597 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23598 (s390_expand_vec_movstr): Update.
23599 (s390_expand_cs_tdsi): Update.
23600 (s390_expand_split_stack_prologue): Update.
23601 * config/sh/sh.c (sh_print_operand): Update.
23602 (expand_cbranchsi4): Update.
23603 (expand_cbranchdi4): Update.
23604 * config/sparc/sparc.c (output_v9branch): Update.
23605 * config/spu/spu.c (get_branch_target): Update.
23606 (ea_load_store_inline): Update.
23607 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23608 * config/tilepro/tilepro.c: Update.
23610 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23612 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23613 (gimplify_save_expr): Likewise.
23615 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23617 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23619 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23621 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23623 * timevar.def (TV_IPA_FNSUMMARY): Define.
23625 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23627 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23628 to back store errata sensitive sequence from being generated.
23629 (sqrtdf2_fix): Likewise.
23631 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23633 * tree-ssa-threadupdate.c (compute_path_counts,
23634 update_joiner_offpath_counts): Use profile_probability.
23636 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23639 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23641 * config/arm/arm-c.c (arm_cpu_builtins): Define
23642 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23644 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23646 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23647 array entries to represent __ieee128 versions of the
23648 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23649 scalar_extract_sig, and scalar_insert_exp built-in functions.
23650 (altivec_resolve_overloaded_builtin): Add special case handling
23651 for the __builtin_scalar_insert_exp function, as represented by
23652 the P9V_BUILTIN_VEC_VSIEDP constant.
23653 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23654 exponent support for __ieee128 argument.
23655 (VSESQP): Add scalar extract signature support for __ieee128
23657 (VSTDCNQP): Add scalar test negative support for __ieee128
23659 (VSIEQP): Add scalar insert exponent support for __int128 argument
23660 with __ieee128 result.
23661 (VSIEQPF): Add scalar insert exponent support for __ieee128
23662 argument with __ieee128 result.
23663 (VSTDCQP): Add scalar test data class support for __ieee128
23665 (VSTDCNQP): Add overload support for scalar test negative with
23666 __ieee128 argument.
23667 (VSTDCQP): Add overload support for scalar test data class
23668 __ieee128 argument.
23669 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23670 UNSPEC_VSX_SXSIGDP.
23671 (UNSPEC_VSX_SIEXPQP): New constant.
23672 (xsxexpqp): New insn for VSX scalar extract exponent quad
23674 (xsxsigqp): New insn for VSX scalar extract significand quad
23676 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23677 precision with floating point argument.
23678 (xststdcqp): New expand for VSX scalar test data class quad
23680 (xststdcnegqp): New expand for VSX scalar test negative quad
23682 (xststdcqp): New insn to match expansions for VSX scalar test data
23683 class quad precision and VSX scalar test negative quad precision.
23684 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23685 special case operand checking to enforce that second operand of
23686 VSX scalar test data class with quad precision argument is a 7-bit
23688 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23689 prototypes and descriptions of __ieee128 versions of
23690 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23691 scalar_test_data_class, and scalar_test_neg built-in functions.
23693 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23695 PR tree-optimization/81162
23696 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23697 replace a negate with an add.
23699 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23701 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23703 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23705 * config/arm/arm-c.c (arm_cpu_builtins): Define
23706 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23708 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23710 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23711 (armv8-r): Set ARM Cortex-R52 as default CPU.
23712 * config/arm/arm-tables.opt: Regenerate.
23713 * config/arm/arm-tune.md: Regenerate.
23714 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23716 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23717 extension for -mcpu=cortex-r52.
23719 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23721 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23722 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23723 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23724 (fp-armv8): Define it as FP_ARMv8 only.
23725 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23726 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23728 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23729 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23730 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23731 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23732 than TARGET_FPU_ARMV8.
23733 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23734 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23735 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23737 * config/arm/neon.md (neon_vrint): Likewise.
23738 (neon_vcvt): Likewise.
23739 (neon_<fmaxmin_op><mode>): Likewise.
23740 (<fmaxmin><mode>3): Likewise.
23741 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23742 * config/arm/predicates.md (arm_cond_move_operator): Check against
23743 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23745 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23747 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23748 to top of function.
23750 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23752 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23753 loop in comment with memset.
23755 2017-07-14 Martin Liska <mliska@suse.cz>
23757 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23758 * dwarf2out.c (is_java): Remove the function.
23759 (output_pubname): Remove usage of the function.
23760 (lower_bound_default): Remove usage of DW_LANG_Java.
23761 (gen_compile_unit_die): Likewise.
23762 * gcc.c: Remove compiler defaults for .java and .zip files.
23763 * gimple-expr.c (remove_suffix): Change as there's no longer
23764 extension than 4-letter one.
23765 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23766 (gimplify_save_expr): Likewise.
23767 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23768 as it's possible even for other languages than Java.
23769 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23770 * lto-opts.c (lto_write_options): Remove reference to Java.
23771 * opts.c (strip_off_ending): Update file extension handling.
23772 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23773 * tree-eh.c (lower_resx): Likewise.
23774 * tree.c (free_lang_data_in_type): Remove dead code.
23775 (find_decls_types_r): Likewise.
23776 (build_common_builtin_nodes): Remove Java from a comment.
23777 (verify_type): Remove dead code.
23778 * varasm.c (assemble_external): Remove Java from a comment.
23780 2017-07-14 Martin Liska <mliska@suse.cz>
23782 * opts.c (finish_options): Add quotes.
23783 (common_handle_option): Likewise.
23785 2017-07-14 Martin Liska <mliska@suse.cz>
23787 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23788 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23789 Remove N_SO_PASCAL.
23790 * dwarf2out.c (lower_bound_default): Do not handle
23792 (gen_compile_unit_die): Likewise.
23793 * gcc.c: Remove default extension binding for GNU Pascal.
23794 * stmt.c: Remove Pascal language from a comment.
23795 * xcoffout.c: Likewise.
23797 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23800 * diagnostic-show-locus.c (fixit_cmp): New function.
23801 (layout::layout): Sort m_fixit_hints.
23802 (column_range::column_range): Assert that the values are valid.
23803 (struct char_span): New struct.
23804 (correction::overwrite): New method.
23805 (struct source_line): New struct.
23806 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23807 calls in terms of classes source_line and char_span, and
23808 correction::overwrite.
23809 (selftest::test_overlapped_fixit_printing_2): New function.
23810 (selftest::diagnostic_show_locus_c_tests): Call it.
23812 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
23814 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
23815 early if there is no lhs.
23817 2017-07-13 Martin Liska <mliska@suse.cz>
23819 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
23820 (gen_reference_type_die): Likewise.
23821 * stor-layout.c: Remove Pascal-related comment.
23823 2017-07-13 Martin Liska <mliska@suse.cz>
23825 * opts.c (finish_options): Add quotes to error messages.
23826 (parse_sanitizer_options): Likewise.
23828 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
23830 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
23832 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
23834 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
23836 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23838 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
23840 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
23842 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23845 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
23846 provides the hardware capability bits, define the macro
23847 __BUILTIN_CPU_SUPPORTS__.
23848 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
23849 if GLIBC does not provide the hardware capability bits. Add a
23850 gcc_unreachable call if the built-in cpu function is neither
23851 __builtin_cpu_is nor __builtin_cpu_supports.
23852 (rs6000_get_function_versions_dispatcher): Change the warning
23853 that an old GLIBC is used which does not export the capability
23854 bits to be an error.
23855 * doc/extend.texi (target_clones attribute): Document the
23856 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
23857 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
23858 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
23859 the macros defined by GCC if the newer GLIBC is available.
23861 2017-07-12 Jeff Law <law@redhat.com>
23863 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
23864 remaining includes slightly.
23865 * config/riscv/riscv-builtins.c: Include profile-count.h.
23867 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23870 * config/avr/avr.c (avr_set_current_function): In diagnostic
23871 messages: Quote keywords and (parts of) identifiers.
23872 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
23875 2017-07-12 Carl Love <cel@us.ibm.com>
23877 * config/rs6000/rs6000-c.c: Add support for built-in functions
23878 vector bool char vec_revb (vector bool char);
23879 vector bool short vec_revb (vector short char);
23880 vector bool int vec_revb (vector bool int);
23881 vector bool long long vec_revb (vector bool long long);
23882 * doc/extend.texi: Update the built-in documentation file for the
23883 new built-in functions.
23885 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23887 * config/s390/s390.md: Remove movcc splitter.
23889 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23891 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
23892 load/store on condition.
23894 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23897 * config/avr/avr.c (avr_encode_section_info)
23898 [progmem && !TREE_READONLY]: Error if progmem object needs
23901 2017-07-11 Michael Collison <michael.collison@arm.com>
23903 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
23906 2017-07-11 Carl Love <cel@us.ibm.com>
23908 * config/rs6000/rs6000-c.c: Add support for builtins
23909 vector unsigned int vec_parity_lsbb (vector signed int);
23910 vector unsigned int vec_parity_lsbb (vector unsigned int);
23911 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
23912 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
23913 vector unsigned long long vec_parity_lsbb (vector signed long long);
23914 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
23915 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
23916 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
23917 * doc/extend.texi: Update the built-in documentation file for the
23918 new built-in functions.
23920 2017-07-11 David Malcolm <dmalcolm@redhat.com>
23922 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
23923 (layout::m_primary_loc): New field.
23924 (layout::layout): Initialize new field. Move location filtering
23925 logic from here to...
23926 (layout::maybe_add_location_range): ...this new method. Add
23927 support for filtering to just the lines already specified by other
23929 (layout::will_show_line_p): New method.
23930 (gcc_rich_location::add_location_if_nearby): New method.
23931 (selftest::test_add_location_if_nearby): New test function.
23932 (selftest::diagnostic_show_locus_c_tests): Call it.
23933 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
23936 2017-07-11 Tom de Vries <tom@codesourcery.com>
23938 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
23939 (bb_first_real_insn): New function.
23940 (nvptx_single): Add extra initialization of broadcasted condition
23943 2017-07-11 Nathan Sidwell <nathan@acm.org>
23945 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
23947 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
23949 * doc/extend.texi (AVR Function Attributes): Remove weblink to
23950 Binutils doc as TEXI will mess them up.
23951 * doc/invoke.texi (AVR Options): Same here.
23953 2017-07-11 Daniel Cederman <cederman@gaisler.com>
23955 * config/sparc/sparc.opt (mfix-ut700): New option.
23956 (mfix-gr712rc): Likewise.
23957 (sparc_fix_b2bst): New variable.
23958 * doc/invoke.texi (SPARC options): Document them.
23959 (ARM options): Fix warnings.
23960 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
23961 instructions to prevent sequences that can trigger the store-store
23962 errata for certain LEON3FT processors.
23963 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
23964 (sparc_option_override): Set sparc_fix_b2bst appropriately.
23965 * config/sparc/sparc.md (fix_b2bst): New attribute.
23966 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
23968 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
23971 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
23973 (*rsqrtsf2_sse): Ditto.
23975 (div<mode>3): Macroize insn from divdf3 and divsf3
23976 using MODEF mode iterator.
23978 2017-07-10 Martin Sebor <msebor@redhat.com>
23980 PR tree-optimization/80397
23981 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
23982 instead of testing for equality to INTEGER_TYPE.
23984 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
23986 * config.gcc: Remove uclibc from arc target spec.
23988 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
23990 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
23992 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23995 * lto-wrapper.c (remove_option): New function.
23996 (merge_and_complain): Merge PIC/PIE options more realistically.
23998 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24000 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24004 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24005 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24006 * config.in: Regenerate.
24007 * configure: Regenerate.
24008 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24009 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24010 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24011 (TARGET_GASISR_PROLOGUES): ...target mask.
24012 * common/config/avr/avr-common.c
24013 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24014 Set -mgas-isr-prologues.
24015 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24016 INSERT_PASS_BEFORE for it.
24017 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24018 * config/avr/avr.c (avr_option_override)
24019 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24020 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24021 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24022 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24023 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24025 (make_avr_pass_pre_proep): New function.
24026 (emit_push_sfr) <treg>: Add argument to function and use it
24027 instead of TMP_REG.
24028 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24029 and set machine->gasisr.yes.
24030 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24031 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24032 __gcc_isr.n_pushed to .L__stack_usage.
24033 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24034 (avr_asm_final_postscan_insn): ...this new static function.
24035 * config/avr/avr.h (machine_function)
24036 <is_no_gccisr, use_L__stack_usage>: New fields.
24037 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24038 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24039 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24040 (gasisr, *gasisr): New expander and insn.
24041 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24042 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24043 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24045 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24047 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24050 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24052 Move jump-tables out of .text again.
24055 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
24056 (ASM_OUTPUT_ADDR_VEC): New function.
24057 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
24058 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
24059 INSN_ADDRESSes as asm comment.
24060 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
24061 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
24062 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
24063 * config/avr/avr.md (*tablejump): Adjust comment.
24064 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
24065 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
24067 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
24068 (avr_output_addr_vec): New proto.
24069 (avr_log_t) <insn_addresses>: New field.
24071 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
24074 * config/i386/i386.c (ix86_function_arg_advance): Set
24075 outgoing_args_on_stack to true if there are outgoing arguments
24077 (ix86_function_arg): Likewise.
24078 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
24079 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
24080 * config/i386/i386.h (machine_function): Add
24081 outgoing_args_on_stack.
24083 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
24085 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
24086 supporting pthreds.
24087 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
24089 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
24091 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
24092 (REAL_H): Remove $(MACHMODE_H).
24093 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
24095 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
24096 $(MACHMODE_H) and double-int.h.
24097 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
24099 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
24102 2017-07-07 Andrew Pinski <apinski@cavium.com>
24104 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
24105 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
24107 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
24109 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
24110 Add warning if GCC was not configured to link against a GLIBC that
24111 exports the hardware capability bits.
24112 (make_resolver_func): Make resolver function private and not a
24113 COMDAT function. Create the name with clone_function_name instead
24114 of make_unique_name.
24117 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
24118 correct operand in doing the split.
24120 2017-07-07 Carl Love <cel@us.ibm.com>
24122 * config/rs6000/rs6000-c: Add support for built-in function
24123 vector unsigned short vec_pack_to_short_fp32 (vector float,
24125 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
24126 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
24127 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
24128 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
24129 (convert_4f32_8i16): Add define_expand.
24130 * doc/extend.texi: Update the built-in documentation file for the
24131 new built-in function.
24133 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24135 * config/sparc/m8.md: New file.
24136 * config/sparc/sparc.md: Include m8.md.
24138 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24140 * config/sparc/sparc.opt: New option -mvis4b.
24141 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
24142 (sparc_option_override): Handle VIS4B.
24143 (enum sparc_builtins): Define
24144 SPARC_BUILTIN_DICTUNPACK{8,16,32},
24145 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
24146 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
24147 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
24148 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
24149 (check_constant_argument): New function.
24150 (sparc_vis_init_builtins): Define builtins
24151 __builtin_vis_dictunpack{8,16,32},
24152 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
24153 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
24154 __builtin_vis_fpcmpde{8,16,32}shl and
24155 __builtin_vis_fpcmpur{8,16,32}shl.
24156 (sparc_expand_builtin): Check that the constant operands to
24157 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
24158 constant and in range.
24159 * config/sparc/sparc-c.c (sparc_target_macros): Handle
24161 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
24162 (SPARC_IMM5_P): Likewise.
24163 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
24164 (enabled): Handle vis4b.
24165 (UNSPEC_DICTUNPACK): New unspec.
24166 (UNSPEC_FPCMPSHL): Likewise.
24167 (UNSPEC_FPUCMPSHL): Likewise.
24168 (UNSPEC_FPCMPDESHL): Likewise.
24169 (UNSPEC_FPCMPURSHL): Likewise.
24170 (cpu_feature): New CPU feature `vis4b'.
24171 (dictunpack{8,16,32}): New insns.
24172 (FPCSMODE): New mode iterator.
24173 (fpcscond): New code iterator.
24174 (fpcsucond): Likewise.
24175 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
24176 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
24177 (fpcmpde{8,16,32}{si,di}shl): Likewise.
24178 (fpcmpur{8,16,32}{si,di}shl): Likewise.
24179 * config/sparc/constraints.md: Define constraints `q' for unsigned
24180 2-bit integer constants and `t' for unsigned 5-bit integer
24182 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
24184 (imm5_operand_dictunpack16): Likewise.
24185 (imm5_operand_dictunpack32): Likewise.
24186 (imm2_operand): Likewise.
24187 * doc/invoke.texi (SPARC Options): Document -mvis4b.
24188 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
24189 ditunpack* and fpcmp*shl builtins.
24191 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24193 * config.gcc: Handle m8 in --with-{cpu,tune} options.
24194 * config.in: Add HAVE_AS_SPARC6 define.
24195 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
24197 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
24199 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
24200 (CPP_CPU_SPEC): Handle m8.
24201 (ASM_CPU_SPEC): Likewise.
24202 * config/sparc/sparc-opts.h (enum processor_type): Add
24204 * config/sparc/sparc.c (m8_costs): New struct.
24205 (sparc_option_override): Handle TARGET_CPU_m8.
24206 (sparc32_initialize_trampoline): Likewise.
24207 (sparc64_initialize_trampoline): Likewise.
24208 (sparc_issue_rate): Likewise.
24209 (sparc_register_move_cost): Likewise.
24210 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
24211 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
24212 (ASM_CPU64_DEFAULT_SPEC): Likewise.
24213 (CPP_CPU_SPEC): Handle M8.
24214 (ASM_CPU_SPEC): Likewise.
24215 (AS_M8_FLAG): Define.
24216 * config/sparc/sparc.md: Add m8 to the cpu attribute.
24217 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
24218 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
24220 * configure: Regenerate.
24221 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
24224 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24226 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
24228 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
24229 ("*movdi_insn_sp32"): Do not set v3pipe.
24230 ("*movsi_insn"): Likewise.
24231 ("*movdi_insn_sp64"): Likewise.
24232 ("*movsf_insn"): Likewise.
24233 ("*movdf_insn_sp32"): Likewise.
24234 ("*movdf_insn_sp64"): Likewise.
24235 ("*zero_extendsidi2_insn_sp64"): Likewise.
24236 ("*sign_extendsidi2_insn"): Likewise.
24237 ("*mov<VM32:mode>_insn"): Likewise.
24238 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24239 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24240 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24241 ("<vlop:code><VL:mode>3"): Likewise.
24242 ("*not_<vlop:code><VL:mode>3"): Likewise.
24243 ("*nand<VL:mode>_vis"): Likewise.
24244 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
24245 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
24246 ("one_cmpl<VL:mode>2"): Likewise.
24247 ("faligndata<VM64:mode>_vis"): Likewise.
24248 ("alignaddrsi_vis"): Likewise.
24249 ("alignaddrdi_vis"): Likweise.
24250 ("alignaddrlsi_vis"): Likewise.
24251 ("alignaddrldi_vis"): Likewise.
24252 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24253 ("bmaskdi_vis"): Likewise.
24254 ("bmasksi_vis"): Likewise.
24255 ("bshuffle<VM64:mode>_vis"): Likewise.
24256 ("cmask8<P:mode>_vis"): Likewise.
24257 ("cmask16<P:mode>_vis"): Likewise.
24258 ("cmask32<P:mode>_vis"): Likewise.
24259 ("pdistn<P:mode>_vis"): Likewise.
24260 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24262 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24264 * config/sparc/sparc.md ("subtype"): New insn attribute.
24265 ("*wrgsr_sp64"): Set insn subtype.
24266 ("*rdgsr_sp64"): Likewise.
24267 ("alignaddrsi_vis"): Likewise.
24268 ("alignaddrdi_vis"): Likewise.
24269 ("alignaddrlsi_vis"): Likewise.
24270 ("alignaddrldi_vis"): Likewise.
24271 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24272 ("fexpand_vis"): Likewise.
24273 ("fpmerge_vis"): Likewise.
24274 ("faligndata<VM64:mode>_vis"): Likewise.
24275 ("bshuffle<VM64:mode>_vis"): Likewise.
24276 ("cmask8<P:mode>_vis"): Likewise.
24277 ("cmask16<P:mode>_vis"): Likewise.
24278 ("cmask32<P:mode>_vis"): Likewise.
24279 ("fchksm16_vis"): Likewise.
24280 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
24281 ("fmean16_vis"): Likewise.
24282 ("fp<plusminus_insn>64_vis"): Likewise.
24283 ("<plusminus_insn>v8qi3"): Likewise.
24284 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24285 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
24286 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
24287 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
24288 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
24289 ("*movqi_insn"): Likewise.
24290 ("*movhi_insn"): Likewise.
24291 ("*movsi_insn"): Likewise.
24292 ("movsi_pic_gotdata_op"): Likewise.
24293 ("*movdi_insn_sp32"): Likewise.
24294 ("*movdi_insn_sp64"): Likewise.
24295 ("movdi_pic_gotdata_op"): Likewise.
24296 ("*movsf_insn"): Likewise.
24297 ("*movdf_insn_sp32"): Likewise.
24298 ("*movdf_insn_sp64"): Likewise.
24299 ("*zero_extendhisi2_insn"): Likewise.
24300 ("*zero_extendqihi2_insn"): Likewise.
24301 ("*zero_extendqisi2_insn"): Likewise.
24302 ("*zero_extendqidi2_insn"): Likewise.
24303 ("*zero_extendhidi2_insn"): Likewise.
24304 ("*zero_extendsidi2_insn_sp64"): Likewise.
24305 ("ldfsr"): Likewise.
24306 ("prefetch_64"): Likewise.
24307 ("prefetch_32"): Likewise.
24308 ("tie_ld32"): Likewise.
24309 ("tie_ld64"): Likewise.
24310 ("*tldo_ldub_sp32"): Likewise.
24311 ("*tldo_ldub1_sp32"): Likewise.
24312 ("*tldo_ldub2_sp32"): Likewise.
24313 ("*tldo_ldub_sp64"): Likewise.
24314 ("*tldo_ldub1_sp64"): Likewise.
24315 ("*tldo_ldub2_sp64"): Likewise.
24316 ("*tldo_ldub3_sp64"): Likewise.
24317 ("*tldo_lduh_sp32"): Likewise.
24318 ("*tldo_lduh1_sp32"): Likewise.
24319 ("*tldo_lduh_sp64"): Likewise.
24320 ("*tldo_lduh1_sp64"): Likewise.
24321 ("*tldo_lduh2_sp64"): Likewise.
24322 ("*tldo_lduw_sp32"): Likewise.
24323 ("*tldo_lduw_sp64"): Likewise.
24324 ("*tldo_lduw1_sp64"): Likewise.
24325 ("*tldo_ldx_sp64"): Likewise.
24326 ("*mov<VM32:mode>_insn"): Likewise.
24327 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24328 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24330 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24332 * config/sparc/sparc.md ("type"): New insn type viscmp.
24333 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
24335 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
24336 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
24337 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24338 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
24340 ("n7_vis_logical_11cycle"): Likewise.
24341 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
24342 * config/sparc/niagara2.md ("niag3_vis": Likewise.
24343 * config/sparc/niagara.md ("niag_vis"): Likewise.
24344 * config/sparc/ultra3.md ("us3_fga"): Likewise.
24345 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
24347 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24349 * config/sparc/sparc.md: New instruction type `bmask'.
24350 (bmaskdi_vis): Use the `bmask' type.
24351 (bmasksi_vis): Likewise.
24352 * config/sparc/ultra3.md (us3_array): Likewise.
24353 * config/sparc/niagara7.md (n7_array): Likewise.
24354 * config/sparc/niagara4.md (n4_array): Likewise.
24355 * config/sparc/niagara2.md (niag2_vis): Likewise.
24356 (niag3_vis): Likewise.
24357 * config/sparc/niagara.md (niag_vis): Likewise.
24359 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24361 * ipa-comdats.c: Remove optimize check from gate.
24362 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
24363 for functions not optimized.
24364 (ipa_fn_summary_read): Skip optimize check.
24365 (ipa_fn_summary_write): Likewise.
24366 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
24368 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
24370 (can_inline_edge_p): Check flag_pcc_struct_return for match.
24371 (check_callers): Give up on caller which is not optimized.
24372 (inline_small_functions): Likewise.
24373 (ipa_inline): Do not give up when not optimizing.
24374 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
24375 away unoptimizes cdtors.
24376 (whole_program_function_and_variable_visibility): Do
24377 ipa_discover_readonly_nonaddressable_vars in LTO mode.
24378 * ipa.c (process_references): Do not check optimize.
24379 (symbol_table::remove_unreachable_nodes): Update optimize check.
24380 (set_writeonly_bit): Update optimize check.
24381 (pass_ipa_cdtor_merge::gate): Do not check optimize.
24382 (pass_ipa_single_use::gate): Remove.
24384 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24386 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
24387 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
24388 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
24389 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
24390 permute_load, permute_store, adjust_extract, adjust_splat,
24391 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
24392 replace_swap_with_copy, dump_swap_insn_table,
24393 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
24394 recombine_lvx_pattern, recombine_stvx_pattern,
24395 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
24396 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
24397 to file rs6000-p8swap.c.
24398 * config/rs6000/rs6000-p8swap.c: New file.
24399 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
24400 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
24401 and rs6000*-*-* targets.
24403 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24405 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
24407 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24409 * lto-wrapper.c (merge_and_complain): Do not merge
24410 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
24411 fsigned_zeros, ftrapping_math, fwrapv.
24412 (append_compiler_options): Do not track these options.
24413 (append_linker_options): Likewie
24415 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24417 * cgraphunit.c (cgraph_node::finalize_function): When
24418 !flag_toplevel_reorde set no_reorder flag.
24419 (varpool_node::finalize_decl): Likewise.
24420 (symbol_table::compile): Drop no toplevel reorder path.
24422 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24424 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
24425 edges; zero probability is not better than uninitialized.
24427 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24429 * asan.h (asan_sanitize_allocas_p): Declare.
24430 * asan.c (asan_sanitize_allocas_p): New function.
24431 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
24432 (handle_builtin_alloca): Likewise.
24433 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
24434 if !asan_sanitize_allocas_p.
24435 * params.def (asan-instrument-allocas): Add new option.
24436 * params.h (ASAN_PROTECT_ALLOCAS): Define.
24437 * opts.c (common_handle_option): Disable allocas sanitization for
24440 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24442 * asan.c: Include gimple-fold.h.
24443 (get_last_alloca_addr): New function.
24444 (handle_builtin_stackrestore): Likewise.
24445 (handle_builtin_alloca): Likewise.
24446 (asan_emit_allocas_unpoison): Likewise.
24447 (get_mem_refs_of_builtin_call): Add new parameter, remove const
24448 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
24449 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
24450 (instrument_builtin_call): Pass gimple iterator to
24451 get_mem_refs_of_builtin_call.
24452 (last_alloca_addr): New global.
24453 * asan.h (asan_emit_allocas_unpoison): Declare.
24454 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
24455 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
24456 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
24457 if function calls alloca.
24458 * gimple-fold.c (replace_call_with_value): Remove static keyword.
24459 * gimple-fold.h (replace_call_with_value): Declare.
24460 * internal-fn.c: Include asan.h.
24461 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
24462 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
24464 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24466 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
24467 (C_SELFTEST_FLAGS): New.
24468 (CPP_SELFTEST_FLAGS): New.
24469 (SELFTEST_DEPS): New, from deps of s-selftest.
24470 (C_SELFTEST_DEPS): New, from deps of s-selftest.
24471 (CPP_SELFTEST_DEPS): New.
24472 (selftest): Add dependency on s-selftest-c++.
24473 (s-selftest): Rename to...
24474 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
24475 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
24476 than SELFTEST_FLAGS.
24477 (selftest-gdb): Rename to...
24478 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
24480 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
24481 (selftest-valgrind): Rename to...
24482 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
24484 (selftest-valgrind): Reintroduce as an alias for
24485 selftest-c-valgrind.
24486 (s-selftest-c++): New.
24487 (selftest-c++-gdb): New.
24488 (selftest-c++-valgrind): New.
24490 2017-07-06 Olivier Hainque <hainque@adacore.com>
24492 * gcc.c (process_command): When deciding if undefined variables
24493 should be ignored when processing specs, accept "gcc -v" as well.
24495 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24497 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
24498 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
24500 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24502 * config/arm/arm-cpus.in (armv8-r): Add new entry.
24503 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
24504 * config/arm/arm-tables.opt: Regenerate.
24505 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
24507 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
24509 2017-07-06 Carl Love <cel@us.ibm.com>
24511 * ChangeLog: Clean up from mid air collision
24513 2017-07-06 Carl Love <cel@us.ibm.com>
24515 * config/rs6000/rs6000-c.c: Add support for built-in functions
24516 vector signed int vec_subc (vector signed int, vector signed int);
24517 vector signed __int128 vec_subc (vector signed __int128,
24518 vector signed __int128);
24519 vector unsigned __int128 vec_subc (vector unsigned __int128,
24520 vector unsigned __int128);
24521 vector signed int vec_sube (vector signed int, vector signed int,
24522 vector signed int);
24523 vector unsigned int vec_sube (vector unsigned int,
24524 vector unsigned int,
24525 vector unsigned int);
24526 vector signed __int128 vec_sube (vector signed __int128,
24527 vector signed __int128,
24528 vector signed__int128);
24529 vector unsigned __int128 vec_sube (vector unsigned __int128,
24530 vector unsigned __int128,
24531 vector unsigned __int128);
24532 vector signed int vec_subec (vector signed int, vector signed int,
24533 vector signed int);
24534 vector unsigned int vec_subec (vector unsigned int,
24535 vector unsigned int,
24536 vector unsigned int);
24537 vector signed __int128 vec_subec (vector signed __int128,
24538 vector signed __int128,
24539 vector signed__int128);
24540 vector unsigned __int128 vec_subec (vector unsigned __int128,
24541 vector unsigned __int128,
24542 vector unsigned __int128);
24543 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
24544 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
24545 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
24546 BU_ALTIVEC_OVERLOAD_X definitions.
24547 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
24548 * doc/extend.texi: Update the built-in documentation file for the new
24549 built-in functions.
24551 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24554 * diagnostic-show-locus.c (layout::layout): Use start and finish
24555 spelling location for the start and finish of each range.
24556 * genmatch.c (linemap_client_expand_location_to_spelling_point):
24557 Add unused aspect param.
24558 * input.c (expand_location_1): Add "aspect" param, and use it
24559 to access the correct part of the location.
24560 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
24562 (expand_location_to_spelling_point): Likewise.
24563 (linemap_client_expand_location_to_spelling_point): Add "aspect"
24564 param, and pass it to expand_location_1.
24566 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24568 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24569 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24570 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24571 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24572 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24573 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24574 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24575 _mm_maskz_getmant_ss): New intrinsics.
24576 (__builtin_ia32_getexpss128_mask): Changed to ...
24577 __builtin_ia32_getexpss128_round ... this.
24578 (__builtin_ia32_getexpsd128_mask): Changed to ...
24579 __builtin_ia32_getexpsd128_round ... this.
24580 * config/i386/i386-builtin-types.def
24581 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24582 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24583 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24584 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24585 __builtin_ia32_getmantss_mask_round): New builtins.
24586 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24587 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24588 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24589 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24590 * config/i386/sse.md
24591 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24592 avx512f_sgetexp<mode><mask_scalar_name>
24593 <round_saeonly_scalar_name> ... this.
24594 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24595 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24596 vgetexp<ssescalarmodesuffix>
24597 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24598 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24599 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24600 avx512f_vgetmant<mode><mask_scalar_name>
24601 <round_saeonly_scalar_name> ... this.
24602 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24603 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24604 vgetmant<ssescalarmodesuffix>
24605 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24606 %0<mask_scalar_operand4>, %1, %2
24607 <round_saeonly_scalar_mask_op4>, %3} ... this.
24608 * config/i386/subst.md (mask_scalar_operand4,
24609 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24610 round_saeonly_scalar_nimm_predicate): New subst attributes.
24612 2017-07-06 Julia Koval <julia.koval@intel.com>
24614 * config/i386/i386.c (ix86_erase_embedded_rounding):
24615 Remove code for old rounding pattern.
24617 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24619 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24621 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24623 * doc/sourcebuild.texi (Test Directives, Variants of
24624 dg-require-support): Add documentation for dg-require-stack-check.
24626 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24628 * config/i386/subst.md (mask_scalar, round_scalar,
24629 round_saeonly_scalar): New meta-templates.
24630 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24631 round_scalar_mask_operand3, round_scalar_mask_op3,
24632 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24633 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24634 round_saeonly_scalar_constraint,
24635 round_saeonly_scalar_prefix): New subst attribute.
24636 * config/i386/sse.md
24637 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24638 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24639 <round_scalar_name> ... this.
24640 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24641 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24642 <round_scalar_name> ... this.
24643 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24644 <sse>_vm<code><mode>3<mask_scalar_name>
24645 <round_saeonly_scalar_name> ... this.
24646 (v<plusminus_mnemonic><ssescalarmodesuffix>
24647 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24648 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24649 v<plusminus_mnemonic><ssescalarmodesuffix>
24650 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24651 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24652 (v<multdiv_mnemonic><ssescalarmodesuffix>
24653 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24654 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24655 v<multdiv_mnemonic><ssescalarmodesuffix>
24656 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24657 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24658 (v<maxmin_float><ssescalarmodesuffix>
24659 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24660 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24661 v<maxmin_float><ssescalarmodesuffix>
24662 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24663 %0<mask_scalar_operand3>, %1, %<iptr>2
24664 <round_saeonly_scalar_mask_op3>} ... this.
24666 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24668 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24669 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24671 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24672 Alan Hayward <alan.hayward@arm.com>
24673 David Sherwood <david.sherwood@arm.com>
24675 * combine.c (simplify_if_then_else): Remove "enum" before
24677 * compare-elim.c (can_eliminate_compare): Likewise.
24678 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24680 (aarch64_lookup_simd_builtin_type): Likewise.
24681 (aarch64_simd_builtin_type): Likewise.
24682 (aarch64_init_simd_builtin_types): Likewise.
24683 (aarch64_simd_expand_args): Likewise.
24684 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24686 (aarch64_reverse_mask): Likewise.
24687 (aarch64_simd_emit_reg_reg_move): Likewise.
24688 (aarch64_gen_adjusted_ldpstp): Likewise.
24689 (aarch64_ccmp_mode_to_code): Likewise.
24690 (aarch64_operands_ok_for_ldpstp): Likewise.
24691 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24692 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24694 (aarch64_min_divisions_for_recip_mul): Likewise.
24695 (aarch64_reassociation_width): Likewise.
24696 (aarch64_get_condition_code_1): Likewise.
24697 (aarch64_simd_emit_reg_reg_move): Likewise.
24698 (aarch64_simd_attr_length_rglist): Likewise.
24699 (aarch64_reverse_mask): Likewise.
24700 (aarch64_operands_ok_for_ldpstp): Likewise.
24701 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24702 (aarch64_gen_adjusted_ldpstp): Likewise.
24703 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24705 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24706 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24707 (arm_lookup_simd_builtin_type): Likewise.
24708 (arm_simd_builtin_type): Likewise.
24709 (arm_init_simd_builtin_types): Likewise.
24710 (arm_expand_builtin_args): Likewise.
24711 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24712 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24713 (ft32_setup_incoming_varargs): Likewise.
24714 (ft32_function_arg): Likewise.
24715 (ft32_function_arg_advance): Likewise.
24716 (ft32_pass_by_reference): Likewise.
24717 (ft32_arg_partial_bytes): Likewise.
24718 (ft32_valid_pointer_mode): Likewise.
24719 (ft32_addr_space_pointer_mode): Likewise.
24720 (ft32_addr_space_legitimate_address_p): Likewise.
24721 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24723 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24724 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24725 (iamcu_alignment): Likewise.
24726 (canonicalize_vector_int_perm): Likewise.
24727 (ix86_noce_conversion_profitable_p): Likewise.
24728 (ix86_mpx_bound_mode): Likewise.
24729 (ix86_operands_ok_for_move_multiple): Likewise.
24730 * config/microblaze/microblaze-protos.h
24731 (microblaze_expand_conditional_branch_reg): Likewise.
24732 * config/microblaze/microblaze.c
24733 (microblaze_expand_conditional_branch_reg): Likewise.
24734 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24736 (rs6000_reassociation_width): Likewise.
24737 (rs6000_invalid_binary_op): Likewise.
24738 (fusion_p9_p): Likewise.
24739 (emit_fusion_p9_load): Likewise.
24740 (emit_fusion_p9_store): Likewise.
24741 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24743 (riscv_hard_regno_mode_ok_p): Likewise.
24744 (riscv_address_insns): Likewise.
24745 (riscv_split_symbol): Likewise.
24746 (riscv_legitimize_move): Likewise.
24747 (riscv_function_value): Likewise.
24748 (riscv_hard_regno_nregs): Likewise.
24749 (riscv_expand_builtin): Likewise.
24750 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24751 (riscv_build_integer): Likewise.
24752 (riscv_split_integer): Likewise.
24753 (riscv_legitimate_constant_p): Likewise.
24754 (riscv_cannot_force_const_mem): Likewise.
24755 (riscv_regno_mode_ok_for_base_p): Likewise.
24756 (riscv_valid_base_register_p): Likewise.
24757 (riscv_valid_offset_p): Likewise.
24758 (riscv_valid_lo_sum_p): Likewise.
24759 (riscv_classify_address): Likewise.
24760 (riscv_legitimate_address_p): Likewise.
24761 (riscv_address_insns): Likewise.
24762 (riscv_load_store_insns): Likewise.
24763 (riscv_force_binary): Likewise.
24764 (riscv_split_symbol): Likewise.
24765 (riscv_force_address): Likewise.
24766 (riscv_legitimize_address): Likewise.
24767 (riscv_move_integer): Likewise.
24768 (riscv_legitimize_const_move): Likewise.
24769 (riscv_legitimize_move): Likewise.
24770 (riscv_address_cost): Likewise.
24771 (riscv_subword): Likewise.
24772 (riscv_output_move): Likewise.
24773 (riscv_canonicalize_int_order_test): Likewise.
24774 (riscv_emit_int_order_test): Likewise.
24775 (riscv_function_arg_boundary): Likewise.
24776 (riscv_pass_mode_in_fpr_p): Likewise.
24777 (riscv_pass_fpr_single): Likewise.
24778 (riscv_pass_fpr_pair): Likewise.
24779 (riscv_get_arg_info): Likewise.
24780 (riscv_function_arg): Likewise.
24781 (riscv_function_arg_advance): Likewise.
24782 (riscv_arg_partial_bytes): Likewise.
24783 (riscv_function_value): Likewise.
24784 (riscv_pass_by_reference): Likewise.
24785 (riscv_setup_incoming_varargs): Likewise.
24786 (riscv_print_operand): Likewise.
24787 (riscv_elf_select_rtx_section): Likewise.
24788 (riscv_save_restore_reg): Likewise.
24789 (riscv_for_each_saved_reg): Likewise.
24790 (riscv_register_move_cost): Likewise.
24791 (riscv_hard_regno_mode_ok_p): Likewise.
24792 (riscv_hard_regno_nregs): Likewise.
24793 (riscv_class_max_nregs): Likewise.
24794 (riscv_memory_move_cost): Likewise.
24795 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24796 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24797 (rl78_addr_space_address_mode): Likewise.
24798 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24800 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24801 (rs6000_reassociation_width): Likewise.
24802 (rs6000_invalid_binary_op): Likewise.
24803 (fusion_p9_p): Likewise.
24804 (emit_fusion_p9_load): Likewise.
24805 (emit_fusion_p9_store): Likewise.
24806 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24807 (ok_for_simple_move_operands): Likewise.
24808 (ok_for_simple_move_strict_operands): Likewise.
24809 (ok_for_simple_arith_logic_operands): Likewise.
24810 (visium_legitimize_reload_address): Likewise.
24811 (visium_select_cc_mode): Likewise.
24812 (output_cbranch): Likewise.
24813 (visium_split_double_move): Likewise.
24814 (visium_expand_copysign): Likewise.
24815 (visium_expand_int_cstore): Likewise.
24816 (visium_expand_fp_cstore): Likewise.
24817 * config/visium/visium.c (visium_pass_by_reference): Likewise.
24818 (visium_function_arg): Likewise.
24819 (visium_function_arg_advance): Likewise.
24820 (visium_libcall_value): Likewise.
24821 (visium_setup_incoming_varargs): Likewise.
24822 (visium_legitimate_constant_p): Likewise.
24823 (visium_legitimate_address_p): Likewise.
24824 (visium_legitimize_address): Likewise.
24825 (visium_secondary_reload): Likewise.
24826 (visium_register_move_cost): Likewise.
24827 (visium_memory_move_cost): Likewise.
24828 (prepare_move_operands): Likewise.
24829 (ok_for_simple_move_operands): Likewise.
24830 (ok_for_simple_move_strict_operands): Likewise.
24831 (ok_for_simple_arith_logic_operands): Likewise.
24832 (visium_function_value_1): Likewise.
24833 (rtx_ok_for_offset_p): Likewise.
24834 (visium_legitimize_reload_address): Likewise.
24835 (visium_split_double_move): Likewise.
24836 (visium_expand_copysign): Likewise.
24837 (visium_expand_int_cstore): Likewise.
24838 (visium_expand_fp_cstore): Likewise.
24839 (visium_split_cstore): Likewise.
24840 (visium_select_cc_mode): Likewise.
24841 (visium_split_cbranch): Likewise.
24842 (output_cbranch): Likewise.
24843 (visium_print_operand_address): Likewise.
24844 * expmed.c (flip_storage_order): Likewise.
24845 * expmed.h (emit_cstore): Likewise.
24846 (flip_storage_order): Likewise.
24847 * genrecog.c (validate_pattern): Likewise.
24848 * hsa-gen.c (gen_hsa_addr): Likewise.
24849 * internal-fn.c (expand_arith_overflow): Likewise.
24850 * ira-color.c (allocno_copy_cost_saving): Likewise.
24851 * lra-assigns.c (find_hard_regno_for_1): Likewise.
24852 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
24853 (process_invariant_for_inheritance): Likewise.
24854 * lra-eliminations.c (move_plus_up): Likewise.
24855 * omp-low.c (lower_oacc_reductions): Likewise.
24856 * simplify-rtx.c (simplify_subreg): Likewise.
24857 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
24858 (TARGET_CHKP_BOUND_MODE): Likewise..
24859 * targhooks.c (default_chkp_bound_mode): Likewise.
24860 (default_setup_incoming_vararg_bounds): Likewise.
24861 * targhooks.h (default_chkp_bound_mode): Likewise.
24862 (default_setup_incoming_vararg_bounds): Likewise.
24863 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
24864 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
24865 (have_whole_vector_shift): Likewise.
24866 * tree-vect-stmts.c (vectorizable_load): Likewise.
24867 * doc/tm.texi: Regenerate.
24869 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24871 Graceful degrade if Binutils PR21472 is not available.
24874 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
24875 .rodata in flash test fails.
24876 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
24877 * confgure: Regenerate.
24878 * config.in: Regenerate.
24879 * config/avr/avr.c (avr_asm_named_section)
24880 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
24881 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
24882 (avr_asm_init_sections): Same.
24884 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24886 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
24887 (fma<VH:mode>4_intrinsic): Likewise.
24888 (*fmsub<VCVTF:mode>4): Likewise.
24889 (*fmsub<VH:mode>4_intrinsic): Likewise.
24891 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24894 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
24895 Don't depend on "optimize > 0".
24896 (out_movhi_r_mr, out_movqi_mr_r): Same.
24897 (out_movhi_mr_r, out_movqi_r_mr): Same.
24898 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
24899 io_address_operand on "optimize > 0".
24901 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24903 * tree-loop-distribution.c: Add general explanantion on the pass.
24904 (generate_loops_for_partition): Mark distributed loop.
24905 (pg_add_dependence_edges): New parameter. Handle alias data
24906 dependence specially and record it in the parameter if asked.
24907 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
24908 (init_partition_graph_vertices, add_partition_graph_edge): New.
24909 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
24910 (free_partition_graph_vdata, build_partition_graph): New.
24911 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
24912 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
24913 (data_ref_segment_size, latch_dominated_by_data_ref): New.
24914 (compute_alias_check_pairs, version_loop_by_alias_check): New.
24915 (version_for_distribution_p, finalize_partitions): New.
24916 (distribute_loop): Handle alias data dependence specially. Factor
24917 out loop fusion code as functions and call these functions.
24919 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24921 * tree-loop-distribution.c (classify_partition): New parameter and
24922 better handle reduction statement.
24923 (rdg_build_partitions): Revise comment.
24924 (distribute_loop): Compute statements in all partitions and pass it
24925 to classify_partition.
24927 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24929 * tree-loop-distribution.c (enum partition_type): New.
24930 (struct partition): New field type.
24931 (partition_merge_into): Add parameter. Update partition type.
24932 (data_dep_in_cycle_p, update_type_for_merge): New functions.
24933 (build_rdg_partition_for_vertex): Compute partition type.
24934 (rdg_build_partitions): Dump partition type.
24935 (distribute_loop): Update calls to partition_merge_into.
24937 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24939 * tree-loop-distribution.c (struct ddr_hasher): New.
24940 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
24942 (classify_partition): Call get_data_dependence.
24943 (pg_add_dependence_edges): Ditto.
24944 (distribute_loop): Release data dependence hash table.
24946 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24948 * tree-loop-distribution.c (ref_base_address): Delete.
24949 (similar_memory_accesses): Rename ...
24950 (share_memory_accesses): ... to this. Check if partitions access
24951 the same memory reference.
24952 (distribute_loop): Call share_memory_accesses.
24954 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24956 * tree-loop-distribution.c (struct partition): New field recording
24957 its data reference.
24958 (partition_alloc, partition_free): Init and release data refs.
24959 (partition_merge_into): Merge data refs.
24960 (build_rdg_partition_for_vertex): Collect data refs for partition.
24961 (pg_add_dependence_edges): Change parameters from vector to bitmap.
24963 (distribute_loop): Remve data refs from vertice data of partition
24966 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24968 * tree-loop-distribution.c (params.h): Include header file.
24969 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
24970 (datarefs_vec): New global var.
24971 (create_rdg_vertices): Use datarefs_vec directly.
24972 (free_rdg): Don't free data references.
24973 (build_rdg): Update use. Don't free data references.
24974 (distribute_loop): Compute global variable for data references.
24975 Bail out if there are too many data references.
24977 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24979 * tree-loop-distribution.c (loop_nest): New global var.
24980 (build_rdg): Use loop directly, rather than loop nest.
24981 (pg_add_dependence_edges): Remove loop nest parameter. Use global
24983 (distribute_loop): Compute global variable loop nest. Update use.
24985 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24987 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
24988 (partition_merge_into): New parameter. Dump reason for fusion.
24989 (distribute_loop): Update use of partition_merge_into.
24991 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24993 * tree-loop-distribution.c (bb_top_order_index): New.
24994 (bb_top_order_index_size, bb_top_order_cmp): New.
24995 (stmts_from_loop): Use topological order.
24996 (pass_loop_distribution::execute): Compute and release topological
24997 order for basic blocks.
24999 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25001 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25004 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25006 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25007 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25008 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25009 * internal-fn.def (LOOP_DIST_ALIAS): New.
25010 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25011 (fold_loop_internal_call): ... this.
25012 (vect_loop_dist_alias_call): New function.
25013 (set_uid_loop_bbs): Call fold_loop_internal_call.
25014 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25017 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25020 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25021 Require dead FLAGS_REG at the beginning of a peephole.
25023 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25026 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25027 arguments in the call to __builtin_ia32_sbb_u32.
25028 (_subborrow_u64): Swap _X and _Y arguments in the call to
25029 __builtin_ia32_sbb_u64.
25031 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25034 * tree-vrp.c (compare_assert_loc): Turn into a function template
25035 with stable template parameter. Only test if a->e is NULL,
25036 !a->e == !b->e has been verified already. Use e == NULL or
25037 e != NULL instead of e or ! e tests. If stable is true, don't use
25038 iterative_hash_expr, on the other side allow a or b or both NULL
25039 and sort the NULLs last.
25040 (process_assert_insertions): Sort using compare_assert_loc<false>
25041 instead of compare_assert_loc, later sort using
25042 compare_assert_loc<true> before calling process_assert_insertions_for
25043 in a loop. Use break instead of continue once seen NULL pointer.
25045 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25047 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25048 Cortex-R7 and Cortex-R8 processors.
25050 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25052 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25053 uninitialized while src is not.
25055 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
25057 * common/config/arm/arm-common.c: Adjust include path for
25059 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
25060 (arm-cpu.h): Create in build directory. Adjust dependency rules.
25061 (arm-cpu-data.h): Likewise.
25062 (arm-cpu-cdata.h): Likewise.
25063 * config/arm/arm-cpu.h: Delete.
25064 * config/arm/arm-cpu-cdata.h: Delete.
25065 * config/arm/arm-cpu-data.h: Delete.
25067 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
25069 * config/arm/arm-cpus.in (cortex-a55): New.
25070 (cortex-a75): Likewise.
25071 (cortex-a75.cortex-a55): Likewise.
25072 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
25074 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
25075 * config/arm/arm-cpu-cdata.h: Regenerate.
25076 * config/arm/arm-cpu-data.h: Regenerate.
25077 * config/arm/arm-cpu.h: Regenerate.
25078 * config/arm/arm-tables.opt: Regenerate.
25079 * config/arm/arm-tune.md: Regenerate.
25081 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25083 * haifa-sched.c (sched_create_recovery_edges): Update profile.
25085 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25087 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
25090 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
25092 PR tree-optimization/81292
25093 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
25094 full_string_p, also call adjust_related_strinfos if the adjustment
25095 is simple, otherwise invalidate related strinfos.
25097 2017-07-04 Martin Liska <mliska@suse.cz>
25100 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
25101 newly created variable as DECL_IGNORED_P.
25103 2017-07-04 Martin Liska <mliska@suse.cz>
25106 * ipa-inline.c (inline_small_functions):
25107 Use xstrdup_for_dump.
25109 2017-07-04 Tom de Vries <tom@codesourcery.com>
25111 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
25113 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25116 * config/darwin.c (darwin_function_switched_text_sections):
25119 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
25121 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
25123 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25125 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
25127 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25129 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
25130 min_profitable_iters, and th as inclusive lower bounds.
25131 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
25132 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
25133 for min_profitable_iters and min_profitable_estimate.
25134 (vect_transform_loop): Treat th as an inclusive lower bound.
25135 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25137 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25140 * config/darwin.c (darwin_function_switched_text_sections):
25141 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
25142 in two pieces, and suppress the use of buf.
25144 2017-07-03 Nathan Sidwell <nathan@acm.org>
25146 * hash-table.h (hash_table_mod1): Fix indentation.
25148 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25150 PR middle-end/81290
25151 * predict.c (force_edge_cold): Be more careful about propagation
25153 * profile-count.h (profile_probability::guessed,
25154 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
25156 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
25158 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
25160 * doc/invoke.texi (rcpc architecture extension): Document it.
25162 2017-07-03 Richard Biener <rguenther@suse.de>
25164 PR tree-optimization/60510
25165 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
25166 the scalar reduction PHI and use it.
25167 (vectorizable_reduction): Properly guard the single_defuse_cycle
25168 path for non-SLP reduction chains where we cannot use it.
25169 Rework reduc_def/index and vector type deduction. Rework
25170 vector operand gathering during reduction op code-gen.
25171 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
25172 chains dissolve the chain and leave it to non-SLP reduction
25175 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25177 * tree-data-ref.h (dr_alignment): Declare.
25178 * tree-data-ref.c (dr_alignment): New function.
25179 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
25180 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
25182 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
25184 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25186 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
25187 and base_misalignment fields.
25188 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
25189 * tree-data-ref.c: Include builtins.h.
25190 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
25191 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
25192 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
25193 * tree-vect-data-refs.c: Include tree-cfg.h.
25194 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
25195 fields instead of calculating an alignment here.
25196 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
25197 innermost_loop_behavior fields.
25199 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25201 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
25203 (DR_STEP_ALIGNMENT): New macro.
25204 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
25205 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
25206 (create_data_ref): Print it.
25207 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
25208 to tell whether the step preserves vector (mis)alignment.
25209 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25210 Move the check for an integer step and generalise to all INTEGER_CST.
25211 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
25212 Print the outer step alignment.
25214 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25216 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
25217 with offset_alignment.
25218 (DR_ALIGNED_TO): Delete.
25219 (DR_OFFSET_ALIGNMENT): New macro.
25220 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
25221 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
25222 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
25223 (create_data_ref): Likewise.
25224 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25225 (vect_analyze_data_refs): Likewise.
25226 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
25227 creating dummy innermost behavior.
25229 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25231 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
25232 with a "innermost_loop_behavior *" and refeence tree.
25233 * tree-data-ref.c (dr_analyze_innermost): Likewise.
25234 (create_data_ref): Update call accordingly.
25235 * tree-predcom.c (find_looparound_phi): Likewise.
25237 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25239 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
25240 fields with dr_wrt_vec_loop.
25241 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
25242 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
25243 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
25244 (vect_dr_behavior): New function.
25245 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25246 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
25247 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
25248 track whether the step preserves the misalignment.
25249 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25250 Use vect_dr_behavior.
25251 (vect_setup_realignment): Update call accordingly.
25252 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
25253 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
25254 call to vect_create_addr_base_for_vector_ref.
25255 (vect_create_cond_for_align_checks): Likewise.
25256 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
25257 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
25258 (vect_recog_mask_conversion_pattern): Likewise.
25259 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
25260 (new_stmt_vec_info): Remove redundant zeroing.
25262 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25264 * common/config/arm/arm-common.c (arm_be8_option): New function.
25265 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
25266 (ISA_ARMv6): Add isa_bit_be8.
25267 * config/arm/arm.h (arm_be8_option): Add prototype.
25268 (BE8_SPEC_FUNCTION): New define.
25269 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
25270 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
25271 (mlittle-endian): Similarly.
25272 (mbe8, mbe32): New options.
25273 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
25274 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
25276 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25278 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
25280 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25282 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
25283 (cleanup_tree_cfg_bb): Use it.
25284 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
25286 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
25288 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25291 * loop-doloop.c (add_test): Update profile.
25293 2017-07-03 Martin Liska <mliska@suse.cz>
25296 * sanopt.c (rewrite_usage_of_param): New function.
25297 (sanitize_rewrite_addressable_params): Likewise.
25298 (pass_sanopt::execute): Call rewrite_usage_of_param.
25300 2017-07-03 Richard Biener <rguenther@suse.de>
25302 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
25303 back to using VIEW_CONVERT_EXPR.
25305 2017-07-03 Martin Liska <mliska@suse.cz>
25308 * doc/extend.texi: Document when a resolver function is
25309 generated for target_clones.
25311 2017-07-03 Martin Liska <mliska@suse.cz>
25313 * asan.c (asan_emit_stack_protection): Unpoison just red zones
25314 and shadow memory of auto variables which are subject of
25315 use-after-scope sanitization.
25316 (asan_expand_mark_ifn): Add do set only when is_poison.
25318 2016-07-03 Richard Biener <rguenther@suse.de>
25320 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
25322 (vect_force_simple_reduction): Record reduction def -> phi mapping.
25323 (vectorizable_reduction): Perform reduction PHI creation when
25324 visiting a reduction PHI and adjust and simplify code generation
25325 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
25326 (vect_transform_loop): Visit reduction PHIs.
25327 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
25328 defs into the SLP tree.
25329 (vect_build_slp_tree): Reduction defs terminate the recursion.
25330 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
25332 (vect_get_vec_defs_for_stmt_copy): Export.
25333 (vect_get_vec_defs): Likewise.
25334 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
25336 (vect_get_vec_defs_for_stmt_copy): Declare.
25337 (vect_get_vec_defs): Likewise.
25339 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25341 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
25342 parameter with a "loop" parameter and use it instead of the
25343 loop containing DR_STMT. Don't check simple_iv when doing
25344 BB analysis. Describe the two analysis modes in the comment.
25346 2017-07-03 Tom de Vries <tom@codesourcery.com>
25348 PR tree-optimization/69468
25349 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
25350 (find_same_succ_bb): Handle ignore_edge_flags.
25352 2017-07-03 Tom de Vries <tom@codesourcery.com>
25354 PR tree-optimization/81192
25355 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
25357 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
25359 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
25361 2017-07-03 Tom de Vries <tom@codesourcery.com>
25363 PR tree-optimization/81192
25364 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
25365 BB_SAME_SUCC (bb) == NULL.
25367 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25369 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
25372 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25374 * dumpfile.c: Include profile-count.h
25375 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
25377 (insert_cond_bb): Update profile.
25378 * tree-cfg.h (insert_cond_bb): Update prototype.
25379 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
25380 * tree-dump.c: Do not include tree-cfg.
25382 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25384 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
25386 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25388 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
25391 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25393 * tree-complex.c (expand_complex_div_wide): update profile.
25395 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25396 Alan Hayward <alan.hayward@arm.com>
25397 David Sherwood <david.sherwood@arm.com>
25399 * Makefile.in (MACHMODE_H): Remove insn-modes.h
25400 (CORETYPES_H): New define.
25401 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
25402 (insn-modes-inline.h, s-modes-inline-h): New rules.
25403 (generated_files): Add insn-modes-inline.h.
25404 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
25405 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
25406 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
25407 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
25408 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
25409 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
25410 (build/gencodes.o, build/genconditions.o): Likewise.
25411 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
25412 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
25413 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
25414 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
25415 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
25416 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
25417 * coretypes.h: Include everything up to real.h for generators.
25418 Include insn-modes.h first. Include wide-int-print.h after
25419 wide-int.h. Include insn-modes-inline.h and then machmode.h.
25420 * machmode.h: Don't include insn-modes.h here.
25421 * function-tests.c: Remove includes of signop.h, machmode.h,
25422 double-int.h and wide-int.h.
25424 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
25426 * optc-save-gen.awk: Likewise.
25427 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
25428 * godump.c: Remove include of wide-int-print.h.
25429 * pretty-print.h: Likewise.
25430 * wide-int-print.cc: Likewise.
25431 * wide-int.cc: Likewise.
25432 * hash-map-tests.c: Remove include of signop.h.
25433 * hash-set-tests.c: Likewise.
25434 * rtl-tests.c: Likewise.
25435 * mkconfig.sh: Remove include of machmode.h.
25436 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
25438 (emit_insn_modes_inline_h): ...this new function. Emit the code
25439 into an insn-modes-inline.h header file, adding appropriate
25440 include guards and end comments.
25441 (emit_insn_modes_c_header): Remove include of machmode.h.
25442 (emit_min_insn_modes_c_header): Include coretypes.h rather than
25444 (main): Handle -i flag and call emit_insn_modes_inline_h when
25447 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25449 * tree-ssa-strlen.c (strinfo): Rename the length field to
25450 nonzero_chars. Add a full_string_p field.
25451 (compare_nonzero_chars, zero_length_string_p): New functions.
25452 (get_addr_stridx): Add an offset_out parameter.
25453 Use compare_nonzero_chars.
25454 (get_stridx): Update accordingly. Use compare_nonzero_chars.
25455 (new_strinfo): Update after above changes to strinfo.
25456 (set_endptr_and_length): Set full_string_p.
25457 (get_string_length): Update after above changes to strinfo.
25458 (unshare_strinfo): Update call to new_strinfo.
25459 (maybe_invalidate): Likewise.
25460 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
25461 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
25462 as a uhwi instead of an shwi. Update after above changes to
25463 strinfo and new_strinfo.
25464 (zero_length_string): Assert that chainsi contains full strings.
25465 Use zero_length_string_p. Update call to new_strinfo.
25466 (adjust_related_strinfos): Update after above changes to strinfo.
25467 Copy full_string_p from origsi.
25468 (adjust_last_stmt): Use zero_length_string_p.
25469 (handle_builtin_strlen): Update after above changes to strinfo and
25470 new_strinfo. Install the lhs as the string length if the previous
25471 entry didn't describe a full string.
25472 (handle_builtin_strchr): Update after above changes to strinfo
25474 (handle_builtin_strcpy): Likewise.
25475 (handle_builtin_strcat): Likewise.
25476 (handle_builtin_malloc): Likewise.
25477 (handle_pointer_plus): Likewise.
25478 (handle_builtin_memcpy): Likewise. Track nonzero characters
25479 that aren't necessarily followed by a nul terminator.
25480 (handle_char_store): Likewise.
25482 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25484 PR tree-optimization/80769
25485 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
25486 for malloc and calloc. Document the new invariant that all related
25487 strinfos have delayed lengths or none do.
25488 (verify_related_strinfos): Move earlier in file.
25489 (set_endptr_and_length): New function, split out from...
25490 (get_string_length): ...here. Also set the lengths of related
25492 (zero_length_string): Assert that chainsi has known (rather than
25494 (adjust_related_strinfos): Likewise.
25496 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25498 PR tree-optimization/81136
25499 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
25500 assert that two references with the same misalignment have the same
25501 compile-time misalignment if those compile-time misalignments
25504 2017-07-01 Andi Kleen <ak@linux.intel.com>
25506 * print-tree.c (print_node): Print all attributes.
25508 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25510 * cfg.c (scale_bbs_frequencies): New function.
25511 * cfg.h (scale_bbs_frequencies): Declare it.
25512 * cfgloopanal.c (single_likely_exit): Cleanup.
25513 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
25515 (scale_loop_profile): Likewise.
25516 (loop_version): Likewise.
25517 (create_empty_loop_on_edge): Update.
25518 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
25519 scale_loop_frequencies, scale_loop_profile, loopify,
25520 loop_version): Update prototypes.
25521 * modulo-sched.c (sms_schedule): Update.
25522 * predict.c (unlikely_executed_edge_p): Also check probability.
25523 (probably_never_executed_edge_p): Fix typo.
25524 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25525 * tree-parloops.c (gen_parallel_loop): Update.
25526 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
25527 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25528 * tree-ssa-loop-split.c (split_loop): Update.
25529 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25530 * tree-vect-loop-manip.c (vect_do_peeling): Update.
25531 (vect_loop_versioning): Update.
25532 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25534 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25536 * trans-mem.c (split_bb_make_tm_edge): Update profile.
25538 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25540 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
25541 to keep profile consistent.
25543 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25545 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
25546 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
25547 * profile-count.h (max_safe_multiplier): Make unsigned.
25548 (profile_count::guessed_zero): New.
25550 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25552 * bb-reorder.c (fix_up_crossing_landing_pad,
25553 fix_crossing_conditional_branches): Use make_single_succ_edge
25554 to keep profile consistent.
25556 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25558 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
25561 2017-07-01 Jakub Jelinek <jakub@redhat.com>
25564 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
25565 the right scopes, make sure cond_jump isn't preserved between multiple
25566 iterations. Search for fallthru edge whenever there are 3+ edges and
25567 use find_fallthru_edge for it.
25569 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25571 Patch by Alexander Monakov <amonakov@ispras.ru>
25572 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25573 probabilities consistently.
25575 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25577 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25578 emit_cmp_and_jump_insns.
25580 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25583 * tree-inline.c (expand_call_inline): Combine profile statuses.
25585 2017-06-30 Andrew Pinski <apinski@cavium.com>
25587 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25588 fold_stmt returned true.
25590 2017-06-30 Nathan Sidwell <nathan@acm.org>
25592 * ggc.h (empty_string): Delete.
25593 * cfgexpand.c (expand_asm_stmt): Use plain "".
25594 * optabs.c (expand_asm_memory_barrier): Likewise.
25595 * stringpool.c (empty_string): Delete.
25596 (digit_vector, digit_string): Delete.
25597 (ggc_alloc_string): Use plain "", don't optimize single digit
25598 strings. Use ggc_alloc_atomic.
25600 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25602 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25603 comparison set and one other set, use the cost of the non-comparison
25606 2017-06-30 Nathan Sidwell <nathan@acm.org>
25608 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25611 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25613 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25614 loops. Remove now unneeded calls to gimple_switch_set_label() that
25615 just set removed labels to NULL_TREE.
25617 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25619 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25620 (set_range_info): ...here. Only call set_range_info_raw if domain
25622 (set_nonzero_bits): Call set_range_info_raw.
25623 * tree-ssanames.h (set_range_info_raw): New.
25625 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25628 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25629 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25630 of nonimmediate_operand and <store_mask_constraint> instead of m
25631 for the input operand. For V8FI iterator, always split if input
25632 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25633 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25634 <store_mask_predicate> instead of register_operand and
25635 <store_mask_constraint> instead of v for the input operand. Make
25636 sure both operands aren't MEMs for if not <mask_applied>.
25638 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25640 * lto-wrapper.c (copy_file) Close both file descriptors before
25643 2017-06-30 Martin Liska <mliska@suse.cz>
25646 * multiple_target.c (create_dispatcher_calls): Make ifunc
25647 also for function that don't have calls or are not referenced.
25649 2017-06-30 Richard Biener <rguenther@suse.de>
25651 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25652 analyze the first scalar stmt. Move vector type computation
25653 for the BB case here from ...
25654 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25655 live operation processing in the SLP case properly.
25657 2017-06-30 Richard Biener <rguenther@suse.de>
25659 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25661 2017-06-30 Martin Liska <mliska@suse.cz>
25664 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25665 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25667 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25669 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25672 2017-06-30 Richard Biener <rguenther@suse.de>
25674 PR tree-optimization/81249
25675 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25676 condition reduction result to original scalar type.
25678 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25680 * profile-count.h (enum profile_quality): Fix typos and whitespace
25683 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25685 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25686 type for branch probabilities.
25688 2017-06-29 Julian Brown <julian@codesourcery.com>
25689 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25691 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25692 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25693 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25694 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25696 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25698 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25699 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25700 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25701 CC usage from generic code to here.
25702 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25703 CC usage into the target macros.
25705 2017-06-29 Maya Rashish <coypu@sdf.org>
25707 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25710 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25712 * arm/arm-builtins.c: Include profile-count.h
25713 * except.c (sjlj_emit_function_enter): Use
25714 profile_probability::unlikely.
25716 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25718 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25719 and tocrel_offset be pointer args rather than implicitly using
25721 (legitimate_constant_pool_address_p, rs6000_emit_move,
25722 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25723 tocrel_offset and use in toc_relative_expr_p call.
25724 (print_operand, print_operand_address): Use static tocrel_base_oac
25725 and tocrel_offset_oac.
25726 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25729 2017-06-29 Maya Rashish <coypu@sdf.org>
25731 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25733 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25735 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25736 objects, take into account only the alignment of 'op0' and 'mode1' if
25739 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25741 * ccmp.c (ccmp_tree_comparison_p): New function.
25742 (ccmp_candidate_p): Update to use above function.
25743 (get_compare_parts): New function.
25744 (expand_ccmp_next): Update to use new functions.
25745 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25747 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25748 take mode as argument.
25749 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25750 * expr.c (expand_expr_real_1): Pass mode as argument.
25752 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25754 * combine.c (combine_instructions): Print insns to dump_file, together
25757 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25759 * asan.c (asan_emit_stack_protection): Update.
25760 (create_cond_insert_point): Update.
25761 * auto-profile.c (afdo_propagate_circuit): Update.
25762 * basic-block.h (struct edge_def): Turn probability to
25763 profile_probability.
25764 (EDGE_FREQUENCY): Update.
25765 * bb-reorder.c (find_traces_1_round): Update.
25766 (better_edge_p): Update.
25767 (sanitize_hot_paths): Update.
25768 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25769 (make_single_succ_edge): Update.
25770 (check_bb_profile): Update.
25771 (dump_edge_info): Update.
25772 (update_bb_profile_for_threading): Update.
25773 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25774 probabilitycount to 0.
25775 * cfgbuild.c (compute_outgoing_frequencies): Update.
25776 * cfgcleanup.c (try_forward_edges): Update.
25777 (outgoing_edges_match): Update.
25778 (try_crossjump_to_edge): Update.
25779 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25780 (expand_gimple_tailcall): Update.
25781 (construct_init_block): Use make_single_succ_edge.
25782 (construct_exit_block): Use make_single_succ_edge.
25783 * cfghooks.c (verify_flow_info): Update.
25784 (redirect_edge_succ_nodup): Update.
25785 (split_edge): Update.
25786 (account_profile_record): Update.
25787 * cfgloopanal.c (single_likely_exit): Update.
25788 * cfgloopmanip.c (scale_loop_profile): Update.
25789 (set_zero_probability): Remove.
25790 (duplicate_loop_to_header_edge): Update.
25791 * cfgloopmanip.h (loop_version): Update prototype.
25792 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25793 (force_nonfallthru_and_redirect): Update.
25794 (update_br_prob_note): Update.
25795 (rtl_verify_edges): Update.
25796 (purge_dead_edges): Update.
25797 (rtl_lv_add_condition_to_bb): Update.
25798 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25799 * cgraphunit.c (init_lowered_empty_function): Update.
25800 (cgraph_node::expand_thunk): Update.
25801 * cilk-common.c: Include profile-count.h
25802 * dojump.c (inv): Remove.
25803 (jumpifnot): Update.
25804 (jumpifnot_1): Update.
25805 (do_jump_1): Update.
25807 (do_jump_by_parts_greater_rtx): Update.
25808 (do_compare_rtx_and_jump): Update.
25809 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25810 do_jump_1. do_compare_rtx_and_jump): Update prototype.
25811 * dwarf2cfi.c: Include profile-count.h
25812 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
25813 (sjlj_emit_dispatch_table): Likewise.
25814 * explow.c: Include profile-count.h
25815 * expmed.c (emit_store_flag_force): Update.
25816 (do_cmp_and_jump): Update.
25817 * expr.c (compare_by_pieces_d::generate): Update.
25818 (compare_by_pieces_d::finish_mode): Update.
25819 (emit_block_move_via_loop): Update.
25820 (store_expr_with_bounds): Update.
25821 (store_constructor): Update.
25822 (expand_expr_real_2): Update.
25823 (expand_expr_real_1): Update.
25824 * expr.h (try_casesi, try_tablejump): Update prototypes.
25825 * gimple-pretty-print.c (dump_probability): Update.
25826 (dump_profile): New.
25827 (dump_gimple_label): Update.
25828 (dump_gimple_bb_header): Update.
25829 * graph.c (draw_cfg_node_succ_edges): Update.
25830 * hsa-gen.c (convert_switch_statements): Update.
25831 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
25832 (find_if_case_1): Update.
25833 (find_if_case_2): Update.
25834 * internal-fn.c (expand_arith_overflow_result_store): Update.
25835 (expand_addsub_overflow): Update.
25836 (expand_neg_overflow): Update.
25837 (expand_mul_overflow): Update.
25838 (expand_vector_ubsan_overflow): Update.
25839 * ipa-cp.c (good_cloning_opportunity_p): Update.
25840 * ipa-split.c (split_function): Use make_single_succ_edge.
25841 * ipa-utils.c (ipa_merge_profiles): Update.
25842 * loop-doloop.c (add_test): Update.
25843 (doloop_modify): Update.
25844 * loop-unroll.c (compare_and_jump_seq): Update.
25845 (unroll_loop_runtime_iterations): Update.
25846 * lra-constraints.c (lra_inheritance): Update.
25847 * lto-streamer-in.c (input_cfg): Update.
25848 * lto-streamer-out.c (output_cfg): Update.
25849 * mcf.c (adjust_cfg_counts): Update.
25850 * modulo-sched.c (sms_schedule): Update.
25851 * omp-expand.c (expand_omp_for_init_counts): Update.
25852 (extract_omp_for_update_vars): Update.
25853 (expand_omp_ordered_sink): Update.
25854 (expand_omp_for_ordered_loops): Update.
25855 (expand_omp_for_generic): Update.
25856 (expand_omp_for_static_nochunk): Update.
25857 (expand_omp_for_static_chunk): Update.
25858 (expand_cilk_for): Update.
25859 (expand_omp_simd): Update.
25860 (expand_omp_taskloop_for_outer): Update.
25861 (expand_omp_taskloop_for_inner): Update.
25862 * omp-simd-clone.c (simd_clone_adjust): Update.
25863 * optabs.c (expand_doubleword_shift): Update.
25864 (expand_abs): Update.
25865 (emit_cmp_and_jump_insn_1): Update.
25866 (expand_compare_and_swap_loop): Update.
25867 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
25868 * predict.c (predictable_edge_p): Update.
25869 (edge_probability_reliable_p): Update.
25870 (set_even_probabilities): Update.
25871 (combine_predictions_for_insn): Update.
25872 (combine_predictions_for_bb): Update.
25873 (propagate_freq): Update.
25874 (estimate_bb_frequencies): Update.
25875 (force_edge_cold): Update.
25876 * profile-count.c (profile_count::dump): Add missing space into dump.
25877 (profile_count::debug): Add newline.
25878 (profile_count::differs_from_p): Explicitly convert to unsigned.
25879 (profile_count::stream_in): Update.
25880 (profile_probability::dump): New member function.
25881 (profile_probability::debug): New member function.
25882 (profile_probability::differs_from_p): New member function.
25883 (profile_probability::differs_lot_from_p): New member function.
25884 (profile_probability::stream_in): New member function.
25885 (profile_probability::stream_out): New member function.
25886 * profile-count.h (profile_count_quality): Rename to ...
25887 (profile_quality): ... this one.
25888 (profile_probability): New.
25889 (profile_count): Update.
25890 * profile.c (compute_branch_probabilities): Update.
25891 * recog.c (peep2_attempt): Update.
25892 * sched-ebb.c (schedule_ebbs): Update.
25893 * sched-rgn.c (find_single_block_region): Update.
25894 (compute_dom_prob_ps): Update.
25895 (schedule_region): Update.
25896 * sel-sched-ir.c (compute_succs_info): Update.
25897 * stmt.c (struct case_node): Update.
25898 (do_jump_if_equal): Update.
25899 (get_outgoing_edge_probs): Update.
25900 (conditional_probability): Update.
25901 (emit_case_dispatch_table): Update.
25902 (expand_case): Update.
25903 (expand_sjlj_dispatch_table): Update.
25904 (emit_case_nodes): Update.
25905 * targhooks.c: Update.
25906 * tracer.c (better_p): Update.
25907 (find_best_successor): Update.
25908 * trans-mem.c (expand_transaction): Update.
25909 * tree-call-cdce.c: Update.
25910 * tree-cfg.c (gimple_split_edge): Upate.
25911 (move_sese_region_to_fn): Upate.
25912 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
25913 * tree-eh.c (lower_resx): Upate.
25914 (cleanup_empty_eh_move_lp): Upate.
25915 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25916 * tree-inline.c (copy_edges_for_bb): Update.
25917 (copy_cfg_body): Update.
25918 * tree-parloops.c (gen_parallel_loop): Update.
25919 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
25920 (gimple_gen_time_profiler): Update.
25921 * tree-ssa-dce.c (remove_dead_stmt): Update.
25922 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25923 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
25924 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
25925 (unloop_loops): Update.
25926 (try_peel_loop): Update.
25927 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25928 * tree-ssa-loop-split.c (connect_loops): Update.
25929 (split_loop): Update.
25930 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25931 (hoist_guard): Update.
25932 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
25933 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
25934 (value_replacement): Update.
25935 * tree-ssa-reassoc.c (branch_fixup): Update.
25936 * tree-ssa-tail-merge.c (replace_block_by): Update.
25937 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
25938 (create_edge_and_update_destination_phis): Update.
25939 (compute_path_counts): Update.
25940 (recompute_probabilities): Update.
25941 (update_joiner_offpath_counts): Update.
25942 (freqs_to_counts_path): Update.
25943 (duplicate_thread_path): Update.
25944 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
25945 (struct switch_conv_info): Update.
25946 (gen_inbound_check): Update.
25947 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25948 (vect_do_peeling): Update.
25949 (vect_loop_versioning): Update.
25950 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25951 (optimize_mask_stores): Update.
25952 * ubsan.c (ubsan_expand_null_ifn): Update.
25953 * value-prof.c (gimple_divmod_fixed_value): Update.
25954 (gimple_divmod_fixed_value_transform): Update.
25955 (gimple_mod_pow2): Update.
25956 (gimple_mod_pow2_value_transform): Update.
25957 (gimple_mod_subtract): Update.
25958 (gimple_mod_subtract_transform): Update.
25959 (gimple_ic): Update.
25960 (gimple_stringop_fixed_value): Update.
25961 (gimple_stringops_transform): Update.
25962 * value-prof.h: Update.
25964 2017-06-29 Carl Love <cel@us.ibm.com>
25966 * config/rs6000/rs6000-c.c: Add support for built-in functions
25967 vector signed int vec_signed (vector float);
25968 vector signed long long vec_signed (vector double);
25969 vector signed int vec_signed2 (vector double, vector double);
25970 vector signed int vec_signede (vector double);
25971 vector signed int vec_signedo (vector double);
25972 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
25973 instruction generator.
25974 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25975 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
25976 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
25978 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
25979 vunsignede_v2df): Add define_expands.
25980 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
25981 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
25982 VEC_UNSIGNEDO): Add definitions.
25983 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25984 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
25985 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
25986 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
25987 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
25988 * config/rs6000/altivec.h (vec_signed, vec_signed2,
25989 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
25990 vec_unsignede, vec_unsignedo): Add builtin defines.
25991 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
25993 * doc/extend.texi: Update the built-in documentation file for the
25994 new built-in functions.
25996 2017-06-29 Richard Biener <rguenther@suse.de>
25998 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
25999 reduction chains to LOOP_VINFO_REDUCTIONS.
26000 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26001 SLP reductions after processing reduction chains.
26003 2017-06-29 Nathan Sidwell <nathan@acm.org>
26005 * builtins.c (fold_builtin_FUNCTION): Use
26006 lang_hooks.decl_printable_name.
26008 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26010 PR middle-end/81194
26011 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26012 with only one label.
26013 * stmt.c (expand_case): Assert NCASES is greater than one.
26015 2017-06-29 Richard Biener <rguenther@suse.de>
26017 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26019 (group_case_labels): Likewise.
26020 (find_taken_edge): Push sanity checking on val to workers...
26021 (find_taken_edge_cond_expr): ... here
26022 (find_taken_edge_switch_expr): ... and here, handle cases
26023 with just a default label.
26024 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26025 (group_case_labels): Likewise.
26026 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26027 group_case_labels does anything cleanup the CFG again.
26029 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26031 PR tree-optimization/81196
26032 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26033 exit condition comparing two IVs.
26035 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26037 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26038 profile to the dummy entry at the end of the list of architectures.
26039 * config/arm/arm-cpu-cdata.h: Regenerated.
26041 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26042 Michael Collison <michael.collison@arm.com>
26045 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26047 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26048 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26049 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26050 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26051 of shift when the shift amount is masked with constant equal to
26052 the size of the mode.
26053 * config/aarch64/predicates.md (subreg_lowpart_operator): New
26056 2017-06-29 Martin Liska <mliska@suse.cz>
26058 * config/i386/i386.opt: Change range from [1,5] to [0,5].
26060 2017-06-29 Yury Gribov <tetra2005@gmail.com>
26063 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
26066 (edge_growth_cache_entry::edge_growth_cache_entry): New
26068 (reset_edge_growth_cache): Update to use constructor.
26070 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26072 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
26073 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26074 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26076 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
26078 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
26079 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
26081 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
26083 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
26084 (*-linux-uclibc*): Add t-uclibc tmake_file.
26085 * config/t-musl: New.
26086 * config/t-uclibc: New.
26088 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
26090 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
26092 (gen_comm_data): Emit architectural setting of arch_prof.
26093 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
26095 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
26096 (armv8-m.base, armv8-m.main): Likewise.
26097 * arm-protos.h (arm_build_target): Add profile field.
26098 (arch_option): Likewise.
26099 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
26101 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
26102 arm_active_target.profile.
26104 2017-06-28 Richard Biener <rguenther@suse.de>
26106 PR middle-end/81227
26107 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
26108 TYPE_OVERFLOW_WRAPS.
26109 * match.pd (negate_expr_p): Likewise.
26110 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
26111 fold_build2, not fold_binary.
26113 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26115 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
26116 Convert memory address to Pmode.
26117 (aarch64_print_operand): Assert MEM operands are always Pmode.
26119 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26122 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
26123 Remove redundant if.
26124 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
26125 * config/arm/aarch-common-protos.h
26126 (aarch_forward_to_shift_is_not_shifted_re): Remove.
26127 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
26129 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
26132 * multiple_target.c (create_dispatcher_calls): Set the default
26133 clone to be static, not public.
26135 2017-06-28 Richard Biener <rguenther@suse.de>
26137 * tree-vect-loop.c (vectorizable_reduction): Move special
26138 cond reduction IV var creation ...
26139 (vect_create_epilog_for_reduction): ... here. Remove induction_index
26140 parameter. Use STMT_VINFO_VECTYPE.
26141 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
26144 2017-06-28 Martin Liska <mliska@suse.cz>
26147 * ipa-visibility.c (non_local_p): Handle visibility.
26149 2017-06-28 Martin Liska <mliska@suse.cz>
26152 * common.opt: Add IntegerRange to various options.
26153 * opt-functions.awk (integer_range_info): New function.
26154 * optc-gen.awk: Add integer_range_info to cl_options struct.
26155 * opts-common.c (decode_cmdline_option): Handle
26156 CL_ERR_INT_RANGE_ARG.
26157 (cmdline_handle_error): Likewise.
26158 * opts.c (print_filtered_help): Show valid interval in
26159 when --help is provided.
26160 * opts.h (struct cl_option): Add range_min and range_max fields.
26161 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
26163 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
26165 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
26166 (x * C EQ/NE y * C): New transformation.
26168 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
26170 * genmultilib (combination_space): Accept '+' in option names.
26172 2017-06-28 Martin Liska <mliska@suse.cz>
26175 * asan.c (instrument_derefs): Bail out inner references
26176 that are hard register variables.
26178 2017-06-28 Jakub Jelinek <jakub@redhat.com>
26181 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
26182 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
26184 2017-06-28 Richard Biener <rguenther@suse.de>
26186 * tree-vectorizer.h (vect_get_vec_defs): Remove.
26187 (vect_get_slp_defs): Adjust.
26188 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
26190 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
26192 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
26193 get_initial_defs_for_reduction instead of vect_get_vec_defs.
26194 (vectorizable_reduction): Adjust.
26195 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
26197 (vect_get_slp_defs): Likewise.
26198 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
26199 (vectorizable_bswap): Adjust.
26200 (vectorizable_call): Likewise.
26201 (vectorizable_conversion): Likewise.
26202 (vectorizable_assignment): Likewise.
26203 (vectorizable_shift): Likewise.
26204 (vectorizable_operation): Likewise.
26205 (vectorizable_store): Likewise.
26206 (vectorizable_condition): Likewise.
26207 (vectorizable_comparison): Likewise.
26209 2017-06-28 Michael Collison <michael.collison@arm.com>
26212 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
26214 (arm_gen_movmemqi): Removed unused variable 'i'.
26215 Convert 'for' loop into 'while' loop.
26216 (arm_expand_prologue): Remove last unnecessary set of insn.
26217 (thumb_pop): Remove unused variable 'pushed_words'.
26218 (thumb_exit): Remove last unnecessary set of regs_to_pop.
26220 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26222 * config/s390/predicates.md: Use s390_rel_address_ok_p.
26223 * config/s390/s390-protos.h: Add prototype of
26224 s390_rel_address_ok_p.
26225 * config/s390/s390.c (s390_got_symbol): New function.
26226 (s390_rel_address_ok_p): New function.
26227 (legitimize_pic_address): Use s390_rel_address_ok_p.
26228 (s390_load_got): Use s390_got_symbol.
26229 (s390_option_override): Issue error if
26230 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
26231 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
26233 * config/s390/s390.opt: New option mpic-data-is-text-relative.
26235 2017-06-27 Andrew Pinski <apinski@cavium.com>
26237 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
26238 (X * copysign (1.0, X)): New pattern.
26239 (X * copysign (1.0, -X)): New pattern.
26240 (copysign (-1.0, CST)): New pattern.
26242 2017-06-27 Joseph Myers <joseph@codesourcery.com>
26244 * genmultilib (combination_space): Remove variable.
26245 Validate reuse rules against regular expression for any sequence
26246 of multilib options in any order.
26248 2017-06-27 Michael Collison <michael.collison@arm.com>
26250 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
26251 call aarch64_split_simd_combine.
26252 * (aarch64_combine_internal<mode>): Delete pattern.
26253 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
26254 Allow register and subreg operands.
26256 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26258 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
26259 specific need, just fallback on defaults.
26260 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
26262 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26263 Olivier Hainque <hainque@adacore.com>
26265 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
26267 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
26268 targets. Pick a default if no particular attempt applied.
26269 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
26272 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26274 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
26275 (x86_64-wrs-vxworks7): Likewise.
26277 2017-06-27 Marek Polacek <polacek@redhat.com>
26280 * ubsan.c (instrument_null): Check get_base_address's result for null.
26282 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26284 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
26286 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26288 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
26289 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
26290 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
26291 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
26292 New function types.
26293 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
26294 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
26295 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
26296 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
26297 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
26298 BUILT_IN_FEUPDATEENV): New builtins.
26299 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
26300 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
26301 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
26302 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
26304 (builtin_structptr_types): Adjust size.
26305 * tree.c (builtin_structptr_types): Add four entries.
26307 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26308 Olivier Hainque <hainque@adacore.com>
26310 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
26311 (TLS_SYM): New local macro, forcing reference to __tls__ on
26312 link command lines for VxWorks 7 RTPs, triggering initialization
26314 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
26315 OS features TLS support, true for RTPs on VxWorks 7.
26316 * config/vxworks.c (vxworks_override_options): Setup emutls
26319 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26321 * predict.c (test_prediction_value_range): Use -1U instead of -1
26322 to avoid narrowing conversion warning.
26323 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
26324 to avoid narrowing conversion warning.
26325 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
26327 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
26329 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26331 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
26332 64bit configurations.
26333 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
26334 (SIZE_TYPE): Likewise.
26335 * config/vxworks.c (vxworks_emutls_var_fields): Use
26336 long_unsigned_type_node instead of unsigned_type_node as the offset
26337 field type, which is "pointer" mode in emutls.c.
26339 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26342 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
26344 PR middle-end/81207
26345 * gimple-fold.c (replace_call_with_call_and_fold): Handle
26346 gimple_vuse copying separately from gimple_vdef copying.
26348 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26350 * value-prof.c (free_hist): Remove call to memset and the enclosing if
26353 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26354 Olivier Hainque <hainque@adacore.com>
26356 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
26357 for all vxworks7 targets.
26358 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
26359 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
26360 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
26361 variations for VX6/VX7 and 32/64bits later on in ...
26362 (VXWORKS_LIB_SPEC): Leverage new macros.
26363 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
26364 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
26366 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26368 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
26369 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
26371 2017-06-26 Carl Love <cel@us.ibm.com>
26373 * config/rs6000/rs6000-c.c: Add support for built-in functions
26374 vector bool char vec_reve (vector bool char);
26375 vector signed char vec_reve (vector signed char);
26376 vector unsigned char vec_reve (vector unsigned char);
26377 vector bool int vec_reve (vector bool int);
26378 vector signed int vec_reve (vector signed int);
26379 vector unsigned int vec_reve (vector unsigned int);
26380 vector bool long long vec_reve (vector bool long long);
26381 vector signed long long vec_reve (vector signed long long);
26382 vector unsigned long long vec_reve (vector unsigned long long);
26383 vector bool short vec_reve (vector bool short);
26384 vector signed short vec_reve (vector signed short);
26385 vector double vec_reve (vector double);
26386 vector float vec_reve (vector float);
26387 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
26388 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
26389 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
26390 (altivec_vreve): New pattern.
26391 * config/rs6000/altivec.h (vec_reve): New define.
26392 * doc/extend.texi (vec_rev): Update the built-in documentation file
26393 for the new built-in functions.
26395 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26397 PR tree-optimization/71815
26398 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
26400 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
26402 (slsr_process_phi): Likewise.
26403 (replace_uncond_cands_and_profitable_phis): Don't replace a
26404 multiply candidate with a stride of 1 (copy or cast).
26405 (phi_incr_cost): Call uses_consumed_by_stmt rather than
26407 (lowest_cost_path): Likewise.
26408 (total_savings): Likewise.
26410 2017-06-26 Richard Biener <rguenther@suse.de>
26413 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
26414 Use def_builtin_pure for all gather builtins.
26416 2017-06-26 Richard Biener <rguenther@suse.de>
26418 PR tree-optimization/81203
26419 * tree-tailcall.c (find_tail_calls): Do not move stmts into
26420 non-dominating BBs.
26422 2017-06-26 Marek Polacek <polacek@redhat.com>
26425 * doc/invoke.texi: Document -Wmultistatement-macros.
26427 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
26429 * doc/sourcebuild.texi (ARM-specific attributes): Document new
26430 arm_neon_ok_no_float_abi effective target.
26432 2017-06-26 Richard Biener <rguenther@suse.de>
26434 PR tree-optimization/80928
26435 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
26436 (copy_bbs): Set BB_DUPLICATED flag early.
26437 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
26439 (execute_on_shrinking_pred): Likewise.
26440 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
26441 BB_DUPLICATED blocks.
26442 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
26443 iterate over all PHIs considering removal of *gsi.
26445 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
26447 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
26450 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26452 * config/rs6000/rs6000-string.c: (expand_block_clear,
26453 do_load_for_compare, select_block_compare_mode,
26454 compute_current_alignment, expand_block_compare,
26455 expand_strncmp_align_check, expand_strn_compare,
26456 expand_block_move, rs6000_output_load_multiple)
26457 Move functions related to string/block move/compare
26458 to a separate file.
26459 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
26460 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
26461 for this function which is now used in two files.
26462 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
26463 * config.gcc: Add rs6000-string.o to extra_objs for
26464 targets powerpc*-*-* and rs6000*-*-*.
26466 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
26469 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
26470 32-bit, since indexed is not valid for DImode.
26471 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
26472 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
26473 (define_peephole2 for Altivec d-form load): Add 32-bit support.
26474 (define_peephole2 for Altivec d-form store): Likewise.
26477 * multiple_target.c (create_dispatcher_calls): Only create the
26478 dispatcher call if the function is the default clone of a
26479 versioned function.
26481 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
26483 PR middle-end/80902
26484 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
26485 a call, force the call to not be a tail call.
26487 2017-06-23 Jeff Law <law@redhat.com>
26489 * doc/contrib.texi: Add entry for Steven Pemberton's work on
26492 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
26494 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
26495 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
26496 handling for early expansion of vector shifts (sl,sr,sra,rl).
26497 (builtin_function_type): Add vector shift right instructions
26498 to the unsigned argument list.
26500 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
26502 rtl-optimizatoin/79286
26503 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
26504 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
26505 trap. PIC register plus a const unspec without offset can never trap.
26507 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
26509 * tree.h (builtin_structptr_type): New type.
26510 (builtin_structptr_types): Declare new array.
26511 * tree.c (builtin_structptr_types): New array.
26512 (free_lang_data, build_common_tree_nodes): Use it.
26514 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
26517 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
26520 2017-06-22 Matt Turner <mattst88@gmail.com>
26522 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
26523 Lake models to skylake case. Assume skylake for unknown
26524 models with clflushopt.
26526 2017-06-22 Jeff Law <law@redhat.com>
26528 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
26529 frame sizes that do not satisfy aarch64_uimm12_shift.
26531 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
26533 * profile-count.h (apply_probability,
26534 apply_scale, probability_in): Fix checks for zero.
26536 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26538 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
26539 * doc/cppdiropts.texi (-I @var{dir}): Document it.
26541 2016-06-22 Richard Biener <rguenther@suse.de>
26543 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26544 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
26545 REDUC_MAX_EXPR support.
26546 (vectorizable_reduction): Likewise.
26547 (vect_create_epilog_for_reduction): Likewise.
26549 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
26551 * match.pd (A / (1 << B) -> A >> B): New.
26552 * generic-match-head.c: Include optabs-tree.h.
26553 * gimple-match-head.c: Likewise.
26554 * optabs-tree.h (target_supports_op_p): New.
26555 * optabs-tree.c (target_supports_op_p): New.
26557 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26559 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
26560 $gcc_cv_ld --help output.
26561 (gcc_cv_ld_demangle): Likewise.
26562 (gcc_cv_ld_eh_frame_hdr): Likewise.
26563 (gcc_cv_ld_pie): Likewise.
26564 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
26565 (gcc_cv_ld_buildid): Likewise.
26566 (gcc_cv_ld_sysroot): Likewise.
26567 (ld_bndplt_support): Likewise.
26568 (ld_pushpopstate_support): Likewise.
26569 * configure: Regenerate.
26570 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26572 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26575 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26576 operands indexes to avoid gap between operands and match_dups.
26578 2017-06-21 Andrew Pinski <apinski@cavium.com>
26580 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26581 Increment Arith_shift and Arith_shift_reg by 1.
26582 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26584 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26585 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26586 (aarch64_strip_extend): Add new argument and test for it.
26587 (aarch64_cheap_mult_shift_p): New function.
26588 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26589 add a cost if it is true.
26590 Update calls to aarch64_strip_extend.
26591 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26593 2017-06-21 Andrew Pinski <apinski@cavium.com>
26595 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26597 (thunderxt88): Likewise.
26598 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26599 (thunderx_prefetch_tune): New variable.
26600 (thunderx2t99_prefetch_tune): Update for the correct values.
26601 (thunderxt88_tunings): New variable.
26602 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26603 generic_prefetch_tune.
26604 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26606 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26608 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26609 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26610 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26611 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26612 (aarch64_atomic_cas<mode>, GPI): Likewise.
26614 2017-06-21 Martin Liska <mliska@suse.cz>
26616 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26617 statements on cold and hot labels.
26618 * predict.c (tree_estimate_probability_bb): Remove the
26619 prediction from this place.
26621 2017-06-21 Martin Liska <mliska@suse.cz>
26623 PR tree-optimization/79489
26624 * gimplify.c (maybe_add_early_return_predict_stmt): New
26626 (gimplify_return_expr): Call the function.
26627 * predict.c (tree_estimate_probability_bb): Remove handling
26629 * predict.def: Update comment about early return predictor.
26630 * gimple-predict.h (is_gimple_predict): New function.
26631 * predict.def: Change default value of early return to 66.
26632 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26634 * passes.def: Put pass_strip_predict_hints to the beginning of
26637 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26639 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26640 FUNCTION_DECL declarations.
26641 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26643 (dwaf2out_decl): Likewise.
26644 * godump.c (go_early_global_decl): Skip call to the real debug hook
26645 for FUNCTION_DECL declarations.
26646 * passes.c (rest_of_decl_compilation): Skip call to the
26647 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26648 -fdump-go-spec is passed.
26650 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26652 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26653 Reorder for compactness.
26654 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26655 (def_builtin_pure, def_builtin_pure2): New functions.
26656 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26658 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26660 * match.pd (nop_convert): New predicate.
26661 ((A +- CST1) +- CST2): Allow some NOP conversions.
26663 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26666 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26667 with ctors/dtors if GOVD_SHARED is set.
26669 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26671 * config/aarch64/aarch64.md (movti_aarch64):
26672 Emit mov rather than orr.
26673 (movtf_aarch64): Likewise.
26674 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26675 Emit mov rather than orr.
26677 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26679 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26680 Swap alternatives, make integer dup more expensive.
26682 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26684 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26685 Return true for non-tls symbols.
26687 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26689 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26690 (cortex-a75): Likewise.
26691 (cortex-a75.cortex-a55): Likewise.
26692 * config/aarch64/aarch64-tune.md: Regenerate.
26693 * doc/invoke.texi (-mtune): Document new values for -mtune.
26695 2017-06-21 Tom de Vries <tom@codesourcery.com>
26697 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26698 stack_size feature.
26699 (Effective-Target Keywords, Other attributes): Suggest using
26700 dg-add-options stack_size feature to get stack limit in stack_size
26701 effective target documentation.
26703 2017-06-21 Julian Brown <julian@codesourcery.com>
26704 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26706 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26707 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26708 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26710 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26711 attribute type list for neon_multiply.
26712 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26713 attribute type list for neon_multiply.
26714 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26715 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26716 attribute type list for neon_multiply.
26717 * config/arm/types.md (crypto_pmull): Add.
26718 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26719 attribute type list.
26721 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26723 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26726 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26728 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26729 to make sure not to dereference a NULL cost_classes_ptr pointer.
26731 2017-06-20 Carl Love <cel@us.ibm.com>
26733 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26734 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26735 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26736 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26737 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26738 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26739 VMULOSW): New enum "unspec" values.
26740 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26741 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26742 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26743 altivec_vmulosw): New patterns.
26744 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26745 VMULOSW): Add definitions.
26747 2017-06-20 Julia Koval <julia.koval@intel.com>
26749 * config/i386/i386.c: Fix rounding expand for new pattern.
26750 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26752 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26754 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26755 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26757 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26759 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26762 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26764 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26765 architecture, then by implementer ID.
26766 * config/aarch64/aarch64-tune.md: Regenerate.
26768 2017-06-20 Richard Biener <rguenther@suse.de>
26770 PR middle-end/81097
26771 * fold-const.c (split_tree): Fold to type before negating.
26773 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26775 * diagnostic-show-locus.c
26776 (selftest::test_fixit_deletion_affecting_newline): New function.
26777 (selftest::diagnostic_show_locus_c_tests): Call it.
26779 2017-06-20 Andreas Schwab <schwab@suse.de>
26782 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26785 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26787 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26788 __ARM_FEATURE_COPROC according to support.
26790 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26792 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26793 Rewritten to avoid overflow for > 32-bit pointers.
26796 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26797 by removing enum keyword.
26798 (ubsan_type_descriptor): Likewise. Formatting fix.
26801 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26802 splitter): Require TARGET_SSE2 in the condition.
26804 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26807 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26808 for doing vector set of SFmode on ISA 3.0.
26809 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26810 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
26812 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
26813 SFmode value into a V4SF variable that was extracted from another
26814 V4SF variable without converting the element to double precision
26815 and back to single precision vector format.
26816 (vsx_insert_extract_v4sf_p9_2): Likewise.
26818 2017-06-19 Jakub Jelinek <jakub@redhat.com>
26820 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
26821 in UWHI to avoid undefined overflow.
26824 * ubsan.h (enum ubsan_encode_value_phase): New.
26825 (ubsan_encode_value): Change second argument to
26826 enum ubsan_encode_value_phase with default value of
26827 UBSAN_ENCODE_VALUE_GENERIC.
26828 * ubsan.c (ubsan_encode_value): Change second argument to
26829 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
26830 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
26831 create_tmp_var_raw instead of create_tmp_var and use a
26833 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
26834 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
26835 ubsan_encode_value callers.
26838 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
26839 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
26840 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
26842 2017-06-19 Richard Biener <rguenther@suse.de>
26844 PR middle-end/81118
26845 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
26846 estimates if we changed anything.
26848 2017-06-19 Richard Biener <rguenther@suse.de>
26850 PR tree-optimization/80887
26851 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
26852 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
26853 simplified lookups, then reset mprts_hook.
26854 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
26856 (try_to_simplify): Likewise.
26858 2017-06-19 Martin Liska <mliska@suse.cz>
26861 * gimplify.c (gimplify_switch_expr):
26862 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
26864 2017-06-19 Martin Liska <mliska@suse.cz>
26866 * doc/install.texi: Document that PGO runs in 4 stages.
26868 2017-06-19 Martin Liska <mliska@suse.cz>
26871 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
26872 to dispatcher function name.
26873 * multiple_target.c (replace_function_decl): New function.
26874 (create_dispatcher_calls): Redirect both edges and references.
26876 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
26878 * profile-count.c (profile_count::dump): Dump quality.
26879 (profile_count::differs_from_p): Update for unsigned val.
26880 * profile-count.h (profile_count_quality): New enum.
26881 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
26883 2017-06-19 Richard Biener <rguenther@suse.de>
26885 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
26886 struct function as arg.
26887 (estimate_numbers_of_iterations): Export overload with loop arg.
26888 (free_numbers_of_iterations_estimates_loop): Use an overload of
26889 free_numbers_of_iterations_estimates instead.
26890 * tree-cfg.c (remove_bb): Adjust.
26891 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
26892 * tree-parloops.c (gen_parallel_loop): Likewise.
26893 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
26895 (tree_unroll_loops_completely): Likewise.
26896 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
26897 Use an overload instead and export.
26898 (estimated_loop_iterations): Adjust.
26899 (max_loop_iterations): Likewise.
26900 (likely_max_loop_iterations): Likewise.
26901 (estimate_numbers_of_iterations): Take struct function as arg
26903 (loop_exits_before_overflow): Adjust.
26904 (free_numbers_of_iterations_estimates_loop): Use an overload.
26905 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
26906 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
26908 2017-06-19 Richard Biener <rguenther@suse.de>
26911 * ipa-prop.c (find_constructor_constant_at_offset): Handle
26912 RANGE_EXPR conservatively.
26914 2017-06-16 Carl Love <cel@us.ibm.com>
26916 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26917 definitions for vec_float, vec_float2, vec_floato,
26918 vec_floate built-ins.
26919 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
26920 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
26922 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
26923 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
26924 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
26925 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
26926 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
26927 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
26928 vec_floato): Add builtin defines.
26929 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
26930 Update the built-in documentation file for the new built-in
26933 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26935 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
26936 (mthumb): Mark as the negative of -marm.
26938 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26940 * doc/invoke.texi (ARM Options, -mcpu): Document supported
26942 (ARM Options, -mtune): Document that this accepts the same
26943 extension options as -mcpu.
26944 (ARM Options, -mfpu): Document addition of -mfpu=auto.
26946 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26948 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
26949 permitted extensions.
26951 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26953 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
26954 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
26955 (armv8-m.main): Add option +nodsp.
26956 * config/arm/arm-cpu-cdata.h: Regenerated.
26958 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26960 * config/arm/t-fuchsia: New file.
26961 * config.gcc (arm*-*-fuchsia*): Use it.
26963 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26965 * config/arm/t-symbian: Rewrite for new option infrastructure.
26967 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26969 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
26970 (MULTILIB_REQUIRED): Likewise.
26972 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26974 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
26975 (MULTILIB_RESUE): Likewise.
26976 (MULTILIB_MATCHES): Likewise.
26977 (MULTLIB_REQUIRED): Likewise.
26979 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26981 * config/arm/t-rtems: Rewrite for new option framework.
26983 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26985 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
26986 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
26987 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
26988 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
26989 * config/arm/t-multilib: ... here.
26990 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
26991 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
26992 armv7-a and armv8*-a when A-profile libraries have not been built.
26993 * config/arm/t-rmprofile: Rewrite.
26995 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26997 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
26998 with a backslash. Remove the backslash after substituting unescaped
27000 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27002 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27004 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27005 the driver to pass through the default setting of -mfloat-abi.
27006 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27008 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27009 (all_feat_combs): New rule.
27010 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27012 * config/arm/t-aprofile: Rewrite.
27014 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27016 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27017 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27018 fpu is not specified by the user/command-line.
27019 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27020 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27021 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27022 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27023 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27024 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27026 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27028 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27029 * config/arm/t-arm-elf: Rewritten.
27031 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27033 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27034 have some floating-point instructions.
27035 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27036 (TARGET_MAYBE_HARD_FLOAT): New macro.
27037 * config/arm/arm-builtins.c (arm_init_builtins): Use
27038 TARGET_MAYBE_HARD_FLOAT.
27039 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27041 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27043 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27044 (configargs.h): Include it.
27045 (arm_print_hint_for_fpu_option): New function.
27046 (arm_parse_fpu_option): New function.
27047 (candidate_extension): New class.
27048 (arm_canon_for_multilib): New function.
27049 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27050 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27051 (ARCH_CANONICAL_SPECS): New macro.
27052 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
27054 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27056 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
27057 are set after handling multilib fragments. Set target_cpu_default2
27060 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27062 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
27064 (arm*-*-*): Set target_cpu_default2 to a quoted string.
27065 * config/arm/parsecpu.awk (check_cpu): Validate any extension
27067 (check_arch): Likewise.
27068 * config/arm/arm.c (arm_configure_build_target): Handle
27069 TARGET_CPU_DEFAULT being a string constant. Scan any feature
27070 options in the default.
27072 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27074 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
27075 when an option is an alias of another.
27076 * config/arm/parsecpu.awk (optalias): New parser token.
27077 (gen_comm_data): Mark non-alias options as such. Emit entries
27078 for extension aliases.
27079 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
27080 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
27081 (armv6kz, armv6zk, armv6t2): Likewise.
27082 (armv7): Make vfpv3-d16 an alias.
27083 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
27085 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
27086 Sort in canonical order.
27087 (armv8-a): Sort in canonical order.
27088 (armv8.1-a, armv8.2-a): Likewise.
27089 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
27091 (cortex-a9): Sort in canonical order.
27092 * config/arm/arm.c (selftests.h): Include it.
27093 (arm_test_cpu_arch_data): New function.
27094 (arm_run_self_tests): New function.
27095 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
27096 (targetm): Move declaration to the end of the file.
27097 * arm-cpu-cdata.h: Regenerated.
27099 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27101 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
27102 call to target_mode_check describing the type of option passed.
27103 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
27104 (arm_target_thumb_only): Use arm_parse_arch_option_name or
27105 arm_parse_cpu_option_name to match parameters against list of
27107 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
27108 arm_arch_core_flags data structure.
27109 * config/arm/arm-cpu_cdata.h: Regenerated.
27111 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27113 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
27115 (arm_print_hint_for_cpu_option): Likewise.
27116 (arm_print_hint_for_arch_option): Likewise.
27117 (arm_parse_cpu_option_name): Likewise.
27118 (arm_parse_arch_option_name): Likewise.
27119 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
27120 of entries in the all_fpus list.
27121 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
27122 (arm_parse_cpu_option_name): Declare.
27123 (arm_parse_arch_option_name): Declare.
27124 (arm_parse_option_features): Declare.
27125 (arm_intialize_isa): Declare.
27126 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
27128 (gen_comm_data): ... here. Make definitions non-static.
27129 * config/arm/arm-cpu-data.h: Regenerated.
27130 * config/arm/arm-cpu-cdata.h: Regenerated.
27132 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27134 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
27135 (cpu_arch_extension): New structure.
27136 (cpu_arch_option, arch_option, cpu_option): New structures.
27137 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
27138 architecture types.
27139 (gen_data): Generate new format data tables.
27140 * config/arm/arm.c (cpu_tune): New structure.
27141 (cpu_option, processors): Delete.
27142 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
27143 (arm_print_hint_for_cpu_option): ... this and ...
27144 (arm_print_hint_for_arch_option): ... this.
27145 (arm_parse_arch_cpu_name): Delete. Replace with ...
27146 (arm_parse_cpu_option_name): ... this and ...
27147 (arm_parse_arch_option_name): ... this.
27148 (arm_unrecognized_feature): Change type of target parameter to
27150 (arm_parse_arch_cpu_features): Delete. Replace with ...
27151 (arm_parse_option_features): ... this.
27152 (arm_configure_build_target): Rework to use new configuration data
27154 (arm_print_tune_info): Rework for new configuration data tables.
27155 * config/arm/arm-cpu-data.h: Regenerated.
27156 * config/arm/arm-cpu.h: Regenerated.
27158 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27160 * Makefile.in (OBJS): Move sbitmap.o from here ...
27161 (OBJS-libcommon): ... to here.
27163 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27165 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
27166 (ISA_ALL_CRYPTO): New macro.
27167 (ISA_ALL_SIMD): New macro
27168 (ISA_ALL_FP): New macro.
27169 * config/arm/arm.c (fpu_bitlist): Update initializer.
27170 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
27172 (arm9e): Add fpu. Add option for nofp
27173 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
27174 (arm926ej-s, arm1026ej-s): Likewise.
27175 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
27176 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
27177 neon-fp16, neon-vfpv4, nofp and nosimd.
27178 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
27179 (cortex-a8): Add fpu. Add option for nofp.
27180 (cortex-a9): Add fpu. Add options for nosimd and nofp.
27181 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
27182 (cortex-r4f): Add fpu.
27183 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
27184 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
27186 (cortex-r8): Likewise.
27187 (cortex-m4): Add fpu. Add option for nofp.
27188 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
27189 (cortex-a17.cortex-a7): Likewise.
27190 (cortex-a32): Add fpu. Add options for crypto and nofp.
27191 (cortex-a35, cortex-a53): Likewise.
27192 (cortex-a57): Add fpu. Add option for crypto.
27193 (cortex-a72, cortex-a73): Likewise.
27194 (exynos-m1): Likewise.
27195 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
27196 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
27197 (cortex-m33): Add fpu. Add option for nofp.
27198 * config/arm/arm-cpu-cdata.h: Regenerated
27199 * config/arm/arm-cpu-data.h: Regenerated.
27201 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27203 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
27204 (armv5te, armv5tej): Likewise.
27205 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
27206 (armv7): Add options fp and vfpv3-d16.
27207 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
27208 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
27210 (armv7ve): Likewise.
27211 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
27212 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
27213 (armv8-a): Add nocrypto option.
27214 (armv8.1-a, armv8.2-a): Likewise.
27215 (armv8-m.main): add options fp, fp.dp and nofp.
27217 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27219 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
27221 (armv8-a+crc): Delete.
27222 (armv8.1-a): Add options simd, crypto and nofp.
27223 (armv8.2-a): Add options fp16, simd, crypto and nofp.
27224 (armv8.2-a+fp16): Delete.
27225 (armv8-m.main): Add option dsp.
27226 (armv8-m.main+dsp): Delete.
27227 (cortex-a8): Add fpu. Add nofp option.
27228 (cortex-a9): Add fpu. Add nofp and nosimd options.
27229 * config/arm/parsecpu.awk (gen_data): Generate option tables and
27230 link to main cpu and architecture data structures.
27231 (gen_comm_data): Only put isa attributes from the main architecture
27233 (option): New statement for architecture and CPU entries.
27234 * arm.c (struct cpu_option): New structure.
27235 (struct processors): Add entry for options.
27236 (arm_unrecognized_feature): New function.
27237 (arm_parse_arch_cpu_name): Ignore any characters after the first
27239 (arm_parse_arch_cpu_feature): New function.
27240 (arm_configure_build_target): Separate out any CPU and architecture
27241 features and parse separately. Don't error out if -mfpu=auto is
27242 used with only an architecture string.
27243 (arm_print_asm_arch_directives): New function.
27244 (arm_file_start): Call it.
27245 * config/arm/arm-cpu-cdata.h: Regenerated.
27246 * config/arm/arm-cpu-data.h: Likewise.
27247 * config/arm/arm-tables.opt: Likewise.
27249 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27251 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
27252 assembler when it is not -mfpu=auto.
27254 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27256 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
27257 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
27258 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
27259 (ASM_CPU_SPEC): Rewrite.
27260 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
27261 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
27262 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
27263 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
27264 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
27265 copied string is NUL-terminated. Also strip any characters prefixed
27267 (arm_rewrite_selected_arch): New function.
27268 (arm_rewrite_march): New function.
27270 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27272 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
27273 (x_arm_cpu_string, x_arm_tune_string): Likewise.
27274 (march, mcpu, mtune): Convert to string-based options.
27275 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
27276 (arm_parse_arch_cpu_name): New function.
27277 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
27278 identify selected architecture or CPU.
27279 (arm_option_save): New function.
27280 (TARGET_OPTION_SAVE): Redefine.
27281 (arm_option_restore): Restore string options.
27282 (arm_option_print): Print string options.
27284 2017-06-16 Martin Sebor <msebor@redhat.com>
27286 PR tree-optimization/80933
27287 PR tree-optimization/80934
27288 * builtins.c (fold_builtin_3): Do not handle bcmp here.
27289 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
27290 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
27291 (gimple_fold_builtin): Call them.
27293 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27295 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
27296 as unlikely; update profile.
27298 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27300 * predict.c (force_edge_cold): Handle declaring edges impossible
27303 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27305 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
27307 (try_unroll_loop_completely): Fix reporting.
27309 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27311 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
27313 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
27316 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
27317 if given a non-constant argument for an intrinsic which requires a
27320 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27322 * profile.c (compare_freqs): New function.
27323 (branch_prob): Sort edge list.
27324 (find_spanning_tree): Assume that the list is priority sorted.
27326 2017-06-16 Richard Biener <rguenther@suse.de>
27328 PR tree-optimization/81090
27329 * passes.def (pass_record_bounds): Remove.
27330 * tree-pass.h (make_pass_record_bounds): Likewise.
27331 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
27332 make_pass_record_bounds): Likewise.
27333 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
27334 not free niter estimates at the beginning but at the end.
27335 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
27337 2017-06-16 Richard Biener <rguenther@suse.de>
27339 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
27340 initializer to workaround ICE in host GCC 4.8.
27342 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27344 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
27346 (clone_inlined_nodes): Update.
27348 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27350 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
27351 prefetch settings, and enable prefetching by default at -O3.
27353 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27355 * config/aarch64/aarch64.c (aarch64_override_options_internal):
27356 Set flag_prefetch_loop_arrays according to tuning data.
27358 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27360 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
27361 New tune structure.
27362 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
27363 [Unrelated to main purpose of the patch] Place the pointer field last
27364 to enable type checking errors when tune structure are wrongly merged.
27365 * config/aarch64/aarch64.c (generic_prefetch_tune,)
27366 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
27367 (thunderx2t99_prefetch_tune): New tune constants.
27368 (tune_params *_tunings): Update all tunings (no functional change).
27369 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
27370 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
27371 from tunings structures.
27373 2017-06-16 Jakub Jelinek <jakub@redhat.com>
27376 * ubsan.c (instrument_null): Add T argument, use it instead
27377 of computing it based on IS_LHS.
27378 (instrument_object_size): Likewise.
27379 (pass_ubsan::execute): Adjust instrument_null and
27380 instrument_object_size callers to pass gimple_get_lhs or
27381 gimple_assign_rhs1 result to it. Use instrument_null instead of
27382 calling get_base_address and instrument_mem_ref. Handle
27383 aggregate call arguments for object-size sanitization.
27385 2017-06-16 Yury Gribov <tetra2005@gmail.com>
27387 PR tree-optimization/81089
27388 * tree-vrp.c (is_masked_range_test): Validate operands of
27391 2017-06-15 Martin Sebor <msebor@redhat.com>
27394 * dumpfile.c (dump_register): Avoid calling memset to initialize
27395 a class with a default ctor.
27396 * gcc.c (struct compiler): Remove const qualification.
27397 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
27398 * hash-table.h: Ditto.
27399 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
27401 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
27402 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
27404 * params.h (struct param_info): Make struct members non-const.
27405 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
27406 with default initialization.
27407 * vec.h (vec_copy_construct, vec_default_construct): New helper
27409 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
27410 with vec_copy_construct.
27411 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
27412 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
27413 * doc/invoke.texi (-Wclass-memaccess): Document.
27415 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27417 * emit-rtl.h (is_leaf): Update comment about local
27418 register allocator.
27420 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27423 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
27424 for a variable to have a section before checking if the section has a
27426 Set section to.persistent if persistent attribute is set.
27427 Warn if .persistent attribute is used on an automatic variable.
27429 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
27431 PR rtl-optimization/80474
27432 * reorg.c (update_block): Do not ignore instructions in a delay slot.
27434 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
27436 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
27439 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
27441 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
27442 (casesi): Emit bounds checking as RTL.
27443 (casesi_internal_mips16_<mode>): Remove bounds checking.
27445 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
27447 * config/xtensa/xtensa.c (xtensa_option_override): Append
27448 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
27449 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
27450 xtensa_doloop_hooks): Define unconditionally.
27451 (xtensa_reorg_loops): Only call reorg_loops in the presence of
27453 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
27454 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
27455 for it in xtensa_option_override.
27456 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
27457 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
27459 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
27461 * doc/cppopts.texi: Document '-' special value to -MF.
27463 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
27465 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
27466 (cortex_a53_fconst): Likewise.
27467 (cortex_a53_fpmul): Likewise.
27468 (cortex_a53_f_load_64): Likewise.
27469 (cortex_a53_f_load_many): Likewise.
27470 (cortex_a53_advsimd_alu): Likewise.
27471 (cortex_a53_advsimd_alu_q): Likewise.
27472 (cortex_a53_advsimd_mul): Likewise.
27473 (cortex_a53_advsimd_mul_q): Likewise.
27474 (fpmac bypass): Add new bypass for fpmac-fpmac case.
27475 Add missing fmul, r2f_cvt and fconst cases.
27477 2017-06-14 Richard Biener <rguenther@suse.de>
27479 PR middle-end/81088
27480 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
27482 (fold_binary_loc): When associating do not treat pre-existing
27483 TREE_OVERFLOW on literal constants as a reason to allow
27484 TREE_OVERFLOW on associated literal constants.
27486 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27488 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
27489 (MASK_FEATURES): New macro.
27490 * config/sparc/sparc.c (sparc_option_override): Remove the special
27491 handling of -mfpu and generalize it to all MASK_FEATURES switches.
27493 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27495 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
27496 a division of 0 if non-call exceptions are enabled.
27498 2017-06-14 Andrew Pinski <apinski@cavium.com>
27499 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27502 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
27503 Improve vector initialization code gen for only variable case.
27505 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27507 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
27509 2017-06-14 Richard Biener <rguenther@suse.de>
27511 PR tree-optimization/81083
27512 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
27515 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27517 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
27518 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
27519 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
27520 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
27521 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
27522 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
27524 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27526 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
27527 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
27529 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27531 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
27533 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27535 * config/rs6000/t-rtems: Don't handle SPE.
27537 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27539 * config/rs6000/t-linux: Don't handle SPE.
27541 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27543 * config/rs6000/eabispe.h: Delete file.
27545 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27547 * config/rs6000/t-spe: Delete file.
27549 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27551 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
27552 (rs6000_legitimate_offset_address_p): Return false for anything in
27553 V2SImode or V2SFmode.
27555 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27557 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
27558 except V2SF and V2SI. Rearrange the vector modes, and add comments.
27559 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
27561 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
27562 (rs6000_legitimate_offset_address_p): Ditto.
27563 (rs6000_emit_move): Ditto.
27564 (rs6000_init_builtins): Remove V4HI_type_node.
27566 2017-06-13 Martin Liska <mliska@suse.cz>
27569 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27570 (gate_asan): Likewise.
27571 * asan.h (asan_no_sanitize_address_p): Remove the function.
27572 (sanitize_flags_p): New function.
27573 * builtins.def: Fix coding style.
27574 * common.opt: Use renamed enum value.
27575 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27576 * doc/extend.texi: Document no_sanitize attribute.
27577 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27578 to SANITIZE_UNDEFINED_NONDEFAULT.
27579 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27580 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27581 Use sanitize_flags_p.
27582 * gimplify.c (gimplify_function_tree): Likewise.
27583 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27584 * opts.c (parse_no_sanitize_attribute): New function.
27585 (common_handle_option): Use renamed enum value.
27586 * opts.h (parse_no_sanitize_attribute): Declare.
27587 * tree.c (sanitize_flags_p): New function.
27588 * tree.h: Declared here.
27589 * tsan.c: Use sanitize_flags_p.
27590 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27591 (instrument_mem_ref): Likewise.
27592 (instrument_bool_enum_load): Likewise.
27593 (do_ubsan_in_current_function): Remove the function.
27594 (pass_ubsan::execute): Use sanitize_flags_p.
27595 * ubsan.h: Remove do_ubsan_in_current_function
27596 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27597 (dump_function_to_file): Use it here.
27599 2017-06-13 Martin Jambor <mjambor@suse.cz>
27601 PR tree-optimization/80803
27602 PR tree-optimization/81063
27603 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27604 (propagate_subaccesses_across_link): Enqueue subtree whenever
27605 necessary instead of relying on the caller.
27607 2017-06-13 Martin Jambor <mjambor@suse.cz>
27609 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27610 that have a first_link.
27611 (sort_and_splice_var_accesses): Do not check first_link before
27613 (subtree_mark_written_and_enqueue): Likewise.
27614 (propagate_all_subaccesses): Likewise and do not stop at first
27615 parent with a first_link.
27617 2017-06-13 Martin Jambor <mjambor@suse.cz>
27619 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27622 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27624 * match.pd: New pattern.
27626 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27628 * tree-vrp.c (is_masked_range_test): New function.
27629 (register_edge_assert_for): Determine ranges for
27632 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27634 PR tree-optimization/67328
27635 * fold-const.c (maskable_range_p): New function.
27636 (build_range_check): Generate bittests if possible.
27638 2017-06-13 Martin Liska <mliska@suse.cz>
27640 * gimple-pretty-print.c (dump_probability): Add new argument.
27641 (dump_edge_probability): Dump both probability and count.
27642 (dump_gimple_label): Likewise.
27643 (dump_gimple_bb_header): Likewise.
27645 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27648 * config/avr/avr-devices.c: Fix indentation.
27649 * config/avr/gen-avr-mmcu-specs.c: Dito.
27651 2017-06-13 Richard Biener <rguenther@suse.de>
27653 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27654 instead get vector type from stmt_info.
27655 (vectorizable_reduction): Adjust. Remove dead code.
27657 2017-06-13 Richard Biener <rguenther@suse.de>
27659 PR middle-end/81065
27660 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27661 case of C * (x * C2 + C3).
27662 (fold_addr_of_array_ref_difference): Properly fold index difference.
27664 2017-06-12 David S. Miller <davem@davemloft.net>
27667 * config/sparc/sparc.md (return expander): Emit frame blockage if
27668 function uses alloca.
27670 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27672 * combine.c (make_field_assignment): Check len rather than the mode
27673 precision when calling force_to_mode.
27675 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27677 Support multilibs and devices that see flash in RAM address range.
27680 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27681 (avr_mcu_t) <flash_pm_offset>: New field.
27682 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27683 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27684 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27685 (AVR_TINY_PM_OFFSET): Remove macro.
27686 * config/avr/avr.opt (-mshort-calls): New option.
27687 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27688 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27689 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27690 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27691 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27692 instead of avr_arch->have_jmp_call.
27693 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27694 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27695 avr_arch->flash_pm_offset to define.
27696 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27697 new field flash_pm_offset. Add entry for avrxmega3.
27698 (avr_texinfo): Add entry for avrxmega3.
27699 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27700 attiny212, attiny214,
27701 attiny412, attiny414, attiny416, attiny417,
27702 attiny814, attiny816, attiny817,
27703 attiny1614, attiny1616, attiny1617,
27704 attiny3214, attiny3216, attiny3217.
27705 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27706 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27707 (avr_print_operand_address) [AVR_TINY]: Same.
27708 (avr_asm_init_sections) <readonly_data_section>: Only patch
27709 callback if avr_arch->flash_pm_offset = 0.
27710 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27711 for rodata if avr_arch->flash_pm_offset != 0.
27712 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27713 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27714 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27715 (m_options): Append opt_rcall.
27716 (m_dirnames): Append dir_rcall.
27717 * config/avr/t-multilib: Regenerate.
27719 * configure.ac [target=avr]: Check whether avrxmega3 default
27720 linker description file works as needed.
27721 * configure: Regenerate.
27722 * doc/avr-mmcu.texi: Regenerate.
27723 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27724 <__AVR_ARCH__>: Document avrxmega3 and 103.
27725 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27726 <__AVR_SHORT_CALLS__>: Document it.
27727 <__AVR_PM_BASE_ADDRESS__>: Document it.
27728 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27729 (AVR Variable Attributes) <progmem>: Document this is
27730 not needed for avrxmega3.
27731 (AVR Named Address Spaces) <__flash>: Dito.
27733 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27735 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27737 2017-06-12 Doug Rupp <rupp@adacore.com>
27739 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27740 Append vxworks-stdint.h to the tm_file list.
27741 * config/vxworks-stdint.h: New file.
27743 2017-06-12 Martin Liska <mliska@suse.cz>
27745 PR tree-optimization/81041
27746 * tree-profile.c (gimple_gen_ic_func_profiler):
27747 Create an extra BB in profile-generate
27748 (gimple_gen_time_profiler): Likewise.
27750 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27752 PR tree-optimization/81003
27753 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27754 (update_range_test): Use it instead of force_gimple_operand_gsi.
27756 2017-06-12 Richard Biener <rguenther@suse.de>
27758 PR tree-optimization/81053
27759 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27760 with backedge value not defined in loop. Simplify def stmt
27763 2017-06-11 Tom de Vries <tom@codesourcery.com>
27766 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27768 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27769 nvptx_cannot_force_const_mem.
27771 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27773 * opts.c (finish_options): Move test for flag_split_stack after
27774 it has been initialized.
27776 2017-06-11 Jason Merrill <jason@redhat.com>
27778 * tree.h (id_equal): New.
27779 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27780 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27781 instead of strcmp of IDENTIFIER_POINTER.
27783 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27785 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27786 (mark_all_inlined_calls_cdtor): Fix formating.
27787 (inline_transform): Rescale profile before inlining.
27789 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27791 * cgraph.h (cgraph_edge::clone): Update prototype.
27792 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27793 (cgraph_node::create_clone): Update.
27794 (cgraph_node::create_version_clone): Update.
27795 * tree-inline.c (copy_bb): Update.
27796 (expand_call_inline): Update.
27798 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27800 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27801 factored out from ...
27802 (rs6000_emit_prologue): ... here.
27804 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27806 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27807 factored out from ...
27808 (rs6000_emit_prologue): ... here.
27810 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27812 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
27814 (handle_missing_profiles): Fix computation of tp_first_run.
27815 (counts_to_freqs): Do not touch freqs when count is 0.
27817 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27819 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
27822 2017-06-10 Tom de Vries <tom@codesourcery.com>
27824 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27825 attributes): Document signal effective target.
27827 2017-06-10 Tom de Vries <tom@codesourcery.com>
27829 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27830 Document effective target stack_size.
27832 2017-06-09 David Malcolm <dmalcolm@redhat.com>
27834 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
27835 to the edit_context if they can be auto-applied.
27837 2017-06-9 Ian Lance Taylor <iant@golang.org>
27839 * opts.c (finish_options): If -fsplit-stack, disable implicit
27840 -forder-blocks-and-partition.
27841 * doc/invoke.texi (Optimize Options): Document that when using
27842 -fsplit-stack -forder-blocks-and-partition is not implicitly
27845 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27847 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
27848 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
27849 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
27850 * builtins.def (abort, trap, unreachable): Declare cold.
27851 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
27852 * tree-core.h (ECF_COLD): New.
27853 * tree.c (set_call_expr_flags): Handle ECF_COLD.
27854 (build_common_builtin_nodes): Mark unreachable and abort as cold.
27856 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27858 * predict.c (unlikely_executed_stmt_p): Cleanup.
27860 2017-06-09 Richard Biener <rguenther@suse.de>
27862 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
27863 model if the ref is always written to.
27865 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27867 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
27869 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27871 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
27874 2017-06-09 Tom de Vries <tom@codesourcery.com>
27877 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
27878 "target cannot support label values" when encountering LABEL_REF.
27880 2017-06-09 Martin Liska <mliska@suse.cz>
27882 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
27883 (gimple_gen_ic_func_profiler): Emit direct comparison
27884 of __gcov_indirect_call_callee with NULL.
27885 (gimple_gen_time_profiler): Change probability from
27886 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
27888 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27890 * profile.c (edge_gcov_counts): Turn to pointer.
27891 (compute_branch_probabilities, compute_branch_probabilities): Update.
27892 (branch_prob): Do not clear edge_gcov_count.
27893 * profile.h (edge_gcov_counts): Turn to pointer.
27894 (edge_gcov_count): Update.
27896 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27898 * gimple.h (gimple_check_failed): Mark cold.
27900 2017-06-09 Richard Biener <rguenther@suse.de>
27902 PR tree-optimization/66623
27903 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
27904 refactor check_reduction into two parts, properly computing
27905 whether we have to check reduction validity for outer loop
27908 2017-06-09 Richard Biener <rguenther@suse.de>
27910 PR tree-optimization/79483
27911 * graphite-scop-detection.c (order): New global.
27912 (get_order): Compute bb to order mapping that satisfies code
27913 generation constraints.
27914 (cmp_pbbs): New helper.
27915 (build_scops): Start domwalk at entry block, sort generated
27918 2017-06-09 Richard Biener <rguenther@suse.de>
27920 PR middle-end/81007
27921 * ipa-polymorphic-call.c
27922 (ipa_polymorphic_call_context::restrict_to_inner_class):
27923 Skip FIELD_DECLs with error_mark_node type.
27924 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
27927 2017-06-09 Martin Liska <mliska@suse.cz>
27929 * predict.c (struct branch_predictor): New struct.
27930 (test_prediction_value_range): New test.
27931 (predict_c_tests): New function.
27932 * selftest-run-tests.c (selftest::run_tests): Run the function.
27933 * selftest.h: Declare new tests.
27935 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
27938 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
27939 gen_add3_insn did not fail.
27940 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
27941 r0, construct that number in a temporary reg and add that reg to r0.
27942 If asked to put the result in r0 as well, fail.
27944 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
27946 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27947 for early expansion of vec_eqv.
27949 2017-06-08 Jakub Jelinek <jakub@redhat.com>
27951 PR middle-end/81005
27952 * ubsan.c (instrument_null): Avoid pointless code temporary.
27953 (pass_ubsan::execute): Instrument aggregate arguments of calls.
27955 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
27959 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
27962 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
27965 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27967 * predict.c (unlikely_executed_edge_p): Move ahead.
27968 (probably_never_executed_edge_p): Use it.
27970 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27972 PR middle-end/79988
27973 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
27974 gimple_call_builtin_p call.
27976 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27978 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
27979 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
27980 rtl_check_failed_type2, rtl_check_failed_code1,
27981 rtl_check_failed_code2, rtl_check_failed_code_mode,
27982 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
27983 rtvec_check_failed_bounds, rtl_check_failed_flag,
27984 _fatal_insn_not_found, _fatal_insn): Likewise.
27985 * tree.h (tree_contains_struct_check_failed,
27986 tree_check_failed, tree_not_check_failed,
27987 tree_class_check_failed, tree_range_check_failed,
27988 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
27989 tree_vec_elt_check_failed, phi_node_elt_check_failed,
27990 tree_operand_check_failed, omp_clause_check_failed,
27991 omp_clause_operand_check_failed, omp_clause_range_check_failed):
27994 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27996 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
27997 flag_branch_probabilities.
27998 * ipa-inline.c (edge_badness): Likewise.
27999 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28000 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28001 * predict.c (maybe_hot_frequency_p): Likewise.
28002 (probably_never_executed): Likewise.
28003 * sched-ebb.c (schedule_ebbs): Likewise.
28004 * sched-rgn.c (find_single_block_region): Likewise.
28005 * tracer.c (tail_duplicate): Likewise.
28007 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28009 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28010 longer requires x_flag_profile_use.
28012 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28014 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28015 instead of flag_reorder_blocks_and_partition.
28016 * dbxout.c (dbxout_function_end): Likewise.
28017 * dwarf2out.c (gen_subprogram_die): Likewise.
28018 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28019 * hw-doloop.c (reorg_loops): Likewise.
28020 * varasm.c (assemble_start_function,
28021 assemble_end_function): Likewise.
28022 (decide_function_section): Do not check for
28023 flag_reorder_blocks_and_partition.
28025 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28027 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28029 (chkp_get_hard_register_fake_addr_expr): Ditto.
28030 (chkp_build_addr_expr): Add check for hard reg case.
28031 (chkp_parse_array_and_component_ref): Ditto.
28032 (chkp_find_bounds_1): Ditto.
28033 (chkp_process_stmt): Don't generate bounds store for
28036 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28038 * predict.c (maybe_hot_bb_p): Do not check profile status.
28039 (maybe_hot_edge_p): Likewise.
28040 (probably_never_executed): Check for zero counts even if profile
28042 (unlikely_executed_edge_p): New function.
28043 (unlikely_executed_stmt_p): New function.
28044 (unlikely_executed_bb_p): New function.
28045 (set_even_probabilities): Use unlikely predicates.
28046 (combine_predictions_for_bb): Likewise.
28047 (predict_paths_for_bb): Likewise.
28048 (predict_paths_leading_to_edge): Likewise.
28049 (determine_unlikely_bbs): New function.
28050 (estimate_bb_frequencies): Use it.
28051 (compute_function_frequency): Use zero counts even if profile is
28053 * profile-count.h: Fix typo.
28055 2017-08-08 Julia Koval <julia.koval@intel.com>
28057 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
28058 _mm512_mask_cvtsepi16_storeu_epi8,
28059 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
28060 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
28061 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
28062 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
28063 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
28064 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
28065 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
28066 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
28067 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
28068 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
28069 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
28070 __builtin_ia32_pmovuswb256mem_mask,
28071 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
28072 __builtin_ia32_pmovwb512mem_mask): New builtins.
28074 2017-08-08 Julia Koval <julia.koval@intel.com>
28076 PR target/73350,80862
28077 * config/i386/subst.md (round): Fix round pattern.
28078 * config/i386/i386.c (ix86_erase_embedded_rounding):
28079 Fix erasing rounding for the fixed pattern.
28081 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28083 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
28085 2017-06-08 Martin Liska <mliska@suse.cz>
28087 PR gcov-profile/80911
28088 * gcov.c (block_info::block_info): New constructor.
28090 2017-06-07 Carl Love <cel@us.ibm.com>
28092 * config/rs6000/rs6000-c: The return type of the following
28093 built-in functions was implemented as int not long long. Fix sign
28094 of return value for the unsigned version of vec_mulo and vec_mule.
28095 vector unsigned long long vec_bperm (vector unsigned long long,
28096 vector unsigned char)
28097 vector signed long long vec_mule (vector signed int,
28099 vector unsigned long long vec_mule (vector unsigned int,
28100 vector unsigned int)
28101 vector signed long long vec_mulo (vector signed int,
28103 vector unsigned long long vec_mulo (vector unsigned int,
28104 vector unsigned int)
28105 * doc/extend.texi: Fix the documentation for the built-in
28108 2017-06-07 Carl Love <cel@us.ibm.com>
28111 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
28114 2017-06-07 Carl Love <cel@us.ibm.com>
28116 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
28117 support, Generate doublehv for signed int/float for BE case only.
28119 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
28121 * doc/invoke.texi (mcx16): Rewrite.
28123 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28125 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
28126 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
28127 *mov<mode>_softfloat, and an anonymous splitter): Use
28128 nonimmediate_operand instead of rs6000_nonimmediate_operand.
28130 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28132 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
28134 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28135 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
28136 (rs6000_debug_reg_global): Adjust.
28137 (rs6000_init_hard_regno_mode_ok): Adjust.
28138 (rs6000_dbx_register_number): Adjust.
28139 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
28140 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28141 Remove SPE_ACC and SPEFSCR.
28142 (REG_ALLOC_ORDER): Ditto.
28143 (FRAME_POINTER_REGNUM): Change to 111.
28144 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
28145 (REG_CLASS_NAMES): Ditto.
28146 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
28147 (REGISTER_NAMES): Ditto.
28148 (ADDITIONAL_REG_NAMES): Ditto.
28149 (rs6000_reg_names): Ditto.
28150 * config/rs6000/rs6000.md: Renumber some register number
28153 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28155 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
28157 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28158 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
28160 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
28161 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28162 Delete the SPE high registers.
28163 (REG_ALLOC_ORDER): Ditto.
28164 (enum reg_class): Remove SPE_HIGH_REGS.
28165 (REG_CLASS_NAMES): Ditto.
28166 (REG_CLASS_CONTENTS): Delete the SPE high registers.
28167 (REGISTER_NAMES): Ditto.
28168 (rs6000_reg_names): Ditto.
28169 * doc/tm.texi.in: Remove SPE as example.
28170 * doc/tm.texi: Regenerate.
28172 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28174 * config/rs6000/8540.md (ppc8540_brinc): Delete.
28175 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
28176 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
28177 * config/rs6000/rs6000.md (type): Remove "brinc".
28179 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28181 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
28182 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
28183 * config/rs6000/linuxspe.h: Delete file.
28184 * config/rs6000/rs6000.md: Don't include spe.md.
28185 * config/rs6000/spe.h: Delete file.
28186 * config/rs6000/spe.md: Delete file.
28187 * config/rs6000/t-rs6000: Remove spe.md.
28189 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28191 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
28192 (reg_or_none500mem_operand): Delete.
28193 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
28194 instead of reg_or_none500mem_operand.
28196 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28198 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28199 handling of SPE flags.
28200 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
28202 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28204 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
28206 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
28207 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
28208 paired_divv2sf3): Similar.
28209 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
28210 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
28211 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
28213 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
28214 Rename the paired_* instruction patterns.
28215 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
28217 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
28218 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
28219 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
28220 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
28221 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
28222 PAIRED_VECTOR_MODE.
28223 (struct machine_function): Delete field spe_insn_chain_scanned_p.
28224 (spe_func_has_64bit_regs_p): Delete.
28225 (spe_expand_predicate_builtin): Delete.
28226 (spe_expand_evsel_builtin): Delete.
28227 (TARGET_DWARF_REGISTER_SPAN): Do not define.
28228 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
28229 (invalid_e500_subreg): Delete.
28230 (rs6000_legitimize_address): Always force_reg op2 as well, for
28231 paired single memory accesses.
28232 (rs6000_member_type_forces_blk): Delete.
28233 (rs6000_spe_function_arg): Delete.
28234 (rs6000_expand_unop_builtin): Delete SPE handling.
28235 (rs6000_expand_binop_builtin): Ditto.
28236 (spe_expand_stv_builtin): Delete.
28237 (bdesc_2arg_spe): Delete.
28238 (spe_expand_builtin): Delete.
28239 (spe_expand_predicate_builtin): Delete.
28240 (spe_expand_evsel_builtin): Delete.
28241 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
28242 (spe_init_builtins): Delete.
28243 (spe_func_has_64bit_regs_p): Delete.
28244 (savres_routine_name): Delete "info" parameter. Adjust callers.
28245 (rs6000_emit_stack_reset): Ditto.
28246 (rs6000_dwarf_register_span): Delete.
28247 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
28248 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
28249 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
28251 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
28253 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
28254 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
28255 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
28256 mulv2sf3, divv2sf3): Delete expanders.
28258 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28260 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
28262 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28264 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
28265 * config/rs6000/rs6000.c: Ditto.
28267 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28269 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
28270 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
28271 comparison_operator.
28273 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28275 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
28276 * config/rs6000/rs6000.opt: Ditto.
28277 * config/rs6000/t-rtems: Ditto.
28279 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28281 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
28282 TARGET_E500_SINGLE by 0, simplify.
28283 * config/rs6000/rs6000.c: Ditto.
28284 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
28285 (spe_build_register_parallel): Delete.
28286 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
28287 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
28288 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
28289 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
28290 (E500_CONVERT): Delete.
28291 * config/rs6000/spe.md: Remove many patterns and all define_constants.
28293 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28295 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
28296 * config/rs6000/dfp.md: Ditto.
28297 (negdd2, *negdd2_fpr): Merge.
28298 (absdd2, *absdd2_fpr): Merge.
28299 (negtd2, *negtd2_fpr): Merge.
28300 (abstd2, *abstd2_fpr): Merge.
28301 * config/rs6000/e500.h: Delete file.
28302 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
28303 TARGET_FPRS by 1 and simplify.
28304 * config/rs6000/rs6000-c.c: Ditto.
28305 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
28306 TARGET_DF_SPE by 0.
28307 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
28309 * config/rs6000/rs6000.md: Ditto.
28310 (floatdidf2, *floatdidf2_fpr): Merge.
28311 (move_from_CR_gt_bit): Delete.
28312 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
28313 (E500_CR_IOR_COMPARE): Delete.
28314 (All patterns that require !TARGET_FPRS): Delete.
28315 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
28317 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28319 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
28321 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28323 * graphds.c (add_edge): Intitialize edge's attached data.
28324 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
28325 pointer parameter. Call pointed function on each edge during
28326 graph traversing. Skip traversing the edge when the function
28328 (graphds_dfs, graphds_scc): Ditto.
28329 (for_each_edge): New parameter. Pass the new parameter to callback
28331 * graphds.h (skip_edge_callback): New function pointer type.
28332 (graphds_dfs, graphds_scc): New function pointer parameter.
28333 (graphds_edge_callback, for_each_edge): New parameter.
28335 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28337 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
28338 out code checking if runtime alias check is possible to below ...
28339 Call the new function.
28340 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
28341 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
28343 2017-06-07 Marek Polacek <polacek@redhat.com>
28346 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
28347 TYPE_OVERFLOW_WRAPS check.
28349 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28351 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
28352 if versioning is required.
28353 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
28354 peeling with the check for versioning.
28356 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28358 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
28359 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
28360 Set true to new parameter if new ssa variable is defined.
28361 (vect_gen_vector_loop_niters): Refactor. Set range information
28362 for the new vector loop bound variable.
28363 (vect_do_peeling): Ditto.
28365 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28367 * tree-affine.c (ssa.h): Include header file.
28368 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
28369 has wrapping overflow behavior.
28371 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28373 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
28375 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28377 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
28378 (tree_to_aff_combination): ... here.
28380 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28382 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
28383 reg_pressure model function.
28384 (ivopts_global_cost_for_size): Delete.
28385 (determine_set_costs, iv_ca_recount_cost): Call new model function
28386 ivopts_estimate_reg_pressure.
28388 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28390 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
28391 expensive than udiv. Remove floating point cases from mod.
28393 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28395 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
28396 Increase idiv cost.
28398 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28400 * config/aarch64/aarch64.md
28401 (copysignsf3): Fix mask generation.
28403 2017-06-07 Jakub Jelinek <jakub@redhat.com>
28405 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
28407 (class dump_manager): Add register_dumps method.
28408 * dumpfile.c: Include langhooks.h.
28409 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
28410 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
28411 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
28412 (dump_manager::dump_register): Start with 512 entries instead of 32.
28413 (dump_manager::register_dumps): New method.
28414 * toplev.c (general_init): Instead of invoking register_dumps
28415 langhook, invoke register_dumps method on the dump manager.
28416 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
28419 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
28421 * doc/md.texi: Clarify the restrictions on a define_insn condition.
28422 Say that # requires an associated define_split to exist, and that
28423 the define_split must be suitable for use after register allocation.
28425 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28427 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
28428 (compute_outgoing_frequencies): Also initialize zero counts.
28429 (find_many_sub_basic_blocks): Do not produce uninitialized profile
28430 around loops; preserve more of profile when nothing changes.
28432 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
28434 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
28436 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
28437 * config/arm/arm-cpu-cdata.h: Regenerate.
28438 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
28439 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
28440 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
28441 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
28442 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
28444 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
28446 * config/arm/t-rmprofile: Likewise.
28447 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
28449 2017-06-06 David S. Miller <davem@davemloft.net>
28452 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
28453 blockage if function uses alloca.
28455 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28457 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
28458 New "uid" fields to hold pretty-print IDs of group and ref.
28459 Memory references are now identified as <group_id>:<ref_id>
28460 instead of using [random] addresses.
28461 (dump_mem_details): Simplify, no functional change.
28462 (dump_mem_ref): Simplify and make output more concise.
28463 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
28464 (find_or_create_group): Initialize group uid.
28465 (record_ref): Initialize ref uid. Improve debug output.
28466 (prune_group_by_reuse, should_issue_prefetch_p,)
28467 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
28468 (mark_nontemporal_store, determine_loop_nest_reuse):
28469 Improve debug output.
28471 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28473 * dbgcnt.def (prefetch): New debug counter.
28474 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
28475 (schedule_prefetches): Stop issueing prefetches if debug counter
28478 2017-06-06 Tom de Vries <tom@codesourcery.com>
28480 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
28481 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
28483 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28485 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
28486 Use aarch64_reg_or_zero predicate for operand 4.
28487 (aarch64_compare_and_swap<mode> define_insn_and_split):
28488 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
28489 (aarch64_store_exclusive<mode>): Likewise for operand 2.
28491 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
28493 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
28494 (arm_compute_save_core_reg_mask): This.
28495 (thumb1_compute_save_reg_mask): Rename into ...
28496 (thumb1_compute_save_core_reg_mask): This.
28497 (arm_compute_save_reg0_reg12_mask): Adapt comment.
28498 (arm_compute_frame_layout): Likewise.
28500 2017-06-06 Richard Biener <rguenther@suse.de>
28502 PR tree-optimization/80974
28503 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
28504 keep or clear leaders SSA info.
28506 2017-06-06 Tom de Vries <tom@codesourcery.com>
28508 * config/nvptx/nvptx.c (split_mode_p): New function.
28509 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
28511 2017-06-06 Tom de Vries <tom@codesourcery.com>
28513 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
28515 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28518 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
28521 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28523 * shrink-wrap.c (handle_simple_exit): Update profile.
28524 (try_shrink_wrapping): Upate profile.
28526 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28528 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
28529 (tree_guess_outgoing_edge_probabilities): New.
28530 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
28531 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
28533 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28535 * ipa-split.c (split_function): Initialize return bb profile.
28537 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28539 * profile.c (compute_branch_probabilities): Also initialize
28540 EXIT_BLOCK profile.
28542 2017-06-06 Richard Biener <rguenther@suse.de>
28544 PR tree-optimization/80928
28545 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
28546 (vect_analyze_loop_operations): Properly guard analysis for
28548 (vect_transform_loop): Likewise.
28549 (vect_analyze_loop_2): Also reset SLP type on PHIs.
28550 (vect_model_induction_cost): Do not cost for pure SLP.
28551 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
28552 of induction in inner loop vectorization.
28553 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
28554 (vect_get_and_check_slp_defs): Handle vect_induction_def.
28555 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
28557 (vect_analyze_slp_cost_1): Cost induction.
28558 (vect_detect_hybrid_slp_stmts): Handle PHIs.
28559 (vect_get_slp_vect_defs): Likewise.
28560 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
28561 (vect_transform_stmt): Handle SLP reductions.
28562 * tree-vectorizer.h (vectorizable_induction): Adjust.
28564 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28566 * config/rs6000/rs6000.c (make_resolver_func): Update
28567 init_lowered_empty_function call.
28569 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28571 * doc/invoke.texi: Document the -fprofile-abs-path option.
28572 * common.opt (fprofile-abs-path): New option.
28573 * gcov-io.h (gcov_write_filename): Declare.
28574 * gcov-io.c (gcov_write_filename): New function.
28575 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28576 * profile.c (output_location): Likewise.
28578 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28580 * shring-wrap.c: Revert accidental commit.
28582 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28584 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28586 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28588 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28590 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28591 profile in callgraph edge.
28592 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28593 (apply_scale): Likewise.
28594 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28595 Also scale profile when inlining function with zero profile.
28596 (initialize_cfun): Update exit block profile even when it is zero.
28597 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28598 when profile is read.
28600 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28602 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28603 (CLONE_*): New constants to define the processors we can generate
28604 code for with the target_clone attribute.
28605 (rs6000_clone_map): New array to identify which clone processors
28606 the current program is running on.
28607 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28608 target_clone attribute.
28609 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28610 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28611 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28612 (cpu_expand_builtin): Add support for target_clone attribute.
28613 (rs6000_valid_attribute_p): Allow "default" attribute.
28614 (get_decl_name): New debug function to simplify printing the
28615 current function name in debugging statements.
28616 (rs6000_clone_priority): New functions to support the target_clone
28617 attribute, and be able to generate code to switch between ISA 2.05
28618 through ISA 3.0 (power6 through power9).
28619 (rs6000_compare_version_priority): Likewise.
28620 (rs6000_get_function_versions_dispatcher): Likewise.
28621 (make_resolver_func): Likewise.
28622 (add_condition_to_bb): Likewise.
28623 (dispatch_function_versions): Likewise.
28624 (rs6000_generate_version_dispatcher_body): Likewise.
28625 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28626 (fusion_gpr_load_p): Fix a spacing issue.
28627 * doc/extend.texi (Common Function Attributes): Document that the
28628 PowerPC supports the target_clone attribute.
28630 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28632 * config/arm/arm.h: explain F symbol found in description of ARM
28633 register allocation in its legend.
28635 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28637 * config/mips/frame-header-opt.c: Include profile-count.h.
28638 * config/riscv/riscv.c: Include profile-count.h
28640 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28642 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28644 (sm_set_flag_if_changed): Add bbs field.
28645 (execute_sm_if_changed_flag_set): Pass BBS.
28646 (execute_sm): Update.
28648 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28650 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28653 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28655 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28656 (peephole2): New peephole2 to emit the above.
28657 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28659 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28661 * config/aarch64/aarch64.c (define_peephole2 above
28662 *sub_<shift>_<mode>): New peephole.
28664 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28666 * config/i386/i386.c (make_resolver_func): Update.
28667 * Makefile.in: Add profile-count.h and profile-count.o
28668 * auto-profile.c (afdo_indirect_call): Update to new API.
28669 (afdo_set_bb_count): Update.
28670 (afdo_propagate_edge): Update.
28671 (afdo_propagate_circuit): Update.
28672 (afdo_calculate_branch_prob): Update.
28673 (afdo_annotate_cfg): Update.
28674 * basic-block.h: Include profile-count.h
28675 (struct edge_def): Turn count to profile_count.
28676 (struct basic_block_def): Likewie.
28677 (REG_BR_PROB_BASE): Move to profile-count.h
28678 (RDIV): Move to profile-count.h
28679 * bb-reorder.c (max_entry_count): Turn to profile_count.
28680 (find_traces): Update.
28681 (rotate_loop):Update.
28682 (connect_traces):Update.
28683 (sanitize_hot_paths):Update.
28684 * bt-load.c (migrate_btr_defs): Update.
28685 * cfg.c (RDIV): Remove.
28686 (init_flow): Use alloc_block.
28687 (alloc_block): Uninitialize count.
28688 (unchecked_make_edge): Uninitialize count.
28689 (check_bb_profile): Update.
28690 (dump_edge_info): Update.
28691 (dump_bb_info): Update.
28692 (update_bb_profile_for_threading): Update.
28693 (scale_bbs_frequencies_int): Update.
28694 (scale_bbs_frequencies_gcov_type): Update.
28695 (scale_bbs_frequencies_profile_count): New.
28696 * cfg.h (update_bb_profile_for_threading): Update.
28697 (scale_bbs_frequencies_profile_count): Declare.
28698 * cfgbuild.c (compute_outgoing_frequencies): Update.
28699 (find_many_sub_basic_blocks): Update.
28700 * cfgcleanup.c (try_forward_edges): Update.
28701 (try_crossjump_to_edge): Update.
28702 * cfgexpand.c (expand_gimple_tailcall): Update.
28703 (construct_exit_block): Update.
28704 * cfghooks.c (verify_flow_info): Update.
28705 (dump_bb_for_graph): Update.
28706 (split_edge): Update.
28707 (make_forwarder_block): Update.
28708 (duplicate_block): Update.
28709 (account_profile_record): Update.
28710 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28711 (get_estimated_loop_iterations): Update.
28712 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28713 (single_likely_exit): Update.
28714 * cfgloopmanip.c (scale_loop_profile): Update.
28716 (set_zero_probability): Update.
28717 (lv_adjust_loop_entry_edge): Update.
28718 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28719 (purge_dead_edges): Update.
28720 (rtl_account_profile_record): Update.
28721 * cgraph.c (cgraph_node::create): Uninitialize count.
28722 (symbol_table::create_edge): Uninitialize count.
28723 (cgraph_update_edges_for_call_stmt_node): Update.
28724 (cgraph_edge::dump_edge_flags): Update.
28725 (cgraph_node::dump): Update.
28726 (cgraph_edge::maybe_hot_p): Update.
28727 * cgraph.h: Include profile-count.h
28728 (create_clone), create_edge, create_indirect_edge): Update.
28729 (cgraph_node): Turn count to profile_count.
28730 (cgraph_edge0: Likewise.
28731 (make_speculative, clone): Update.
28732 (create_edge): Update.
28733 (init_lowered_empty_function): Update.
28734 * cgraphclones.c (cgraph_edge::clone): Update.
28735 (duplicate_thunk_for_node): Update.
28736 (cgraph_node::create_clone): Update.
28737 * cgraphunit.c (cgraph_node::analyze): Update.
28738 (cgraph_node::expand_thunk): Update.
28739 * final.c (dump_basic_block_info): Update.
28740 * gimple-streamer-in.c (input_bb): Update.
28741 * gimple-streamer-out.c (output_bb): Update.
28742 * graphite.c (print_global_statistics): Update.
28743 (print_graphite_scop_statistics): Update.
28744 * hsa-brig.c: Include basic-block.h.
28745 * hsa-dump.c: Include basic-block.h.
28746 * hsa-gen.c (T sum_slice): Update.
28747 (convert_switch_statements):Update.
28748 * hsa-regalloc.c: Include basic-block.h.
28749 * ipa-chkp.c (chkp_produce_thunks): Update.
28750 * ipa-cp.c (struct caller_statistics): Update.
28751 (init_caller_stats): Update.
28752 (gather_caller_stats): Update.
28753 (ipcp_cloning_candidate_p): Update.
28754 (good_cloning_opportunity_p): Update.
28755 (get_info_about_necessary_edges): Update.
28756 (dump_profile_updates): Update.
28757 (update_profiling_info): Update.
28758 (update_specialized_profile): Update.
28759 (perhaps_add_new_callers): Update.
28760 (decide_about_value): Update.
28761 (ipa_cp_c_finalize): Update.
28762 * ipa-devirt.c (struct odr_type_warn_count): Update.
28763 (struct decl_warn_count): Update.
28764 (struct final_warning_record): Update.
28765 (possible_polymorphic_call_targets): Update.
28766 (ipa_devirt): Update.
28767 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28768 * ipa-icf.c (sem_function::merge): Update.
28769 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28770 * ipa-inline.c (compute_uninlined_call_time): Update.
28771 (compute_inlined_call_time): Update.
28772 (want_inline_small_function_p): Update.
28773 (want_inline_self_recursive_call_p): Update.
28774 (edge_badness): Update.
28775 (lookup_recursive_calls): Update.
28776 (recursive_inlining): Update.
28777 (inline_small_functions): Update.
28778 (dump_overall_stats): Update.
28779 (dump_inline_stats): Update.
28780 * ipa-profile.c (ipa_profile_generate_summary): Update.
28781 (ipa_propagate_frequency): Update.
28782 (ipa_profile): Update.
28783 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28784 * ipa-utils.c (ipa_merge_profiles): Update.
28785 * loop-doloop.c (doloop_modify): Update.
28786 * loop-unroll.c (report_unroll): Update.
28787 (unroll_loop_runtime_iterations): Update.
28788 * lto-cgraph.c (lto_output_edge): Update.
28789 (lto_output_node): Update.
28790 (input_node): Update.
28791 (input_edge): Update.
28792 (merge_profile_summaries): Update.
28793 * lto-streamer-in.c (input_cfg): Update.
28794 * lto-streamer-out.c (output_cfg): Update.
28795 * mcf.c (create_fixup_graph): Update.
28796 (adjust_cfg_counts): Update.
28797 (sum_edge_counts): Update.
28798 * modulo-sched.c (sms_schedule): Update.
28799 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28800 * predict.c (maybe_hot_count_p): Update.
28801 (probably_never_executed): Update.
28802 (dump_prediction): Update.
28803 (combine_predictions_for_bb): Update.
28804 (propagate_freq): Update.
28805 (handle_missing_profiles): Update.
28806 (counts_to_freqs): Update.
28807 (rebuild_frequencies): Update.
28808 (force_edge_cold): Update.
28809 * predict.h: Include profile-count.h
28810 (maybe_hot_count_p, counts_to_freqs): UPdate.
28811 * print-rtl-function.c: Do not include cfg.h
28812 * print-rtl.c: Include basic-block.h
28813 * profile-count.c: New file.
28814 * profile-count.h: New file.
28815 * profile.c (is_edge_inconsistent): Update.
28816 (correct_negative_edge_counts): Update.
28817 (is_inconsistent): Update.
28818 (set_bb_counts): Update.
28819 (read_profile_edge_counts): Update.
28820 (compute_frequency_overlap): Update.
28821 (compute_branch_probabilities): Update; Initialize and deinitialize
28823 (branch_prob): Update.
28824 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
28825 (edge_gcov_count): New.
28826 (bb_gcov_count): New.
28827 * shrink-wrap.c (try_shrink_wrapping): Update.
28828 * tracer.c (better_p): Update.
28829 * trans-mem.c (expand_transaction): Update.
28830 (ipa_tm_insert_irr_call): Update.
28831 (ipa_tm_insert_gettmclone_call): Update.
28832 * tree-call-cdce.c: Update.
28833 * tree-cfg.c (gimple_duplicate_sese_region): Update.
28834 (gimple_duplicate_sese_tail): Update.
28835 (gimple_account_profile_record): Update.
28836 (execute_fixup_cfg): Update.
28837 * tree-inline.c (copy_bb): Update.
28838 (copy_edges_for_bb): Update.
28839 (initialize_cfun): Update.
28840 (freqs_to_counts): Update.
28841 (copy_cfg_body): Update.
28842 (expand_call_inline): Update.
28843 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
28844 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
28845 (try_unroll_loop_completely): Update.
28846 (try_peel_loop): Update.
28847 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
28848 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
28849 * tree-ssa-loop-split.c (connect_loops): Update.
28850 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
28851 * tree-ssa-reassoc.c (branch_fixup): Update.
28852 * tree-ssa-tail-merge.c (replace_block_by): Update.
28853 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
28854 (compute_path_counts): Update.
28855 (update_profile): Update.
28856 (recompute_probabilities): Update.
28857 (update_joiner_offpath_counts): Update.
28858 (estimated_freqs_path): Update.
28859 (freqs_to_counts_path): Update.
28860 (clear_counts_path): Update.
28861 (ssa_fix_duplicate_block_edges): Update.
28862 (duplicate_thread_path): Update.
28863 * tree-switch-conversion.c (case_bit_test_cmp): Update.
28864 (struct switch_conv_info): Update.
28865 * tree-tailcall.c (decrease_profile): Update.
28866 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
28867 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
28868 * value-prof.c (check_counter): Update.
28869 (gimple_divmod_fixed_value): Update.
28870 (gimple_mod_pow2): Update.
28871 (gimple_mod_subtract): Update.
28872 (gimple_ic_transform): Update.
28873 (gimple_stringop_fixed_value): Update.
28874 * value-prof.h (gimple_ic): Update.
28876 2017-06-02 Carl Love <cel@us.ibm.com>
28878 * config/rs6000/rs6000-c: Add support for built-in functions
28879 vector double vec_doublee (vector signed int);
28880 vector double vec_doublee (vector unsigned int);
28881 vector double vec_doublee (vector float);
28882 vector double vec_doubleh (vector signed int);
28883 vector double vec_doubleh (vector unsigned int);
28884 vector double vec_doubleh (vector float);
28885 vector double vec_doublel (vector signed int);
28886 vector double vec_doublel (vector unsigned int);
28887 vector double vec_doublel (vector float);
28888 vector double vec_doubleo (vector signed int);
28889 vector double vec_doubleo (vector unsigned int);
28890 vector double vec_doubleo (vector float);.
28891 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
28892 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
28894 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
28895 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
28896 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
28898 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
28899 vec_doublel, vec_doubleh.
28900 * doc/extend.texi: Update the built-in documentation file for the
28901 new built-in functions.
28903 2017-06-02 David Malcolm <dmalcolm@redhat.com>
28906 * ipa-inline-analysis.c (free_growth_caches): Set
28907 edge_removal_hook_holder to NULL after removing it.
28909 2017-06-02 Sudakshina Das <sudi.das@arm.com>
28911 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
28912 comparision with zero.
28914 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28915 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28916 for early expansion of vec_min and vec_max builtins.
28917 (builtin_function_type): Add min/max unsigned variants to those
28918 identified as having unsigned arguments.
28920 2017-06-02 Olivier Hainque <hainque@adacore.com>
28922 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
28924 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28926 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
28927 Use VALL_F16 iterator rather than VALL.
28929 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28931 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
28932 Emit CBNZ inside loop when doing a strong exchange and comparing
28933 against zero. Generate the CC flags after the loop.
28935 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
28937 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
28938 (dl_section_ref): New.
28939 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
28940 On AIX, append an expression to subtract the size of the
28941 section length to dl_section_ref.
28943 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28945 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28946 for early expansion of vector absolute builtins.
28948 2017-06-02 Richard Biener <rguenther@suse.de>
28950 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
28951 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
28953 2017-06-02 Richard Biener <rguenther@suse.de>
28955 PR tree-optimization/80948
28956 * tree-tailcall.c (find_tail_calls): Track stmts to move in
28957 stmt order as well.
28959 2017-06-02 Richard Biener <rguenther@suse.de>
28961 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
28963 * tree-vect-stmts.c (process_use): Do not mark backedge defs
28964 for inductions as relevant.
28966 2017-06-02 Richard Biener <rguenther@suse.de>
28968 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
28969 (vectorizable_induction): ... this. Remove dead code.
28971 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
28973 * builtins. (expand_builtin_alloca): Remove second parameter and
28974 infer its value from the first parameter instead.
28975 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
28977 2017-06-02 Jakub Jelinek <jakub@redhat.com>
28979 PR rtl-optimization/80903
28980 * loop-doloop.c (add_test): Unshare sequence.
28982 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
28984 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
28986 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
28988 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
28990 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
28991 xlogue_layout::get_instance, logue_layout::xlogue_layout,
28992 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
28993 (xlogue_layout::get_stub_rtx): Make static.
28994 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
28995 (xlogue_layout::compute_stub_managed_regs): Rename to...
28996 (xlogue_layout::count_stub_managed_regs): ...this.
28997 (xlogue_layout::is_stub_managed_reg): New function.
28998 (xlogue_layout::m_stub_names): Rename to...
28999 (xlogue_layout::s_stub_names): ...this, make static.
29000 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29001 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29002 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29003 xlogue_layout::s_stub_names): Instantiate statics.
29004 (stub_managed_regs): Remove.
29005 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29006 (disable_call_ms2sysv_xlogues): Rename to...
29007 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29008 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29010 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29011 change after reload_completed.
29012 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29014 (ix86_expand_prologue): Likewise.
29015 (ix86_expand_epilogue): Likewise.
29016 (ix86_expand_split_stack_prologue): Likewise.
29017 (ix86_compute_frame_layout): Remove frame parameter ...
29018 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29019 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29021 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29022 (ix86_frame): Move from here ...
29023 * config/i386/i386.h (ix86_frame): ... to here.
29024 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29025 complete ix86_frame data structure instead. Remove some_ld_name.
29027 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29029 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29030 symbols that hold a DECL_VALUE_EXPR.
29032 2017-06-01 Martin Jambor <mjambor@suse.cz>
29034 PR tree-optimization/80898
29035 * tree-sra.c (process_subtree_disqualification): Removed.
29036 (disqualify_candidate): Do not acll
29037 process_subtree_disqualification.
29038 (subtree_mark_written_and_enqueue): New function.
29039 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29040 RHS has been disqualified and re-queue LHS if necessary. Apart
29041 from that, ignore disqualified RHS.
29043 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29045 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29046 address fetch for z10 or later.
29048 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29050 * config/arc/arc.md (tst_movb): Add guard when splitting.
29052 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29054 * config/arc/arc.c (arc_can_eliminate): Test against
29055 arc_frame_pointer_needed.
29057 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29059 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
29060 to prevent store reordering.
29061 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
29062 (type): Add block type.
29063 (stack_tie): Define special instruction to be used in
29066 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29068 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
29069 constraint. It is not valid for the pattern.
29070 (noncommutative_binary_comparison): Likewise.
29072 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29074 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
29077 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29079 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
29080 be used by the reg-alloc.
29082 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29084 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
29085 reg-alloc when having mul64 or mul32x16 instructions.
29086 (mulsidi3): Likewise.
29087 (umulsidi3): Likewise.
29088 (mulsi32x16): New pattern.
29089 (mulsi64): Likewise.
29090 (mulsidi64): Likewise.
29091 (umulsidi64): Likewise.
29092 (MUL32x16_REG): Define.
29093 (mul64_600): Use MUL32x16_REG.
29094 (mac64_600): Likewise.
29095 (umul64_600): Likewise.
29096 (umac64_600): Likewise.
29098 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29100 * config/arc/arc.md (mulsi3_700): Make it commutative.
29102 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
29104 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
29106 (*sign_extendsidi2_insn): Likewise for movstosw.
29108 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29110 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
29111 the type of the input discriminant value. Convert the
29112 discriminant value of signedness vary.
29114 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
29116 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
29117 Add to -Wall section.
29119 2017-06-01 Richard Biener <rguenther@suse.de>
29121 PR middle-end/66313
29122 * fold-const.c (fold_plusminus_mult_expr): If the factored
29123 factor may be zero use a wrapping type for the inner operation.
29124 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
29125 and handle moved defs.
29126 (process_assignment): Properly guard the unary op case. Return a
29127 tri-state indicating that moving the stmt before the call may allow
29128 to continue. Pass through to_move.
29129 (find_tail_calls): Handle moving unrelated defs before
29132 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
29135 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
29136 splitter result in the canonical way.
29138 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
29140 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
29141 also for 32bit target. Update insn attributes.
29142 (zero-extendsidi2 splitter): Allow all registers for operand 1.
29144 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
29146 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
29147 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
29148 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
29149 (_mm_maskz_min_ss): New intrinsics.
29151 2017-05-31 Martin Liska <mliska@suse.cz>
29153 * tree-vect-loop.c (vect_create_epilog_for_reduction):
29154 Change comment style to one we normally use.
29155 (vectorizable_reduction): Likewise.
29156 (vectorizable_induction): Likewise.
29157 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
29158 (vectorizable_call): Likewise.
29159 (vectorizable_simd_clone_call): Likewise.
29160 (vectorizable_conversion): Likewise.
29161 (vectorizable_assignment): Likewise.
29162 (vectorizable_shift): Likewise.
29163 (vectorizable_operation): Likewise.
29164 (vectorizable_store): Likewise.
29165 (vectorizable_load): Likewise.
29166 * tree-vectorizer.h: Likewise.
29168 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
29170 * passes.c (emergency_dump_function): New.
29171 * tree-pass.h (emergency_dump_function): Declare.
29172 * plugin.c (plugins_internal_error_function): Remove.
29173 * plugin.h (plugins_internal_error_function): Remove declaration.
29174 * toplev.c (internal_error_function): New static function. Use it...
29175 (general_init): ...here.
29177 2017-05-31 Graham Markall <graham.markall@embecosm.com>
29179 * config/arc/arc.c (arc_print_operand): Handle constant operands.
29180 (arc_rtx_costs): Add costs for new patterns.
29181 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
29182 * config/arc/predicates.md: Add _1_2_3_operand predicate.
29184 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29186 * tree-ssa-strlen.c (get_next_strinfo): New function.
29187 (get_stridx_plus_constant): Use it.
29188 (zero_length_string): Likewise.
29189 (adjust_related_strinfos): Likewise.
29190 (adjust_last_stmt): Likewise.
29192 2017-05-31 Richard Biener <rguenther@suse.de>
29195 * config/i386/i386.c (ix86_expand_builtin): Remove assert
29196 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
29198 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29200 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
29201 loop_vinfo argument and use of dependence distance vectors.
29202 Check instead whether the two references differ only in their
29203 initial value and assume that they have the same alignment if the
29204 difference is a multiple of the vector alignment.
29205 (vect_analyze_data_refs_alignment): Update call accordingly.
29207 2017-05-31 Martin Liska <mliska@suse.cz>
29210 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
29212 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29214 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
29215 (create_intersect_range_checks): Move from ...
29216 * tree-data-ref.c (create_intersect_range_checks_index)
29217 (create_intersect_range_checks): ... to here.
29218 (create_runtime_alias_checks): New function factored from ...
29219 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
29220 here. Call above function.
29221 * tree-data-ref.h (create_runtime_alias_checks): New function.
29223 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29225 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
29226 segment length for dr_b and compute it in wide_int.
29228 2017-05-31 Richard Biener <rguenther@suse.de>
29230 PR tree-optimization/80906
29231 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
29232 and pass through iv_map.
29233 (copy_bb_and_scalar_dependences): Adjust.
29234 (translate_pending_phi_nodes): Likewise.
29235 (copy_loop_close_phi_args): Handle code-generating IVs instead
29238 2017-05-30 David Malcolm <dmalcolm@redhat.com>
29240 * diagnostic-color.c (color_dict): Add "type-diff".
29241 (parse_gcc_colors): Update comment.
29242 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
29243 -fdiagnostics-show-template-tree and -fno-elide-type.
29244 (GCC_COLORS): Add type-diff to example.
29246 (-fdiagnostics-show-template-tree): New.
29247 (-fno-elide-type): New.
29248 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
29249 the pp_format_decoder callback. Call any m_format_postprocessor's
29251 (pretty_printer::pretty_printer): Initialize
29252 m_format_postprocessor.
29253 (pretty_printer::~pretty_printer): Delete any
29254 m_format_postprocessor.
29255 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
29256 (class format_postprocessor): New class.
29257 (struct pretty_printer::format_decoder): Document the new parameters.
29258 (struct pretty_printer::m_format_postprocessor): New field.
29259 * tree-diagnostic.c (default_tree_printer): Update for new
29260 bool and const char ** params.
29261 * tree-diagnostic.h (default_tree_printer): Likewise.
29263 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
29265 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
29266 (lwa_operand): Delete rs6000_gen_cell_microcode test.
29267 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29268 rs6000_gen_cell_microcode code.
29269 (rs6000_final_prescan_insn): Delete.
29270 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
29271 "warn-cell-microcode" entries.
29272 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
29273 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
29274 throughout. Change cc_reg_not_micro_cr0_operand to
29275 cc_reg_not_cr0_operand throughout.
29276 (*extendhi<mode>2_noload): Delete.
29277 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
29278 (mwarn-cell-microcode): Delete.
29279 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
29280 -mgen-cell-microcode and -mwarn-cell-microcode.
29282 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
29285 * config/i386/constraints.md (Yd): New constraint.
29287 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
29288 and (?Yd, r) alternatives. Update insn attributes.
29289 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
29290 and (?*Yd, r) alternatives. Update insn attributes.
29291 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
29293 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
29295 * gimplify.c (gimplify_modify_expr): Don't create a
29296 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
29299 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
29301 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
29303 2017-05-30 Richard Biener <rguenther@suse.de>
29305 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
29306 and reduc_def fields.
29307 (STMT_VINFO_REDUC_TYPE): New define.
29308 (STMT_VINFO_REDUC_DEF): Likewise.
29309 (vect_force_simple_reduction): Adjust prototype.
29310 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
29311 (vect_is_simple_reduction): Remove check_reduction argument.
29312 (vect_force_simple_reduction): Adjust and set
29313 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29314 (vectorizable_reduction): Do not re-do reduction analysis
29315 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29316 * tree-parloops.c (gather_scalar_reductions): Adjust.
29318 2017-05-30 Richard Biener <rguenther@suse.de>
29320 PR middle-end/80901
29321 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
29324 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
29326 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
29327 Introduce unknown_misalignment parameter and remove vf.
29328 (vect_peeling_hash_get_lowest_cost):
29329 Pass unknown_misalignment parameter.
29330 (vect_enhance_data_refs_alignment):
29331 Fix unsupportable data ref treatment.
29333 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29335 * tree-vect-data-refs.c (vect_get_data_access_cost):
29336 Workaround for SLP handling.
29337 (vect_enhance_data_refs_alignment):
29338 Compute costs for doing no peeling at all, compare to the best
29339 peeling costs so far and avoid peeling if cheaper.
29341 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29343 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
29344 Return peeling info and set costs to zero for unlimited cost
29346 (vect_enhance_data_refs_alignment): Also inspect all datarefs
29347 with unknown misalignment. Compute and costs for unknown
29348 misalignment, compare them to the costs for known misalignment
29349 and choose the cheapest for peeling.
29351 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29353 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
29354 (vect_get_peeling_costs_all_drs): Create function.
29355 (vect_peeling_hash_get_lowest_cost):
29356 Use vect_get_peeling_costs_all_drs.
29357 (vect_peeling_supportable): Create function.
29358 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
29360 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29362 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
29363 DR_HAS_NEGATIVE_STEP.
29364 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
29365 (vect_enhance_data_refs_alignment): Use.
29366 (vect_duplicate_ssa_name_ptr_info): Use.
29367 * tree-vectorizer.h (dr_misalignment): Use.
29368 (known_alignment_for_access_p): Use.
29370 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
29373 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
29375 (has_section_name): New function.
29377 2017-05-30 Martin Liska <mliska@suse.cz>
29380 * auto-profile.c (get_function_decl_from_block): Fix
29383 2017-05-30 Richard Biener <rguenther@suse.de>
29385 PR middle-end/80876
29386 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
29388 2017-05-30 Martin Liska <mliska@suse.cz>
29390 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
29391 * dumpfile.h (struct dump_file_info): Remove ctors.
29393 2017-05-30 Martin Liska <mliska@suse.cz>
29395 * predict.def: Fix GNU coding style.
29397 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29399 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29400 Mark 'to' argument with ATTRIBUTE_UNUSED.
29402 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29404 * config/xtensa/xtensa.c (xtensa_emit_call): Use
29405 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
29406 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
29409 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
29411 * doc/install.texi (Options specification): Restore entry of
29412 --enable-sjlj-exceptions.
29414 2017-05-27 Michael Eager <eager@eagercon.com>
29417 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
29419 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
29421 * config/microblaze/microblaze.h
29422 (FIXED_REGISTERS): Update in macro.
29423 (CALL_USED_REGISTERS): Update in macro.
29425 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
29427 * doc/install.texi: Add links to macOS binary distributions.
29429 2017-05-27 Jakub Jelinek <jakub@redhat.com>
29433 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29435 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29437 2017-05-26 Martin Liska <mliska@suse.cz>
29439 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
29441 2017-05-26 Martin Liska <mliska@suse.cz>
29443 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
29444 always leading ';; '.
29445 (dump_bb_info): Likewise.
29446 (brief_dump_cfg): Likewise.
29447 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
29448 * dumpfile.c: Remove usage of TDF_VERBOSE.
29449 * dumpfile.h (enum dump_kind): Likewise.
29450 (dump_gimple_bb_header): Do not use TDF_COMMENT.
29451 * print-tree.c (debug_verbose): Remove.
29452 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
29453 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
29454 * tree-diagnostic.c (default_tree_printer): Replace
29455 TDF_DIAGNOSTIC with TDF_SLIM.
29457 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29459 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
29460 in parameter loop, rather than loop_vinfo.
29461 (create_intersect_range_checks): Ditto.
29462 (vect_create_cond_for_alias_checks): Update call to above functions.
29464 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29466 PR tree-optimization/80815
29467 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
29468 for merging runtime alias checks. Handle negative DR_STEPs.
29470 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29472 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
29474 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
29475 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
29476 out code pruning runtime alias checks.
29477 * tree-data-ref.c (prune_runtime_alias_test_list): New function
29478 factored out from above.
29479 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29481 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29483 (prune_runtime_alias_test_list): New decalaration.
29485 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29487 * tree-vect-data-refs.c (compare_tree): Rename and move ...
29488 * tree-data-ref.c (data_ref_compare_tree): ... to here.
29489 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
29490 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
29491 (operator==, comp_dr_with_seg_len_pair): Ditto.
29492 (vect_prune_runtime_alias_test_list): Ditto.
29494 2017-05-26 Martin Liska <mliska@suse.cz>
29497 * params.def: Bound partial-inlining-entry-probability param.
29499 2017-05-26 Marek Polacek <polacek@redhat.com>
29502 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
29505 2017-05-26 Richard Biener <rguenther@suse.de>
29507 PR tree-optimization/80842
29508 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
29511 2017-05-26 Richard Biener <rguenther@suse.de>
29513 PR tree-optimization/80844
29514 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
29516 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
29518 * doc/md.texi (Machine Constraints): Update x86 family
29519 machine constraints section to match 'config/i386/constraints.md'.
29521 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
29523 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
29525 2017-05-25 Nathan Sidwell <nathan@acm.org>
29527 * doc/invoke.texi (--enable-languages): Update documentation.
29529 2017-05-25 Martin Liska <mliska@suse.cz>
29531 * dumpfile.c: Add TDF_FOLDING.
29532 * dumpfile.h (enum dump_kind): Likewise.
29533 * genmatch.c (dt_simplify::gen_1): Use it.
29535 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29537 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
29539 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29541 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29542 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
29544 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29546 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
29547 * match.pd (X == C): Rewrite it here.
29548 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
29549 with_certain_nonzero_bits2): New predicates.
29550 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
29552 2017-05-24 Nathan Sidwell <nathan@acm.org>
29554 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
29557 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
29560 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29562 * config/powerpcspe: New port. Files are copied from the rs6000
29563 port, with "rs6000" in filenames replaced by "powerpcspe".
29565 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29567 PR rtl-optimization/80754
29568 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29570 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29572 * config/sparc/sparc.md (length): Return the correct value for -mflat
29573 sibcalls to match output_sibcall.
29575 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29579 * config/rs6000/rs6000.c (struct machine_function): Add new field
29581 (rs6000_get_separate_components): Init that field, use it.
29582 (rs6000_components_for_bb): Use the field.
29584 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29586 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29588 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29590 PR middle-end/80823
29591 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29593 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29596 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29597 address against address_operand predicate.
29598 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29600 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29602 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29603 parameters passed indirectly.
29605 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29607 * config/i386/i386.md (*movdi_internal): Remove SSE4
29608 alternative 18 (?r, *v). Update insn attributes.
29609 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29610 Update insn attributes.
29611 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29612 Update insn attributes.
29613 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29614 alternative 1 (r, v). Remove isa attribute.
29615 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29616 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29617 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29619 2017-05-23 Tom de Vries <tom@codesourcery.com>
29621 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29624 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29626 * cgraphunit.c (symbol_table::process_new_functions): Update.
29627 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29628 (inline_generate_summary): Rename to ...
29629 (ipa_fn_summary_generate): ... this one.
29630 (inline_read_summary): Rename to ...
29631 (ipa_fn_summary_read): ... this one.
29632 (inline_write_summary): Rename to ...
29633 (ipa_fn_summary_write): ... this one.
29634 (inline_free_summary): Rename to ...
29635 (ipa_free_fn_summary): ... this one.
29636 (pass_data_local_fn_summary, pass_local_fn_summary,
29637 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29638 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29639 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29640 make_pass_ipa_fn_summary): New.
29641 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29642 inline_write_summary, inline_free_summary): Remove.
29643 (ipa_free_fn_summary) : New.
29644 * ipa-inline.c (ipa_inline): Update.
29645 (pass_ipa_inline): Do not generate summaries.
29646 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29648 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29649 and add pass_ipa_fn_summary.
29650 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29652 (make_pass_inline_parameters): Remove.
29654 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29656 * omp-low.c (struct omp_context): Remove "default_kind" member.
29659 * omp-offload.c (execute_oacc_device_lower): Remove the
29660 parallelism dimensions function attributes for unparallelized
29661 OpenACC kernels constructs.
29663 2017-05-23 Martin Liska <mliska@suse.cz>
29665 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29667 (cgraph_edge::make_speculative): Likewise.
29668 (cgraph_edge::resolve_speculation): Likewise.
29669 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29670 (cgraph_node::dump): Likewise.
29671 * cgraph.h: Likewise.
29672 * cgraphunit.c (analyze_functions): Likewise.
29673 (symbol_table::compile): Likewise.
29674 * ipa-cp.c (print_all_lattices): Likewise.
29675 (determine_versionability): Likewise.
29676 (initialize_node_lattices): Likewise.
29677 (ipcp_verify_propagated_values): Likewise.
29678 (estimate_local_effects): Likewise.
29679 (update_profiling_info): Likewise.
29680 (create_specialized_node): Likewise.
29681 (perhaps_add_new_callers): Likewise.
29682 (decide_about_value): Likewise.
29683 (decide_whether_version_node): Likewise.
29684 (identify_dead_nodes): Likewise.
29685 (ipcp_store_bits_results): Likewise.
29686 * ipa-devirt.c (dump_targets): Likewise.
29687 (ipa_devirt): Likewise.
29688 * ipa-icf.c (sem_item::dump): Likewise.
29689 (sem_function::equals): Likewise.
29690 (sem_variable::equals): Likewise.
29691 (sem_item_optimizer::read_section): Likewise.
29692 (sem_item_optimizer::execute): Likewise.
29693 (congruence_class::dump): Likewise.
29694 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29695 (dump_inline_summary): Likewise.
29696 (estimate_node_size_and_time): Likewise.
29697 (inline_analyze_function): Likewise.
29698 * ipa-inline-transform.c (inline_call): Likewise.
29699 * ipa-inline.c (report_inline_failed_reason): Likewise.
29700 (want_early_inline_function_p): Likewise.
29701 (edge_badness): Likewise.
29702 (update_edge_key): Likewise.
29703 (inline_small_functions): Likewise.
29704 * ipa-profile.c (ipa_profile): Likewise.
29705 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29706 (ipa_make_edge_direct_to_target): Likewise.
29707 (remove_described_reference): Likewise.
29708 (ipa_impossible_devirt_target): Likewise.
29709 (propagate_controlled_uses): Likewise.
29710 (ipa_print_node_params): Likewise.
29711 (ipcp_transform_function): Likewise.
29712 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29713 (propagate_pure_const): Likewise.
29714 * ipa-reference.c (generate_summary): Likewise.
29715 (read_write_all_from_decl): Likewise.
29716 (propagate): Likewise.
29717 (ipa_reference_read_optimization_summary): Likewise.
29718 * ipa-utils.c (ipa_merge_profiles): Likewise.
29719 * ipa.c (walk_polymorphic_call_targets): Likewise.
29720 (symbol_table::remove_unreachable_nodes): Likewise.
29721 (ipa_single_use): Likewise.
29722 * passes.c (execute_todo): Likewise.
29723 * predict.c (drop_profile): Likewise.
29724 * symtab.c (symtab_node::get_dump_name): New function.
29725 (symtab_node::dump_name): Likewise.
29726 (symtab_node::dump_asm_name): Likewise.
29727 (symtab_node::dump_references): Likewise.
29728 (symtab_node::dump_referring): Likewise.
29729 (symtab_node::dump_base): Likewise.
29730 (symtab_node::debug_symtab): Likewise.
29731 * tree-sra.c (convert_callers_for_node): Likewise.
29732 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29733 * value-prof.c (init_node_map): Likewise.
29735 2017-05-23 Martin Liska <mliska@suse.cz>
29737 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29738 and symtab_node::debug_symtab to symbol_table::debug.
29739 * cgraphunit.c (analyze_functions): Use the renamed function.
29740 (symbol_table::compile): Likewise.
29741 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29742 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29743 * passes.c (execute_todo): Likewise.
29744 * symtab.c (symbol_table::dump): New function.
29745 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29747 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29749 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29750 that nonconst implies exec.
29752 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29754 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29755 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29756 (inline_edge_summary_vec): Turn into ...
29757 (ipa_call_summaries): ... this one.
29758 (redirect_to_unreachable, edge_set_predicate,
29759 evaluate_properties_for_edge, inline_summary_alloc,
29760 reset_ipa_call_summary, reset_inline_summary,
29761 inline_summary_t::duplicate): Update.
29762 (inline_edge_duplication_hook): Turn to ...
29763 (ipa_call_summary_t::duplicate): ... this one.
29764 (inline_edge_removal_hook): Turn to ...
29765 (ipa_call_summary_t::remove): ... this one.
29766 (dump_inline_edge_summary): Turn to ...
29767 (dump_ipa_call_summary): ... this one.
29768 (estimate_function_body_sizes): Update.
29769 (inline_update_callee_summaries): Update.
29770 (remap_edge_change_prob): Update.
29771 (remap_edge_summaries): Update.
29772 (inline_merge_summary): Update.
29773 (do_estimate_edge_time): Update.
29774 (inline_generate_summary): Update.
29775 (inline_read_section): Update.
29776 (inline_read_summary): Update.
29777 (inline_free_summary): Update.
29778 * ipa-inline.c (can_inline_edge_p): Update.
29779 (compute_inlined_call_time): Update.
29780 (want_inline_small_function_p): Update.
29781 (edge_badness): Update.
29782 (early_inliner): Update.
29783 * ipa-inline.h (inline_edge_summary): Turn to ...
29784 (ipa_call_summary): ... this one.
29785 (ipa_call_summary_t): New class.
29786 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29787 (ipa_call_summaries): New.
29788 (inline_edge_summary): Remove.
29789 (estimate_edge_growth): Update.
29790 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29791 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29792 * ipa-split.c (execute_split_functions): Update.
29793 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29795 2017-05-23 Tom de Vries <tom@codesourcery.com>
29797 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29798 attributes): Document rdrand effective target.
29800 2017-05-23 Tom de Vries <tom@codesourcery.com>
29802 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29803 attributes): Sort alphabetically.
29805 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29807 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29809 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29812 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
29813 V2DF/V2DI splat into two separate patterns, one that handles
29814 registers, and the other that only handles memory. Drop support
29815 for splatting from a GPR on ISA 2.07 and then splitting the
29816 splat into direct move and splat.
29817 (vsx_splat_<mode>_reg): Likewise.
29818 (vsx_splat_<mode>_mem): Likewise.
29820 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
29822 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
29824 2017-05-22 Jakub Jelinek <jakub@redhat.com>
29826 PR middle-end/80809
29827 * omp-low.c (finish_taskreg_remap): New function.
29828 (finish_taskreg_scan): If unit size of ctx->record_type
29829 is non-constant, unshare the size expression and replace
29830 decls in it with possible outer var refs.
29832 PR middle-end/80809
29833 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
29834 GOVD_SHARED rather than GOVD_PRIVATE with it.
29835 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
29836 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
29838 PR middle-end/80853
29839 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
29840 as last argument to build_outer_var_ref for pointer bases of array
29841 section reductions.
29843 2017-05-19 Martin Sebor <msebor@redhat.com>
29845 * print-tree.c (print_node): Print DECL_READ_P flag.
29847 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29849 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
29850 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
29851 * cgraph.c: Likewise.
29852 * cgraphunit.c: Likewise.
29853 * gengtype.c: Likewise.
29854 * ipa-cp.c: Likewise.
29855 * ipa-devirt.c: Likewise.
29856 * ipa-icf.c: Likewise.
29857 * ipa-predicate.c: Likewise.
29858 * ipa-profile.c: Likewise.
29859 * ipa-prop.c: Likewise.
29860 * ipa-split.c: Likewise.
29862 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
29863 edge_predicate_pool, dump_inline_hints,
29864 inline_summary::account_size_time, redirect_to_unreachable,
29865 edge_set_predicate, set_hint_predicate,
29866 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
29867 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
29868 inline_summary_t::remove, remap_hint_predicate_after_duplication,
29869 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
29870 ipa_call_summary_t::remove, initialize_growth_caches,
29871 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
29872 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
29873 mark_modified, unmodified_parm_1, unmodified_parm,
29874 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
29875 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
29876 compute_bb_predicates, will_be_nonconstant_expr_predicate,
29877 will_be_nonconstant_predicate, record_modified_bb_info,
29878 get_minimal_bb, record_modified, param_change_prob,
29879 phi_result_unknown_predicate, predicate_for_phi_result,
29880 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
29881 estimate_function_body_sizes, compute_inline_parameters,
29882 compute_inline_parameters_for_curren, pass_data_inline_parameters,
29883 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
29884 inline_update_callee_summaries, remap_edge_change_prob,
29885 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
29886 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
29887 inline_analyze_function, inline_summary_t::insert,
29888 inline_generate_summary, read_ipa_call_summary, inline_read_section,
29889 inline_read_summary, write_ipa_call_summary, inline_write_summary,
29890 inline_free_summary): Move to ipa-fnsummary.h
29891 (predicate_t): Remove.
29892 * ipa-fnsummary.c: New file.
29893 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
29894 (enum inline_hints_vals, inline_hints, agg_position_info,
29895 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
29896 inline_summaries, ipa_call_summary, ipa_call_summary_t,
29897 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
29898 dump_inline_summary, dump_inline_hints, inline_generate_summary,
29899 inline_read_summary, inline_write_summary, inline_free_summary,
29900 inline_analyze_function, initialize_inline_failed,
29901 inline_merge_summary, inline_update_overall_summary,
29902 compute_inline_parameters): Move to ipa-fnsummary.h
29903 * ipa-fnsummary.h: New file.
29904 * ipa-inline-transform.h: Include ipa-inline.h.
29905 * ipa-inline.c: LIkewise.
29907 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29909 * ipa-inline.c (edge_badness): Use inlined_time instead of
29910 inline_summaries->get.
29912 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29914 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
29916 2017-05-22 Nathan Sidwell <nathan@acm.org>
29918 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
29919 (fdump-lang): Document 'raw' option.
29920 * dumpfile.h (TDI_tu): Delete.
29921 * dumpfile.c (dump_files): Remove translation-unit.
29922 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
29924 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
29926 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
29927 command option from $(AWK) call.
29928 * config/avr/genmultilib.awk: Simplify and rewrite so that it
29929 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
29930 [FORMAT]: Remove handling of variable.
29931 * config/avr/t-multilib: Regenerate.
29933 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29935 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
29937 (dump_inline_summary): Do not print self_time.
29938 (estimate_function_body_sizes): Do not set self_time.
29939 (compute_inline_parameters): Likewise.
29940 (inline_read_section, inline_write_summary): Do not stream self_time.
29941 * ipa-inline.h (inline_summary): Drop self_time.
29943 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29945 * ipa-inline-analysis.c (account_size_time): Rename to ...
29946 (inline_summary::account_size_time): ... this one.
29947 (reset_ipa_call_summary): Turn to ...
29948 (ipa_call_summary::reset): ... this one.
29949 (reset_inline_summary): Turn to ...
29950 (inline_summary::reset): ... this one.
29951 (inline_summary_t::remove): Update.
29952 (inline_summary_t::duplicate): Update.
29953 (ipa_call_summary_t::remove): Update.
29954 (dump_inline_summary): Update.
29955 (estimate_function_body_sizes): Update.
29956 (compute_inline_parameters): Update.
29957 (estimate_node_size_and_time): Update.
29958 (inline_merge_summary): Update.
29959 (inline_update_overall_summary): Update.
29960 (inline_read_section): Update.
29961 (inline_write_summary): Update.
29962 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
29963 add account_size_time and reset member functions.
29964 (ipa_call_summary): Add reset function.
29965 * ipa-predicate.h (predicate::operator &): Constify.
29967 2017-05-22 Richard Biener <rguenther@suse.de>
29969 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
29971 2017-05-19 Jason Merrill <jason@redhat.com>
29973 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
29975 2017-05-19 Marek Polacek <polacek@redhat.com>
29978 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
29979 TYPE_OVERFLOW_WRAPS checks.
29981 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
29983 * tree-core.h (enum omp_clause_default_kind): Add
29984 "OMP_CLAUSE_DEFAULT_PRESENT".
29985 * tree-pretty-print.c (dump_omp_clause): Handle it.
29986 * gimplify.c (enum gimplify_omp_var_data): Add
29987 "GOVD_MAP_FORCE_PRESENT".
29988 (gimplify_adjust_omp_clauses_1): Map it to
29989 "GOMP_MAP_FORCE_PRESENT".
29990 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
29992 * gimplify.c (oacc_default_clause): Clarify.
29994 2017-05-19 Nathan Sidwell <nathan@acm.org>
29996 LANG_HOOK_REGISTER_DUMPS
29997 * toplev.c (general_init): Call register dump lang hook.
29998 * doc/invoke.texi: Document -fdump-lang option family.
29999 * dumpfile.c (dump_files): Remove class dump here.
30000 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30001 * dumpfile.h (tree_dump_index): Remove TDI_class.
30002 * langhooks-def.h (lhd_register_dumps): Declare.
30003 (LANG_HOOKS_REGISTER_DUMPS): Define.
30004 (LANG_HOOKS_INITIALIZER): Add it.
30005 * langhooks.c (lhd_register_dumps): Define.
30006 * langhooks.h (struct lang_hooks): Add register_dumps.
30008 2017-05-19 Nathan Sidwell <nathan@acm.org>
30010 * context.h (context::set_passes): New.
30011 * context.c (context::context): Do not create pass manager.
30012 * toplev.c (general_init): Create pass manager here.
30014 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30016 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30017 use this splitter if two add or or instructions would also work for
30018 the constant we want to generate.
30020 2017-05-19 Richard Biener <rguenther@suse.de>
30023 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30024 predicate evaluation.
30026 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30028 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30030 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30033 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30035 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30036 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30037 (inline_edge_summary_vec): Turn into ...
30038 (ipa_call_summaries): ... this one.
30039 (redirect_to_unreachable, edge_set_predicate,
30040 evaluate_properties_for_edge, inline_summary_alloc,
30041 reset_ipa_call_summary, reset_inline_summary,
30042 inline_summary_t::duplicate): Update.
30043 (inline_edge_duplication_hook): Turn to ...
30044 (ipa_call_summary_t::duplicate): ... this one.
30045 (inline_edge_removal_hook): Turn to ...
30046 (ipa_call_summary_t::remove): ... this one.
30047 (dump_inline_edge_summary): Turn to ...
30048 (dump_ipa_call_summary): ... this one.
30049 (estimate_function_body_sizes): Update.
30050 (inline_update_callee_summaries): Update.
30051 (remap_edge_change_prob): Update.
30052 (remap_edge_summaries): Update.
30053 (inline_merge_summary): Update.
30054 (do_estimate_edge_time): Update.
30055 (inline_generate_summary): Update.
30056 (inline_read_section): Update.
30057 (inline_read_summary): Update.
30058 (inline_free_summary): Update.
30059 * ipa-inline.c (can_inline_edge_p): Update.
30060 (compute_inlined_call_time): Update.
30061 (want_inline_small_function_p): Update.
30062 (edge_badness): Update.
30063 (early_inliner): Update.
30064 * ipa-inline.h (inline_edge_summary): Turn to ...
30065 (ipa_call_summary): ... this one.
30066 (ipa_call_summary_t): New class.
30067 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30068 (ipa_call_summaries): New.
30069 (inline_edge_summary): Remove.
30070 (estimate_edge_growth): Update.
30071 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30072 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30073 * ipa-split.c (execute_split_functions): Update.
30074 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30076 2017-05-19 Richard Biener <rguenther@suse.de>
30078 PR middle-end/80764
30079 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
30081 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
30083 * config/rs6000/rs6000.c (struct machine_function): Add field
30084 fpr_is_wrapped_separately.
30085 (rs6000_get_separate_components): Use 64 components. Handle the
30086 new FPR components.
30087 (rs6000_components_for_bb): Handle the FPR components.
30088 (rs6000_emit_prologue_components): Handle the FPR components.
30089 (rs6000_emit_epilogue_components): Handle the FPR components.
30090 (rs6000_set_handled_components): Handle the FPR components.
30091 (rs6000_emit_prologue): Don't output prologue code for those FPRs
30092 that are already separately shrink-wrapped.
30093 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
30094 that are already separately shrink-wrapped.
30096 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30099 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
30102 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
30103 (define_peephole2 for Altivec d-form load): Add peepholes to catch
30104 cases where the register allocator uses a move and an offsettable
30105 memory operation to/from a FPR register on ISA 2.06/2.07.
30106 (define_peephole2 for Altivec d-form store): Likewise.
30108 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
30111 * config/i386/mmx.md (*mov<mode>_internal): Enable
30112 alternatives 11, 12, 13 and 14 also for 32bit targets.
30113 Remove alternatives 15, 16, 17 and 18.
30114 * config/i386/sse.md (vec_concatv2di): Change
30115 alternative (!x, *y) to (x, ?!*Yn).
30117 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
30119 * dumpfile.h (enum dump_kind): Remove stray comma.
30121 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30123 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
30124 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
30125 predicate::num_conditions
30126 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
30127 (CHANGED): turn into predicate::changed.
30128 (agg_position_info): Move to ipa-predicate.h
30129 (add_condition, predicate::add_clause, predicate::operator &=,
30130 predicate::or_with, predicate::evaluate, predicate::probability,
30131 dump_condition, dump_clause, predicate::dump,
30132 predicate::remap_after_duplication, predicate::remap_after_inlining,
30133 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
30134 (evaluate_conditions_for_known_args): Update.
30135 (set_cond_stmt_execution_predicate): Update.
30136 * ipa-inline.h: Include ipa-predicate.h
30137 (condition, inline_param_summary, conditions, agg_position_info,
30138 predicate): Move to ipa-predicate.h
30139 * ipa-predicate.c: New file.
30140 * ipa-predicate.h: New file.
30142 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
30144 * final.c (leaf_function_p): Check we are not in a sequence.
30146 2017-05-18 Martin Liska <mliska@suse.cz>
30148 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
30149 * dumpfile.c (dump_register): Use new enum dump_kind.
30150 (get_dump_file_name): Likewise.
30151 (dump_enable_all): Likewise.
30152 (dump_switch_p_1): Likewise.
30153 (enable_rtl_dump_file): Remove usage of TDF_RTL.
30154 * dumpfile.h (enum dump_kind): New enum type.
30155 (struct dump_file_info): Create constructor and
30156 format fields and comments.
30157 * passes.c (pass_manager::register_one_dump_file):
30159 * statistics.c (statistics_early_init): Likewise.
30160 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
30161 TDF_TREE with TDF_SLIM.
30162 (gather_memory_references_ref): Likewise.
30164 2017-05-18 Martin Liska <mliska@suse.cz>
30166 * vec.h (struct vnull): Use it.
30168 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30170 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
30171 (true_predicate, false_predicate, true_predicate_p,
30172 false_predicate_p): Remove.
30173 (single_cond_predicate, not_inlined_predicate): Turn to member function
30175 (add_condition): Update.
30176 (add_clause): Turn to...
30177 (predicate::add_clause): ... this one; update; allow passing NULL
30179 (and_predicates): Turn to ...
30180 (predicate::operator &=): ... this one.
30181 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
30182 (or_predicates): Turn to ...
30183 (predicate::or_with): ... this one.
30184 (evaluate_predicate): Turn to ...
30185 (predicate::evaluate): ... this one.
30186 (predicate_probability): Turn to ...
30187 (predicate::probability): ... this one.
30188 (dump_condition): Update.
30189 (dump_predicate): Turn to ...
30190 (predicate::dump): ... this one.
30191 (account_size_time): Update.
30192 (edge_set_predicate): Update.
30193 (set_hint_predicate): UPdate.
30194 (evaluate_conditions_for_known_args): Update.
30195 (evaluate_properties_for_edge): Update.
30196 (remap_predicate_after_duplication): Turn to...
30197 (predicate::remap_after_duplication): ... this one.
30198 (remap_hint_predicate_after_duplication): Update.
30199 (inline_summary_t::duplicate): UPdate.
30200 (dump_inline_edge_summary): Update.
30201 (dump_inline_summary): Update.
30202 (set_cond_stmt_execution_predicate): Update.
30203 (set_switch_stmt_execution_predicate): Update.
30204 (compute_bb_predicates): Update.
30205 (will_be_nonconstant_expr_predicate): Update.
30206 (will_be_nonconstant_predicate): Update.
30207 (phi_result_unknown_predicate): Update.
30208 (predicate_for_phi_result): Update.
30209 (array_index_predicate): Update.
30210 (estimate_function_body_sizes): Update.
30211 (estimate_node_size_and_time): Update.
30212 (estimate_ipcp_clone_size_and_time): Update.
30213 (remap_predicate): Rename to ...
30214 (predicate::remap_after_inlining): ... this one.
30215 (remap_hint_predicate): Update.
30216 (inline_merge_summary): Update.
30217 (inline_update_overall_summary): Update.
30218 (estimate_size_after_inlining): Update.
30219 (read_predicate): Rename to ...
30220 (predicate::stream_in): ... this one.
30221 (read_inline_edge_summary): Update.
30222 (write_predicate): Rename to ...
30223 (predicate::stream_out): ... this one.
30224 (write_inline_edge_summary): Update.
30225 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
30226 (clause_t): Turn to uint32_t
30227 (predicate): Turn to class; implement constructor and operators
30229 (size_time_entry): Update.
30230 (inline_summary): Update.
30231 (inline_edge_summary): Update.
30233 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
30235 * fold-const.c (fold_binary_loc): Move transformation...
30236 * match.pd (C - X CMP X): ... here.
30238 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
30240 * config/sparc/sparc.c (sparc_option_override): Set function
30241 alignment for -mcpu=niagara7 to 64 to match the I$ line.
30242 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
30244 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
30246 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
30248 2017-05-18 Marek Polacek <polacek@redhat.com>
30251 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
30252 (pass_ubsan::execute): Call gimple_assign_single_p instead of
30253 gimple_assign_load_p.
30255 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
30257 PR middle-end/80692
30258 * real.c (do_compare): Give decimal_do_compare preference over
30259 comparing just the signs.
30261 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
30263 * doc/md.texi (Canonicalization of Instructions): Describe the
30264 canonical form of instructions that inherently set a condition
30267 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
30269 PR middle-end/80775
30270 * tree-cfg.c: Move deletion of unreachable case statements to after
30271 the merging of consecutive case labels.
30273 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
30275 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
30276 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
30277 restoring of callee-saved registers.
30279 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
30281 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
30282 * config/visium/visium.c (single_set_and_flags): Likewise.
30283 * config/visium/visium.md (Substitutions): Likewise.
30285 2017-05-17 Martin Liska <mliska@suse.cz>
30287 * cfg.c: Introduce dump_flags_t type and
30288 use it instead of int type.
30290 * cfghooks.c: Likewise.
30291 * cfghooks.h (struct cfg_hooks): Likewise.
30292 * cfgrtl.c: Likewise.
30293 * cfgrtl.h: Likewise.
30294 * cgraph.c (cgraph_node::get_body): Likewise.
30295 * coretypes.h: Likewise.
30296 * domwalk.c: Likewise.
30297 * domwalk.h: Likewise.
30298 * dumpfile.c (struct dump_option_value_info): Likewise.
30299 (dump_enable_all): Likewise.
30300 (dump_switch_p_1): Likewise.
30301 (opt_info_switch_p): Likewise.
30302 * dumpfile.h (enum tree_dump_index): Likewise.
30303 (struct dump_file_info): Likewise.
30304 * genemit.c: Likewise.
30305 * generic-match-head.c: Likewise.
30306 * gengtype.c (open_base_files): Likewise.
30307 * gimple-pretty-print.c: Likewise.
30308 * gimple-pretty-print.h: Likewise.
30309 * graph.c (print_graph_cfg): Likewise.
30310 * graphite-scop-detection.c (dot_all_sese): Likewise.
30311 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
30312 * loop-unroll.c (report_unroll): Likewise.
30313 * passes.c (pass_manager::register_one_dump_file): Likewise.
30314 * print-tree.c: Likewise.
30315 * statistics.c: Likewise.
30316 * tree-cfg.c: Likewise.
30317 * tree-cfg.h: Likewise.
30318 * tree-dfa.c: Likewise.
30319 * tree-dfa.h: Likewise.
30320 * tree-dump.c (dump_function): Likewise.
30321 * tree-dump.h (struct dump_info): Likewise.
30322 * tree-pretty-print.c: Likewise.
30323 * tree-pretty-print.h: Likewise.
30324 * tree-ssa-live.c: Likewise.
30325 * tree-ssa-live.h: Likewise.
30326 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
30327 * tree-vect-loop.c: Likewise.
30328 * tree-vect-slp.c: Likewise.
30330 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
30331 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30333 PR tree-optimization/80457
30334 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
30335 of all arguments to a statement as scalar_to_vec operations.
30336 (vectorizable_call): Adjust call to vect_model_simple_cost for
30338 (vectorizable_conversion): Likewise.
30339 (vectorizable_assignment): Likewise.
30340 (vectorizable_shift): Likewise.
30341 (vectorizable_operation): Likewise.
30342 (vectorizable_comparison): Likewise.
30343 (vect_is_simple_cond): Record the def types for operands.
30344 (vectorizable_condition): Likewise, call vect_model_simple_cost.
30345 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
30346 for statement argument count.
30348 2017-05-16 Carl Love <cel@us.ibm.com>
30350 * config/rs6000/rs6000-c: Add support for built-in functions
30351 vector unsigned long long vec_bperm (vector unsigned long long,
30352 vector unsigned char)
30353 vector signed long long vec_mule (vector signed int,
30355 vector unsigned long long vec_mule (vector unsigned int,
30356 vector unsigned int)
30357 vector signed long long vec_mulo (vector signed int,
30359 vector unsigned long long vec_mulo (vector unsigned int,
30360 vector unsigned int)
30361 vector signed char vec_sldw (vector signed char,
30362 vector signed char,
30364 vector unsigned char vec_sldw (vector unsigned char,
30365 vector unsigned char,
30367 vector signed short vec_sldw (vector signed short,
30368 vector signed short,
30370 vector unsigned short vec_sldw (vector unsigned short,
30371 vector unsigned short,
30373 vector signed int vec_sldw (vector signed int,
30376 vector unsigned int vec_sldw (vector unsigned int,
30377 vector unsigned int,
30379 vector signed long long vec_sldw (vector signed long long,
30380 vector signed long long,
30382 vector unsigned long long vec_sldw (vector unsigned long long,
30383 vector unsigned long long,
30385 * config/rs6000/rs6000-c: Add support for built-in functions
30386 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
30387 * config/rs6000/altivec.h: Add defintion for vec_sldw.
30388 * doc/extend.texi: Update the built-in documentation for the
30389 new built-in functions.
30391 2017-05-16 Marek Polacek <polacek@redhat.com>
30395 * tree.c (save_expr): Don't fold the expression.
30397 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
30399 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
30400 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
30401 and (?*y,m). Update insn attributes.
30403 2017-05-16 Martin Liska <mliska@suse.cz>
30405 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
30406 flags argument of print_gimple_stmt, print_gimple_expr,
30407 print_generic_stmt and print_generic_expr.
30408 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
30409 * coretypes.h: Likewise.
30410 * except.c (dump_eh_tree): Likewise.
30411 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
30412 * gimple-pretty-print.h: Likewise.
30413 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
30414 (backprop::push_to_worklist): Likewise.
30415 (backprop::pop_from_worklist): Likewise.
30416 (backprop::process_use): Likewise.
30417 (backprop::intersect_uses): Likewise.
30418 (note_replacement): Likewise.
30419 * gimple-ssa-store-merging.c
30420 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
30421 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
30422 (pass_store_merging::execute): Likewise.
30423 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
30424 (ssa_base_cand_dump_callback): Likewise.
30425 (dump_incr_vec): Likewise.
30426 (replace_refs): Likewise.
30427 (replace_mult_candidate): Likewise.
30428 (create_add_on_incoming_edge): Likewise.
30429 (create_phi_basis): Likewise.
30430 (insert_initializers): Likewise.
30431 (all_phi_incrs_profitable): Likewise.
30432 (introduce_cast_before_cand): Likewise.
30433 (replace_one_candidate): Likewise.
30434 * gimplify.c (gimplify_expr): Likewise.
30435 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
30436 (set_rename): Likewise.
30437 (rename_uses): Likewise.
30438 (copy_loop_phi_nodes): Likewise.
30439 (add_close_phis_to_merge_points): Likewise.
30440 (copy_loop_close_phi_args): Likewise.
30441 (copy_cond_phi_args): Likewise.
30442 (graphite_copy_stmts_from_block): Likewise.
30443 (translate_pending_phi_nodes): Likewise.
30444 * graphite-poly.c (print_pdr): Likewise.
30445 (dump_gbb_cases): Likewise.
30446 (dump_gbb_conditions): Likewise.
30447 (print_scop_params): Likewise.
30448 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
30449 (build_cross_bb_scalars_use): Likewise.
30450 (gather_bbs::before_dom_children): Likewise.
30451 * hsa-dump.c (dump_hsa_immed): Likewise.
30452 * ipa-cp.c (print_ipcp_constant_value): Likewise.
30453 (get_replacement_map): Likewise.
30454 * ipa-inline-analysis.c (dump_condition): Likewise.
30455 (estimate_function_body_sizes): Likewise.
30456 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
30457 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
30458 * ipa-prop.c (ipa_dump_param): Likewise.
30459 (ipa_print_node_jump_functions_for_edge): Likewise.
30460 (ipa_modify_call_arguments): Likewise.
30461 (ipa_modify_expr): Likewise.
30462 (ipa_dump_param_adjustments): Likewise.
30463 (ipa_dump_agg_replacement_values): Likewise.
30464 (ipcp_modif_dom_walker::before_dom_children): Likewise.
30465 * ipa-pure-const.c (check_stmt): Likewise.
30466 (pass_nothrow::execute): Likewise.
30467 * ipa-split.c (execute_split_functions): Likewise.
30468 * omp-offload.c (dump_oacc_loop_part): Likewise.
30469 (dump_oacc_loop): Likewise.
30470 * trans-mem.c (tm_log_emit): Likewise.
30471 (tm_memopt_accumulate_memops): Likewise.
30472 (dump_tm_memopt_set): Likewise.
30473 (dump_tm_memopt_transform): Likewise.
30474 * tree-cfg.c (gimple_verify_flow_info): Likewise.
30475 (print_loop): Likewise.
30476 * tree-chkp-opt.c (chkp_print_addr): Likewise.
30477 (chkp_gather_checks_info): Likewise.
30478 (chkp_get_check_result): Likewise.
30479 (chkp_remove_check_if_pass): Likewise.
30480 (chkp_use_outer_bounds_if_possible): Likewise.
30481 (chkp_reduce_bounds_lifetime): Likewise.
30482 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
30483 (chkp_mark_completed_bounds): Likewise.
30484 (chkp_register_incomplete_bounds): Likewise.
30485 (chkp_mark_invalid_bounds): Likewise.
30486 (chkp_maybe_copy_and_register_bounds): Likewise.
30487 (chkp_build_returned_bound): Likewise.
30488 (chkp_get_bound_for_parm): Likewise.
30489 (chkp_build_bndldx): Likewise.
30490 (chkp_get_bounds_by_definition): Likewise.
30491 (chkp_generate_extern_var_bounds): Likewise.
30492 (chkp_get_bounds_for_decl_addr): Likewise.
30493 * tree-chrec.c (chrec_apply): Likewise.
30494 * tree-data-ref.c (dump_data_reference): Likewise.
30495 (dump_subscript): Likewise.
30496 (dump_data_dependence_relation): Likewise.
30497 (analyze_overlapping_iterations): Likewise.
30498 * tree-inline.c (expand_call_inline): Likewise.
30499 (tree_function_versioning): Likewise.
30500 * tree-into-ssa.c (dump_defs_stack): Likewise.
30501 (dump_currdefs): Likewise.
30502 (dump_names_replaced_by): Likewise.
30503 (dump_update_ssa): Likewise.
30504 (update_ssa): Likewise.
30505 * tree-object-size.c (pass_object_sizes::execute): Likewise.
30506 * tree-parloops.c (build_new_reduction): Likewise.
30507 (try_create_reduction_list): Likewise.
30508 (ref_conflicts_with_region): Likewise.
30509 (oacc_entry_exit_ok_1): Likewise.
30510 (oacc_entry_exit_single_gang): Likewise.
30511 * tree-pretty-print.h: Likewise.
30512 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
30513 (get_scalar_evolution): Likewise.
30514 (add_to_evolution): Likewise.
30515 (get_loop_exit_condition): Likewise.
30516 (analyze_evolution_in_loop): Likewise.
30517 (analyze_initial_condition): Likewise.
30518 (analyze_scalar_evolution): Likewise.
30519 (instantiate_scev): Likewise.
30520 (number_of_latch_executions): Likewise.
30521 (gather_chrec_stats): Likewise.
30522 (final_value_replacement_loop): Likewise.
30523 (scev_const_prop): Likewise.
30524 * tree-sra.c (dump_access): Likewise.
30525 (disqualify_candidate): Likewise.
30526 (create_access): Likewise.
30527 (reject): Likewise.
30528 (maybe_add_sra_candidate): Likewise.
30529 (create_access_replacement): Likewise.
30530 (analyze_access_subtree): Likewise.
30531 (analyze_all_variable_accesses): Likewise.
30532 (sra_modify_assign): Likewise.
30533 (initialize_constant_pool_replacements): Likewise.
30534 (find_param_candidates): Likewise.
30535 (decide_one_param_reduction): Likewise.
30536 (replace_removed_params_ssa_names): Likewise.
30537 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
30538 * tree-ssa-copy.c (dump_copy_of): Likewise.
30539 (copy_prop_visit_cond_stmt): Likewise.
30540 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
30541 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
30542 (record_equivalences_from_stmt): Likewise.
30543 * tree-ssa-dse.c (compute_trims): Likewise.
30544 (delete_dead_call): Likewise.
30545 (delete_dead_assignment): Likewise.
30546 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
30547 (forward_propagate_into_cond): Likewise.
30548 (pass_forwprop::execute): Likewise.
30549 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
30550 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
30552 (move_computations_worker): Likewise.
30553 (execute_sm): Likewise.
30554 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
30555 (remove_exits_and_undefined_stmts): Likewise.
30556 (remove_redundant_iv_tests): Likewise.
30557 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
30558 (adjust_iv_update_pos): Likewise.
30559 * tree-ssa-math-opts.c (bswap_replace): Likewise.
30560 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
30561 (value_replacement): Likewise.
30562 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
30563 * tree-ssa-pre.c (print_pre_expr): Likewise.
30564 (get_representative_for): Likewise.
30565 (create_expression_by_pieces): Likewise.
30566 (insert_into_preds_of_block): Likewise.
30567 (eliminate_insert): Likewise.
30568 (eliminate_dom_walker::before_dom_children): Likewise.
30569 (eliminate): Likewise.
30570 (remove_dead_inserted_code): Likewise.
30571 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30572 * tree-ssa-reassoc.c (get_rank): Likewise.
30573 (eliminate_duplicate_pair): Likewise.
30574 (eliminate_plus_minus_pair): Likewise.
30575 (eliminate_not_pairs): Likewise.
30576 (undistribute_ops_list): Likewise.
30577 (eliminate_redundant_comparison): Likewise.
30578 (update_range_test): Likewise.
30579 (optimize_range_tests_var_bound): Likewise.
30580 (optimize_vec_cond_expr): Likewise.
30581 (rewrite_expr_tree): Likewise.
30582 (rewrite_expr_tree_parallel): Likewise.
30583 (linearize_expr): Likewise.
30584 (break_up_subtract): Likewise.
30585 (linearize_expr_tree): Likewise.
30586 (attempt_builtin_powi): Likewise.
30587 (attempt_builtin_copysign): Likewise.
30588 (transform_stmt_to_copy): Likewise.
30589 (transform_stmt_to_multiply): Likewise.
30590 (dump_ops_vector): Likewise.
30591 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30592 (print_scc): Likewise.
30593 (set_ssa_val_to): Likewise.
30594 (visit_reference_op_store): Likewise.
30595 (visit_use): Likewise.
30596 (sccvn_dom_walker::before_dom_children): Likewise.
30597 (run_scc_vn): Likewise.
30598 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30600 (expr_hash_elt::print): Likewise.
30601 (const_and_copies::pop_to_marker): Likewise.
30602 (const_and_copies::record_const_or_copy_raw): Likewise.
30603 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30604 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30605 (dump_predicates): Likewise.
30606 (find_uninit_use): Likewise.
30607 (warn_uninitialized_phi): Likewise.
30608 (pass_late_warn_uninitialized::execute): Likewise.
30609 * tree-ssa.c (verify_vssa): Likewise.
30610 (verify_ssa): Likewise.
30611 (maybe_optimize_var): Likewise.
30612 * tree-vrp.c (dump_value_range): Likewise.
30613 (dump_all_value_ranges): Likewise.
30614 (dump_asserts_for): Likewise.
30615 (register_edge_assert_for_2): Likewise.
30616 (vrp_visit_cond_stmt): Likewise.
30617 (vrp_visit_switch_stmt): Likewise.
30618 (vrp_visit_stmt): Likewise.
30619 (vrp_visit_phi_node): Likewise.
30620 (simplify_cond_using_ranges_1): Likewise.
30621 (fold_predicate_in): Likewise.
30622 (evrp_dom_walker::before_dom_children): Likewise.
30623 (evrp_dom_walker::push_value_range): Likewise.
30624 (evrp_dom_walker::pop_value_range): Likewise.
30625 (execute_early_vrp): Likewise.
30627 2017-05-16 Richard Biener <rguenther@suse.de>
30629 * dwarf2out.c (loc_list_from_tree_1): Do not create
30630 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30632 2017-05-16 Richard Biener <rguenther@suse.de>
30634 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30636 (note_variable_value_in_expr): If we resolved the decl ref
30637 do not push to the stack.
30639 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30641 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30642 operations in fast-math mode.
30643 (vaddq_f16): Likewise.
30644 (vmul_f16): Likewise.
30645 (vmulq_f16): Likewise.
30646 (vsub_f16): Likewise.
30647 (vsubq_f16): Likewise.
30648 * config/arm/neon.md (add<mode>3): New.
30650 (fma:<VH:mode>3): New. Also remove outdated comment.
30653 2017-05-16 Martin Liska <mliska@suse.cz>
30657 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30658 (odr_types_equivalent_p): Likewise.
30660 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30662 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30664 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30667 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30668 non-interunit SSE move alternatives with '?'.
30669 (zero-extendsidi peephole2): New peephole to skip intermediate
30670 general register in SSE zero-extend sequence.
30672 2017-05-15 Jeff Law <law@redhat.com>
30674 * reorg.c (relax_delay_slots): Create a new variable to hold
30675 the temporary target rather than clobbering TARGET_LABEL.
30677 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30678 missing argument to extract_bit_field call.
30679 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30681 2017-05-15 Martin Liska <mliska@suse.cz>
30684 * gcc.c (process_command): Do not allow empty argument of -o option.
30686 2017-05-15 Renlin Li <renlin.li@arm.com>
30688 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30689 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30690 * config/aarch64/constraints.md (Usf): Add long call check.
30691 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30692 (call_value): Likewise.
30693 (sibcall): Likewise.
30694 (sibcall_value): Likewise.
30696 (call_value_insn): New.
30697 (sibcall_insn): Update rtx pattern.
30698 (sibcall_value_insn): Likewise.
30699 (call_internal): Remove.
30700 (call_value_internal): Likewise.
30701 (sibcall_internal): Likewise.
30702 (sibcall_value_internal): Likewise.
30703 (call_reg): Likewise.
30704 (call_symbol): Likewise.
30705 (call_value_reg): Likewise.
30706 (call_value_symbol): Likewise.
30708 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30711 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30713 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30715 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30716 compatible with CCGOCmode and with CCZmode.
30718 2017-05-14 Martin Sebor <msebor@redhat.com>
30720 PR middle-end/77671
30721 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30722 (gimple_fold_builtin_snprintf): Same.
30723 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30724 (gimple_fold_builtin_snprintf): Same.
30725 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30726 of character types.
30727 (is_call_safe): New function.
30728 (try_substitute_return_value): Call it.
30729 (try_simplify_call): New function.
30730 (pass_sprintf_length::handle_gimple_call): Call it.
30732 2017-05-14 Martin Sebor <msebor@redhat.com>
30734 PR middle-end/80669
30735 * builtins.c (expand_builtin_stpncpy): Simplify.
30737 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30739 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30740 * config/i386/i386.h
30741 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30742 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30743 (struct machine_function): Add new members call_ms2sysv,
30744 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30745 (struct machine_frame_state): New fields sp_realigned and
30746 sp_realigned_offset.
30747 * config/i386/i386.c
30748 (enum xlogue_stub): New enum.
30749 (enum xlogue_stub_sets): New enum.
30750 (class xlogue_layout): New class.
30751 (struct ix86_frame): New fields stack_realign_allocate_offset,
30752 stack_realign_offset and outlined_save_offset. Modify comments to
30753 detail stack layout when using out-of-line stubs.
30754 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30755 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30756 -mcall-ms2sysv-xlogues.
30757 (stub_managed_regs): New static variable.
30758 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30759 registers managed by out-of-line stub.
30760 (disable_call_ms2sysv_xlogues): New function.
30761 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30762 m->call_ms2sysv when appropriate and compute frame layout for
30764 (sp_valid_at, fp_valid_at): New inline functions.
30765 (choose_basereg): New function.
30766 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30768 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30769 Use align parameter of choose_baseaddr to generated aligned SSE movs
30771 (pro_epilogue_adjust_stack): Modify to track
30772 machine_frame_state::sp_realigned.
30773 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30774 (ix86_nsaved_sseregs): Likewise.
30775 (ix86_emit_save_regs): Likewise.
30776 (ix86_emit_save_regs_using_mov): Likewise.
30777 (ix86_emit_save_sse_regs_using_mov): Likewise.
30778 (get_scratch_register_on_entry): Likewise.
30779 (gen_frame_set): New function.
30780 (gen_frame_load): Likewise.
30781 (gen_frame_store): Likewise.
30782 (emit_outlined_ms2sysv_save): Likewise.
30783 (emit_outlined_ms2sysv_restore): Likewise.
30784 (ix86_expand_prologue): Modify stack re-alignment code and call
30785 emit_outlined_ms2sysv_save when appropriate.
30786 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30787 parameter rtx_insn *insn, which allows the function to be used to only
30788 generate the notes.
30789 (ix86_expand_epilogue): Modify validity checks of frame and stack
30790 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30791 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30792 * config/i386/predicates.md
30793 (save_multiple): New predicate.
30794 (restore_multiple): Likewise.
30795 * config/i386/sse.md
30796 (save_multiple<mode>): New pattern.
30797 (save_multiple_realign<mode>): Likewise.
30798 (restore_multiple<mode>): Likewise.
30799 (restore_multiple_and_return<mode>): Likewise.
30800 (restore_multiple_leave_return<mode>): Likewise.
30801 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30803 2017-05-14 Julia Koval <julia.koval@intel.com>
30805 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30806 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30807 (__builtin_ia32_xsetbv): New builtins.
30808 * config/i386/i386.c (ix86_expand_special_args_builtin):
30810 (ix86_expand_builtin): Special expand for new intrinsics.
30811 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
30812 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
30813 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
30815 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30817 * cfganal.c (inverted_post_order_compute): Change argument type
30819 * cfganal.h (inverted_post_order_compute): Adjust prototype.
30820 * df-core.c (rest_of_handle_df_initialize): Adjust.
30821 (rest_of_handle_df_finish): Likewise.
30822 (df_analyze_1): Likewise.
30823 (df_analyze): Likewise.
30824 (loop_inverted_post_order_compute): Change argument to be a vec *.
30825 (df_analyze_loop): Adjust.
30826 (df_get_n_blocks): Likewise.
30827 (df_get_postorder): Likewise.
30828 * df.h (struct df_d): Change field to be a vec.
30829 * lcm.c (compute_laterin): Adjust.
30830 (compute_available): Likewise.
30831 * lra-lives.c (lra_create_live_ranges_1): Likewise.
30832 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
30833 * tree-ssa-pre.c (compute_antic): Likewise.
30835 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30837 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
30838 (depth_first_search::depth_first_search): Change structure init
30839 function to this constructor.
30840 (depth_first_search::add_bb): Rename function to this member.
30841 (depth_first_search::execute): Likewise.
30842 (flow_dfs_compute_reverse_finish): Adjust.
30844 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30846 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
30847 (longest_simple_path): Likewise.
30848 * shrink-wrap.c (spread_components): Likewise.
30849 (disqualify_problematic_components): Likewise.
30850 (emit_common_heads_for_components): Likewise.
30851 (emit_common_tails_for_components): Likewise.
30852 (insert_prologue_epilogue_for_components): Likewise.
30854 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30856 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
30859 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30861 * df-core.c (df_set_blocks): Start using auto_bitmap.
30862 (df_compact_blocks): Likewise.
30863 * df-problems.c (df_rd_confluence_n): Likewise.
30864 * df-scan.c (df_insn_rescan_all): Likewise.
30865 (df_process_deferred_rescans): Likewise.
30866 (df_update_entry_block_defs): Likewise.
30867 (df_update_exit_block_uses): Likewise.
30868 (df_entry_block_bitmap_verify): Likewise.
30869 (df_exit_block_bitmap_verify): Likewise.
30870 (df_scan_verify): Likewise.
30871 * lra-constraints.c (lra_constraints): Likewise.
30872 (undo_optional_reloads): Likewise.
30873 (lra_undo_inheritance): Likewise.
30874 * lra-remat.c (calculate_gen_cands): Likewise.
30875 (do_remat): Likewise.
30876 * lra-spills.c (assign_spill_hard_regs): Likewise.
30877 (spill_pseudos): Likewise.
30878 * tree-ssa-pre.c (bitmap_set_and): Likewise.
30879 (bitmap_set_subtract_values): Likewise.
30881 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30883 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
30884 management with auto_bitmap.
30885 (fix_inter_tick): Likewise.
30886 (fix_recovery_deps): Likewise.
30887 * ira.c (add_store_equivs): Likewise.
30888 (find_moveable_pseudos): Likewise.
30889 (split_live_ranges_for_shrink_wrap): Likewise.
30890 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
30891 (rtx_reuse_manager::seen_def_p): Likewise.
30892 (rtx_reuse_manager::set_seen_def): Likewise.
30893 * print-rtl.h (class rtx_reuse_manager): Likewise.
30895 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30897 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
30899 (migrate_btr_def): Likewise.
30900 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
30901 * df-core.c (loop_post_order_compute): Likewise.
30902 (loop_inverted_post_order_compute): Likewise.
30903 * hsa-common.h: Likewise.
30904 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
30905 * init-regs.c (initialize_uninitialized_regs): Likewise.
30906 * ipa-inline.c (resolve_noninline_speculation): Likewise.
30907 (inline_small_functions): Likewise.
30908 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
30909 * ira.c (combine_and_move_insns): Likewise.
30910 (build_insn_chain): Likewise.
30911 * loop-invariant.c (find_invariants): Likewise.
30912 * lower-subreg.c (propagate_pseudo_copies): Likewise.
30913 * predict.c (tree_predict_by_opcode): Likewise.
30914 (predict_paths_leading_to): Likewise.
30915 (predict_paths_leading_to_edge): Likewise.
30916 (estimate_loops_at_level): Likewise.
30917 (estimate_loops): Likewise.
30918 * shrink-wrap.c (try_shrink_wrapping): Likewise.
30919 (spread_components): Likewise.
30920 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
30921 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
30922 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
30923 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
30924 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
30925 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
30926 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
30927 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
30928 (mark_threaded_blocks): Likewise.
30929 (thread_through_all_blocks): Likewise.
30930 * tree-ssa.c (verify_ssa): Likewise.
30931 (execute_update_addresses_taken): Likewise.
30932 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
30934 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30936 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
30938 (post_order_compute): Likewise.
30939 (inverted_post_order_compute): Likewise.
30940 (pre_and_rev_post_order_compute_fn): Likewise.
30942 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30944 * genrecog.c (int_set::int_set): Explicitly construct our
30945 auto_vec base class.
30946 * vec.h (auto_vec::auto_vec): New constructor.
30948 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30950 * bitmap.h (class auto_bitmap): New constructor taking
30951 bitmap_obstack * argument.
30953 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30955 * bitmap.h (class auto_bitmap): Change type of m_bits to
30956 bitmap_head, and adjust ctor / dtor and member operators.
30958 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
30960 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
30961 when returned register mode doesn't match original mode.
30963 2017-05-12 Jeff Law <law@redhat.com>
30964 Jakub Jelinek <jakub@redhat.com>
30966 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
30967 we look for cc setter after the compare-elim changes.
30968 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
30969 within the vector to match what compare-elim now expects.
30970 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
30971 (xorsi3_flags, one_cmplsi2_flags): Likewise.
30973 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
30974 after the compare-elim changes.
30975 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
30976 the vector to match what compare-elim now expects.
30977 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
30978 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
30979 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
30980 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
30981 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
30983 * config/visium/visium.c (single_set_and_flags): Fix where
30984 we look for cc setter after the compare-elim changes.
30985 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
30986 with the vector to match what compare-elim now expects.
30987 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
30988 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
30989 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
30990 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
30991 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
30992 (neg<mode>2_insn_set_overflow): Likewise.
30994 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
30996 PR middle-end/79794
30997 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
30998 maybe_expand_insn call, set ops[0].target. If still set after call,
30999 set alt_rtl. Add extra arg to recursive calls.
31000 (extract_bit_field): Add alt_rtl argument. Pass to
31002 * expmed.h (extract_bit_field): Fix prototype.
31003 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31004 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31005 to extract_bit_field_calls.
31006 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31007 Pass alt_rtl to extract_bit_field calls.
31008 * calls.c (store_unaligned_arguments_into_psuedos)
31009 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31010 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31012 * optabs.h (struct expand_operand): Add target bitfield.
31014 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31016 * compare-elim.c (try_eliminate_compare): Canonicalize
31017 operation with embedded compare to
31018 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31019 (set (reg) (operation)].
31021 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31023 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31026 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31027 cost of adding a carry flag for ADC instruction.
31028 [case MINUS]: Ignore the cost of subtracting a carry flag
31029 for SBB instruction.
31031 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31033 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31035 * config/rs6000/bmiintrin.h: New file.
31036 * config/rs6000/bmi2intrin.h: New file.
31037 * config/rs6000/x86intrin.h: New file.
31039 2017-05-12 Jeff Law <law@redhat.com>
31041 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31044 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31046 PR middle-end/80707
31047 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31049 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31051 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31052 early expansion of vector divide builtins.
31053 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
31054 builtins identified as having unsigned arguments.
31056 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31058 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
31059 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
31060 expansion of vector logical operations (and, andc, or, xor,
31063 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31065 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
31066 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
31068 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31070 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31071 early GIMPLE expansion of vector multiplies.
31073 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31075 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
31076 TARGET_HAVE_MOVT conditional.
31077 (movt splitter): Likewise.
31079 2017-05-12 Richard Biener <rguenther@suse.de>
31081 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
31082 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31083 Fold all stmts not inplace.
31085 2017-05-12 Richard Biener <rguenther@suse.de>
31087 PR tree-optimization/80713
31088 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
31089 inserted_exprs bit for not removed stmts.
31091 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
31093 PR middle-end/69921
31094 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
31095 parallelized" attribute for parallelized OpenACC kernels.
31096 * omp-offload.c (execute_oacc_device_lower): Use it.
31098 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
31099 Set "oacc kernels" attribute.
31100 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
31101 parameter. Adjust all users.
31102 (oacc_fn_attrib_kernels_p): Remove function.
31103 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
31104 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
31105 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
31106 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
31107 assert "oacc kernels" attribute is set.
31109 2017-05-11 Carl Love <cel@us.ibm.com>
31111 * config/rs6000/rs6000-c: Add support for built-in functions
31112 vector unsigned char vec_popcnt (vector signed char)
31113 vector unsigned char vec_popcnt (vector unsigned char)
31114 vector unsigned short vec_popcnt (vector signed short)
31115 vector unsigned short vec_popcnt (vector unsigned short)
31116 vector unsigned int vec_popcnt (vector signed int)
31117 vector unsigned int vec_popcnt (vector unsigned int)
31118 vector unsigned long long vec_popcnt (vector signed long long)
31119 vector unsigned long long vec_popcnt (vector unsigned long long)
31120 vector signed long long vec_slo (vector signed long long,
31121 vector signed char)
31122 vector signed long long vec_slo (vector signed long long,
31123 vector unsigned char)
31124 vector unsigned long long vec_slo (vector unsigned long long,
31125 vector signed char)
31126 vector unsigned long long vec_slo (vector unsigned long long,
31127 vector unsigned char)
31128 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
31129 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
31130 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
31131 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
31132 * doc/extend.texi: Update the built-in documentation file for the
31133 new built-in functions.
31135 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
31137 * attribs.h (sorted_attr_string): Move machine independent
31138 functions for target clone support from the i386 port to common
31139 code. Rename ix86_function_versions to common_function_versions.
31140 Rename make_name to make_unique_name.
31141 (common_function_versions): Likewise.
31142 (make_unique_name): Likewise.
31143 (make_dispatcher_decl): Likewise.
31144 (is_function_default_version): Likewise.
31145 * attribs.c (attr_strcmp): Likewise.
31146 (sorted_attr_string): Likewise.
31147 (common_function_versions): Likewise.
31148 (make_unique_name): Likewise.
31149 (make_dispatcher_decl): Likewise.
31150 (is_function_default_version): Likewise.
31151 * config/i386/i386.c (attr_strcmp): Likewise.
31152 (sorted_attr_string): Likewise.
31153 (ix86_function_versions): Likewise.
31154 (make_name): Likewise.
31155 (make_dispatcher_decl): Likewise.
31156 (is_function_default_version): Likewise.
31157 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
31159 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31162 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
31163 Account for direct move costs for vec_construct of integer
31166 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
31169 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
31170 (UNSPEC_STX_ATOMIC): Ditto.
31171 (loaddi_via_sse): New insn.
31172 (storedi_via_sse): Ditto.
31173 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
31174 Update corresponding peephole2 patterns.
31175 (atomic_storedi_fpu): Ditto.
31177 2017-05-11 Julia Koval <julia.koval@intel.com>
31179 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
31180 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
31182 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
31183 (__builtin_ia32_rsqrt14ss_mask): New builtins.
31184 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
31186 2017-05-11 Nathan Sidwell <nathan@acm.org>
31188 * graphite-poly.c: Include dumpfile.h.
31190 * dumpfle.h (dump_function): Declare here ...
31191 * tree-dump.h (dump_function): ... not here.
31192 * dumpfile.c: #include tree-cfg.h.
31193 (dump_function): Move here from ...
31194 * tree-dump.c (dump_function): ... here.
31195 * gimplify.c: #include splay-tree.h, not tree-dump.h.
31196 * graphite-poly.c: Don't include tree-dump.h.
31197 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
31198 * print-tree.c: Likewise.
31199 * stor-layout.c: Likewise.
31200 * tree-nested.c: Likewise.
31202 * dumpfile.c (dump_start): Use TDF_FLAGS.
31203 (dump_enable_all): Fix TDF_KIND check thinko.
31205 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
31207 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31208 array entries to represent two legal parameterizations of the
31209 overloaded __builtin_cmpb function, as represented by the
31210 P6_OV_BUILTIN_CMPB constant.
31211 (altivec_resolve_overloaded_builtin): Add special case handling
31212 for the __builtin_cmpb function, as represented by the
31213 P6_OV_BUILTIN_CMPB constant.
31214 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
31215 (BU_P6_64BIT_2): New macro.
31216 (BU_P6_OVERLOAD_2): New macro
31217 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
31218 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
31219 (CMPB): Add overload support to represent both 32-bit and 64-bit
31220 compare-bytes function.
31221 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
31222 support for TARGET_CMPB.
31223 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
31224 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31225 documentation of the __builtin_cmpb overloaded built-in function.
31227 2017-05-11 Richard Biener <rguenther@suse.de>
31229 PR tree-optimization/80705
31230 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
31231 bases are not vectorizable.
31233 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31235 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
31236 when counting register pressure.
31238 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31240 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
31241 (iv_ca_more_deps): Renamed to ...
31242 (iv_ca_compare_deps): ... this.
31243 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
31245 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31247 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
31249 (determine_group_iv_costs): ... here.
31250 (find_inv_vars_cb): Record inv var if it's not recorded before.
31252 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31254 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
31255 (get_shiftadd_cost): Ditto.
31257 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31259 * tree-ssa-address.c: Include header file.
31260 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
31262 (add_to_parts): Refactor.
31263 (addr_to_parts): New parameter. Update use of move_hint_to_base.
31264 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
31267 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31269 PR tree-optimization/53090
31270 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
31272 (extract_cond_operands): Detect condition with IV on both sides
31273 and return COMP_IV_EXPR_2.
31274 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
31275 (rewrite_use_compare): Simplify by removing call to function
31276 extract_cond_operands.
31278 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31280 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
31281 (extract_cond_operands): Detect condition comparing against non-
31282 invariant bound and return appropriate enum value.
31283 (find_interesting_uses_cond): Update use of extract_cond_operands.
31284 Handle its return value accordingly.
31285 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
31287 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31289 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
31290 nonlinear iv_use computation in loop invariant sensitive way.
31292 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31294 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
31295 (find_iv_candidates): Call relate_compare_use_with_all_cands.
31297 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31299 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
31300 (dump_cand): Support iv_cand.inv_exprs.
31301 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
31303 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
31306 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31308 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
31310 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
31311 as local function. Include necessary header files.
31312 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
31314 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31316 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
31318 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31320 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
31321 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
31322 RSHIFT_EXPR and BIT_NOT_EXPR.
31324 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31326 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
31327 (adjust_setup_cost): New parameter supporting round up adjustment.
31328 (struct address_cost_data): Delete.
31329 (force_expr_to_var_cost): Don't bound cost with spill_cost.
31330 (split_address_cost, ptr_difference_cost): Delete.
31331 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
31332 (struct ainc_cost_data): New struct.
31333 (get_address_cost_ainc): New function.
31334 (get_address_cost, get_computation_cost): Reimplement.
31335 (determine_group_iv_cost_address): Record inv_expr for all uses of
31337 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
31338 (iv_ca_has_deps): Reimplemented to ...
31339 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
31341 (iv_ca_extend): Call iv_ca_more_deps.
31343 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31345 * tree-ssa-address.c (struct mem_address): Move to header file.
31346 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
31347 * tree-ssa-address.h (struct mem_address): Move from C file.
31348 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
31350 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31352 * tree-affine.h (aff_combination_type): New interface.
31353 (aff_combination_zero_p): Remove static.
31354 (aff_combination_const_p): New interface.
31355 (aff_combination_singleton_var_p): New interfaces.
31357 2017-05-11 Richard Biener <rguenther@suse.de>
31359 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31360 Skip unreachable blocks and destinations.
31361 (eliminate): Move stmt removal and fixup ...
31362 (fini_eliminate): ... here. Skip inserted exprs.
31363 (pass_pre::execute): Move fini_pre after fini_eliminate.
31364 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
31365 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
31366 PRE to get rid of dead code that has invalid SSA form and
31367 split critical edges again.
31369 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31371 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31373 2017-05-11 Richard Biener <rguenther@suse.de>
31375 * passes.c (execute_function_todo): Verify loops if they are
31376 said to be up-to-date.
31377 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
31378 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
31380 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
31383 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
31384 handle calling assemble_external ourself.
31387 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
31388 modes with zero size. Enhance comment.
31390 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31392 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
31393 built-ins for vec_xl and vec_xst with short and char pointer
31396 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31398 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
31399 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
31400 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
31401 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
31402 (_mm_maskz_min_round_ss): New intrinsics.
31403 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
31404 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
31405 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
31406 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
31407 (__builtin_ia32_minss_mask_round): New builtins.
31408 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31409 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31410 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
31412 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
31413 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
31415 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
31418 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31420 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
31421 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
31422 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
31423 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
31424 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
31425 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
31426 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
31427 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31428 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31429 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
31430 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
31431 (__builtin_ia32_mulss_mask_round): New builtins.
31432 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31433 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31434 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
31436 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
31437 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31439 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31442 2017-05-10 Julia Koval <julia.koval@intel.com>
31444 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
31445 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
31446 (_mm256_setr_m128i): New intrinsics.
31448 2017-05-10 Julia Koval <julia.koval@intel.com>
31450 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
31451 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
31452 (_mm_maskz_rcp14_ss): New intrinsics.
31453 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
31454 (__builtin_ia32_rcp14ss_mask): New builtins.
31455 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
31457 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
31459 PR tree-optimization/51513
31460 * tree-cfg.c (gimple_seq_unreachable_p): New function.
31461 (assert_unreachable_fallthru_edge_p): Use it.
31462 (group_case_labels_stmt): Likewise.
31463 * tree-cfg.h: Prototype it.
31464 * stmt.c: Include cfghooks.h and tree-cfg.h.
31465 (emit_case_dispatch_table) <gap_label>: New local variable.
31466 Use it to fill dispatch table gaps.
31467 Test for default_label before updating probabilities.
31468 (expand_case) <default_label>: Remove unneeded initialization.
31469 Test for unreachable default case statement and remove its edge.
31470 Set default_label accordingly.
31471 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
31473 2017-05-10 Carl Love <cel@us.ibm.com>
31475 * config/rs6000/rs6000-c: Add support for built-in functions
31476 vector signed char vec_neg (vector signed char)
31477 vector signed short int vec_neg (vector short int)
31478 vector signed int vec_neg (vector signed int)
31479 vector signed long long vec_neg (vector signed long long)
31480 vector float vec_neg (vector float)
31481 vector double vec_neg (vector double)
31482 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
31484 * config/rs6000/altivec.h: Add define for vec_neg
31485 * doc/extend.texi: Update the built-in documentation for the
31486 new built-in functions.
31488 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31490 PR tree-optimization/77644
31491 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
31493 2017-05-10 Nathan Sidwell <nathan@acm.org>
31495 * dumpfile.h (TDI_lang_all): New.
31496 (TDF_KIND): New. Renumber others
31497 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
31499 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
31501 (get_dump_file_name): Adjust suffix generation.
31502 (dump_enable_all): Use TDF_KIND.
31503 * doc/invoke.texi (-fdump-lang-all): Document.
31505 * dumpfile.h: Tabify.
31507 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
31510 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
31511 Move member access before delete.
31513 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
31515 * tree-inline.c (expand_call_inline): Split block at stmt
31518 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
31521 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
31522 are now unused after splitting mov{sf,sd}_hardfloat.
31523 (f32_lr2): Likewise.
31524 (f32_lm): Likewise.
31525 (f32_lm2): Likewise.
31526 (f32_li): Likewise.
31527 (f32_li2): Likewise.
31528 (f32_lv): Likewise.
31529 (f32_sr): Likewise.
31530 (f32_sr2): Likewise.
31531 (f32_sm): Likewise.
31532 (f32_sm2): Likewise.
31533 (f32_si): Likewise.
31534 (f32_si2): Likewise.
31535 (f32_sv): Likewise.
31536 (f32_dm): Likewise.
31537 (f32_vsx): Likewise.
31538 (f32_av): Likewise.
31539 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
31540 For movsf, order stores so the VSX stores occur before the GPR
31541 store which encourages the register allocator to use a traditional
31542 FPR instead of a GPR. For movsd, order the stores so that the GPR
31543 store comes before the VSX stores to allow the power6 to work.
31544 This is due to the power6 not having a 32-bit integer store
31545 instruction from a FPR.
31546 (movsf_hardfloat): Likewise.
31547 (movsd_hardfloat): Likewise.
31549 2017-05-09 Martin Sebor <msebor@redhat.com>
31551 PR translation/80280
31552 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
31555 PR translation/80280
31556 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
31557 data member added in r247778.
31558 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
31560 2017-05-09 Nathan Sidwell <nathan@acm.org>
31562 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
31564 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31567 2017-05-09 Marek Polacek <polacek@redhat.com>
31569 * doc/invoke.texi: Fix typo.
31571 2017-05-09 Richard Biener <rguenther@suse.de>
31573 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31574 (vrp_val_is_min): Likewise.
31575 (set_value_range_to_value): Likewise.
31576 (set_value_range_to_nonnegative): Likewise.
31577 (gimple_assign_nonzero_p): Likewise.
31578 (gimple_stmt_nonzero_p): Likewise.
31579 (vrp_int_const_binop): Likewise. Remove unreachable case.
31580 (adjust_range_with_scev): Adjust comments.
31581 (compare_range_with_value): Likewise.
31582 (extract_range_from_phi_node): Likewise.
31583 (test_for_singularity): Likewise.
31585 2017-05-09 Richard Biener <rguenther@suse.de>
31587 * tree-vrp.c (get_single_symbol): Add assert that we don't
31588 get overflowed constants as invariant part.
31589 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31590 checks. Use wi::cmp instead of recursing for integer constants.
31591 (compare_values): Just ignore whether we assumed undefined
31592 overflow instead of failing the compare.
31593 (extract_range_for_var_from_comparison_expr): Add comment before the
31594 TREE_NO_WARNING sets.
31595 (test_for_singularity): Likewise.
31596 (extract_range_from_comparison): Do not disable optimization
31597 when we assumed undefined overflow.
31598 (extract_range_basic): Remove init of unused var.
31600 2017-05-09 Richard Biener <rguenther@suse.de>
31602 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31603 (extract_range_from_multiplicative_op_1): Adjust.
31604 (extract_range_from_binary_expr_1): Use int_const_binop.
31606 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31609 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31610 rs6000_store_data_bypass_p in seven define_bypass directives and
31611 in several comments.
31612 * config/rs6000/rs6000-protos.h: Add prototype for
31613 rs6000_store_data_bypass_p function.
31614 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31615 function implements slightly different (rs6000-specific) semantics
31616 than store_data_bypass_p, returning false rather than aborting
31617 with assertion error when arguments do not satisfy the
31618 requirements of store data bypass.
31619 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31620 rs6000_store_data_bypass_p.
31622 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31624 * config/xtensa/xtensa-protos.h
31625 (xtensa_initial_elimination_offset): New declaration.
31626 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31627 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31628 macro definition, add case for FRAME_POINTER_REGNUM when
31629 FRAME_GROWS_DOWNWARD.
31630 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31631 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31632 xtensa_initial_elimination_offset.
31634 2017-05-08 Nathan Sidwell <nathan@acm.org>
31636 * doc/invoke.texi: Alphabetize -fdump options.
31638 2017-05-08 Martin Sebor <msebor@redhat.com>
31640 PR translation/80280
31641 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31643 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31645 * target.def (compute_frame_layout): New optional target hook.
31646 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31647 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31648 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31650 * reload1.c (verify_initial_elim_offsets): Likewise.
31651 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31652 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31653 (arm_get_frame_offsets): Split up into this ...
31654 (arm_compute_frame_layout): ... and this function.
31656 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31658 * config/aarch64/constraints.md (Usa): New constraint.
31659 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31661 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31663 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31664 with_multilib_list after it has been checked.
31666 2017-05-08 Richard Biener <rguenther@suse.de>
31668 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31669 (bitmap_set_subtract_values): Likewise.
31671 2017-05-08 Richard Biener <rguenther@suse.de>
31673 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31674 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31676 (gimple_stmt_nonzero_warnv_p): Rename to ...
31677 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31679 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31680 (extract_range_basic): Adjust, do not disable propagation on
31681 strict overflow sensitive simplification.
31682 (vrp_visit_cond_stmt): Likewise.
31684 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31686 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31687 body size unconditionally.
31689 2017-05-07 Jeff Law <law@redhat.com>
31692 2017-05-06 Jeff Law <law@redhat.com>
31693 PR tree-optimization/78496
31694 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31697 PR tree-optimization/78496
31698 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31699 (simplify_stmt_using_ranges): Call it.
31700 (vrp_dom_walker::before_dom_children): Extract equivalences
31701 from an ASSERT_EXPR with an equality comparison against a
31704 2017-05-06 Jeff Law <law@redhat.com>
31706 PR tree-optimization/78496
31707 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31710 PR tree-optimization/78496
31711 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31712 (simplify_stmt_using_ranges): Call it.
31713 (vrp_dom_walker::before_dom_children): Extract equivalences
31714 from an ASSERT_EXPR with an equality comparison against a
31717 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31719 * lra-constraints.c (lra_copy_reg_equiv): New function.
31720 (split_reg): Use it to copy equivalence information from the
31721 original register to the spill register.
31723 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31725 PR rtl-optimization/75964
31726 * simplify-rtx.c (simplify_const_relational_operation): Remove
31727 invalid handling of comparisons of integer ABS.
31729 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31731 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31732 initialize to zero.
31733 (init_regs): Remove declaration.
31734 (function_arg_advance_32): Initialize error_p as boolean variable.
31736 2017-05-05 Nathan Sidwell <nathan@acm.org>
31738 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31739 lines. Use for (;;).
31741 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31743 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31744 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31745 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31746 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31747 VF=2 that require versioning.
31749 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31751 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31754 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31756 * diagnostic.h (diagnostic_override_option_index): Convert from
31757 macro to inline function.
31759 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31761 * diagnostic.c (last_module_changed_p): New function.
31762 (set_last_module): New function.
31763 (diagnostic_report_current_module): Convert macro usage to
31764 the above functions.
31765 * diagnostic.h (diagnostic_context::last_module): Strengthen
31766 from const line_map * to const line_map_ordinary *.
31767 (diagnostic_last_module_changed): Delete macro.
31768 (diagnostic_set_last_module): Delete macro.
31770 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31772 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31773 with diagnostic_report_diagnostic.
31774 (diagnostic_n_impl_richloc): Likewise.
31775 * diagnostic.h (report_diagnostic): Delete macro.
31776 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31777 with diagnostic_report_diagnostic.
31778 * substring-locations.c (format_warning_va): Likewise.
31780 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31782 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31783 save/restor of format_spec. Move option-printing code to...
31784 (print_option_information): ...this new function, and
31785 reimplement by simply printing to the pretty_printer,
31786 rather than appending to the format string.
31788 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31790 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31791 handling logic into...
31792 (update_effective_level_from_pragmas): ...this new function.
31794 2017-05-04 Andrew Waterman <andrew@sifive.com>
31796 * config/riscv/riscv.opt (mstrict-align): New option.
31797 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31798 (SLOW_UNALIGNED_ACCESS): Define.
31799 (riscv_slow_unaligned_access): Declare.
31800 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31802 (riscv_slow_unaligned_access): New variable.
31803 (rocket_tune_info): Set slow_unaligned_access to true.
31804 (optimize_size_tune_info): Set slow_unaligned_access to false.
31805 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31806 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31807 (riscv_option_override): Set riscv_slow_unaligned_access.
31808 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31810 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
31812 * config/riscv/riscv.md: Unify indentation.
31814 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
31819 * config/rs6000/rs6000.md (u code attribute): Add FIX and
31821 (extendsi<mode>2): Add support for doing sign extension via
31822 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
31823 don't have ISA 3.0 instructions.
31824 (extendsi<mode>2 splitter): Likewise.
31825 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
31826 generate the normal insns since SImode can now go in vector
31827 registers. Disallow the special UNSPECs needed for previous
31828 machines to hide SImode being used. Add new insns
31829 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
31830 (fix_trunc<mode>si2_stfiwx): Likewise.
31831 (fix_trunc<mode>si2_internal): Likewise.
31832 (fixuns_trunc<mode>si2): Likewise.
31833 (fixuns_trunc<mode>si2_stfiwx): Likewise.
31834 (fctiw<u>z_<mode>_smallint): Likewise.
31835 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
31836 of floating point to 32-bit integer from doing a direct move to
31837 the GPR registers to do a store.
31838 (fctiwz_<mode>): Break long line.
31840 2017-05-05 Bin Cheng <bin.cheng@arm.com>
31842 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
31843 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
31844 (addr_list, addr_offset_valid_p): New.
31845 (split_address_groups): Check offset validity with above function.
31846 (gt-tree-ssa-loop-ivopts.h): Include header file.
31848 2017-05-05 Nathan Sidwell <nathan@acm.org>
31850 * config.gcc (arm*-*-*): Add missing 'fi'.
31852 2017-05-05 Steve Ellcey <sellcey@cavium.com>
31854 * doc/invoke.texi (-fopt-info): Explicitly say order of options
31855 included in -fopt-info does not matter.
31856 * doc/optinfo.texi (-fopt-info): Fix description of default
31857 behavour. Explicitly say order of options included in -fopt-info
31860 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
31862 * config.gcc: Allow combinations of aprofile and rmprofile values for
31863 --with-multilib-list.
31864 * config/arm/t-multilib: New file.
31865 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
31866 variables. Remove setting of ISA and floating-point ABI in
31867 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
31868 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
31869 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
31870 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
31872 * config/arm/t-rmprofile: Likewise except for the matches changes.
31873 * doc/install.texi (--with-multilib-list): Document the combination of
31874 aprofile and rmprofile values and warn about pitfalls in doing that.
31876 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31878 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
31879 (movdi_aarch64): Likewise.
31881 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31883 PR tree-optimization/80632
31884 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
31886 (build_arrays): Initialize it for virtual phis.
31887 (fix_phi_nodes): Use it for virtual phis.
31889 PR tree-optimization/80558
31890 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
31891 [x, y] op z into [x op, y op z] for op & or | if conditions
31894 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
31895 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31898 * config/arm/arm.md (use_literal_pool): Remove.
31899 (64-bit immediate split): No longer takes cost into consideration
31900 if arm_disable_literal_pool is enabled.
31901 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
31902 used when arm_disable_literal_pool is enabled.
31903 (arm_max_const_double_inline_cost): Remove use of
31904 arm_disable_literal_pool.
31905 (push_minipool_fix): Add assert.
31906 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
31907 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
31908 (no_literal_pool_sf_immediate): New.
31910 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31912 PR tree-optimization/80613
31913 * tree-ssa-dce.c (propagate_necessity): Remove cases for
31914 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
31916 2017-05-05 Richard Biener <rguenther@suse.de>
31918 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
31920 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
31922 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
31923 of this flag from insn conditions due to removal from r247495.
31925 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31927 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
31929 (arm_early_store_addr_dep_ptr): Likewise.
31930 * config/arm/aarch-common-protos.h
31931 (arm_early_load_addr_dep_ptr): Add prototype.
31932 (arm_early_store_addr_dep_ptr): Likewise.
31933 * config/arm/cortex-a53.md: Add new bypasses.
31935 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31937 * tree.c (next_type_uid): Change type to unsigned.
31938 (type_hash_canon): Decrement back next_type_uid if
31939 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
31940 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
31943 2017-05-04 Martin Sebor <msebor@redhat.com>
31945 * builtins.c: Fix a trivial typo in a comment.
31947 PR middle-end/79234
31948 * builtins.c (check_sizes): Adjust to handle reading past the end.
31949 Avoid printing excessive upper bound of ranges. Use %E to print
31950 tree nodes instead of converting them to %wu.
31951 (expand_builtin_memchr): New function.
31952 (compute_dest_size): Rename...
31953 (compute_objsize): ...to this.
31954 (expand_builtin_memcpy): Adjust.
31955 (expand_builtin_mempcpy): Adjust.
31956 (expand_builtin_strcat): Adjust.
31957 (expand_builtin_strcpy): Adjust.
31958 (check_strncat_sizes): Adjust.
31959 (expand_builtin_strncat): Adjust.
31960 (expand_builtin_strncpy): Adjust and simplify.
31961 (expand_builtin_memset): Adjust.
31962 (expand_builtin_bzero): Adjust.
31963 (expand_builtin_memcmp): Adjust.
31964 (expand_builtin): Handle memcmp.
31965 (maybe_emit_chk_warning): Check strncat just once.
31967 2017-05-04 Martin Sebor <msebor@redhat.com>
31969 PR preprocessor/79214
31970 PR middle-end/79222
31971 PR middle-end/79223
31972 * builtins.c (check_sizes): Add inlining context and issue
31973 warnings even when -Wno-system-headers is set.
31974 (check_strncat_sizes): Same.
31975 (expand_builtin_strncat): Same.
31976 (expand_builtin_memmove): New function.
31977 (expand_builtin_stpncpy): Same.
31978 (expand_builtin): Handle memmove and stpncpy.
31980 2017-05-04 Bin Cheng <bin.cheng@arm.com>
31982 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
31983 which is not used any more.
31985 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31987 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
31989 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31991 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
31992 (cortexa53_tunings): Likewise.
31993 (cortexa57_tunings): Likewise.
31994 (cortexa72_tunings): Likewise.
31995 (cortexa73_tunings): Likewise.
31997 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31999 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32000 Set loop alignment to 8.
32002 2017-05-04 Martin Sebor <msebor@redhat.com>
32004 PR translation/80280
32005 * builtins.c (expand_builtin_object_size): Add missing quoting to
32006 %D and like directives.
32007 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32008 (hsa_type_for_tree_type): Same.
32009 (verify_function_arguments): Same.
32010 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32011 * varasm.c (get_section): Same.
32014 2017-05-04 Martin Sebor <msebor@redhat.com>
32016 PR translation/80280
32017 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32019 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32021 * config/aarch64/aarch64.c (generic_addrcost_table):
32022 Change HI/TI mode setting.
32024 2017-05-04 Martin Jambor <mjambor@suse.cz>
32026 PR tree-optimization/80622
32027 * tree-sra.c (comes_initialized_p): New function.
32028 (build_accesses_from_assign): Only set write lazily when
32029 comes_initialized_p is false.
32030 (analyze_access_subtree): Use comes_initialized_p.
32031 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32032 instead of testing for PARM_DECL.
32034 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32036 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32037 constraint on operand 0 to allow more general addressing modes.
32038 Adjust output template.
32039 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32041 * config/aarch64/aarch64-protos.h
32042 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32043 * config/aarch64/constraints.md (Dp): New address constraint.
32044 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32047 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32049 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32050 update use of estimate_ipcp_clone_size_and_time.
32051 (estimate_local_effects): Update use of
32052 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32053 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
32054 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
32055 Return nonspecialized time.
32057 2017-05-04 Richard Biener <rguenther@suse.de>
32059 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
32060 for the last VUSE which def dominates the PHI. Directly call
32062 (get_continuation_for_phi_1): Remove.
32064 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
32066 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
32067 to explain the use of truncating division. Cap the number of
32068 iterations to the maximum given by nb_iterations_upper_bound,
32071 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32073 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
32074 * configure: Regenerate.
32075 * config.in: Regenerate.
32076 * config/i386/driver-mingw32.c: new file.
32077 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
32078 * config.host: Link driver-mingw32.o on MinGW host.
32079 * doc/install.texi: Document new --enable-mingw-wildcard configure
32082 2017-05-04 Marek Polacek <polacek@redhat.com>
32084 PR tree-optimization/80612
32085 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
32087 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32088 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
32090 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
32091 (movt splitter): Likewise.
32092 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
32093 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
32094 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
32095 block for Thumb-1 with MOVT.
32096 (thumb2_legitimate_address_p): Move code block ...
32097 (can_avoid_literal_pool_for_label_p): ... into this new function.
32098 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
32100 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
32101 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
32102 "M-profile targets with the MOVT instruction".
32104 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32106 * config/arm/arm-builtins.c (arm_init_builtins): Rename
32107 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
32108 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
32110 2017-05-04 Martin Liska <mliska@suse.cz>
32112 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
32113 variable cond_code.
32115 2017-05-04 Richard Biener <rguenther@suse.de>
32117 * tree.c (array_at_struct_end_p): Handle arrays at struct
32118 end with flexarrays more conservatively. Refactor and treat
32119 arrays of arrays or aggregates more strict. Fix
32120 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
32121 * tree.c (array_at_struct_end_p): Adjust prototype.
32122 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
32123 * gimple-fold.c (get_range_strlen): Likewise.
32124 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
32126 2017-05-04 Richard Biener <rguenther@suse.de>
32128 PR tree-optimization/31130
32129 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
32131 (supports_overflow_infinity): Likewise.
32132 (is_negative_overflow_infinity): Likewise.
32133 (is_positive_overflow_infinity): Likewise.
32134 (is_overflow_infinity): Likewise.
32135 (stmt_overflow_infinity): Likewise.
32136 (overflow_infinity_range_p): Likewise.
32137 (usable_range_p): Remove as always returning true.
32138 (make_overflow_infinity): Remove.
32139 (negative_overflow_infinity): Likewise.
32140 (positive_overflow_infinity): Likewise.
32141 (avoid_overflow_infinity): Likewise.
32142 (set_value_range): Adjust accordingly.
32143 (set_value_range_to_nonnegative): Likewise, remove now unused
32144 overflow_infinity arg.
32145 (vrp_operand_equal_p): Adjust.
32146 (update_value_range): Likewise.
32147 (range_int_cst_singleton_p): Likewise.
32148 (operand_less_p): Likewise.
32149 (compare_values_warnv): Likewise.
32150 (extract_range_for_var_from_comparison_expr): Likewise.
32151 (vrp_int_const_binop): Likewise.
32152 (zero_nonzero_bits_from_vr): Likewise.
32153 (extract_range_from_multiplicative_op_1): Likewise.
32154 (extract_range_from_binary_expr_1): Likewise.
32155 (extract_range_from_unary_expr): Likewise.
32156 (extract_range_from_comparison): Likewise.
32157 (extract_range_basic): Likewise.
32158 (adjust_range_with_scev): Likewise.
32159 (compare_ranges): Likewise.
32160 (compare_range_with_value): Likewise.
32161 (dump_value_range): Likewise.
32162 (test_for_singularity): Likewise, remove strict_overflow_p parameter
32164 (simplify_cond_using_ranges): Adjust.
32166 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32168 * brig-builtins.def: Added a builtin for class_f64.
32169 * builtin-types.def: Added a builtin type needed by class_f64.
32171 2017-05-03 Jason Merrill <jason@redhat.com>
32173 * timevar.def: Add TV_CONSTEXPR.
32175 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32177 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
32179 2017-05-03 Martin Jambor <mjambor@suse.cz>
32181 * ipa-prop.c (ipa_update_after_lto_read): Removed.
32182 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
32183 * ipa-cp.c (ipcp_propagate_stage): Do not call
32184 ipa_update_after_lto_read.
32185 * ipa-inline.c (ipa_inline): Likewise.
32187 2017-05-03 Martin Jambor <mjambor@suse.cz>
32189 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
32190 tag. Added a default constructor and a destructor.
32191 (ipa_edge_args_sum_t): New class;
32192 (ipa_edge_args_sum): Declare.
32193 (ipa_edge_args_vector): Remove declaration.
32194 (IPA_EDGE_REF): Use ipa_edge_args_sum.
32195 (ipa_free_edge_args_substructures): Remove declaration.
32196 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
32197 (ipa_edge_args_info_available_for_edge_p): Likewise.
32198 * ipa-prop.c (ipa_edge_args_vector): Removed.
32199 (edge_removal_hook_holder): Likewise.
32200 (edge_duplication_hook_holder): Likewise.
32201 (ipa_edge_args_sum): New variable.
32202 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
32203 ipa_edge_args_vector.
32204 (ipa_free_edge_args_substructures): Likewise.
32205 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
32206 ipa_edge_args_vector.
32207 (ipa_edge_removal_hook): Turned into method
32208 ipa_edge_args_sum_t::remove.
32209 (ipa_edge_duplication_hook): Turned into method
32210 ipa_edge_args_sum_t::duplicate.
32211 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
32212 registering edge hooks.
32213 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
32214 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
32215 ipa_edge_args_sum instead of ipa_edge_args_vector.
32216 * ipa-profile.c (ipa_profile): Likewise.
32218 2017-05-03 Martin Jambor <mjambor@suse.cz>
32220 * symbol-summary.h (function_summary): New method exists.
32221 (function_summary::symtab_removal): Deallocate through release.
32222 (call_summary): New class.
32223 (gt_ggc_mx): New overload.
32224 (gt_pch_nx): Likewise.
32225 (gt_pch_nx): Likewise.
32227 2017-05-03 Jeff Law <law@redhat.com>
32229 PR tree-optimization/78496
32230 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
32231 from simplify_cond_using_ranges. Split off code to walk
32232 backwards through casts into ...
32233 (simplify_cond_using_ranges_2): New function.
32234 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
32235 (execute_vrp): After identifying jump threads, call
32236 simplify_cond_using_ranges_2.
32238 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
32241 * ipa-inline.h (inline_summary): Add ctor.
32242 (create_ggc): Do not use ggc_cleared_alloc.
32244 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
32245 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32247 * gcc.c (handle_braces): Support escaping in switch matching
32249 * doc/invoke.texi (Spec Files): Document it.
32250 Remove superfluous @code markup in items.
32252 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32254 * diagnostic-show-locus.c (struct column_range): New struct.
32255 (get_affected_columns): New function.
32256 (get_printed_columns): New function.
32257 (struct correction): New struct.
32258 (correction::ensure_capacity): New function.
32259 (correction::ensure_terminated): New function.
32260 (struct line_corrections): New struct.
32261 (line_corrections::~line_corrections): New dtor.
32262 (line_corrections::add_hint): New function.
32263 (layout::print_trailing_fixits): Reimplement in terms of the new
32265 (selftest::test_overlapped_fixit_printing): New function.
32266 (selftest::diagnostic_show_locus_c_tests): Call it.
32268 2017-05-03 Nathan Sidwell <nathan@acm.org>
32270 Canonicalize canonical type hashing
32271 * tree.h (type_hash_canon_hash): Declare.
32272 * tree.c (type_hash_list, attribute_hash_list): Move into
32273 type_hash_canon_hash.
32274 (build_type_attribute_qual_variant): Break out hash code calc into
32275 type_hash_canon_hash.
32276 (type_hash_canon_hash): New. Generic type hash computation.
32277 (build_range_type_1, build_array_type_1, build_function_type,
32278 build_method_type_directly, build_offset_type, build_complex_type,
32279 make_vector_type): Call it.
32281 2017-05-03 Richard Biener <rguenther@suse.de>
32283 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
32284 When all DRs have unknown misaligned do not always peel
32285 when there is a store but apply the same costing model as if
32286 there were only loads.
32288 2017-05-03 Richard Biener <rguenther@suse.de>
32291 PR tree-optimization/80492
32292 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32293 compare_base_decls returning dont-know properly.
32295 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
32297 * config/arm/iterators.md (CCSI): New mode iterator.
32298 (arch): New mode attribute.
32299 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
32300 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
32301 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
32302 code iterator for success result mode.
32303 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
32304 the corresponding new insn generators.
32306 2017-05-03 Bin Cheng <bin.cheng@arm.com>
32309 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32310 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32312 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
32314 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
32315 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
32316 (DDR_A): Wrap DDR argument in brackets.
32317 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
32318 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
32319 (DDR_REVERSED_P): Likewise.
32321 2017-05-03 Jakub Jelinek <jakub@redhat.com>
32323 PR tree-optimization/79472
32324 * tree-switch-conversion.c (struct switch_conv_info): Add
32325 contiguous_range and default_case_nonstandard fields.
32326 (collect_switch_conv_info): Compute contiguous_range and
32327 default_case_nonstandard fields, don't clear final_bb if
32328 contiguous_range and only the default case doesn't have the required
32330 (check_all_empty_except_final): Set default_case_nonstandard instead
32331 of failing if contiguous_range and the default case doesn't have empty
32333 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
32334 and only the default case doesn't have the required constants. Skip
32336 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
32337 if default_case_nonstandard.
32338 (build_constructors): Build constant 1 just once. Assert that default
32339 values aren't inserted in between cases if contiguous_range. Skip
32341 (build_arrays): Skip virtual phis.
32342 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
32343 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
32344 Handle virtual phis.
32345 (gen_inbound_check): Handle default_case_nonstandard case.
32346 (process_switch): Adjust check_final_bb caller. Call
32347 gather_default_values with the first non-default case instead of
32348 default case if default_case_nonstandard.
32350 2017-05-02 Nathan Sidwell <nathan@acm.org>
32352 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
32353 check. Fix formatting.
32355 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
32357 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
32358 errors when comparing specialized and unspecialized times.
32360 2017-05-02 David Malcolm <dmalcolm@redhat.com>
32362 * diagnostic-show-locus.c
32363 (layout::should_print_annotation_line_p): Make private.
32364 (layout::print_annotation_line): Make private.
32365 (layout::annotation_line_showed_range_p): Make private.
32366 (layout::show_ruler): Make private.
32367 (layout::print_source_line): Make private. Pass in line and
32368 line_width, rather than calling location_get_source_line. Drop
32370 (layout::print_leading_fixits): New method.
32371 (layout::print_any_fixits): Rename to...
32372 (layout::print_trailing_fixits): ...this, and make private.
32373 Don't print newline fixits.
32374 (diagnostic_show_locus): Move logic for printing one row into...
32375 (layout::print_line): ...this new function. Move the
32376 location_get_source_line call and error-handling from
32377 print_source_line to here. Call print_leading_fixits, and rename
32378 print_any_fixits to print_trailing_fixits.
32379 (selftest::test_fixit_insert_containing_newline): Update now that
32380 newlines are partially supported.
32381 (selftest::test_fixit_insert_containing_newline_2): New test.
32382 (selftest::test_fixit_replace_containing_newline): Update comments.
32383 (selftest::diagnostic_show_locus_c_tests): Call the new test.
32384 * edit-context.c (class added_line): New class.
32385 (class edited_line): Describe newline handling in comment.
32386 (edited_line::actually_edited_p): New method.
32387 (edited_line::print_content): Delete redundant decl.
32388 (edited_line::m_predecessors): New field.
32389 (edited_file::print_content): Call edited_line::print_content.
32390 (edited_file::print_diff): Update to support newlines.
32391 (edited_file::print_diff_hunk): Likewise.
32392 (edited_file::print_run_of_changed_lines): New function.
32393 (edited_file::print_diff_line): Convert to...
32394 (print_diff_line): ...this.
32395 (edited_file::get_effective_line_count): New function.
32396 (edited_line::edited_line): Initialize new field m_predecessors.
32397 (edited_line::~edited_line): Clean up m_predecessors.
32398 (edited_line::apply_fixit): Handle newlines.
32399 (edited_line::get_effective_line_count): New function.
32400 (edited_line::print_content): New function.
32401 (edited_line::print_diff_lines): New function.
32402 (selftest::test_applying_fixits_insert_containing_newline): New
32404 (selftest::test_applying_fixits_replace_containing_newline): New
32406 (selftest::insert_line): New function.
32407 (selftest::test_applying_fixits_multiple_lines): Add example of
32409 (selftest::edit_context_c_tests): Call the new tests.
32411 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32413 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
32414 parameter cand. Update dump information.
32415 (get_computation_cost): Update uses.
32417 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32419 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
32420 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
32421 (get_computation_at, rewrite_use_address): Update use of
32422 get_computation_aff.
32424 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32426 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
32427 (get_computation): Delete.
32428 (get_computation_cost): Implement like get_computation_cost_at.
32429 Use get_computation_at.
32430 (get_computation_cost_at): Delete.
32431 (rewrite_use_nonlinear_expr): Use get_computation_at.
32432 (rewrite_use_compare, remove_unused_ivs): Ditto.
32434 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32436 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
32438 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32440 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
32441 (ivopts_global_cost_for_size): Rename parameter and update uses.
32442 (iv_ca_recount_cost): Update uses.
32443 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
32444 candidates seperately in n_invs and n_cands.
32445 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
32447 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32449 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
32450 (find_inv_vars_cb): New.
32451 (find_depends): Renamed to ...
32452 (find_inv_vars): ... this.
32453 (add_candidate_1, force_var_cost): Call find_inv_vars.
32454 (split_address_cost, determine_group_iv_cost_cond): Ditto.
32456 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32458 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
32459 inv_vars. Add inv_exprs.
32460 (struct iv_cand): Rename depends_on to inv_vars.
32461 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
32462 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
32463 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
32464 (dump_cand): Dump inv_vars.
32465 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
32466 (record_invariant, find_depends, add_candidate_1): Ditto.
32467 (set_group_iv_cost, force_var_cost): Ditto.
32468 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
32469 (get_computation_cost_at, get_computation_cost): Ditto.
32470 (determine_group_iv_cost_generic): Ditto.
32471 (determine_group_iv_cost_address): Ditto.
32472 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
32473 (determine_group_iv_costs): Ditto.
32474 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
32475 (iv_ca_set_remove_invariants): Renamed to ...
32476 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
32477 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
32478 (iv_ca_set_add_invariants): Renamed to ...
32479 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
32480 (iv_ca_set_cp): Use iv_ca_set_add_invs.
32481 (iv_ca_has_deps): Support inv_vars and inv_exprs.
32482 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
32483 (create_new_ivs): Remove useless dump.
32485 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32487 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
32489 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
32490 (iv_ca_set_no_cp, create_new_iv): Ditto.
32492 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32494 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32496 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32498 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
32499 function tree_check2.
32501 2017-05-02 Martin Liska <mliska@suse.cz>
32503 * doc/gcov.texi: Add missing preposition.
32504 * gcov.c (function_info::function_info): Properly fill up
32505 all member variables.
32507 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32509 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
32511 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32513 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
32515 2017-05-02 Martin Liska <mliska@suse.cz>
32518 * lto-streamer-in.c (lto_read_tree_1): Remove
32519 LTO_STREAMER_DEBUG.
32520 * lto-streamer.c (struct tree_hash_entry): Likewise.
32521 (struct tree_entry_hasher): Likewise.
32522 (tree_entry_hasher::hash): Likewise.
32523 (tree_entry_hasher::equal): Likewise.
32524 (lto_streamer_init): Likewise.
32525 (lto_orig_address_map): Likewise.
32526 (lto_orig_address_get): Likewise.
32527 (lto_orig_address_remove): Likewise.
32528 * lto-streamer.h: Likewise.
32529 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
32530 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
32532 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
32534 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
32535 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
32536 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
32537 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
32538 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
32539 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
32540 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
32541 (mm_maskz_sub_ss): New intrinsics.
32542 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32543 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32544 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
32545 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
32546 (__builtin_ia32_subss_mask_round): New builtins.
32547 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32548 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32549 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
32551 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
32552 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32554 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32557 2017-05-02 Martin Jambor <mjambor@suse.cz>
32559 PR tree-optimization/78687
32560 * tree-sra.c (access): New field parent.
32561 (process_subtree_disqualification): New function.
32562 (disqualify_candidate): Call it.
32563 (build_accesses_from_assign): Reset write flag if creating an
32565 (build_access_subtree): Fill in parent field and also prpagate
32566 down grp_write flag.
32567 (create_artificial_child_access): New parameter set_grp_write, set
32568 grp_write to its value.
32569 (propagate_subaccesses_across_link): Also propagate grp_write flag
32571 (propagate_all_subaccesses): Push the closest parent back to work
32572 queue if add_access_to_work_queue returned true.
32574 2017-05-02 Richard Biener <rguenther@suse.de>
32576 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32577 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32578 -fstrict-overflow documentation.
32579 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32580 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32581 flag_strict_overflow.
32582 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32583 * lto-opts.c (lto_write_options): Do not stream it.
32584 * lto-wrapper.c (merge_and_complain): Do not handle it.
32585 * opts.c (default_options_table): Do not set -fstrict-overflow.
32586 (finish_options): Likewise do not clear it when sanitizing.
32587 * simplify-rtx.c (simplify_const_relational_operation): Do not
32588 test flag_strict_overflow.
32590 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32592 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32593 using enabled attribute.
32594 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32595 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32596 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32597 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32598 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32599 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32600 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32601 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32602 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32603 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32605 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32607 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32609 2017-05-02 Richard Biener <rguenther@suse.de>
32611 PR tree-optimization/80591
32613 2017-04-10 Richard Biener <rguenther@suse.de>
32615 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32618 2017-05-02 Richard Biener <rguenther@suse.de>
32620 PR tree-optimization/80549
32621 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32622 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32623 headers if they do not have a preheader.
32625 2017-05-02 Martin Liska <mliska@suse.cz>
32628 * common.opt: Fix typo.
32629 * doc/invoke.texi: Likewise.
32631 2017-05-01 Jan Beulich <jbeulich@suse.com>
32633 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32634 swapping, add (x,x,m,x,n) alternative.
32636 2017-05-01 Nathan Sidwell <nathan@acm.org>
32638 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32639 unnecessary unadjusted_alignment check.
32641 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32644 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32646 * gimplify.c (gimplify_cilk_detach): New function.
32647 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32648 * tree-core.h: Document EXPR_CILK_SPAWN.
32649 * tree.h (EXPR_CILK_SPAWN): Define.
32651 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32653 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32654 to use new fixit_hint representation, using the "replace" logic.
32655 (get_line_span_for_fixit_hint): Likewise.
32656 (layout::print_any_fixits): Likewise.
32657 (selftest::test_one_liner_many_fixits): Rename to...
32658 (selftest::test_one_liner_many_fixits_1): ...this, and update
32659 comment and expected output to reflect that the multiple fix-it
32660 hints are now consolidated into one insertion.
32661 (selftest::test_one_liner_many_fixits_2): New test.
32662 (selftest::test_diagnostic_show_locus_one_liner): Update for
32664 (selftest::test_fixit_consolidation): Update for fix-it API
32666 * diagnostic.c (print_parseable_fixits): Likewise.
32667 * edit-context.c (edited_line::m_line_events): Convert from
32668 auto_vec <line_event *> to auto_vec <line_event>.
32669 (class line_event): Convert from abstract base class to a concrete
32670 class, taking over the role of replace_event.
32671 (class insert_event): Delete.
32672 (class replace_event): Rename to class line_event. Convert to
32674 (edit_context::add_fixits): Reimplement.
32675 (edit_context::apply_insert): Delete.
32676 (edit_context::apply_replace): Rename to...
32677 (edit_context::apply_fixit): ...this. Convert to half-open range.
32678 (edited_file::apply_insert): Delete.
32679 (edited_file::apply_replace): Rename to...
32680 (edited_file::apply_fixit): ...this.
32681 (edited_line::~edited_line): Drop deletion of events.
32682 (edited_line::apply_insert): Delete.
32683 (edited_line::apply_replace): Rename to...
32684 (edited_line::apply_fixit): ...this. Convert to half-open range.
32685 Update for change to type of m_line_events.
32686 * edit-context.h (edit_context::apply_insert): Delete.
32687 (edit_context::apply_replace): Rename to...
32688 (edit_context::apply_fixit): ...this.
32690 2017-05-01 Martin Sebor <msebor@redhat.com>
32692 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32695 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32698 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32699 __get_cpuid_max returns 0.
32700 (__get_cpuid_count): Ditto.
32702 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32704 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32705 replacement expression is another instance of one of its arguments.
32707 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32710 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32711 check for stack push/pop autoinc.
32712 * config/i386/i386.c (ix86_agi_dependent): Return false
32713 if the only reason why modified_in_p returned true is that
32714 addr is SP based and set_insn is a push or pop.
32716 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32718 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32721 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32724 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32725 (account_size_time): Use two predicates - exec_pred and
32727 (evaluate_conditions_for_known_args): Compute both clause and
32729 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32730 (inline_summary_t::duplicate): Update.
32731 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32733 (compute_inline_parameters): Likewise.
32734 (estimate_edge_size_and_time): Update caluclation of time.
32735 (estimate_node_size_and_time): Compute both time and nonspecialized
32737 (estimate_ipcp_clone_size_and_time): Update.
32738 (inline_merge_summary): Update.
32739 (do_estimate_edge_time): Update.
32740 (do_estimate_edge_size): Update.
32741 (do_estimate_edge_hints): Update.
32742 (inline_read_section, inline_write_summary): Stream both new predicates.
32743 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32745 (compute_inlined_call_time): Cleanup.
32746 (big_speedup_p): Update.
32747 (edge_badness): Update.
32748 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32749 (size_time_entry): Replace predicate by exec_predicate and
32750 nonconst_predicate.
32751 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32752 (estimate_edge_time): Return also nonspec_time.
32753 (reset_edge_growth_cache): Update.
32755 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32757 PR rtl-optimization/80491
32758 * ifcvt.c (noce_process_if_block): When looking for x setter
32759 with missing else_bb, don't check only the insn right before
32760 cond_earliest, but look for the last insn that x is modified in
32761 within the same bb.
32763 PR rtl-optimization/80491
32764 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32766 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32768 PR tree-optimization/80487
32769 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32771 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32773 PR tree-optimization/79697
32774 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32775 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32776 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32778 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32779 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32781 2017-04-28 Martin Sebor <msebor@redhat.com>
32783 PR tree-optimization/80523
32784 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32785 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32787 (maybe_warn, format_directive, parse_directive): Use new functions.
32788 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32790 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32792 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32794 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32796 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32797 target_header_dir): Set correctly.
32798 * configure: Regenerated.
32799 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32800 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32801 instead of SYSTEM_HEADER_DIR.
32803 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32805 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32806 (estimate_local_effects): Likewise.
32807 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32808 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32809 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32810 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
32811 do_estimate_edge_time, estimate_edge_time): Likewise.
32812 * ipa-inline-analysis.c (estimate_node_size_and_time,
32813 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
32814 (estimate_time_after_inlining): Remove.
32816 2017-04-28 Martin Liska <mliska@suse.cz>
32818 * doc/gcov.texi: Enhance documentation of gcov.
32820 2017-04-28 Martin Liska <mliska@suse.cz>
32822 * doc/gcov.texi: Sort options in alphabetic order.
32823 * doc/gcov-dump.texi: Likewise.
32824 * doc/gcov-tool.texi: Likewise.
32825 * gcov.c (print_usage): Likewise.
32826 * gcov-dump.c (print_usage): Likewise.
32827 * gcov-tool.c (print_merge_usage_message): Likewise.
32828 (print_rewrite_usage_message): Likewise.
32829 (print_overlap_usage_message): Likewise.
32831 2017-04-28 Martin Liska <mliska@suse.cz>
32833 PR gcov-profile/53915
32834 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
32836 2017-04-28 Martin Liska <mliska@suse.cz>
32838 PR gcov-profile/79891
32839 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
32840 is marked by compiler as living on a line.
32841 (get_cycles_count): Remove usage of the union.
32842 (output_intermediate_file): Likewise.
32843 (find_source): Fix GNU coding style.
32844 (accumulate_line_counts): Remove old non-all block mode.
32845 (output_lines): Remove usage of the union.
32846 * profile.c (output_location): Include all BBs, even if
32847 belonging to a same line (and file) as a previous BB.
32849 2017-04-28 Martin Liska <mliska@suse.cz>
32851 * gcov.c (process_args): Handle new argument 'w'.
32852 (read_graph_file): Assign ID to BBs.
32853 (output_branch_count): Display BB # if verbose flag is set.
32854 (output_lines): Likewise for arcs.
32855 (print_usage): Add '--verbose' option help.
32856 * doc/gcov.texi: Document --verbose (-w) option.
32858 2017-04-28 Martin Liska <mliska@suse.cz>
32860 * gcov.c (struct block_location_info): New struct.
32861 (process_file): Fill up the new structure.
32862 (read_graph_file): Replace usage of encoding by the newly added
32864 (add_line_counts): Likewise.
32865 (accumulate_line_counts): Remove usage of the union.
32866 (function_info::function_info): New function.
32867 (function_info::~function_info): Likewise.
32868 (process_file): Call delete instead of release_function.
32869 (release_function): Release the function.
32870 (release_structures): Call delete instead of release_function.
32871 (solve_flow_graph): Replace usage of num_blocks.
32872 (find_exception_blocks): Likewise.
32873 (output_lines): Fix GNU coding style.
32875 2017-04-28 Martin Liska <mliska@suse.cz>
32878 * coverage.c (coverage_remove_note_file): New function.
32879 * coverage.h: Declare the function.
32880 * toplev.c (finalize): Clean if an error has been seen.
32882 2017-04-28 Martin Liska <mliska@suse.cz>
32884 PR gcov-profile/80031
32885 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
32886 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
32887 * gcov.c (read_graph_file): Read just number of blocks.
32888 * profile.c (branch_prob): Do not stream 0 flags per a basic
32891 2017-04-28 Martin Liska <mliska@suse.cz>
32893 * gcov-dump.c (tag_*): Add new argument to declarations.
32894 (dump_gcov_file): Likewise.
32895 (tag_blocks): Add and use new argument depth.
32896 (tag_arcs): Likewise.
32897 (tag_lines): Likewise.
32898 (tag_counters): Likewise.
32899 (tag_summary): Likewise.
32900 (dump_working_sets): Use depth to do a proper indentation.
32902 2017-04-28 Jakub Jelinek <jakub@redhat.com>
32905 * cgraph.h (symtab_node::debug_symtab): No longer inline.
32906 * symtab.c (symtab_node::debug_symtab): Move definition here.
32908 2017-04-28 Richard Biener <rguenther@suse.de>
32910 * lto-streamer.h (LTO_major_version): Bump to 7.
32912 2017-04-28 Richard Biener <rguenther@suse.de>
32914 * tree-vrp.c (assert_info): New struct.
32915 (add_assert_info): New helper.
32916 (register_edge_assert_for_2): Refactor to add asserts to a vector
32918 (register_edge_assert_for_1): Likewise.
32919 (register_edge_assert_for): Likewise.
32920 (finish_register_edge_assert_for): New helper actually registering
32921 asserts where live on edge.
32922 (find_conditional_asserts): Adjust.
32923 (find_switch_asserts): Likewise.
32924 (evrp_dom_walker::try_find_new_range): Generalize.
32925 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
32927 2017-04-27 Marek Polacek <polacek@redhat.com>
32930 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
32931 arg10 and arg11 to itype.
32933 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
32935 * doc/extend.texi (Object Size Checking): Improve grammar.
32937 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
32940 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
32941 that the logic for permitting reciprocal estimates matches that
32944 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32947 * tree.c (type_cache_hasher::equal): Only compare
32948 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
32949 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
32950 non-aggregate element types.
32951 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
32952 about the flag on ARRAY_TYPEs in the comment, formatting fix.
32954 2017-04-27 Richard Biener <rguenther@suse.de>
32956 PR middle-end/80533
32957 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
32958 stripping ARRAY_REFs from MEM_EXPR make sure we're not
32959 keeping a reference to a trailing array.
32961 2017-04-27 Richard Biener <rguenther@suse.de>
32963 PR middle-end/80539
32964 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
32965 being in loop-closed SSA form conservatively.
32966 (chrec_fold_multiply_poly_poly): Likewise.
32968 2017-04-27 Tamar Christina <tamar.christina@arm.com>
32970 PR middle-end/79665
32971 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
32972 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
32974 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32977 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
32978 (aarch64_function_arg_alignment): Return unsigned int again, but still
32979 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
32980 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32981 Don't emit -Wpsabi note.
32982 (aarch64_function_arg_boundary): Likewise.
32983 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32986 2017-04-26 Nathan Sidwell <nathan@acm.org>
32988 * tree.h (crc32_unsigned_n): Declare.
32989 (crc32_unsigned, crc32_unsigned): Make inline.
32990 * tree.c (crc32_unsigned_bits): Replace with ...
32991 (crc32_unsigned_n): ... this.
32992 (crc32_unsigned, crc32_byte): Remove.
32993 (crc32_string): Remove unnecessary braces.
32995 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32997 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
32998 * ipa-inline-analysis.c (MAX_TIME): Remove.
32999 (account_size_time): Use sreal for time.
33000 (dump_inline_summary): Update.
33001 (estimate_function_body_sizes): Update.
33002 (estimate_edge_size_and_time): Update.
33003 (estimate_calls_size_and_time): Update.
33004 (estimate_node_size_and_time): Update.
33005 (inline_merge_summary): Update.
33006 (inline_update_overall_summary): Update.
33007 (estimate_time_after_inlining): Update.
33008 (inline_read_section): Update.
33009 (inline_write_summary): Update.
33010 * ipa-inline.c (compute_uninlined_call_time): Update.
33011 (compute_inlined_call_time): Update.
33012 (recursive_inlining): Update.
33013 (inline_small_functions): Update.
33014 (dump_overall_stats): Update.
33015 * ipa-inline.h: Include sreal.h.
33016 (size_time_entry): Turn time to sreal.
33017 (inline_summary): Turn self_time nad time to sreal.
33019 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33021 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33023 (sreal::stream_out, sreal::stream_in): New.
33024 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33026 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33028 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33031 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33034 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33036 (dimode_scalar_chain::compute_convert_gain): Ditto.
33037 (dimode_scalar_chain::make_vector_copies): Ditto.
33038 (dimode_scalar_chain::convert_reg): Ditto.
33039 (dimode_scalar_chain::convert_insn): Ditto.
33040 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33041 (VI248_AVX512BW_1): New mode iterator.
33042 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33043 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33046 2017-04-25 Martin Sebor <msebor@redhat.com>
33048 PR tree-optimization/80497
33049 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33050 constants are representable in HOST_WIDE_INT.
33051 (parse_directive): Ditto.
33053 2017-04-25 Martin Sebor <msebor@redhat.com>
33056 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
33057 (new_zero_array): Adjust signature.
33058 (dom_info::dom_init): Used unsigned rather that size_t.
33059 (dom_info::dom_info): Same.
33061 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33062 Jakub Jelinek <jakub@redhat.com>
33065 * config/arm/arm.c: Include gimple.h.
33066 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33067 returns negative, increment ncrn only if it returned positive.
33068 (arm_needs_doubleword_align): Return int instead of bool,
33069 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
33070 members, but if there is any such non-FIELD_DECL
33071 > PARM_BOUNDARY aligned decl, return -1 instead of false.
33072 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33073 returns negative, increment nregs only if it returned positive.
33074 (arm_setup_incoming_varargs): Likewise.
33075 (arm_function_arg_boundary): Emit -Wpsabi note if
33076 arm_needs_doubleword_align returns negative, return
33077 DOUBLEWORD_ALIGNMENT only if it returned positive.
33079 2017-04-25 Marek Polacek <polacek@redhat.com>
33082 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33083 first argument to type.
33085 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
33088 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
33089 type checks to test for compatibility instead of equality.
33091 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33092 Jakub Jelinek <jakub@redhat.com>
33095 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
33097 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
33098 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
33099 the alignment computation, but return their maximum in warn_alignment.
33100 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33101 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
33103 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
33104 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33107 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33109 * config/arc/simdext.md (dmpyh): Fix typo.
33111 2017-04-25 Richard Biener <rguenther@suse.de>
33113 PR tree-optimization/80492
33114 * alias.c (compare_base_decls): Handle registers with asm
33115 specification conservatively.
33116 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33117 compare_base_decls returning dont-know properly.
33119 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33121 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
33122 (legitimate_offset_address_p): New function.
33123 (arc_legitimate_address_p): Use above function.
33125 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33127 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
33129 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33131 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
33132 ACCH registers whenever they are available.
33134 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33136 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
33137 double regs fix when not used.
33139 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33141 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
33143 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
33144 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
33146 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33148 * config/arc/arc.c (arc_output_addsi): Check for h-register class
33149 when emitting short ADD instructions.
33151 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33153 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
33155 (cmpsi_cc_c_insn): Likewise.
33156 (cbranchsi4_scratch): Compute proper instruction length using
33157 compact_hreg_operand.
33158 * config/arc/predicates.md (compact_hreg_operand): New predicate.
33160 2017-04-25 Richard Biener <rguenther@suse.de>
33162 PR middle-end/80509
33163 * passes.c (pass_manager::pass_manager): Initialize
33164 m_name_to_pass_map.
33166 2017-04-25 Richard Biener <rguenther@suse.de>
33168 PR tree-optimization/79201
33169 * tree-ssa-sink.c (statement_sink_location): Handle calls.
33171 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33174 * config/s390/vector.md: Split MEM->GPR vector moves for
33175 non-s_operand addresses.
33177 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33180 * config/s390/predicates.md (reload_const_wide_int_operand): New
33182 * config/s390/s390.md ("movti"): Remove d/P alternative.
33183 ("movti_bigconst"): New pattern definition.
33185 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33188 * s390-protos.h (s390_expand_cs_hqi): Removed.
33189 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
33190 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
33191 modes as well as CCZ1mode and CCZmode.
33192 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
33193 signature of s390_emit_compare_and_swap.
33194 (s390_expand_cs_hqi): Likewise, make static.
33195 (s390_expand_cs_tdsi): Generate an explicit compare before trying
33196 compare-and-swap, in some cases.
33197 (s390_expand_cs): Wrapper function.
33198 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
33200 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
33201 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
33202 patterns for small and large integers. Forbid symref memory operands.
33203 Move expander to s390.c. Require cc register.
33204 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
33205 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
33206 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
33207 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
33208 symref memory operands. Remove CC mode and call s390_match_ccmode
33210 ("atomic_exchange<mode>"): Allow and implement all integer modes.
33212 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33214 * config/s390/s390.md (define_peephole2): New peephole to help
33215 combining the load-and-test pattern with volatile memory.
33217 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33219 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
33220 with CCZmode for TARGET_Z196.
33222 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33224 PR rtl-optimization/80501
33225 * combine.c (make_compound_operation_int): Set subreg_code to SET
33226 even for AND with mask of the sign bit of mode.
33228 PR rtl-optimization/80500
33229 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
33230 sum's initial value.
33232 2017-04-25 Julian Brown <julian@codesourcery.com>
33233 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33235 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
33237 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
33239 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
33241 2017-04-25 Julian Brown <julian@codesourcery.com>
33242 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33244 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
33245 (thunderx2t99_sha): New Reservation.
33247 2017-04-25 Julian Brown <julian@codesourcery.com>
33248 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33250 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
33251 type for 1-element load.
33253 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
33255 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
33257 2017-04-24 Martin Jambor <mjambor@suse.cz>
33259 PR tree-optimization/80293
33260 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
33261 char arrays not totally scalarizable if it is false.
33262 (analyze_all_variable_accesses): Pass correct value in the new
33263 parameter. Add a statistics counter.
33265 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
33267 PR middle-end/79931
33268 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
33270 2017-04-24 Richard Biener <rguenther@suse.de>
33272 PR tree-optimization/80494
33273 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
33274 out for complex types.
33276 2017-04-24 Richard Biener <rguenther@suse.de>
33278 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
33279 * tree-ssa-sccvn.c (print_scc): Print SCC size.
33280 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
33281 (DFS): Adjust and never fail.
33282 (sccvn_dom_walker::fail): Remove.
33283 (sccvn_dom_walker::before_dom_children): Adjust.
33284 (run_scc_vn): Likewise and never fail.
33285 * tree-ssa-pre.c (pass_pre::execute): Adjust.
33286 (pass_fre::execute): Likewise.
33288 2017-04-24 Richard Biener <rguenther@suse.de>
33290 PR tree-optimization/79725
33291 * tree-ssa-sink.c (statement_sink_location): Return whether
33292 failure reason was zero uses. Move that check later.
33293 (sink_code_in_bb): Deal with zero uses by removing the stmt
33296 2017-04-24 Richard Biener <rguenther@suse.de>
33299 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
33300 pointer-based references.
33302 2017-04-24 Richard Biener <rguenther@suse.de>
33305 * pass_manager.h (pass_manager::operator new): Remove.
33306 (pass_manager::operator delete): Likewise.
33307 * passes.c (pass_manager::operator new): Remove.
33308 (pass_manager::operator delete): Likewise.
33309 (pass_manager::pass_manager): Zero individual pass members.
33311 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
33314 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
33315 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
33316 Check "XEXP (src, 1)" operand here.
33317 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
33318 Check "XEXP (src, 1)" operand here.
33319 (dimode_scalar_chain::make_vector_copies): Detect count register
33320 of a shift instruction. Zero extend count register from QImode
33321 to DImode to satisfy vector shift pattern count operand predicate.
33322 Substitute vector shift count operand with a DImode copy.
33323 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
33326 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
33328 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
33329 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33330 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
33331 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33332 (UNSPEC_NOREX_MEM): Remove definition.
33334 2017-04-21 Richard Biener <rguenther@suse.de>
33336 PR tree-optimization/79547
33337 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
33338 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
33339 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
33340 without any constraints.
33342 2017-04-21 Richard Biener <rguenther@suse.de>
33344 PR tree-optimization/78847
33345 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
33347 2017-04-21 Richard Biener <rguenther@suse.de>
33349 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
33350 (build_distinct_type_copy): Likewise.
33351 (build_variant_type_copy): Likewise.
33352 * tree.c (build_qualified_type): Pass down mem-stat info.
33353 (build_distinct_type_copy): Likewise.
33354 (build_variant_type_copy): Likewise.
33356 2017-04-21 Richard Biener <rguenther@suse.de>
33358 PR tree-optimization/80237
33359 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
33361 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
33362 for a simplified result.
33364 2016-04-21 Richard Biener <rguenther@suse.de>
33366 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
33367 sth as strict as a simple_iv but a chrec without symbols and an
33368 operand defined in the loop we are peeling (and not some subloop).
33369 (propagate_constants_for_unrolling): Propagate all constants.
33371 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33374 * config/i386/i386.c (print_reg): Remove assert for disalowed
33375 regno values, call output_operand_lossage instead.
33377 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33380 * config/i386/constraints.md (Yc): New register constraint.
33381 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
33382 Use Yc constraint for alternative 2 of operand 0. Remove
33383 preferred_for_speed attribute.
33385 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33387 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
33388 lastprivate clauses in SIMT case.
33390 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
33392 * doc/invoke.texi (-Wextra-semi): Document new warning option.
33394 2017-04-20 Richard Biener <rguenther@suse.de>
33396 PR tree-optimization/57796
33397 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
33398 as N scalar stores.
33399 (vect_model_load_cost): Cost gathers as N scalar loads.
33401 2017-04-20 Richard Biener <rguenther@suse.de>
33403 * ggc-page.c (ggc_allocated_p): Rename to ...
33404 (safe_lookup_page_table_entry): ... this and return the lookup
33406 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
33408 2017-04-20 Richard Biener <rguenther@suse.de>
33410 PR tree-optimization/80453
33411 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
33412 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
33413 from the conditions.
33414 (vn_phi_eq): Pass them down.
33415 (vn_phi_lookup): Record them.
33416 (vn_phi_insert): Likewise.
33418 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
33420 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
33421 uninitialized variable warning to avoid buffer overrun.
33423 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33426 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
33427 is suppressed for '{ 0 }' in C.
33429 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33431 * BASE-VER: Set to 8.0.0.
33433 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
33435 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
33436 priority .init_array and .fini_array section with SECTION_NOTYPE
33439 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33441 PR middle-end/80423
33442 * tree.h (build_array_type): Add typeless_storage default argument.
33443 * tree.c (type_cache_hasher::equal): Also compare
33444 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
33445 (build_array_type): Add typeless_storage argument, set
33446 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
33448 (build_nonshared_array_type): Adjust build_array_type_1 caller.
33449 (build_array_type): Likewise. Add typeless_storage argument.
33451 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33452 Jakub Jelinek <jakub@redhat.com>
33454 PR tree-optimization/80426
33455 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
33456 operation on symbolic operands, also compute the overflow for the
33457 invariant part when the operation degenerates into a negation.
33459 2017-04-19 Jakub Jelinek <jakub@redhat.com>
33462 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
33463 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
33466 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
33468 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
33471 * config/avr/avr.c (tree.h): Include it.
33472 (cgraph.h): Include it.
33473 (avr_encode_section_info): Don't warn for uninitialized progmem
33474 variable if it's just an alias.
33476 2017-04-19 Richard Biener <rguenther@suse.de>
33479 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
33480 when needed by AutoPGO.
33482 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
33485 * doc/lto.texi: Remove an extra 'that'.
33487 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
33489 PR rtl-optimization/80429
33490 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
33491 are only used in debug insns.
33493 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33494 Vladimir Makarov <vmakarov@redhat.com>
33496 * config/sparc/predicates.md (input_operand): Add comment. Return
33497 true for any memory operand when LRA is in progress.
33498 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
33500 2017-04-18 Jeff Law <law@redhat.com>
33503 * mips.md ({return,simple_return}_internal): Do not overwrite
33506 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33508 PR tree-optimization/80443
33509 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
33510 instead of adding 1, subtract -1 and similarly instead of subtracting
33513 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
33515 PR rtl-optimization/80357
33516 * haifa-sched.c (tmp_bitmap): New variable.
33517 (model_recompute): Handle duplicate use records.
33518 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
33519 (free_global_sched_pressure_data): Free it.
33521 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
33524 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33525 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33526 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33527 instead of SYSTEM_HEADER_DIR.
33529 2017-04-18 Jeff Law <law@redhat.com>
33531 PR middle-end/80422
33532 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
33533 predecessors after walking up the insn chain.
33535 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33538 * dwarf2out.c (modified_type_die): Try harder not to emit internal
33539 sizetype type into debug info.
33541 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33544 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
33545 unneeded test for TARGET_UPPER_REGS_SF.
33546 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
33548 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33551 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
33552 instead of gsi_after_labels.
33554 2017-04-18 Jeff Law <law@redhat.com>
33556 * regcprop.c (maybe_mode_change): Avoid creating copies of the
33560 2017-04-13 Jeff Law <law@redhat.com>
33561 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33562 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33564 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33567 * config/avr/avr.c (intl.h): Include it.
33568 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33570 2017-04-18 Martin Liska <mliska@suse.cz>
33572 PR gcov-profile/78783
33573 * gcov-tool.c (gcov_output_files): Validate that destination
33574 file is either removed by the tool or by a user.
33576 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33577 Guy Benyei <guybe@mellanox.com>
33579 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33580 block, and do not negate it, the stored id is already negative.
33582 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33584 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33586 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33589 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33590 masks of options that should be turned off if the VSX vector
33591 options are turned off.
33592 (OTHER_P8_VECTOR_MASKS): Likewise.
33593 (OTHER_VSX_VECTOR_MASKS): Likewise.
33594 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33595 rs6000_disable_incompatible_switches to validate no type switches
33597 (rs6000_incompatible_switch): New function to disallow turning on
33598 other vector options if -mno-vsx, -mno-power8-vector, or
33599 -mno-power9-vector are specified.
33601 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33603 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33605 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33607 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33608 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33609 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33610 (ARG_POINTER_CFA_OFFSET): Likewise.
33612 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33614 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33615 conditions to take advantage of various optimizations.
33617 2017-04-13 Jeff Law <law@redhat.com>
33619 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33620 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33621 (zero_extendsidi2_dext): Likewise.
33623 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33626 * fold-const.c (fold_ternary_loc): Revert
33627 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33630 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33632 PR rtl-optimization/80343
33633 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33634 new scratch pseudo.
33636 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33639 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33640 to ubsan_encode_value.
33642 2017-04-13 Jeff Law <law@redhat.com>
33644 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33645 appearing in DEBUG_INSNs.
33647 2017-04-13 Martin Liska <mliska@suse.cz>
33649 PR gcov-profile/80413
33650 * gcov-io.c (gcov_write_string): Copy to buffer just when
33651 allocated size is greater than zero.
33653 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33656 * dwarf2out.c (decls_for_scope): Ignore declarations of
33657 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33659 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33662 * ipa-visibility.c (non_local_p): Fix typos.
33663 (localize_node): When localizing symbol in same comdat group,
33664 dissolve the group only when we know external symbols are going
33666 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33668 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33670 PR tree-optimization/79390
33671 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33672 order does not result in usable sequence, retry with reversed operand
33678 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33679 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33680 op0 instead of fold_convert_loc (loc, type, arg0).
33682 2017-04-12 Jeff Law <law@redhat.com>
33684 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33685 has a delay slot in the generated code.
33687 * config/cris/cris.md (cris_preferred_reload_class): Return
33688 GENNONACR_REGS rather than GENERAL_REGS.
33690 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33693 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33694 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33695 signedness of the result type.
33697 2017-04-12 Richard Biener <rguenther@suse.de>
33698 Jeff Law <law@redhat.com>
33700 PR tree-optimization/80359
33701 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33702 trim stores to TARGET_MEM_REFs.
33704 2017-04-12 Richard Biener <rguenther@suse.de>
33706 PR tree-optimization/79390
33707 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33708 threading case even more.
33710 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33713 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33714 for quad_address_p for TImode, instead of just not indexed_address.
33716 2017-04-12 Richard Biener <rguenther@suse.de>
33717 Bernd Edlinger <bernd.edlinger@hotmail.de>
33719 PR middle-end/79671
33720 * alias.c (component_uses_parent_alias_set_from): Handle
33721 TYPE_TYPELESS_STORAGE.
33722 (get_alias_set): Likewise.
33723 * tree-core.h (tree_type_common): Add typeless_storage flag.
33724 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33725 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33726 for types containing members with TYPE_TYPELESS_STORAGE.
33727 (place_field): Likewise.
33728 (layout_type): Likewise for ARRAY_TYPE.
33729 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33730 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33731 TYPE_TYPELESS_STORAGE.
33732 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33734 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33737 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33738 first argument to type.
33740 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33744 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33745 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33746 (rs6000_expand_binop_builtin): Likewise.
33747 (rs6000_expand_ternop_builtin): Likewise; also add missing
33748 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33749 vshasigma built-ins.
33750 * doc/extend.texi: Document that vec_xxpermdi's third argument
33751 must be a constant.
33753 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33755 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33756 Use shift_const cost parameter when calculating gain of STV shifts.
33758 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33760 PR rtl-optimization/70478
33761 * lra-constraints.c (process_alt_operands): Check memory for
33762 disfavoring memory insn operand.
33764 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33766 PR middle-end/80100
33767 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33768 left shift in unsigned HOST_WIDE_INT type.
33770 PR rtl-optimization/80385
33771 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33772 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33775 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33776 if they have any depend clauses.
33778 2017-04-11 Martin Liska <mliska@suse.cz>
33781 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33782 * ipa-split.c (split_function): Create a local comdat symbol
33783 if caller is in a comdat group.
33785 2017-04-11 Martin Liska <mliska@suse.cz>
33788 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33791 2017-04-11 Martin Sebor <msebor@redhat.com>
33793 PR middle-end/80364
33794 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33795 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33797 (directive::set_width, directive::set_precision, format_character):
33799 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33802 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33805 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33806 conflict, set target->arch_name instead of target->cpu_name.
33808 2017-04-11 Richard Biener <rguenther@suse.de>
33810 PR tree-optimization/80374
33811 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
33812 build_zero_cst, remove fold_convertible_p check again.
33814 2017-04-11 Martin Liska <mliska@suse.cz>
33817 * ubsan.c (instrument_object_size): Do not instrument register
33820 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33823 * config/i386/i386-builtin-types.def
33824 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
33825 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
33826 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
33827 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
33828 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
33829 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
33830 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
33831 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
33832 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
33833 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
33834 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
33835 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
33836 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
33837 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
33838 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
33839 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
33840 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
33841 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
33842 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
33843 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
33844 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
33845 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
33846 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
33847 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
33848 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
33849 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
33850 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
33851 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
33852 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
33853 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
33854 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
33855 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
33856 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
33857 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
33858 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
33859 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
33860 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
33861 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
33862 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
33863 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
33864 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
33865 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
33866 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
33867 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
33868 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
33869 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
33871 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
33872 flag to second_arg_count, handle 4 argument function type _COUNT
33873 aliases, handle second_arg_count on second argument rather than last.
33875 2017-04-10 Jeff Law <law@redhat.com>
33877 PR tree-optimization/80374
33878 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
33879 record anything if we can not convert integer_zero_node to the
33882 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33885 * config/rs6000/rs6000.c (rs6000_option_override_internal):
33886 Enhance special handling given to the TARGET_P9_MINMAX option in
33887 relation to certain other options.
33889 2017-04-10 Bin Cheng <bin.cheng@arm.com>
33891 PR tree-optimization/80153
33892 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
33893 remove POINTER_PLUS_EXPR's base part directly, rather than through
33896 2017-04-10 Richard Biener <rguenther@suse.de>
33897 Bin Cheng <bin.cheng@arm.com>
33899 PR tree-optimization/80153
33900 * tree-affine.c (aff_combination_to_tree): Get base pointer from
33901 the first element of pointer type aff_tree. Build result expr in
33903 (add_elt_to_tree): Convert to type unconditionally. Remove other
33904 fold_convert calls.
33905 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
33906 (rewrite_use_nonlinear_expr): Check invariant using iv information.
33908 2017-04-10 Richard Biener <rguenther@suse.de>
33910 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33913 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
33915 PR rtl-optimization/70478
33916 * lra-constraints.c (curr_small_class_check): New.
33917 (update_and_check_small_class_inputs): New.
33918 (process_alt_operands): Update curr_small_class_check. Disfavor
33919 alternative insn memory operands. Check available regs for small
33922 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33925 * config/mips/mips.opt (-mvirt): Update description.
33926 * doc/invoke.texi (-mvirt): Likewise.
33928 2017-04-10 Richard Biener <rguenther@suse.de>
33930 PR middle-end/80362
33931 * fold-const.c (fold_binary_loc): Look at unstripped ops when
33932 looking for NEGATE_EXPR in -A / -B to A / B folding.
33934 2017-04-10 Martin Liska <mliska@suse.cz>
33936 PR gcov-profile/80224
33937 * gcov.c (print_usage): Fix usage string.
33938 (get_gcov_intermediate_filename): Remove.
33939 (output_gcov_file): Use both for normal and intermediate format.
33940 (generate_results): Do not initialize special file for
33941 intermediate format.
33943 2017-04-10 Richard Biener <rguenther@suse.de>
33945 PR tree-optimization/80304
33946 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
33949 2017-04-10 Nathan Sidwell <nathan@acm.org>
33952 * config/rs6000/rs6000.c (rs6000_vector_type): New.
33953 (rs6000_init_builtins): Use it.
33955 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33957 * config/arm/arm.md (<mrc>): Add mode to SET source.
33958 (<mrrc>): Likewise.
33960 2017-04-10 Richard Biener <rguenther@suse.de>
33962 PR middle-end/80344
33963 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
33965 2017-04-10 Jakub Jelinek <jakub@redhat.com>
33968 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
33969 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
33970 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
33971 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
33972 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
33973 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
33974 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
33975 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
33976 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
33977 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
33978 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
33979 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
33980 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
33981 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
33982 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
33983 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
33984 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
33985 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
33986 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
33987 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
33988 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
33989 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
33990 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
33992 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
33994 PR rtl-optimization/70478
33995 * lra-constraints.c: Reverse the last patch.
33997 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
33999 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34000 Add comment for WCHAR_T.
34002 2017-04-08 Martin Liska <mliska@suse.cz>
34005 2017-04-07 Martin Liska <mliska@suse.cz>
34008 * ipa-split.c (split_function): Add function part to a same comdat
34011 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34014 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34016 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34018 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34019 (vec_pairing): ...to this.
34020 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34021 (rs6000_sched_init): Adjust for name change.
34022 (struct rs6000_sched_context): Likewise.
34023 (rs6000_init_sched_context): Likewise.
34024 (rs6000_set_sched_context): Likewise.
34026 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34032 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34034 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34035 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34036 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34038 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34040 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34042 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34044 PR rtl-optimization/70703
34045 * ira-color.c (update_conflict_hard_regno_costs): Use
34046 int64_t instead of HOST_WIDE_INT.
34048 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34050 PR rtl-optimization/70478
34051 * lra-constraints.c (process_alt_operands): Disfavor alternative
34052 insn memory operands.
34054 2017-04-07 Jeff Law <law@redhat.com>
34056 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
34057 CALL and NOTE_INSN_CALL_ARG_LOCATION.
34059 2017-04-07 Martin Liska <mliska@suse.cz>
34062 * config/aarch64/aarch64.c (aarch64_process_target_attr):
34063 Show error message instead of an ICE.
34065 2017-04-07 Martin Liska <mliska@suse.cz>
34068 * ipa-split.c (split_function): Add function part to a same comdat
34071 2017-04-07 Richard Biener <rguenther@suse.de>
34073 PR middle-end/80341
34074 * tree.c (get_unwidened): Also handle ! for_type case for
34076 * convert.c (do_narrow): Split out from ...
34077 (convert_to_integer_1): ... here. Do not pass final truncation
34078 type to get_unwidened for TRUNC_DIV_EXPR.
34080 2017-04-07 Richard Biener <rguenther@suse.de>
34082 * tree-affine.c (wide_int_ext_for_comb): Take type rather
34084 (aff_combination_const): Adjust.
34085 (aff_combination_scale): Likewise.
34086 (aff_combination_add_elt): Likewise.
34087 (aff_combination_add_cst): Likewise.
34088 (aff_combination_convert): Likewise.
34089 (add_elt_to_tree): Likewise. Remove unused argument.
34090 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
34092 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
34094 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
34096 * config/arm/arm.c (arm_default_short_enums): Use
34097 ARM_DEFAULT_SHORT_ENUMS.
34098 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
34100 2017-04-06 Jakub Jelinek <jakub@redhat.com>
34103 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
34104 members with redundant out-of-class redeclaration.
34106 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34109 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
34110 * config/i386/i386.md (*zero_extendsidi2):
34111 Add (?*x,*x) and (?*v,*v) alternatives.
34113 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34116 * config/i386/i386.c (ix86_expand_builtin)
34117 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
34118 mode from insn data. Convert operands to insn operand mode.
34119 Copy operands that don't satisfy insn predicate to a register.
34121 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
34123 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
34126 2017-04-06 Richard Biener <rguenther@suse.de>
34128 PR tree-optimization/80334
34129 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
34130 preserve alignment of accesses.
34132 2017-04-06 Richard Biener <rguenther@suse.de>
34134 PR tree-optimization/80262
34135 * tree-sra.c (build_ref_for_offset): Preserve address-space
34137 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
34138 Drop useless address-space information on MEM_REF offsets.
34140 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
34142 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
34144 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34146 PR rtl-optimization/70703
34147 * ira-color.c (update_conflict_hard_regno_costs): Use
34148 HOST_WIDE_INT instead of long.
34150 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
34153 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
34154 not defined for x86_64 target. Add -mmmx target option when __SSE2__
34156 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
34157 for x86_64 target. Handle -m3dnowa option.
34159 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34161 PR rtl-optimization/70703
34162 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
34163 (update_conflict_hard_regno_costs): Use long instead of unsigned
34164 arithmetic for cost calculation.
34166 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34167 Bernd Edlinger <bernd.edlinger@hotmail.de>
34170 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
34173 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
34176 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
34177 ptr_mode with Pmode throughout.
34178 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
34179 into probe_stack_range and use DImode.
34181 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
34184 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
34185 call_eh_return is true.
34187 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34189 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
34190 Initialize last_match_fntype_index.
34192 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34195 * tree-nvr.c: Include internal-fn.h.
34196 (pass_return_slot::execute): Ignore internal calls without
34199 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34200 Richard Biener <rguenther@suse.de>
34203 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
34204 captures used multiple times, except for the last use.
34205 * generic-match-head.c: Include gimplify.h.
34207 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34209 PR tree-optimization/79390
34210 * target.h (struct noce_if_info): Declare.
34211 * targhooks.h (default_noce_conversion_profitable_p): Declare.
34212 * target.def (noce_conversion_profitable_p): New target hook.
34213 * ifcvt.h (struct noce_if_info): New type, moved from ...
34214 * ifcvt.c (struct noce_if_info): ... here.
34215 (noce_conversion_profitable_p): Renamed to ...
34216 (default_noce_conversion_profitable_p): ... this. No longer
34218 (noce_try_store_flag_constants, noce_try_addcc,
34219 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
34220 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
34221 instead of noce_conversion_profitable_p.
34222 * config/i386/i386.c: Include ifcvt.h.
34223 (ix86_option_override_internal): Don't override
34224 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
34225 (ix86_noce_conversion_profitable_p): New function.
34226 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
34227 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
34228 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
34229 * doc/tm.texi: Regenerated.
34231 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34233 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
34236 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
34239 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
34240 instructions for small multiply cores.
34242 2017-04-04 Jeff Law <law@redhat.com>
34244 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
34246 (mips_expand_vec_perm_const): Initialize elements in orig_perm
34247 that are not set by the loop over the elements.
34249 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34252 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
34253 int mode, convert_modes it to mode as unsigned, otherwise use
34254 lowpart_subreg to mode rather than SImode.
34255 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
34256 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
34257 Use DImode instead of SImode for the shift count operand.
34258 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
34261 2017-04-04 Richard Biener <rguenther@suse.de>
34263 PR middle-end/80281
34264 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
34265 arithmetic done for the negate or the plus. Simplify.
34266 (A - (-B) -> A + B): Likewise.
34267 * fold-const.c (split_tree): Make sure to not negate pointers.
34269 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
34271 PR rtl-optimization/60818
34272 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
34273 a compare of comparisons with the thing compared if this results
34274 in a different machine mode.
34276 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
34278 * alias.c (base_alias_check): Fix typo in comment.
34279 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
34280 * cgraphunit.c (symbol_table::compile): Likewise.
34281 * collect2.c (maybe_run_lto_and_relink): Likewise.
34282 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
34283 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
34284 * config/avr/avr.c (avr_map_op_t): Likewise.
34285 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
34286 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
34287 * config/epiphany/epiphany.md (movcc): Likewise.
34288 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
34289 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
34291 * config/mips/mips.c (mips_save_restore_reg): Likewise.
34292 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
34293 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
34294 * config/sh/sh.c (sh_rtx_costs): Likewise.
34295 * fold-const.c (fold_truth_andor): Likewise.
34296 * genautomata.c (collapse_flag): Likewise.
34297 * gengtype.h (struct type::u::s): Likewise.
34298 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
34299 * input.c (FORMAT_AMOUNT): Likewise.
34300 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
34301 (known_aggs_to_agg_replacement_list): Likewise.
34302 * ipa-inline-analysis.c: Likewise.
34303 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
34304 * ipa-polymorphic-call.c
34305 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
34306 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
34307 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
34309 * modulo-sched.c (apply_reg_moves): Likewise.
34310 * omp-expand.c (build_omp_regions_1): Likewise.
34311 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
34312 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
34313 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
34314 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
34315 * value-prof.c: Likewise.
34316 * var-tracking.c (val_reset): Likewise.
34318 2017-04-03 Richard Biener <rguenther@suse.de>
34320 PR tree-optimization/80275
34321 * fold-const.c (split_address_to_core_and_offset): Handle
34324 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
34326 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
34327 descriptors is at least equal to that of functions.
34329 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34331 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
34333 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34336 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
34337 (mov<IMOD4:mode>): New expander.
34338 (*mov<IMOD4:mode>_internal): New insn and split pattern.
34340 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
34342 PR rtl-optimization/79405
34343 * fwprop.c (propagations_left): New variable.
34344 (forward_propagate_into): Decrement it.
34345 (fwprop_init): Initialize it.
34346 (fw_prop): If the variable has reached zero, stop propagating.
34347 (fwprop_addr): Ditto.
34349 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34352 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
34353 a FUNCTION_DECL, pass it as decl instead of origin to
34356 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
34358 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
34361 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
34364 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
34365 TARGET_VSX_SMALL_INTEGER.
34367 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34369 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
34370 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
34372 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34374 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
34375 extraction from odd-numbered MSA register.
34377 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34379 PR middle-end/80173
34380 * expmed.c (store_bit_field_1): Don't attempt to create
34381 a word subreg out of hard registers wider than word if they
34382 have HARD_REGNO_NREGS of 1 for their mode.
34384 PR middle-end/80163
34385 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
34386 conversions to integer types wider than word and pointer.
34389 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
34390 (rtx_equal_for_cselib_p): Pass 0 to it.
34391 * cselib.c (cselib_hasher::equal): Likewise.
34392 (rtx_equal_for_cselib_1): Add depth argument. If depth
34393 is 128, don't look up VALUE locs and punt. Increment
34394 depth in recursive calls when walking VALUE locs.
34396 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
34398 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
34399 (make_gcov_file_name): Use the canonical path name for generating
34401 (read_line): Fix handling of files with ascii null bytes.
34403 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
34405 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
34406 to initialise a vector register instead
34407 of using a const_int.
34409 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34411 PR translation/80189
34412 * gimplify.c (omp_default_clause): Use %qs instead of %s in
34413 diagnostic messages.
34415 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
34418 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
34419 (dfp_diex_<mode>): Update mode of operand 1.
34420 * doc/extend.texi (dxex, dxexq): Document change to return type.
34421 (diex, diexq): Document change to argument type.
34423 2017-03-30 Martin Jambor <mjambor@suse.cz>
34426 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
34427 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
34428 it reflects the signature changes performed at the callee side.
34429 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
34430 to cgraph_build_function_type_skip_args.
34431 (build_function_decl_skip_args): Adjust call to the above function.
34433 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34436 * config/i386/sse.md
34437 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
34438 register as dest whenever it is a MEM not rtx_equal_p to the
34439 corresponding dup operand, and when forcing into reg move the
34440 reg into the memory afterwards.
34441 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
34442 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
34443 for the force_reg mode.
34444 (avx512vl_vextractf128<mode>): Use register as dest either
34445 always when a MEM, or when it is a MEM not rtx_equal_p to the
34446 corresponding dup operand, or even not when it is a CONST_VECTOR
34447 depending on the mode and lo vs. hi.
34448 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
34450 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
34451 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
34452 Likewise. Require that operands[2] is even.
34453 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
34454 Remove extraneous parens. Require that operands[2] is a multiple
34456 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
34457 operands[0] is a MEM if <mask_applied>, the predicates/constraints
34458 disallow memory then.
34460 2017-03-30 Richard Biener <rguenther@suse.de>
34462 PR tree-optimization/77498
34463 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
34464 to non-constants over backedges.
34466 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
34468 PR rtl-optimization/80233
34469 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
34470 as last_combined_insn. Do not test for BARRIER_P separately.
34472 2017-03-29 Andreas Schwab <schwab@suse.de>
34475 * calls.c (prepare_call_address): Convert funexp to Pmode before
34476 copying to temp reg.
34478 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34480 PR tree-optimization/80158
34481 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
34482 Handle possible future case of more than one alternate
34484 (replace_rhs_if_not_dup): Likewise.
34485 (replace_one_candidate): Likewise.
34487 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
34489 PR rtl-optimization/80193
34490 * ira.c (ira): Do not check allocation for LRA.
34492 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
34494 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
34495 (nvptx_output_simt_exit): Declare.
34496 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
34497 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
34498 (init_softstack_frame): Move initialization of crtl->is_leaf to...
34499 (nvptx_declare_function_name): ...here. Emit declaration of local
34500 memory space buffer for omp_simt_enter insn.
34501 (nvptx_output_unisimt_switch): New.
34502 (nvptx_output_softstack_switch): New.
34503 (nvptx_output_simt_enter): New.
34504 (nvptx_output_simt_exit): New.
34505 * config/nvptx/nvptx.h (struct machine_function): New fields
34506 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
34507 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
34508 (UNSPECV_SIMT_EXIT): Ditto.
34509 (omp_simt_enter_insn): New insn.
34510 (omp_simt_enter): New expansion.
34511 (omp_simt_exit): New insn.
34512 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
34514 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
34515 (expand_GOMP_SIMT_ENTER_ALLOC): New.
34516 (expand_GOMP_SIMT_EXIT): New.
34517 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
34518 (GOMP_SIMT_ENTER_ALLOC): Ditto.
34519 (GOMP_SIMT_EXIT): Ditto.
34520 * target-insns.def (omp_simt_enter): New insn.
34521 (omp_simt_exit): Ditto.
34522 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
34524 (lower_rec_simd_input_clauses): Implement SIMT privatization.
34525 (lower_rec_input_clauses): Likewise.
34526 (lower_lastprivate_clauses): Handle SIMT privatization.
34528 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
34529 (ompdevlow_adjust_simt_enter): New.
34530 (find_simtpriv_var_op): New.
34531 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
34532 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
34534 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
34535 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
34536 (copy_decl_for_dup_finish): Ditto.
34538 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
34540 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
34543 * config/i386/i386.c (ix86_option_override_internal): Always
34544 allow -mpreferred-stack-boundary=3 for 64-bit targets.
34546 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34548 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
34550 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34552 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
34553 mark new edge's irreducible flag accordign to it.
34554 (vect_do_peeling): Check loop preheader edge's irreducible flag
34555 and pass it to function slpeel_add_loop_guard.
34557 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
34559 PR tree-optimization/80218
34560 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34561 Update block frequencies and counts.
34563 2017-03-28 Richard Biener <rguenther@suse.de>
34565 PR tree-optimization/78644
34566 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34567 of a simplification result we may not use it at all.
34569 2017-03-28 Richard Biener <rguenther@suse.de>
34572 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34573 without arguments, generate default definition of a SSA name.
34575 2017-03-28 Richard Biener <rguenther@suse.de>
34577 PR middle-end/80222
34578 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34579 TYPE_REF_CAN_ALIAS_ALL references.
34580 * fold-const.c (fold_indirect_ref_1): Likewise.
34582 2017-03-28 Martin Liska <mliska@suse.cz>
34585 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34586 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34588 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34589 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34591 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34592 (EXTRA_SPECS): Define.
34593 (SUBTARGET_EXTRA_SPECS): Likewise.
34594 (SUBTARGET_CPP_SPEC): Likewise.
34595 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34596 SUBTARGET_EXTRA_SPECS.
34597 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34599 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34601 * config/arc/simdext.md (vst64_insn): Update pattern.
34602 (vld32wh_insn): Likewise.
34603 (vld32wl_insn): Likewise.
34604 (vld64_insn): Likewise.
34605 (vld32_insn): Likewise.
34607 2017-03-28 Marek Polacek <polacek@redhat.com>
34610 * fold-const.c (fold_comparison): Use protected_set_expr_location
34611 instead of SET_EXPR_LOCATION.
34613 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34615 * tree.c (add_expr): Avoid name lookup warning.
34617 2017-03-27 Jeff Law <law@redhat.com>
34619 PR tree-optimization/80216
34620 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34621 function name. Limit recursion depth.
34622 (record_temporary_equivalences): Corresponding changes.
34624 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34626 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34629 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34632 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34634 * cfgcleanup.c (reg_note_cfa_p): New array.
34635 (insns_have_identical_cfa_notes): New function.
34636 (old_insns_match_p): Don't cross-jump in between /f
34637 and non-/f instructions. If both i1 and i2 are frame related,
34638 verify all CFA notes, their order and content.
34640 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34643 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34644 HImode and SImode with zero extend to DImode to one insn.
34645 (bswap<mode>2_extenddi): Likewise.
34646 (bswapsi2_extenddi): Likewise.
34647 (bswaphi2_extendsi): Likewise.
34648 (bswaphi2): Combine bswap HImode and SImode into one insn.
34649 Separate memory insns from swapping register.
34650 (bswapsi2): Likewise.
34651 (bswap<mode>2): Likewise.
34652 (bswaphi2_internal): Delete, no longer used.
34653 (bswapsi2_internal): Likewise.
34654 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34655 store, and gpr<-gpr swap insns.
34656 (bswap<mode>2_store): Likewise.
34657 (bswaphi2_reg): Register only splitter, combine with the splitter.
34658 (bswaphi2 splitter): Likewise.
34659 (bswapsi2_reg): Likewise.
34660 (bswapsi2 splitter): Likewise.
34661 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34662 the insns into load, store, and register/register insns.
34663 (bswapdi2_ldbrx): Likewise.
34664 (bswapdi2_load): Likewise.
34665 (bswapdi2_store): Likewise.
34666 (bswapdi2_reg): Likewise.
34668 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34670 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34671 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34673 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34676 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34678 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34679 special handling for target option conflicts between dform
34680 options (-mpower9-dform, -mpower9-dform-vector,
34681 -mpower9-dform-scalar) and -mno-direct-move.
34683 2017-03-27 Richard Biener <rguenther@suse.de>
34685 PR tree-optimization/80181
34686 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34688 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34690 * config/arc/predicates.md (move_double_src_operand): Replace the
34691 call to move_double_src_operand with a call to address_operand.
34693 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34695 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34696 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34697 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34699 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34701 * config/arc/predicates.md (long_immediate_loadstore_operand):
34702 Consider scaled addresses cases.
34704 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34706 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34707 restored when in interrupt.
34708 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34709 doesn't have delay slot.
34711 2017-03-27 Richard Biener <rguenther@suse.de>
34714 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34715 inlined thunk clones.
34717 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34720 * asan.c (instrument_derefs): Copy over last operand from
34721 original COMPONENT_REF to the new COMPONENT_REF with
34722 DECL_BIT_FIELD_REPRESENTATIVE.
34723 * ubsan.c (instrument_object_size): Likewise.
34725 2017-03-27 Richard Biener <rguenther@suse.de>
34727 PR tree-optimization/80170
34728 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34729 sure DR/SCEV didnt fold in constants we do not see when looking
34730 at the reference base alignment.
34732 2017-03-27 Richard Biener <rguenther@suse.de>
34734 PR middle-end/80171
34735 * gimple-fold.c (fold_ctor_reference): Properly guard against
34736 NULL return value from canonicalize_constructor_val.
34738 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34741 * config/i386/i386.c (ix86_expand_builtin)
34742 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34743 flags reg setting and flags reg using instructions.
34744 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34745 clobbering instructions to zero extend op2.
34747 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34749 * doc/install.texi (Configuration) <--with-aix-soname>:
34750 Update link to AIX ld.
34752 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34754 PR rtl-optimization/80160
34755 PR rtl-optimization/80159
34756 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34757 reg_alternate_class into account.
34759 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34762 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34763 to consider in curr_insn_transform.
34765 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34767 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34768 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34769 and emit_mode_inner.
34771 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34773 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34774 argument to the overloaded builtin variants. Use the new flag to
34775 deprecate certain builtin variants.
34776 * config/s390/s390-builtin-types.def: Add new builtin types.
34777 * config/s390/s390-builtins.h: Support new flags field for
34778 overloaded builtins.
34779 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34780 (s390_macro_to_expand): Enable vector float data type.
34781 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34782 builtins by incrementing the __VEC__ version number.
34783 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34785 (s390_resolve_overloaded_builtin): Emit error messages depending
34786 on the builtin flags.
34787 * config/s390/s390.c (s390_expand_builtin): Support additional
34788 flags argument. Change error message to match the messages
34789 emitted in s390-c.c.
34790 * config/s390/s390.md: New UNSPEC_* constants.
34791 (op_type): Add new instruction types.
34792 * config/s390/vecintrin.h: Add new builtins and test data class
34794 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34795 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34796 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34797 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34798 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34799 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34801 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34802 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34803 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34804 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34806 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34807 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34808 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34809 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
34811 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
34812 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
34813 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
34814 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
34815 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
34816 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
34817 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
34819 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
34820 ("vec_scatter_element<V_HW_4:mode>_DI")
34821 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
34822 ("vec_fpint<mode>", "vflls")
34823 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
34824 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
34825 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
34826 ("*vec_cmphe<mode>_cc"): ... these.
34828 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
34829 mode constant instead of magic value.
34831 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34833 * config/s390/s390.c (s390_expand_vec_compare): Support other
34834 vector floating point modes than just V2DF.
34835 (s390_expand_vcond): Likewise.
34836 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
34837 (s390_cannot_change_mode_class): Prevent mode changes between TF
34838 and V1TF in vector registers.
34839 * config/s390/s390.md (DF, SF): New mode attributes.
34840 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
34841 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
34842 SFmode support for VRs.
34843 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
34845 (VFT, VF_HW): New mode iterators.
34846 (vw, sdx): New mode attributes.
34847 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
34848 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
34849 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
34850 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
34851 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
34852 also the new vector floating point modes. Renaming to ...
34854 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
34855 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
34856 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
34857 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
34858 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
34859 ("vec_unordered<mode>"): ... these.
34861 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
34862 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
34863 ("*vec_extendv2df"): New insn definitions.
34865 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34867 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
34868 ("mulditi3_2", "*muldi3_sign"): New patterns.
34869 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
34870 rename the pattern definition.
34872 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34874 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
34876 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
34878 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34880 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
34881 instruction if possible.
34882 * config/s390/vector.md (vec_halfnumelts): New mode
34884 ("*vec_vllezlf<mode>"): New pattern.
34886 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34888 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
34889 ("popcountv4si2", "popcountv2di2"): Rename to ...
34890 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
34891 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
34893 ("popcount<mode>2_vxe"): New pattern.
34895 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34897 * common/config/s390/s390-common.c (processor_flags_table): Add
34899 * config.gcc: Add arch12.
34900 * config/s390/driver-native.c (s390_host_detect_local_cpu):
34901 Default to arch12 for unknown CPU model numbers.
34902 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
34903 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
34904 PROCESSOR_max sanity check.
34905 * config/s390/s390-opts.h (enum processor_type): Add
34907 * config/s390/s390.c (processor_table): Add arch12.
34908 (s390_expand_builtin): Add check for B_VXE flag.
34909 (s390_issue_rate): Add PROCESSOR_ARCH12.
34910 (s390_get_sched_attrmask): Likewise.
34911 (s390_get_unit_mask): Likewise.
34912 (s390_sched_score): Enable z13 scheduling for arch12.
34913 (s390_sched_reorder): Likewise.
34914 (s390_sched_variable_issue): Likewise.
34915 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
34917 (s390_tune_attr): Use z13 scheduling also for arch12.
34918 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
34919 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
34920 (TARGET_VXE_P): New macros.
34921 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
34922 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
34923 * config/s390/s390.opt: Add arch12 as processor_type.
34925 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34927 * config/s390/s390.md
34928 ("fixuns_truncdddi2", "fixuns_trunctddi2")
34929 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
34930 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
34932 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
34933 Rename expanders to ...
34935 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
34936 ("fixuns_truncdddi2_emu"): ... these.
34938 ("fixuns_trunc<mode>si2_emu"): New expander.
34940 ("*fixuns_truncdfdi2_z13"): Rename to ...
34941 ("*fixuns_truncdfdi2_vx"): ... this.
34943 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34945 * config/s390/2964.md: Remove the single element vector compare
34946 instructions which are no longer used.
34947 * config/s390/s390.c (s390_select_ccmode): Remove handling of
34949 (s390_canonicalize_comparison): Remove handling of DFmode
34951 (s390_expand_vec_compare_scalar): Remove function.
34952 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
34953 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
34955 ("*cmp<mode>_ccs"): Add wfcdb instruction.
34957 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34959 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
34961 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
34962 will anyway by matched by mov<mode>_64dfp.
34964 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34966 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
34967 vlef/vstef. Add missing operand to vleif.
34969 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34971 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
34972 pair for all vector types with 64 bit elements.
34973 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
34974 * config/s390/vector.md (V_HW_64): ... here.
34975 (V_128_NOSINGLE): New mode iterator.
34976 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
34977 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
34978 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
34979 ("*vec_load_pairv2di"): Change to ...
34980 ("*vec_load_pair<mode>"): ... this one.
34982 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34984 * config/s390/constraints.md: Add comments.
34985 (jKK): Reject element sizes > 8 bytes.
34986 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
34988 * config/s390/s390.md: Add the s_operand checks formerly in
34989 s390_split_ok_p to various splitters where they are still
34991 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
34992 for 128 bit vectors. Plus two splitters.
34994 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34996 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
34999 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35002 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35003 error if the boundary argument is not constant.
35005 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35007 PR rtl-optimization/80112
35008 * loop-doloop.c (doloop_condition_get): Don't check condition
35009 if cmp isn't SET with IF_THEN_ELSE src.
35011 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35013 PR tree-optimization/80158
35014 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35015 replacing a candidate statement, also replace it for the
35016 candidate's alternate interpretation.
35017 (replace_rhs_if_not_dup): Likewise.
35018 (replace_one_candidate): Likewise.
35020 2017-03-24 Richard Biener <rguenther@suse.de>
35022 PR tree-optimization/80167
35023 * graphite-isl-ast-to-gimple.c
35024 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35026 (translate_isl_ast_to_gimple::get_rename): Likewise.
35028 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35030 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35031 handling of certain combinations of target options, including the
35032 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35033 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35035 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35038 * config/arm/arm.md (*load_multiple): Add reload_completed to
35039 matching condition.
35041 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35042 Richard Biener <rguenth@suse.de>
35044 PR tree-optimization/79908
35045 PR tree-optimization/80136
35046 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35047 been cast away, gimplify_and_add suffices.
35049 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35051 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35053 2017-03-23 Richard Biener <rguenther@suse.de>
35055 PR tree-optimization/80032
35056 * gimplify.c (gimple_push_cleanup): Forced unconditional
35057 cleanups still have to go to the conditional_cleanups
35060 2017-03-22 Jakub Jelinek <jakub@redhat.com>
35062 PR tree-optimization/80072
35063 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
35065 (next_operand_entry_id): Change type to unsigned int.
35066 (sort_by_operand_rank): Make sure to return the right return value
35067 even if unsigned fields are bigger than INT_MAX.
35068 (struct oecount): Change cnt and id type to unsigned int.
35069 (oecount_hasher::equal): Formatting fix.
35070 (oecount_cmp): Make sure to return the right return value
35071 even if unsigned fields are bigger than INT_MAX.
35072 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
35075 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
35076 TREE_READONLY on result if writing it more than once.
35079 * doc/invoke.texi (-fsanitize=thread): Document that with
35080 -fnon-call-exceptions atomics are not able to throw
35084 * tsan.c: Include tree-eh.h.
35085 (instrument_builtin_call): Call maybe_clean_eh_stmt or
35086 maybe_clean_or_replace_eh_stmt where needed.
35087 (instrument_memory_accesses): Add cfg_changed argument.
35088 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
35089 if it returned true.
35090 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
35092 PR rtl-optimization/63191
35093 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
35094 wrapper function, moved the whole old content into ...
35095 (ix86_delegitimize_address_1): ... this. New inline function.
35096 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
35097 true as last argument instead of ix86_delegitimize_address.
35099 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35101 * config/aarch64/aarch64.c (generic_branch_cost): Copy
35102 cortexa57_branch_cost.
35104 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35106 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
35108 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35111 * doc/md.texi (Constraints): Document wA constraint.
35112 * config/rs6000/constraints.md (wA): New.
35113 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
35114 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
35115 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
35116 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
35118 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
35121 * gimplify.c (is_oacc_declared): New function.
35122 (oacc_default_clause): Use it to set default flags for acc declared
35123 variables inside parallel regions.
35124 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
35125 declared variables.
35126 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
35127 declare attribute to any decl as necessary.
35129 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
35132 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
35133 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
35134 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
35135 (arm_arch_lpae): This.
35136 * config/arm/arm.c (arm_arch7ve): Rename into ...
35137 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
35138 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
35141 2017-03-22 Martin Liska <mliska@suse.cz>
35144 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
35145 error message instead of an ICE.
35147 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35149 * doc/extend.texi (6.11 Additional Floating Types): Revise.
35151 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
35153 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35155 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35158 2017-03-21 Martin Sebor <msebor@redhat.com>
35160 * doc/extend.texi: Use "cannot" instead of "can't."
35161 * doc/hostconfig.texi: Same.
35162 * doc/install.texi: Same.
35163 * doc/invoke.texi: Same.
35164 * doc/loop.texi: Same.
35165 * doc/md.texi: Same.
35166 * doc/objc.texi: Same.
35167 * doc/rtl.texi: Same.
35168 * doc/tm.texi: Same.
35169 * doc/tm.texi.in: Same.
35170 * doc/trouble.texi: Same.
35172 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
35175 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
35176 (collect_checksum_attributes): Set it.
35177 (die_checksum_ordered): Use it.
35179 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35181 PR tree-optimization/79908
35182 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
35183 change: For a VA_ARG whose LHS has been cast away, use
35184 force_gimple_operand to construct the side effects.
35186 2017-03-21 David Malcolm <dmalcolm@redhat.com>
35188 PR translation/80001
35189 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
35190 more amenable to translation.
35191 (oacc_loop_auto_partitions): Likewise.
35193 2017-03-21 Marek Polacek <polacek@redhat.com>
35194 Martin Sebor <msebor@redhat.com>
35196 PR tree-optimization/80109
35197 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
35198 on INTEGRAL_TYPE_P.
35200 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35201 Segher Boessenkool <segher@kernel.crashing.org>
35204 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
35205 check reg_used_between_p between insn and one of succ or succ2
35206 depending on if succ is artificial insn not inserted into insn
35209 2017-03-21 Martin Liska <mliska@suse.cz>
35211 PR gcov-profile/80081
35212 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
35213 * doc/gcc.texi: Include gcov-dump stuff.
35214 * doc/gcov-dump.texi: New file.
35216 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
35218 PR rtl-optimization/79150
35219 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
35220 conditional jump, if the jump is the last insn of the loop.
35222 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35223 Richard Biener <rguenth@suse.de>
35225 PR tree-optimization/79908
35226 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35227 been cast away, use force_gimple_operand to construct the side
35230 2017-03-21 Martin Liska <mliska@suse.cz>
35232 PR libfortran/79956
35233 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
35236 2017-03-21 Brad Spengler <spender@grsecurity.net>
35239 * plugin.c (htab_hash_plugin): New function.
35240 (add_new_plugin): Use it and adjust.
35241 (parse_plugin_arg_opt): Adjust.
35242 (init_one_plugin): Likewise.
35244 2017-03-21 Richard Biener <rguenther@suse.de>
35246 PR tree-optimization/80032
35247 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
35248 if set force the cleanup to happen unconditionally.
35249 (gimplify_target_expr): Push inserted clobbers with force_uncond
35250 to avoid them being removed by control-dependent DCE.
35252 2017-03-21 Richard Biener <rguenther@suse.de>
35254 PR tree-optimization/80122
35255 * tree-inline.c (copy_bb): Do not expans va-arg packs or
35256 va_arg_pack_len when the inlined call stmt requires pack
35258 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
35260 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35263 * tsan.c (instrument_builtin_call): If the memory model argument
35264 is not a constant, assume it is valid.
35267 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
35270 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35272 PR rtl-optimization/79910
35273 * combine.c (can_combine_p): Do not allow combining an I0 or I1
35274 if its dest is used by an insn before I2 (other than the combined
35275 insns themselves, which are properly handled already).
35277 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35280 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35282 * combine.c (record_used_regs): New static function.
35283 (try_combine): Handle situations where there is an additional
35284 instruction between I2 and I3 which needs to have a LOG_LINK
35288 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35290 * combine.c (try_combine): Delete redundant i1 test. Call
35291 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35293 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35296 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
35297 alternatives 13/14.
35299 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35301 PR tree-optimization/80054
35302 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
35303 the optimization if a PHI or any of its arguments is not dominated
35304 by the candidate's basis. Use gphi* rather than gimple* as
35306 (replace_profitable_candidates): Clean up a gimple* variable that
35307 should be a gphi* variable.
35309 2017-03-20 Martin Sebor <msebor@redhat.com>
35312 * doc/extend.texi (attribute constructor): Document present limitation.
35314 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
35317 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
35318 __POWER9_VECTOR__ #ifdef control, change template definition to
35319 use Power9-specific built-in function.
35320 (vec_any_eq): Likewise.
35321 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
35322 to control outcomes from this test.
35323 (vector_ae_<mode>p): For VEC_F modes, likewise.
35325 2017-03-20 Ian Lance Taylor <iant@google.com>
35327 * config/i386/i386.c (ix86_function_regparm): Save an extra
35328 register for -fsplit-stack with DECL_STATIC_CHAIN.
35330 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35333 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
35334 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
35336 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35338 * config/riscv/riscv.c (riscv_print_operand): Use "fence
35340 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
35343 2017-03-20 Marek Polacek <polacek@redhat.com>
35346 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
35348 2017-03-20 Richard Biener <rguenther@suse.de>
35350 PR tree-optimization/80113
35351 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
35352 allocate extra SSA name for PHI def.
35353 (add_close_phis_to_outer_loops): Likewise.
35354 (add_close_phis_to_merge_points): Likewise.
35355 (copy_loop_close_phi_args): Likewise.
35356 (copy_cond_phi_nodes): Likewise.
35358 2017-03-20 Martin Liska <mliska@suse.cz>
35360 PR middle-end/79753
35361 * tree-chkp.c (chkp_build_returned_bound): Do not build
35362 returned bounds for a LHS that's not a BOUNDED_P type.
35364 2017-03-20 Martin Liska <mliska@suse.cz>
35368 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
35369 COMPLEX_CST and VECTOR_CST.
35371 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35374 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
35375 target operand. A new splitter adds the clobber statement in case
35376 the target operand is dead anyway.
35378 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
35380 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
35381 to age-old versions of binutils and glibc.
35383 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
35385 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
35387 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35389 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
35391 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35393 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
35394 requirement for binutils 2.13.
35396 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35398 * combine.c (try_combine): Delete redundant i1 test. Call
35399 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35401 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
35403 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
35404 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
35406 <riscv64-*-elf>: Re-arrange section
35407 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
35408 <riscv32-*-linux>: Likewise.
35409 <riscv64-*-elf>: Likewise
35410 <riscv64-*-linux>: Likewise.
35412 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
35415 * aarch64.opt(verbose-cost-dump): Fix typo.
35417 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
35420 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
35421 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
35423 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35425 * reload.c (find_reloads): When reloading a nonoffsettable address,
35426 use RELOAD_OTHER for it and its address reloads.
35428 PR rtl-optimization/79910
35429 * combine.c (record_used_regs): New static function.
35430 (try_combine): Handle situations where there is an additional
35431 instruction between I2 and I3 which needs to have a LOG_LINK
35434 2017-03-17 Jeff Law <law@redhat.com>
35436 PR tree-optimization/71437
35437 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
35438 conditional in the hash table first.
35439 (vrp_dom_walker::before_dom_children): Extract condition from
35440 ASSERT_EXPR. Record condition, its inverion and any implied
35441 conditions as well.
35443 2017-03-17 Marek Polacek <polacek@redhat.com>
35444 Markus Trippelsdorf <markus@trippelsdorf.de>
35446 PR tree-optimization/80079
35447 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
35450 2017-03-17 Richard Biener <rguenther@suse.de>
35452 PR middle-end/80075
35453 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
35454 Properly verify the LHS before the RHS possibly claims to be
35456 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
35459 2017-03-17 Martin Jambor <mjambor@suse.cz>
35461 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
35462 (List of -O2 options): Likewise.
35463 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
35466 2017-03-17 Tom de Vries <tom@codesourcery.com>
35468 * gcov-dump.c (print_usage): Print bug_report_url.
35470 2017-03-17 Richard Biener <rguenther@suse.de>
35472 PR middle-end/80050
35473 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
35474 (parser::peek): Likewise.
35476 2017-03-17 Richard Biener <rguenther@suse.de>
35478 PR tree-optimization/80048
35479 * sese.c (free_sese_info): Properly release rename_map and
35480 copied_bb_map elements.
35482 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
35484 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
35485 Add linked-list forward and backlinks. Insert on
35486 construction, remove on destruction.
35487 (class pass_store_merging): Add m_stores_head field.
35488 (pass_store_merging::terminate_and_process_all_chains):
35489 Iterate over m_stores_head list.
35490 (pass_store_merging::terminate_all_aliasing_chains):
35492 (pass_store_merging::execute): Check for debug stmts first.
35493 Push new chains onto the m_stores_head stack.
35495 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
35498 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
35499 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
35500 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
35502 2017-03-16 Jeff Law <law@redhat.com>
35504 PR tree-optimization/71437
35505 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
35506 member function. Implementation moved into after_dom_children
35507 member function and into the threader's thread_outgoing_edges
35509 (dom_opt_dom_walker::after_dom_children): Simplify by moving
35510 some code into new thread_outgoing_edges.
35511 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
35512 definition. Simplify marker handling (do it here). Assume we always
35513 have the available expression and the const/copies tables.
35514 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
35516 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
35517 * tree-vrp.c (equiv_stack): No longer file scoped.
35518 (vrp_dom_walker): New class.
35519 (vrp_dom_walker::before_dom_children): New member function.
35520 (vrp_dom_walker::after_dom_children): Likewise.
35521 (identify_jump_threads): Setup domwalker. Use it rather than
35522 walking edges in a random order by hand. Simplify setup/finalization.
35523 (finalize_jump_threads): Remove.
35524 (vrp_finalize): Do not call identify_jump_threads here.
35525 (execute_vrp): Do it here instead and call thread_through_all_blocks
35528 PR tree-optimization/71437
35529 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
35531 (simplify_stmt_for_jump_threading): Add basic_block argument. All
35533 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
35534 (dom_opt_dom_walker::thread_across_edge): Remove
35535 handle_dominating_asserts argument. All callers changed.
35536 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
35537 changes. Remove calls to lhs_of_dominating_assert. Other
35538 uses of handle_dominating_asserts turn into unconditional code
35539 (simplify_control_stmt_condition_1): Likewise.
35540 (simplify_control_stmt_condition): Likewise.
35541 (thread_through_normal_block, thread_across_edge): Likewise.
35542 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
35543 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
35544 object if it is not an SSA_NAME.
35545 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
35546 before calling into the VRP specific simplifiers.
35547 (identify_jump_threads): Remove handle_dominating_asserts
35550 2017-03-16 Jakub Jelinek <jakub@redhat.com>
35553 * tree-diagnostic.c (default_tree_printer): No longer static.
35554 * tree-diagnostic.h (default_tree_printer): New prototype.
35556 2017-03-16 Tamar Christina <tamar.christina@arm.com>
35558 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
35559 Change ins into fmov.
35561 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35563 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
35564 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
35565 Use h_con constraint for operand 1.
35566 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35567 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35569 2017-03-15 Jeff Law <law@redhat.com>
35571 PR tree-optimization/71437
35572 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35573 (record_temporary_equivalences): Use it.
35575 PR tree-optimization/71437
35576 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35577 tree-ssa-scopedtables.
35578 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35579 (record_conditions, record_cond, vuse_eq): Likewise.
35580 (record_edge_info): Adjust to API tweak of record_conditions.
35581 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35582 (record_temporary_equivalences, optimize_stmt): Likewise.
35583 (eliminate_redundant_computations): Likewise.
35584 (record_equivalences_from_stmt): Likewise.
35585 * tree-ssa-scopedtables.c: Include options.h and params.h.
35586 (vuse_eq): New function, moved from tree-ssa-dom.c
35587 (build_and_record_new_cond): Likewise.
35588 (record_conditions): Likewise. Accept vector of conditions rather
35589 than edge_equivalence structure for first argument.
35590 for the first argument.
35591 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35592 from tree-ssa-dom.c.
35593 (avail_exprs_stack::record_cond): Likewise.
35594 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35595 from tree-ssa-dom.c.
35596 (avail_exprs_stack): Add new member functions lookup_avail_expr
35598 (record_conditions): Declare.
35600 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35603 * lra-constraints.c (process_alt_operands): Increase reject for
35604 reloading an input/output operand.
35606 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35609 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35610 insns to convert from signed/unsigned char/short to IEEE 128-bit
35612 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35614 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35617 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35618 subreg of inner mode for values already in registers.
35620 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35622 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35623 iteration reg is used after the loop.
35625 2017-03-14 Martin Sebor <msebor@redhat.com>
35627 PR tree-optimization/79800
35628 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35629 precision in negative-positive range.
35630 (format_floating): Call non-const overload with adjusted precision.
35632 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35635 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35638 2017-03-14 Martin Sebor <msebor@redhat.com>
35640 PR middle-end/80020
35641 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35642 * builtins.def (aligned_alloc): Use it.
35645 * Makefile.in (GTFILES): Add calls.c.
35646 * calls.c: Include "gt-calls.h".
35648 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35650 PR rtl-optimization/79728
35651 * regs.h (struct target_regs): New field
35652 x_contains_allocatable_regs_of_mode.
35653 (contains_allocatable_regs_of_mode): New macro.
35654 * reginfo.c (init_reg_sets_1): Initialize it, and change
35655 contains_reg_of_mode so it includes global regs as well.
35656 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35657 rather than contains_regs_of_mode.
35659 2017-03-14 Martin Liska <mliska@suse.cz>
35661 * doc/invoke.texi: Document options that can't be combined with
35662 -fcheck-pointer-bounds.
35664 2017-03-14 Martin Liska <mliska@suse.cz>
35666 PR middle-end/79831
35667 * doc/invoke.texi (-Wchkp): Document the option.
35669 2017-03-14 Martin Liska <mliska@suse.cz>
35671 * Makefile.in: Install gcov-dump.
35673 2017-03-14 Martin Liska <mliska@suse.cz>
35675 * multiple_target.c (expand_target_clones): Bail out for
35676 an invalid attribute.
35678 2017-03-14 Richard Biener <rguenther@suse.de>
35680 * alias.c (struct alias_set_entry): Pack properly.
35681 * cfgloop.h (struct loop): Likewise.
35682 * cse.c (struct set): Likewise.
35683 * ipa-utils.c (struct searchc_env): Likewise.
35684 * loop-invariant.c (struct invariant): Likewise.
35685 * lra-remat.c (struct cand): Likewise.
35686 * recog.c (struct change_t): Likewise.
35687 * rtl.h (struct address_info): Likewise.
35688 * symbol-summary.h (function_summary): Likewise.
35689 * tree-loop-distribution.c (struct partition): Likewise.
35690 * tree-object-size.c (struct object_size_info): Likewise.
35691 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35692 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35693 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35694 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35695 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35696 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35697 (struct _stmt_vec_info): Likewise.
35699 2017-03-14 Martin Liska <mliska@suse.cz>
35702 * multiple_target.c (create_dispatcher_calls): Check that
35703 a target can create a function dispatcher.
35705 2017-03-14 Martin Liska <mliska@suse.cz>
35708 * multiple_target.c (expand_target_clones): Drop local.local
35709 flag for default implementation.
35711 2017-03-14 Richard Biener <rguenther@suse.de>
35713 PR tree-optimization/80030
35714 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35716 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35718 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35719 gcc_fallthrough() instead of __attribute__((fallthrough));
35721 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35723 * doc/gcc.texi: Remove "up" link to (DIR).
35724 * doc/gccint.texi: Ditto.
35726 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35728 * doc/install.texi (Specific) <avr>: Remove reference to
35731 2017-03-13 Jeff Law <law@redhat.com>
35733 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35734 attribute rather than comments.
35736 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35737 match_scratch operand is highest.
35739 2017-03-13 Martin Liska <mliska@suse.cz>
35741 PR middle-end/78339
35742 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35743 is a CHKP clone, use original declaration.
35745 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35747 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35748 (arc_conditional_register_usage): Use a different allocation order
35749 when optimizing for size.
35750 * common/config/arc/arc-common.c (arc_option_optimization_table):
35751 Section anchors default on when optimizing for size.
35753 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35755 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35757 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35759 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35760 * config/arc/arc.md (cpu_facility): Add cd variant.
35761 (*movqi_insn): Add code density variant.
35762 (*movhi_insn): Likewise.
35763 (*movqi_insn): Likewise.
35764 (*addsi3_mixed): Likewise.
35765 (subsi3_insn): Likewise.
35767 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35769 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35771 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35773 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35774 expressions with MINUS and UNARY ops.
35776 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35779 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35781 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35782 between vec_select and vector argument.
35783 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35784 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35785 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35786 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35787 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35788 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35790 2017-03-13 Richard Biener <rguenther@suse.de>
35793 * params.def (vect-max-peeling-for-alignment): Fix typo.
35795 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35797 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35798 issue that only occurred with binutils below 2.18.
35800 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35802 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35803 refer to binutils 2.11/2.12 minimum.
35805 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35807 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35808 ftp.kernel.org and simplify binutils requirement.
35810 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
35812 * doc/invoke.texi (Warning Options): Fix spelling of link-time
35814 (Optimize Options): Ditto. Also remove redundancy.
35816 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35818 PR translation/79848
35819 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
35821 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
35822 to G_ to avoid double translation.
35824 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35826 PR translation/79923
35827 * auto-profile.c (get_combined_location): Convert leading
35828 character of diagnostics to lower case and remove trailing period.
35829 (read_profile): Likewise for various diagnostics.
35830 * config/arm/arm.c (arm_option_override): Remove trailing period
35831 from various diagnostics.
35832 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
35833 (msp430_expand_delay_cycles): Likewise.
35835 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35838 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
35839 full command-line argument, rather than just "str".
35840 (aarch64_validate_march): Likewise.
35841 (aarch64_validate_mtune): Likewise.
35843 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
35845 PR rtl-optimization/78911
35846 * lra-assigns.c (must_not_spill_p): New function.
35847 (spill_for): Use it.
35849 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35851 PR tree-optimization/79981
35852 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
35853 ATOMIC_COMPARE_EXCHANGE ifn result.
35854 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
35855 IFN_ATOMIC_COMPARE_EXCHANGE.
35857 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35860 * opts.c (parse_sanitizer_options): Add missing question mark to
35861 "did you mean" message.
35863 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35865 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
35867 (VMULEUH_UNS): Likewise.
35868 (VMULOUB_UNS): Likewise.
35869 (VMULOUH_UNS): Likewise.
35870 * config/rs6000/rs6000.c (builtin_function_type): Remove
35871 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
35873 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35876 * read-rtl-function.c (function_reader::read_rtx_operand): Update
35877 x with result of extra_parsing_for_operand_code_0.
35878 (function_reader::extra_parsing_for_operand_code_0): Convert
35879 return type from void to rtx, returning x. When reading
35880 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
35881 larger size containing struct block_symbol.
35883 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
35885 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
35886 -mfloat128-hardware without -m64.
35888 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
35891 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
35892 entries to the case statement that marks unsigned arguments to
35893 overloaded functions.
35895 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
35897 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
35898 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
35900 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
35903 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
35904 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
35906 2017-03-10 Martin Liska <mliska@suse.cz>
35910 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
35911 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
35914 2017-03-10 Olivier Hainque <hainque@adacore.com>
35916 * tree-switch-conversion (array_value_type): Start by resetting
35917 candidate type to it's main variant.
35919 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35921 PR rtl-optimization/79909
35922 * combine.c (try_combine): Use simplify_replace_rtx on individual
35923 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
35924 of the whole CALL_INSN_FUNCTION_USAGE.
35926 PR tree-optimization/79972
35927 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
35928 get_range_info on SSA_NAMEs. Formatting fixes.
35930 2017-03-10 Richard Biener <rguenther@suse.de>
35931 Jakub Jelinek <jakub@redhat.com>
35933 PR tree-optimization/77975
35934 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
35935 edge to be constant.
35936 (get_val_for): For constant x return it. Formatting fix.
35937 (loop_niter_by_eval): Avoid pointless looping if the next iteration
35938 would use the same bases as the current one.
35940 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35942 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
35943 instead of vec_select for V1TImode.
35944 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
35946 (VSX_LE_128): Add V1TI to this mode iterator.
35947 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
35948 (*vsx_le_perm_store_<mode>): Likewise.
35949 (pre-reload splitter for VSX stores): Likewise.
35950 (post-reload splitter for VSX stores): Likewise.
35951 (*vsx_xxpermdi2_le_<mode>): Likewise.
35952 (*vsx_lxvd2x2_le_<mode>): Likewise.
35953 (*vsx_stxvd2x2_le_<mode>): Likewise.
35955 2017-03-09 Michael Eager <eager@eagercon.com>
35957 Correct failures with --enable-checking=yes,rtl.
35959 * config/microblaze/microblaze.c (microblaze_expand_shift):
35960 Replace GET_CODE test with CONST_INT_P and INTVAL test with
35961 test for const0_rtx.
35962 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
35963 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
35965 2017-03-09 Richard Biener <rguenther@suse.de>
35967 PR tree-optimization/79977
35968 * graphite-scop-detection.c (scop_detection::merge_sese):
35969 Handle the case of extra exits to blocks dominating the entry.
35971 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
35973 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
35976 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
35978 PR rtl-optimization/79949
35979 * lra-constraints.c (process_alt_operands): Check memory when
35980 trying to predict a cycle. Print about the overall increase.
35982 2017-03-09 Richard Biener <rguenther@suse.de>
35984 PR middle-end/79971
35985 * gimple-expr.c (useless_type_conversion_p): Preserve
35986 TYPE_SATURATING for fixed-point types.
35988 2017-03-09 Richard Biener <rguenther@suse.de>
35991 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
35992 alignment of BLKmode params.
35994 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35997 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
35998 (VALL_NO_V2Q): Likewise.
36000 * config/aarch64/aarch64-simd.md
36001 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36003 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36004 VALL_NO_V2Q mode iterator.
36005 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36007 2017-03-09 Martin Liska <mliska@suse.cz>
36009 PR tree-optimization/79631
36010 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36011 tree_int_cst_sign_bit just for INTEGER constants.
36013 2017-03-09 Martin Liska <mliska@suse.cz>
36017 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36020 2017-03-09 Marek Polacek <polacek@redhat.com>
36023 * tree.c (inchash::add_expr): Handle TREE_VEC.
36025 2017-03-09 Martin Liska <mliska@suse.cz>
36028 (chkp_narrow_size_and_offset): New function.
36029 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36030 (void chkp_parse_bit_field_ref): New function.
36031 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36032 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36034 2017-03-09 Martin Liska <mliska@suse.cz>
36037 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36038 (chkp_find_bounds_1): Remove gcc_unreachable.
36040 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36043 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36044 BUILT_IN_SYNC*, determine the access type from the size suffix and
36045 always build a MEM_REF with that type. Handle forgotten
36046 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36049 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36050 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36051 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36052 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36053 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
36054 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
36055 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
36056 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
36057 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
36058 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
36059 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
36060 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
36061 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
36062 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
36063 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
36064 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
36065 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
36066 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
36067 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
36068 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
36069 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
36070 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
36071 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
36072 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
36073 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
36074 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
36075 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
36076 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
36077 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
36078 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
36079 definitions outside of __OPTIMIZE__ guarded section.
36082 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
36083 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
36084 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
36085 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
36088 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36090 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
36091 ("vfenez<mode>"): Add missing constraints.
36093 2017-03-08 Martin Sebor <msebor@redhat.com>
36096 * config/nds32/nds32.c (nds32_option_override):
36097 Fix misspelled diagnostic.
36099 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36102 * gimplify.c (gimplify_omp_for): Replace index var in outer
36103 taskloop statement with an artificial variable and add
36104 OMP_CLAUSE_PRIVATE clause for it.
36106 2017-03-08 Richard Biener <rguenther@suse.de>
36108 PR tree-optimization/79955
36109 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
36110 for accesses that are completely outside of the variable.
36112 2017-03-08 Andrew Haley <aph@redhat.com>
36114 PR tree-optimization/79943
36115 * tree-ssa-loop-split.c (compute_new_first_bound): When
36116 calculating the new upper bound, (END-BEG) should be added, not
36119 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36121 * config/avr/avr.md (setmemhi): Make sure match_dup
36122 operand number comes before match_scratch.
36124 2017-03-08 Richard Biener <rguenther@suse.de>
36126 PR tree-optimization/79920
36127 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
36128 with ncopies == 1 to ...
36129 (vect_transform_slp_perm_load): ... here. Properly compute
36130 all element loads by iterating VF times over the group. Do
36131 not handle ncopies (computed in a broken way) in
36132 vect_create_mask_and_perm.
36134 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36137 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
36138 is a uniform vector, use uniform_vector_p return value instead of
36139 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
36141 2017-03-07 Marek Polacek <polacek@redhat.com>
36143 PR middle-end/79809
36144 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
36145 (alloca_call_type): Likewise.
36147 2017-03-07 Martin Liska <mliska@suse.cz>
36149 * gcov.c (process_args): Put comment to correct location.
36151 2017-03-07 Martin Liska <mliska@suse.cz>
36153 PR middle-end/68270
36154 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
36155 Use array_at_struct_end_p instead of DECL_CHAIN (field).
36156 (chkp_narrow_bounds_for_field): Likewise.
36157 (chkp_parse_array_and_component_ref): Pass one more argument to
36160 2017-03-07 Richard Biener <rguenther@suse.de>
36162 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
36165 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
36167 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
36168 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
36170 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36173 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
36174 to end of description.
36175 (PARAM_MAX_STORES_TO_MERGE): Likewise.
36177 2017-03-07 Jakub Jelinek <jakub@redhat.com>
36179 PR rtl-optimization/79901
36180 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
36182 (*avx512f_<code><mode>3<mask_name>): ... this.
36183 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
36184 iterator instead of VI8_AVX2_AVX512BW.
36186 PR rtl-optimization/79901
36187 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
36188 min/max expander, expand it using expand_vec_cond_expr.
36191 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
36194 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36197 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
36198 to void * for PCH reasons.
36199 * dwarf2out.c (output_loc_operands, output_die): Cast
36200 v.val_vec.array to unsigned char *.
36202 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
36205 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
36208 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
36210 PR rtl-optimization/79571
36211 * lra-constraints.c (process_alt_operands): Calculate static
36212 reject and subtract it from overall when only addresses will be
36215 2017-03-06 Julia Koval <julia.koval@intel.com>
36218 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
36219 incoming stack boundary to 128 for 64-bit targets.
36221 2017-03-06 Richard Biener <rguenther@suse.de>
36223 PR tree-optimization/79894
36224 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
36225 to NULL after folding it.
36227 2017-03-06 Richard Biener <rguenther@suse.de>
36229 PR tree-optimization/79824
36230 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
36231 check disabling peeling for gaps.
36233 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
36235 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
36236 attributes): Document gettimeofday.
36238 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36240 * config/s390/s390.c (s390_option_override_internal): Set
36241 PARAM_MIN_VECT_LOOP_BOUND
36243 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36245 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
36246 * config/s390/s390.md: Likewise.
36248 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36251 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
36252 (<avx2_avx512>_perm<mode>): Rename to ...
36253 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
36255 (<avx512>_perm<mode>_mask): Rename to ...
36256 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
36258 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
36259 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
36260 instead of VI8F_256_512.
36261 (avx512f_perm<mode>): New define_expand.
36262 (avx512f_perm<mode>_mask): Likewise.
36263 (avx512f_perm<mode>_1<mask_name>): New define_insn.
36264 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
36266 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36268 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
36269 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
36271 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
36273 2017-03-06 Martin Liska <mliska@suse.cz>
36276 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
36277 when having a SSA NAME w/o VAR_DECL assigned to it.
36279 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36281 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
36282 msa_dpsub_<su>_d): Fix MODE for vec_select.
36284 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36286 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
36288 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
36290 2017-03-06 Richard Biener <rguenther@suse.de>
36292 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
36293 * plugin.c (register_plugin_info): Likewise.
36294 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
36296 2017-03-05 Jakub Jelinek <jakub@redhat.com>
36298 * config/i386/sse.md (sse_storehps, sse_storelps,
36299 avx_<castmode><avxsizesuffix>_<castmode>,
36300 avx512f_<castmode><avxsizesuffix>_<castmode>,
36301 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
36302 in condition that at least one operand is not a MEM.
36304 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36306 PR middle-end/79805
36307 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
36308 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
36310 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
36311 gimple_call_nothrow_p flag based on whether original builtin can throw.
36312 If it can, emit following stmts on the fallthrough edge.
36313 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
36314 don't create new bb if inserting just debug stmts on the edge, try to
36315 insert them on the fallthru bb or just reset debug stmts.
36317 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
36320 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
36321 restore recog_data (including the operand rtxes inside it) around
36322 the call to get_insn_template.
36324 2017-03-03 Martin Sebor <msebor@redhat.com>
36326 PR tree-optimization/79699
36327 * context.c (context::~context): Free MPFR caches to avoid
36328 a memory leak on program exit.
36330 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36332 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
36333 Use wide_int::ulow () instead of .elt (0).
36335 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36337 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
36338 (*pushxf): Limit oF constraint to 32bit targets and add oC
36339 constraint for 64bit targets.
36340 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
36341 (*pushdf): Change rmF constraint to rmC.
36343 2017-03-03 Martin Liska <mliska@suse.cz>
36345 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
36346 Remove unused variable.
36348 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36351 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
36352 is a memory operand, increase num_memory.
36353 (ix86_expand_args_builtin): Likewise.
36355 2017-03-03 Jan Hubicka <jh@suse.cz>
36358 * ipa-devirt.c (maybe_record_node): Properly handle
36359 __cxa_pure_virtual visibility.
36361 2017-03-03 Martin Liska <mliska@suse.cz>
36363 PR tree-optimization/79803
36364 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
36366 (pass_loop_prefetch::execute): Disabled optimization if an
36367 assumption about L1 cache size is not met.
36369 2017-03-03 Martin Liska <mliska@suse.cz>
36371 PR rtl-optimization/79574
36372 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
36373 (hash_scan_set): Likewise.
36374 (dump_hash_table): Likewise.
36375 (hoist_code): Likewise.
36377 2017-03-03 Richard Biener <rguenther@suse.de>
36379 * fixed-value.c (fixed_from_string): Restore use of elt (1)
36380 in place of uhigh ().
36381 (fixed_convert_from_real): Likewise.
36383 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36386 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
36388 2017-03-03 Richard Biener <rguenther@suse.de>
36390 PR middle-end/79818
36391 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
36392 TYPE_OVERFLOW_UNDEFINED check.
36394 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36396 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
36398 (vector_ae_<mode>_p): Likewise.
36399 (vector_nez_<mode>_p): Likewise.
36400 (vector_ne_v2di_p): Likewise.
36401 (vector_ae_v2di_p): Likewise.
36402 (vector_ne_<mode>_p): Likewise.
36403 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
36405 (vsx_tsqrt<mode>2_fe): Likewise.
36407 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
36410 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
36412 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36414 PR rtl-optimization/79780
36415 * cprop.c (one_cprop_pass): When second and further conditional trap
36416 in a single basic block is turned into an unconditional trap, turn it
36417 into a deleted note to avoid RTL verification failures.
36419 2017-03-02 Richard Biener <rguenther@suse.de>
36421 * fold-const.c (const_binop): Use ulow () instead of elt (0).
36423 2017-03-02 Richard Biener <rguenther@suse.de>
36425 PR tree-optimization/79345
36427 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
36428 param and abort the walk, returning -1 if it is hit.
36429 (walk_aliased_vdefs): Take a limit param and pass it on.
36430 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
36431 defaulting to 0 and return a signed int.
36432 * tree-ssa-uninit.c (struct check_defs_data): New struct.
36433 (check_defs): New helper.
36434 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
36435 about uninitialized memory.
36436 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
36437 bogus uninitialized warning.
36438 (fixed_convert_from_real): Likewise.
36440 2017-03-02 Bin Cheng <bin.cheng@arm.com>
36442 PR tree-optimization/66768
36443 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
36444 iv_use if base object can't be determined.
36446 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36448 PR tree-optimization/79345
36449 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
36450 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
36451 (get_pattern_stats): Initialize it.
36452 * genemit.c (gen_expand): Verify match_scratch numbers come after
36453 match_operand/match_dup numbers.
36454 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
36455 match_scratch numbers.
36456 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
36458 * config/s390/s390.md (trunctdsd2): Likewise.
36460 2017-03-02 Richard Biener <rguenther@suse.de>
36462 * wide-int.h (wide_int_storage::operator=): Implement in terms
36465 2017-03-02 Richard Biener <rguenther@suse.de>
36467 PR tree-optimization/79777
36468 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
36469 the to insert expression to sth existing.
36471 2017-03-01 Martin Sebor <msebor@redhat.com>
36473 PR middle-end/79692
36474 * gimple-ssa-sprintf.c
36475 (directive::known_width_and_precision): New function.
36476 (format_integer): Use it.
36477 (get_mpfr_format_length): Consider the full range of precision
36478 when computing %g output with the # flag. Set the likely byte
36479 count to 3 rather than 1 when precision is indeterminate.
36480 (format_floating): Correct the lower bound of precision.
36482 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36484 * doc/invoke.texi: Document default code model for 64-bit Linux.
36486 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36489 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
36490 udiv rather than div since input pattern is unsigned.
36492 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
36494 * config/i386/i386.c (print_reg): Warn for values of
36495 unsupported size in integer register.
36497 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
36500 * config/rs6000/predicates.md (current_file_function_operand): Do
36501 not allow self calls to be local if the function is replaceable.
36503 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
36506 * config/rs6000/altivec.h (vec_ctz and others): Change the
36507 preprocessor macro that controls conditional compilation from
36508 _ARCH_PWR9 to __POWER9_VECTOR__.
36509 (vec_all_ne): Change parameterization of __altivec_scalar_pred
36510 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
36511 control (instead of _ARCH_PWR9 control) so that template
36512 definition uses power9-specific function.
36513 (vec_any_eq): Likewise.
36514 (vec_all_ne): Change macro definition to use a power9-specific
36515 expansion under #ifdef __POWER9_VECTOR__ control (instead of
36516 _ARCH_PWR9 control).
36517 (vec_any_eq) Likewise.
36518 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
36519 expansion for CMPNEF to remove support for xvcmpnesp instruction.
36520 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
36521 support for xvcmpnedp instruction.
36522 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
36523 macro expansion so that Power9 implementation of vec_all_ne does
36524 not use the AltiVec predicate framework.
36525 (VCMPNEH_P): Likewise.
36526 (VCMPNEW_P): Likewise.
36527 (VCMPNED_P): Likewise.
36528 (VCMPNEFP_P): Likewise.
36529 (VCMPNEDP_P): Likewise.
36530 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
36531 implementation of vec_any_eq to not use AltiVec predicate
36533 (VCMPAEH_P): Likewise.
36534 (VCMPAEW_P): Likewise.
36535 (VCMPAED_P): Likewise.
36536 (VCMPAEFP_P): Likewise.
36537 (VCMPAEDP_P): Likewise.
36538 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
36539 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
36540 not use the AltiVec predicate framework.
36541 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
36542 of vec_any_eq to not use AltiVec predicate framework.
36543 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36544 support for predefined __POWER9_VECTOR__ macro to indicate that
36545 Power9 instruction selection is enabled.
36546 (altivec_overloaded_builtins): Remove extraneous
36547 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
36548 function argument types RS6000_BTI_bool_V16QI and
36549 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
36550 entry for overloaded function argument types RS6000_BTI_bool_V4SI
36551 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
36552 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
36553 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
36554 Power9 for implementations of vec_cmpne. Change the signature for
36555 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
36556 (representing vec_all_ne) to remove the previously described first
36557 argument of type RS6000_BTI_INTSI, as this was an artifact of
36558 reliance on the AltiVec predicate framework, which is no longer
36559 used in the implementation of these functions. Add
36560 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
36561 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
36562 since, unlike the AltiVec predicate framework implementation, we
36563 do not share function descriptors between vec_alle and vec_anyeq.
36564 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
36565 set of modes that receive special treatment even when
36566 TARGET_P9_VECTOR is true. The special treatment emits code that
36567 does not depend on Power9 instructions.
36568 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36569 define_expand to not rely on AltiVec predicate framework.
36570 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36572 (vector_ne_v2di_p): Change this define_expand to not rely on
36573 AltiVec predicate framework.
36574 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36576 (vector_ne_<mode>_p): Change this define_expand to not rely on
36577 AltiVec predicate framework.
36578 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36580 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36581 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36582 define_insn pattern.
36583 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36584 define_insn pattern because the xvcmpne<VSs>. instruction is not
36586 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36587 instruction is not supported.
36589 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36591 * config/nvptx/nvptx.c: Include intl.h.
36593 2017-03-01 Martin Jambor <mjambor@suse.cz>
36596 * ipa-prop.h (ipa_bits): Removed field known.
36597 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36598 to pointers. Adjusted their comments to warn about their sharing.
36599 (ipcp_transformation_summary): Change bits to a vector of pointers.
36600 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36601 (ipa_get_ipa_bits_for_value): Declare.
36602 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36603 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36604 (ipa_bits_hash_table): Likewise.
36605 (ipa_vr_ggc_hash_traits): Likewise.
36606 (ipa_vr_hash_table): Likewise.
36607 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36608 being pointers and vr_known being removed.
36609 (ipa_set_jf_unknown): Likewise.
36610 (ipa_get_ipa_bits_for_value): New function.
36611 (ipa_set_jfunc_bits): Likewise.
36612 (ipa_get_value_range): New overloaded functions.
36613 (ipa_set_jfunc_vr): Likewise.
36614 (ipa_compute_jump_functions_for_edge): Use the above functions to
36615 construct bits and vr parts of jump functions.
36616 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36617 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36619 (ipcp_grow_transformations_if_necessary): Also allocate
36620 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36622 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36623 them. Fix too long lines.
36624 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36625 vr_known being removed.
36626 (ipa_read_jump_function): Use new setter functions to construct bits
36627 and vr parts of jump functions or set them to NULL.
36628 (write_ipcp_transformation_info): Adjust for bits being pointers.
36629 (read_ipcp_transformation_info): Likewise.
36630 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36632 Include gt-ipa-prop.h.
36633 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36635 (ipcp_store_bits_results): Likewise.
36636 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36637 Do not write to existing jump functions but use a temporary instead.
36639 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36642 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36643 attempt to use its first operand as BIT_FIELD_REF base.
36645 2017-03-01 Richard Biener <rguenther@suse.de>
36647 PR middle-end/79721
36648 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36649 interpolating formula in wrapping arithmetic.
36650 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36652 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36654 PR tree-optimization/79734
36655 * tree-vect-generic.c (expand_vector_condition): Optimize
36656 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36657 Handle VEC_COND_EXPR where comparison has different inner width from
36658 type's inner width.
36660 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36662 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36663 markup, and similar issues. Remove @opindex entries for things
36664 that aren't options. Add missing -mmpy-option entries.
36666 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36668 PR tree-optimization/79737
36669 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36670 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36671 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36672 instead of byte_size. Formatting fix.
36673 (shift_bytes_in_array_right): Formatting fix.
36675 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36678 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36679 condition on optimize for the leaf function test.
36681 2017-02-28 Martin Liska <mliska@suse.cz>
36684 * read-rtl-function.c (function_reader::handle_unknown_directive):
36685 Bail out when one uses -flto.
36687 2017-02-28 Martin Liska <mliska@suse.cz>
36689 * common.opt: Replace space with tabular for options of <number>
36691 * config/i386/i386.opt: Show <number> value for
36692 -mlarge-data-threshold.
36693 * opts.c (print_filtered_help): Do not display number in hexadecimal
36696 2017-02-28 Martin Liska <mliska@suse.cz>
36698 * common.opt: Fix --help=option -Q for options which are of
36701 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36703 * config/i386/i386.c (print_reg): Error out for values
36704 of 8-bit size in invalid integer register.
36706 2017-02-28 Martin Sebor <msebor@redhat.com>
36708 PR tree-optimization/79691
36709 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36711 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36714 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36715 gcc_unreachable with output_operand_lossage.
36717 2017-02-28 Richard Biener <rguenther@suse.de>
36719 PR tree-optimization/79740
36720 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36722 (visit_nary_op): Insert the nary into the hashtable if we
36723 pattern-matched sth.
36724 * tree-ssa-pre.c (eliminate_insert): Robustify.
36726 2017-02-28 Richard Biener <rguenther@suse.de>
36728 PR middle-end/79731
36729 * fold-const.c (decode_field_reference): Reject out-of-bound
36732 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36734 * config/i386/i386.c: Include intl.h.
36735 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36736 instead of just cond ? "..." : "...".
36737 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36738 * coverage.c (read_counts_file): Likewise.
36739 * omp-offload.c: Include intl.h.
36740 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36741 of just cond ? "..." : "...".
36742 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36743 of just cond ? "..." : "...".
36745 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36748 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36750 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36751 'tune for' CPU name.
36752 * config/arm/arm-cpu-data.h: Regenerated.
36754 2017-02-28 Richard Biener <rguenther@suse.de>
36756 PR tree-optimization/79732
36757 * tree-inline.c (expand_call_inline): Do not shadow var.
36759 2017-02-28 Richard Biener <rguenther@suse.de>
36761 PR tree-optimization/79723
36762 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36763 address-space properly.
36765 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36767 * doc/optinfo.texi (Optimization groups): Fix option used for
36769 * doc/invoke.texi (-fopt-info): Document "omp".
36770 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36771 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36772 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36773 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36774 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36776 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36778 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36779 OPTGROUP_OMP with "omp".
36781 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36784 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36785 for arithmetic shift of unsigned V2DI.
36787 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36789 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36790 arc/linux.h headers.
36791 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36792 (LINK_SPEC): Likewise.
36793 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36794 (EXTRA_SPECS): Likewise.
36795 (STARTFILE_SPEC): Likewise.
36796 (ENDFILE_SPEC): Likewise.
36797 (LIB_SPEC): Likewise.
36798 (TARGET_SDATA_DEFAULT): Likewise.
36799 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36800 (MULTILIB_DEFAULTS): Likewise.
36801 (DWARF2_UNWIND_INFO): Likewise.
36802 * config/arc/big.h: New file.
36803 * config/arc/elf.h: Likewise.
36804 * config/arc/linux.h: Likewise.
36805 * config/arc/t-uClibc: Remove.
36807 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36809 PR tree-optimization/77536
36810 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
36811 (tree_transform_and_unroll_loop): Use above function to compute the
36812 estimated niter of unrolled loop and use it when scaling profile.
36813 Also use count info rather than frequency if it's non-zero.
36814 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
36815 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
36816 (vect_transform_loop): Call above function.
36818 2017-02-27 Richard Biener <rguenther@suse.de>
36820 PR tree-optimization/45397
36821 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
36822 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
36823 (visit_nary_op): Add pattern matching for CSEing sign-changed
36824 or truncated operations with wider ones.
36826 2017-02-27 Richard Biener <rguenther@suse.de>
36828 PR tree-optimization/79690
36829 * tree-vect-stmts.c (vectorizable_store): Use vector type
36830 built from the DR with address-space.
36832 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
36834 * doc/invoke.texi (Optimize Options): Refine the description
36835 of asan-use-after-return.
36837 2017-02-25 Alan Modra <amodra@gmail.com>
36839 PR rtl-optimization/79584
36840 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
36841 base, not ad->base_term, the reg within base. Remove assertion
36842 that ad->base == ad->base_term. Replace gen_int_mode using
36843 bogus mode with const0_rtx.
36845 2017-02-25 Jakub Jelinek <jakub@redhat.com>
36847 PR middle-end/79396
36848 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
36849 FMA_EXPR like tcc_binary or tcc_unary.
36851 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
36854 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
36856 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
36857 (output_loc_operands): Handle DW_OP_call_ref and
36858 DW_OP_GNU_variable_value.
36859 (struct variable_value_struct): New type.
36860 (struct variable_value_hasher): Likewise.
36861 (variable_value_hash): New variable.
36862 (string_types): Remove.
36863 (copy_loc_descr): New function.
36864 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
36865 (prepend_loc_descr_to_each): New function.
36866 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
36867 instead of add_loc_descr_to_each if the first argument is single
36868 location list and the second has multiple.
36869 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
36870 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
36871 when looking for variable value which doesn't have other location info.
36872 (loc_list_from_tree): Formatting fix.
36873 (gen_array_type_die): Simplify DW_AT_string_length handling.
36874 (adjust_string_types): Remove.
36875 (gen_subprogram_die): Don't call adjust_string_types nor test/set
36876 string_types. Call resolve_variable_values.
36877 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
36878 (resolve_addr_in_expr): Likewise. Add A argument.
36879 (copy_deref_exprloc): Remove deref argument. Adjust for the
36880 original expression being DW_OP_GNU_variable_value with optionally
36881 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
36882 optionally after it.
36883 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
36884 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
36885 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
36886 (variable_value_hasher::hash, variable_value_hasher::equal): New
36888 (resolve_variable_value_in_expr, resolve_variable_value,
36889 resolve_variable_values, note_variable_value_in_expr,
36890 note_variable_value): New functions.
36891 (dwarf2out_early_finish): Call note_variable_value on all toplevel
36894 2017-02-24 Jakub Jelinek <jakub@redhat.com>
36897 * opts.h (handle_generated_option): Add GENERATED_P argument.
36898 * opts-common.c (handle_option): Adjust function comment.
36899 (handle_generated_option): Add GENERATED_P argument, pass it to
36901 (control_warning_option): Pass false to handle_generated_option
36903 * opts.c (maybe_default_option): Pass true to handle_generated_option
36905 * optc-gen.awk: Likewise.
36907 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36909 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
36910 a REG, look at the REG it is a SUBREG of.
36911 (splitter for cmpeqsi_t): Ditto.
36913 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36915 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
36916 the special USEs with the pattern of the insn, not the insn itself.
36918 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
36921 * doc/invoke.texi: Document -mload-store-pairs.
36923 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36924 Sandra Loosemore <sandra@codesourcery.com>
36926 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
36927 argument isn't a CONST_INT.
36928 (nios2_alternate_compare_const): Assert op is a CONST_INT.
36929 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
36930 (nios2_validate_compare): Bypass alternate compare logic if *op2
36931 is not a CONST_INT.
36932 (ldstwm_operation_p): Return false if first_base is not a REG or
36933 if first_offset is not a CONST_INT.
36935 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36937 * config/cris/cris.md: Use correct operand in a define_peephole2.
36939 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36941 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
36943 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36945 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
36946 this_insn if it is an INSN or JUMP_INSN.
36947 (force_offsettable): Look at base, not at addr.
36948 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
36949 on things that aren't necessarily CONST_INTs.
36951 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
36953 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
36954 -mfpmath=sse is the default also for x86-32 targets with SSE2
36955 instruction set when @option{-ffast-math} is enabled
36957 2017-02-24 Jeff Law <law@redhat.com>
36959 PR rtl-optimizatoin/79286
36960 * ira.c (update_equiv_regs): Drop may_trap_p exception to
36963 2017-02-24 Richard Biener <rguenther@suse.de>
36965 PR tree-optimization/79389
36966 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
36969 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
36971 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
36972 function comment to reflect reality.
36973 (loop_exits_before_overflow): Fix typo in function description.
36975 2017-02-24 Richard Biener <rguenther@suse.de>
36977 PR tree-optimization/79389
36978 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
36979 properly that a threading opportunity exists. Detect conditional
36980 copy/constant propagation opportunities.
36982 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
36984 * config/visium/visium.md (type): Add trap.
36985 (b): New mode attribute.
36986 (*btst): Rename into...
36987 (*btst<mode>): ...this and adjust.
36988 (*cbranchsi4_btst_insn): Rename into...
36989 (*cbranch<mode>4_btst_insn): ...this and adjust.
36990 (trap): New define_insn.
36992 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36994 PR tree-optimization/79389
36995 * ifcvt.c (struct noce_if_info): Add rev_cond field.
36996 (noce_reversed_cond_code): New function.
36997 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
36998 reversed_comparison_code. Formatting fix.
36999 (noce_try_store_flag): Test rev_cond != NULL in addition to
37000 reversed_comparison_code.
37001 (noce_try_store_flag_constants): Likewise.
37002 (noce_try_store_flag_mask): Likewise.
37003 (noce_try_addcc): Use rev_cond if non-NULL instead of
37004 reversed_comparison_code.
37005 (noce_try_cmove_arith): Likewise. Formatting fixes.
37006 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37007 (noce_find_if_block): Initialize rev_cond.
37008 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37009 instead of false as last argument never attempt to reverse it
37012 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37014 PR tree-optimization/79663
37015 * tree-predcom.c (combine_chains): Process refs in reverse order
37016 only for ZERO length chains, and add explaining comment.
37018 2017-02-23 Jeff Law <law@redhat.com>
37020 PR tree-optimization/79578
37021 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37022 in call to operand_equal_p.
37024 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37027 * config/i386/cpuid.h: Fix another undefined behavior.
37029 2017-02-23 Richard Biener <rguenther@suse.de>
37031 PR tree-optimization/79683
37032 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37033 vector types for data-refs.
37035 2017-02-23 Martin Liska <mliska@suse.cz>
37037 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37039 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37041 PR middle-end/79665
37042 * internal-fn.c (get_range_pos_neg): Moved to ...
37043 * tree.c (get_range_pos_neg): ... here. No longer static.
37044 * tree.h (get_range_pos_neg): New prototype.
37045 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37046 are known to be in between 0 and signed maximum inclusive, try to
37047 expand both unsigned and signed divmod and use the cheaper one from
37050 2017-02-22 Jeff Law <law@redhat.com>
37052 PR tree-optimization/79578
37053 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37054 to compare base operands.
37056 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
37059 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
37060 gpc_reg_operand instead of fpr_reg_operand.
37062 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
37064 * config/mips/mips.c (mips_return_in_memory): Force FP
37065 vector types to be returned in memory for o32 ABI.
37067 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37069 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
37070 instead of DW_TAG_member for static data member declarations and don't
37071 set no_linkage_name for static inline data members.
37072 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
37075 2017-02-22 Martin Liska <mliska@suse.cz>
37077 * doc/invoke.texi: Replace inequality signs with square brackets
37080 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
37083 * lra-constraints.c (simplify_operand_subreg): Handle
37084 WORD_REGISTER_OPERATIONS targets.
37086 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37089 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
37090 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
37091 elimination by swapping fld*.
37093 2017-02-22 Richard Biener <rguenther@suse.de>
37095 PR tree-optimization/79673
37096 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
37097 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
37098 irrelevant address-space qualifiers and avoiding a
37099 ADDR_SPACE_CONVERT_EXPR from fold_convert.
37101 2017-02-22 Richard Biener <rguenther@suse.de>
37103 PR tree-optimization/79666
37104 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
37105 to not symbolically negate if that may introduce undefined
37108 2017-02-22 Martin Liska <mliska@suse.cz>
37111 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
37112 * data-streamer-out.c (streamer_write_gcov_count_stream):
37114 * value-prof.c (stream_out_histogram_value): Make assert more
37115 precise based on type of counter.
37117 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
37120 * config/i386/i386.md (standard_x87sse_constant_load splitter):
37121 Use nonimmediate_operand instead of memory_operand for operand 1.
37122 (float-extend standard_x87sse_constant_load splitter): Ditto.
37124 2017-02-21 Jeff Law <law@redhat.com>
37126 PR tree-optimization/79621
37127 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
37128 blocks with edges to themselves.
37130 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37133 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
37134 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
37135 Use gimple_call_builtin_p.
37138 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
37139 on temporarily removed DEBUG_INSNs.
37141 PR tree-optimization/79649
37142 * tree-loop-distribution.c (classify_partition): Give up on
37143 non-generic address space loads/stores.
37145 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
37147 * doc/loop.texi (Loop manipulation): Remove nonexistent
37148 tree_ssa_loop_version from the documentation.
37149 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
37151 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37154 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
37155 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
37156 * config/rs6000/rs6000.c: Include except.h.
37157 (rs6000_expand_split_stack_prologue): Call
37158 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
37160 2017-02-21 Martin Jambor <mjambor@suse.cz>
37163 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
37164 have been analyzed.
37166 2017-02-21 Martin Jambor <mjambor@suse.cz>
37168 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
37169 for backward compatibility only.
37170 * doc/invoke.texi (Option Summary): Remove all references to
37171 -fipa-cp-alignment.
37173 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
37177 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37179 * lra-constraints.c (curr_insn_transform): Handle
37180 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37182 2017-02-21 Martin Liska <mliska@suse.cz>
37184 * config/i386/i386.opt: Replace -masm-dialect with -masm.
37186 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
37188 PR translation/79638
37189 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
37191 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
37194 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
37195 (arm_function_ok_for_sibcall): Return false for an indirect call by
37196 descriptor if all the argument registers are used.
37197 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
37198 alignment of the function.
37200 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37202 PR tree-optimization/61441
37203 * simplify-rtx.c (simplify_const_unary_operation): For
37204 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
37205 the sNaN unmodified.
37207 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
37209 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
37210 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
37211 instead of SYSTEM_HEADER_DIR.
37213 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
37214 Martin Liška <mliska@suse.cz>
37216 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
37217 Fix typos and grammar, use active voice, and clarify.
37219 2017-02-20 Marek Polacek <polacek@redhat.com>
37221 PR middle-end/79537
37222 * gimplify.c (gimplify_expr): Handle unused *&&L;.
37225 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
37227 2017-02-20 Jakub Jelinek <jakub@redhat.com>
37230 * config/i386/i386.c (ix86_expand_builtin): Handle
37231 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
37232 ix86_builtins_isa[fcode].isa as a requirement of those
37233 flags and any other flag in the bitmask.
37234 (ix86_init_mmx_sse_builtins): Use 0 instead of
37235 ~OPTION_MASK_ISA_64BIT as mask.
37236 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
37237 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
37238 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
37239 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
37241 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37244 * lra-constraints.c (split_reg): Check requested split mode
37245 is supported by the register.
37247 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37249 * lra-constraints.c (simplify_operand_subreg): Remove early
37252 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37255 * lra-constraints.c (curr_insn_transform): Tighten condition
37256 for converting SUBREG reloads from OP_OUT to OP_INOUT.
37258 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37261 * lra-constraints.c (curr_insn_transform): Handle
37262 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37264 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
37267 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
37269 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
37271 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
37274 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
37277 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37279 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
37282 2017-02-18 Jakub Jelinek <jakub@redhat.com>
37284 * final.c (last_columnnum, override_columnnum): New variables.
37285 (final_start_function): Set last_columnnum, pass it to begin_prologue
37286 hook and pass 0 to dwarf2out_begin_prologue.
37287 (final_scan_insn): Update override_columnnum. Pass last_columnnum
37288 to source_line debug hook.
37289 (notice_source_line): Compute last_columnnum and for debug_column_info
37290 return true on column changes.
37291 * debug.h (struct gcc_debug_hooks): Add column argument to
37292 source_line and begin_prologue hooks.
37293 (debug_nothing_int_charstar_int_bool): Remove prototype.
37294 (debug_nothing_int_int_charstar,
37295 debug_nothing_int_int_charstar_int_bool): New prototypes.
37296 (dwarf2out_begin_prologue): Add column argument.
37297 * debug.c (do_nothing_debug_hooks): Adjust source_line and
37298 begin_prologue hooks.
37299 (debug_nothing_int_charstar_int_bool): Remove.
37300 (debug_nothing_int_int_charstar,
37301 debug_nothing_int_int_charstar_int_bool): New functions.
37302 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
37303 through to dwarf2out_source_line.
37304 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
37305 (dwarf2out_source_line): Add column argument, emit it if requested.
37306 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
37308 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37309 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37310 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
37311 through to dwarf2out_begin_prologue.
37312 (vmsdbgout_source_line): Add column argument, pass it through to
37313 dwarf2out_source_line.
37314 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
37315 dbxout_source_line caller.
37316 (dbxout_source_line): Add column argument.
37318 * common.opt (gno-column-info, gcolumn-info): New options.
37319 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
37320 (check_die): Also test for multiple DW_AT_decl_column attributes.
37321 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
37322 DW_AT_decl_column if requested.
37323 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
37325 (gen_variable_die): Likewise.
37326 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
37327 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
37330 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
37331 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
37332 (ix86_handle_option): Handle OPT_m3dnowa.
37333 * doc/invoke.texi (-m3dnowa): Document.
37334 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
37335 -m3dnowa instead of -m3dnow -march=athlon.
37338 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
37339 instead of gcc_assert for K, r and R code checks. Formatting fixes.
37341 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37344 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
37345 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
37346 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
37347 generator for vsx_xxpermdi_<mode>_be.
37348 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
37349 force big-endian semantics.
37350 (vsx_xxpermdi_<mode>_be): New define_expand with same
37351 implementation as previous version of vsx_xxpermdi_<mode>.
37353 2017-02-17 Jakub Jelinek <jakub@redhat.com>
37355 PR tree-optimization/79327
37356 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
37357 variable, its initialization and use.
37359 2017-02-17 Julia Koval <julia.koval@intel.com>
37361 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
37362 (OPTION_MASK_ISA_PKU_UNSET): New.
37363 (ix86_handle_option): Handle -mrdpid.
37364 * config/i386/cpuid.h (bit_RDPID): New.
37365 * config/i386/driver-i386.c (host_detect_local_cpu):
37366 Detect RDPID feature.
37367 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
37368 * config/i386/i386-c.c (ix86_target_macros_internal):
37370 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
37371 (ix86_valid_target_attribute_inner_p): Add "rdpid".
37372 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
37373 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
37374 * config/i386/i386.md (define_insn "rdpid"): New.
37375 * config/i386/i386.opt Add -mrdpid.
37376 * config/i386/immintrin.h (_rdpid_u32): New.
37378 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
37380 PR rtl-optimization/79541
37381 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
37382 instead of transforming it into USE.
37384 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
37386 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
37387 If HONOR_SNANS (SFmode) force the input to a register.
37388 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
37389 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
37390 an frsp or similar insn.
37392 2017-02-17 Martin Liska <mliska@suse.cz>
37394 PR rtl-optimization/79577
37395 * params.def (selsched-max-sched-times): Increase minimum to 1.
37397 2017-02-17 Martin Liska <mliska@suse.cz>
37399 PR rtl-optimization/79574
37400 * gcse.c (want_to_gcse_p): Prevent integer overflow.
37402 2017-02-17 Martin Liska <mliska@suse.cz>
37404 PR tree-optimization/79529
37405 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
37406 ssa_defined_default_def_p to handle cases which are implicitly
37408 * tree-ssa.c (ssa_defined_default_def_p): New function.
37409 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
37410 which are implicitly defined.
37411 * tree-ssa.h (ssa_defined_default_def_p): Declare.
37413 2017-02-17 Richard Biener <rguenther@suse.de>
37415 PR middle-end/79576
37416 * params.def (max-ssa-name-query-depth): Limit to 10.
37418 2017-02-17 Richard Biener <rguenther@suse.de>
37420 PR tree-optimization/79552
37421 * tree-ssa-structalias.c (visit_loadstore): Properly verify
37424 2017-02-17 Richard Biener <rguenther@suse.de>
37427 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
37429 2017-02-17 Marek Polacek <polacek@redhat.com>
37431 PR middle-end/79536
37432 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
37433 (fold_negate_expr): New wrapper.
37435 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
37437 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
37438 Correct terminology and de-emphasize pre-standard behavior.
37440 2017-02-16 Alan Modra <amodra@gmail.com>
37442 PR rtl-optimization/79286
37443 * ira.c (def_dominates_uses): New function.
37444 (update_equiv_regs): Don't create an equivalence for insns that
37445 may trap where the register def does not dominate the use.
37447 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
37449 PR rtl-optimization/78127
37450 * lra.c (lra): Call lra_eliminate before finish the loop after
37453 2017-02-16 Richard Biener <rguenther@suse.de>
37455 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
37457 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
37458 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
37459 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
37460 (isl_val_int_from_wi): New function.
37461 (extract_affine_gmp): Rename to ...
37462 (extract_affine_wi): ... this, take a widest_int.
37463 (extract_affine_int): Just wrap extract_affine_wi.
37464 (add_param_constraints): Use isl_val_int_from_wi.
37465 (add_loop_constraints): Likewise, and extract_affine_wi.
37467 2017-02-15 Jeff Law <law@redhat.com>
37469 PR middle-end/79521
37470 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
37471 ira_init_register_move_cost_if_necessary.
37473 2017-02-15 Martin Sebor <msebor@redhat.com>
37475 PR middle-end/32003
37476 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
37477 removed in a prior commit.
37479 2017-02-15 Bin Cheng <bin.cheng@arm.com>
37481 PR tree-optimization/79347
37482 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
37483 counters during peeling.
37485 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
37487 * Makefile.in (site.exp): Remove "set ISLVER".
37489 2017-02-15 Jakub Jelinek <jakub@redhat.com>
37492 * real.c (real_from_integer): Call real_convert even for decimal.
37494 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
37497 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
37499 2017-02-14 Andrew Pinski <apinski@cavium.com>
37501 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
37502 cores and change the partno/implementer to be correct.
37503 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
37504 the 'B" as the implementer.
37505 * config/aarch64/aarch64-tune.md: Regenerate.
37507 2017-02-14 Carl Love <cel@us.ibm.com>
37509 * config/rs6000/rs6000.c: Add case statement entry to make the
37510 xvcvuxdsp built-in argument unsigned.
37511 * config/rs6000/vsx.md: Fix the source and return operand types so they
37512 match the instruction definitions from the ISA document. Fix typo
37513 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
37516 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
37519 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
37520 member early_clobber_alts.
37521 * lra-lives.c (reg_early_clobber_p): New.
37522 (process_bb_lives): Use it.
37523 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
37524 (debug_operand_data): Initialize early_clobber_alts.
37525 (setup_operand_alternative): Set up early_clobber_alts.
37526 (collect_non_operand_hard_regs): Ditto. Pass early clobber
37527 alternatives to new_insn_reg.
37528 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
37530 (lra_update_insn_regno_info): Pass the new arg.
37532 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37534 PR middle-end/79505
37535 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
37536 (new_oacc_loop_raw): Don't clear already cleared fields.
37539 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
37540 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
37541 _mm512_prefetch_i64gather_ps): New inline functions and macros.
37543 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
37546 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
37548 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
37551 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
37552 the extra instruction to the right place to store 128-bit constant
37555 2017-02-14 Martin Sebor <msebor@redhat.com>
37557 PR middle-end/79448
37558 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
37559 warning for strings of unknown length.
37561 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
37563 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
37565 2017-02-14 Jeff Law <law@redhat.com>
37568 * ira-costs.c (scan_one_insn): Initialize register move costs
37569 for pseudos seen in USE/CLOBBER insns.
37571 PR tree-optimization/79095
37572 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37573 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37574 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37575 if the operands are known to be not equal, then the resulting range
37577 (intersect_ranges): If the new range is ~[0,0] and the old range is
37578 wide, then prefer ~[0,0].
37579 * tree-vrp.c (overflow_comparison_p_1): New function.
37580 (overflow_comparison_p): New function.
37581 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37582 if NAME is used in an overflow test.
37583 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37584 overflow check that can be expressed as an equality test, then adjust
37585 ops to be that equality test.
37587 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37589 * config/s390/s390-builtin-types.def: Remove flags argument.
37590 * config/s390/s390.c (s390_init_builtins): Likewise.
37592 2017-02-14 Martin Liska <mliska@suse.cz>
37594 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37595 vector. Fix trailing white spaces.
37597 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37599 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37602 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37604 PR rtl-optimization/68664
37605 * config/arm/arm.c (arm_sched_can_speculate_insn):
37606 New function. Declare prototype.
37607 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37609 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37611 PR rtl-optimization/68664
37612 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37614 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37616 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37618 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37619 max skip bytes for function, loop and jump.
37621 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37623 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37624 ABS_EXPR for gimple dump.
37626 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37629 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37631 PR tree-optimization/79408
37632 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37633 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37634 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37635 also if rhs1 is INTEGER_CST.
37637 2017-02-14 Richard Biener <rguenther@suse.de>
37639 PR middle-end/79432
37640 * tree-into-ssa.c (insert_phi_nodes): When the function can
37641 have abnormal edges rewrite SSA names with broken use-def
37642 dominance out of SSA and register them for PHI insertion.
37644 2017-02-13 Martin Sebor <msebor@redhat.com>
37646 PR middle-end/79496
37647 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37648 clearing info.nowrite flag when snprintf size argument is a range.
37650 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37652 * cprop.c (cprop_jump): Add missing space in string literal.
37653 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37654 (get_constraint_for_component_ref): Likewise.
37655 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37656 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37657 * lra-constraints.c (process_alt_operands): Likewise.
37658 * ipa-inline.c (inline_small_functions): Likewise.
37659 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37660 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37661 * trans-mem.c (diagnose_tm_1_op): Likewise.
37662 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37663 (grid_parallel_clauses_gridifiable): Likewise.
37665 * config/nvptx/mkoffload.c (process): Add space in between
37668 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37669 "MOD4_SSE_REGS" and "ALL_REGS".
37671 * spellcheck.c (test_data): Add , in between "foo" and "food".
37673 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37676 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37677 boundary crossing check and subsequent code generation agree.
37679 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37681 * config/aarch64/aarch64.c (has_memory_op): Delete.
37682 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37685 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37687 PR rtl-optimization/79388
37688 PR rtl-optimization/79450
37689 * combine.c (distribute_notes): When removing TEM_INSN for which
37690 corresponding dest has last value recorded, invalidate that last
37693 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37695 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37696 of explicit '@'. Add missing assembly comment marker on branch costs
37699 2017-02-13 Nathan Sidwell <nathan@acm.org>
37701 * gengtype-lex.l (<in_struct>): Add '/'.
37703 2017-02-13 Martin Liska <mliska@suse.cz>
37706 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37708 2017-02-13 Richard Biener <rguenther@suse.de>
37710 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37712 * configure: Re-generate.
37713 * config.in: Likewise.
37714 * graphite-dependences.c: Simplify as if
37715 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37716 * graphite-isl-ast-to-gimple.c: Likewise.
37717 * graphite-optimize-isl.c: Likewise.
37718 * graphite-poly.c: Likewise.
37719 * graphite-sese-to-poly.c: Likewise.
37720 * graphite.h: Likewise.
37721 * toplev.c: Include isl/version.h and use isl_version () for
37722 printing the ISL version.
37723 * doc/install.texi: Update ISL requirement.
37725 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37727 * doc/standards.texi (Standards): Update reference to
37730 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37732 * doc/extend.texi (Named Address Spaces): sourceware.org now
37734 * doc/install.texi (Binaries): Ditto.
37737 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37739 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37740 standard terminology "stringize"/"stringizing" throughout.
37741 * doc/cppinternals.texi: Likewise.
37743 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37745 * doc/extend.texi: Fix some spelling mistakes and typos.
37746 * doc/invoke.texi: Likewise.
37748 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37751 * params.def (inline-min-speedup) Change from 10 to 8.
37753 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37755 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37758 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37761 * ipa-inline-analysis.c (get_minimal_bb): New function.
37762 (record_modified): Use it.
37763 (remap_edge_change_prob): Handle also ancestor functions.
37765 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37767 * doc/contrib.texi (Contributors): Remove broken link into
37768 the Mauve CVS repository.
37770 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37772 PR middle-end/79454
37773 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37774 result computation whenever lhs doesn't have vector mode, not
37775 just when it has BLKmode.
37777 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37779 * doc/makefile.texi (profiledbootstrap): Refer to the
37780 installation instructions only in textual form.
37782 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37785 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37787 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37789 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37790 (Specific): Update mingw-w64 reference.
37792 (Specific): Remove broken link to Renesas RX processor.
37794 2017-02-10 Richard Biener <rguenther@suse.de>
37796 * toplev.c (process_options): Do not mention obsolete graphite
37797 options when printing sorry message about missing graphite support.
37798 Mention -floop-nest-optimize.
37800 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37802 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37803 (vtst_p16): Likewise.
37804 (vtstq_p8): Likewise.
37805 (vtstq_p16): Likewise.
37807 (vtstq_p64): Likewise.
37808 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37809 (vset_lane_p64): New.
37810 (vsetq_lane_p64): New.
37812 2017-02-10 Jakub Jelinek <jakub@redhat.com>
37814 PR tree-optimization/79411
37815 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
37816 stmt operands are SSA_NAMEs used in abnormal phis.
37817 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
37820 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37823 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
37826 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37828 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
37830 2017-02-09 Jakub Jelinek <jakub@redhat.com>
37832 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
37836 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
37837 not arbitrary TREE_CONSTANT.
37840 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
37841 "omp declare target link" attribute unless is_global_var.
37842 * omp-offload.c (find_link_var_op): Likewise.
37844 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
37845 Chung-Lin Tang <cltang@codesourcery.com>
37847 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
37849 (gimplify_adjust_omp_clauses): Don't delete TILE.
37850 (gimplify_omp_for): Deal with TILE.
37851 * internal-fn.c (expand_GOACC_TILE): New function.
37852 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
37854 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
37855 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
37857 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
37858 avoid DIV for outermost collapse var.
37859 (expand_oacc_for): Insert tile element loop as needed. Adjust.
37860 Remove out of date comments, fix whitespace.
37861 * omp-general.c (omp_extract_for_data): Deal with tiling.
37862 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
37863 adjust OLF_DIM_BASE value.
37864 (struct omp_for_data): Add tiling field.
37865 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
37866 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
37867 for auto loops. Remove default auto determining, moved to
37868 oacc_loop_fixed_partitions.
37869 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
37870 stmts, add e_mask field.
37871 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
37872 (oacc_thread_numbers): Use oacc_dim_call.
37873 (oacc_xform_tile): New.
37874 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
37875 (finish_oacc_loop): Adjust for ifns vector.
37876 (oacc_loop_discover_walk): Append loop abstraction sites to list,
37877 add case for GOACC_TILE fns.
37878 (oacc_loop_xform_loop): Delete.
37879 (oacc_loop_process): Iterate over call list directly, and add
37880 handling for GOACC_TILE fns.
37881 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
37883 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
37884 vector partitioning to outer loops. Assign 2 partitions to loops
37885 when available. Add TILE handling.
37886 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
37887 (execite_oacc_device_lower): Process GOACC_TILE fns,
37888 ignore unknown specs.
37889 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
37890 * tree.c (omp_clause_num_ops): Adjust TILE ops.
37891 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
37893 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
37895 * configure.ac (ACX_BUGURL): Update.
37896 * configure: Regenerate.
37898 2017-02-09 Richard Biener <rguenther@suse.de>
37900 PR tree-optimization/69823
37901 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
37902 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
37904 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37906 * config/arc/arc-c.def: Add __NPS400__ definition.
37907 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
37908 (TARGET_NPS400): Define.
37910 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37912 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
37914 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
37915 pointer, arch_info.
37916 (arc_cpu_types): Fill the arch_info field with a pointer into the
37917 arc_arch_types table.
37918 (arc_selected_cpu): Declare.
37919 * config/arc/arc.c (arc_selected_cpu): Make global.
37920 (arc_selected_arch): Delete.
37921 (arc_base_cpu): Delete.
37922 (arc_override_options): Remove references to deleted variables,
37923 update access to arch information.
37924 (ARC_OPT): Update access to arch information.
37925 (ARC_OPTX): Likewise.
37926 * config/arc/arc.h (arc_base_cpu): Remove declaration.
37927 (TARGET_ARC600): Update access to arch information.
37928 (TARGET_ARC601): Likewise.
37929 (TARGET_ARC700): Likewise.
37930 (TARGET_EM): Likewise.
37931 (TARGET_HS): Likewise.
37932 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
37935 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
37938 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
37939 condition/operands for integer GE/LE/GEU/LEU operations.
37941 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
37943 PR translation/79397
37944 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
37947 2017-02-08 Martin Jambor <mjambor@suse.cz>
37950 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
37951 whether allocation happened.
37952 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
37953 nothing was allocated.
37955 2017-02-08 Jakub Jelinek <jakub@redhat.com>
37957 PR tree-optimization/79408
37958 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
37959 constant, but SSA_NAME with a known integer range, use the minimum
37960 of that range instead of op1 to determine if modulo can be replaced
37961 with its first operand.
37963 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37965 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
37967 2017-02-08 Richard Biener <rguenther@suse.de>
37969 PR tree-optimization/71824
37970 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37971 Check all loops contained in the merged region.
37973 2017-02-07 Andrew Pinski <apinski@cavium.com>
37975 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
37977 2017-02-07 Andrew Pinski <apinski@cavium.com>
37979 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
37980 (thunderxt88): Likewise.
37981 (thunderxt81): Disable LSE and change v8.1 to v8.
37982 (thunderxt83): Likewise.
37984 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37985 Richard Biener <rguenther@suse.de>
37987 PR middle-end/79399
37988 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
37989 type from int to size_t.
37990 * ira-costs.c (struct_costs_size): Change type from int to size_t.
37992 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37994 PR rtl-optimization/79386
37995 * cprop.c (bypass_conditional_jumps): Initialize
37996 bypass_last_basic_block already before splitting bbs after
37997 unconditional traps...
37998 (bypass_conditional_jumps): ... rather than here.
38001 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38002 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38003 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38004 fix -masm=intel patterns.
38006 2017-02-07 Richard Biener <rguenther@suse.de>
38008 PR tree-optimization/79256
38009 PR middle-end/79278
38010 * builtins.c (get_object_alignment_2): Use min_align_of_type
38011 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38012 and ADJUST_FIELD_ALIGN.
38014 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38016 * doc/tm.texi: Regenerate.
38017 * stor-layout.c (layout_decl): Adjust.
38018 (update_alignment_for_field): Likewise.
38019 (place_field): Likewise.
38020 (min_align_of_type): Likewise.
38021 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38022 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38023 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38024 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38025 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38026 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38027 * config/i386/i386.c (x86_field_alignment): Likewise.
38028 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38029 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38030 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38031 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38032 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38033 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38037 2017-01-30 Richard Biener <rguenther@suse.de>
38039 PR tree-optimization/79256
38040 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38041 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38044 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38046 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38047 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38048 builtins to SImode and emit a zero-extend, if necessary.
38050 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38052 * docs/invoke.texi (RISC-V Options): Alphabetize.
38054 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38056 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
38059 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38061 * config/riscv/riscv.c: New file.
38062 * common/config/riscv/riscv-common.c: Likewise.
38063 * config.gcc: Likewise.
38064 * config/riscv/constraints.md: Likewise.
38065 * config/riscv/elf.h: Likewise.
38066 * config/riscv/generic.md: Likewise.
38067 * config/riscv/linux.h: Likewise.
38068 * config/riscv/multilib-generator: Likewise.
38069 * config/riscv/peephole.md: Likewise.
38070 * config/riscv/pic.md: Likewise.
38071 * config/riscv/predicates.md: Likewise.
38072 * config/riscv/riscv-builtins.c: Likewise.
38073 * config/riscv/riscv-c.c: Likewise.
38074 * config/riscv/riscv-ftypes.def: Likewise.
38075 * config/riscv/riscv-modes.def: Likewise.
38076 * config/riscv/riscv-opts.h: Likewise.
38077 * config/riscv/riscv-protos.h: Likewise.
38078 * config/riscv/riscv.h: Likewise.
38079 * config/riscv/riscv.md: Likewise.
38080 * config/riscv/riscv.opt: Likewise.
38081 * config/riscv/sync.md: Likewise.
38082 * config/riscv/t-elf-multilib: Likewise.
38083 * config/riscv/t-linux: Likewise.
38084 * config/riscv/t-linux-multilib: Likewise.
38085 * config/riscv/t-riscv: Likewise.
38086 * configure.ac: Likewise.
38087 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
38088 Waterman as RISC-V maintainers.
38089 * doc/install.texi: Add RISC-V entries.
38090 * doc/invoke.texi: Add RISC-V options section.
38091 * doc/md.texi: Add RISC-V constraints section.
38092 * configure: Regenerated.
38094 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
38097 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
38098 false values to be constant vectors with all 0 or all 1 bits set.
38099 (vcondu<mode><mode>): Likewise.
38100 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
38102 (fpmask_comparison_operator): Update comment.
38103 (vecint_comparison_operator): New predicate.
38104 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
38105 vector conditionals when the true and false values are constant
38106 vectors with all 0 bits or all 1 bits set.
38108 2017-02-06 Martin Sebor <msebor@redhat.com>
38110 PR tree-optimization/79376
38111 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
38113 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
38115 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
38116 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
38117 to simplify split condition.
38119 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38121 * omp-expand.c (oxpand_omp_atomic_fetch_op,
38122 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
38125 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
38127 PR rtl-optimization/68664
38128 * target.def (can_speculate_insn): New hook.
38129 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
38130 * doc/tm.texi: Regenerate.
38131 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
38132 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
38133 (rs6000_sched_can_speculate_insn): New function.
38135 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38137 PR tree-optimization/79284
38138 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
38139 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
38140 vectorizable_mask_load_store, vectorizable_operation,
38141 vect_is_simple_cond, get_same_sized_vectype): Use it instead
38142 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
38143 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
38144 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
38145 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
38146 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
38147 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
38148 is_gimple_assign (stmt). Replace another such test with
38149 is_gimple_assign (stmt).
38151 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38154 * config/avr/avr.c (rtl-iter.h): Include it.
38155 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
38156 (avr_legitimate_combined_insn): ...and implementation.
38158 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
38160 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
38161 * config/s390/s390.c (s390_const_operand_ok)
38162 (s390_canonicalize_comparison, s390_extract_part)
38163 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
38164 (s390_contiguous_bitmask_p, s390_rtx_costs)
38165 (legitimize_pic_address): Likewise.
38166 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
38167 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
38168 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
38169 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
38170 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
38172 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38174 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
38175 REGNO($0) == REGNO($1).
38177 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38179 * config/s390/linux.h(SIZE_TYPE): Add comment.
38181 2017-02-06 Julian Brown <julian@codesourcery.com>
38182 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38183 Virendra Pathak <virendra.pathak@broadcom.com>
38185 * config/aarch64/aarch64-cores.def: Change the scheduler
38187 * config/aarch64/aarch64.md: Include thunderx2t99.md.
38188 * config/aarch64/thunderx2t99.md: New file.
38190 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38192 * doc/standards.texi (Go Language): Update link to language
38195 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38197 * tree-eh.c (lower_resx): Sanitize profile.
38198 (cleanup_empty_eh_move_lp): Likewise.
38200 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38203 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
38205 * cfgloopmanip.h (loop_version): Update prototype.
38206 * modulo-sched.c (sms_schedule): Update call of loop_version.
38207 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
38208 * tree-parloops.c (gen_parallel_loop): Likewise.
38209 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
38210 * tree-ssa-loop-split.c (split_loop): Likewise.
38211 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
38212 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
38214 2017-02-05 Martin Liska <mliska@suse.cz>
38217 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
38219 (print_operand_address): Initialize a struct to zero.
38221 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38223 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
38224 garbage collector only in textual form.
38226 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38228 * doc/extend.texi (x86 specific memory model extensions for
38229 transactional memory): Simplify a phrase.
38231 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
38234 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
38235 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
38236 (atomic_storedi_1): Likewise.
38238 2017-02-04 Jakub Jelinek <jakub@redhat.com>
38240 PR tree-optimization/79338
38241 * tree-parloops.c (gather_scalar_reductions): Don't call
38242 vect_analyze_loop_form for loop->inner before destroying loop's
38245 2017-02-03 Martin Sebor <msebor@redhat.com>
38247 PR tree-optimization/79327
38248 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
38249 when precision has resulted in leading zeros.
38250 (format_integer): Adjust the likely counter to assume an unknown
38251 argument that may be zero is non-zero.
38253 2017-02-03 Jason Merrill <jason@redhat.com>
38256 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
38257 avoid copying non-taken branch.
38259 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38261 PR tree-optimization/79340
38262 * tree-vect-loop.c (vectorizable_reduction): Release
38263 vec_defs elements after safe_splicing them into other vectors.
38266 PR tree-optimization/79327
38267 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
38268 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
38270 (format_integer): Use wide_int_to_tree instead of build_int_cst
38271 + to_?hwi. If argmin is NULL, just set argmin and argmax to
38272 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
38273 of shortest and longest sequence.
38275 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
38277 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
38278 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
38280 2017-02-03 Walter Lee <walt@tilera.com>
38283 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
38284 after initial stackframe link reg save.
38285 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
38287 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38290 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
38291 wu for stxssp alternative.
38293 2017-02-03 Martin Sebor <msebor@redhat.com>
38295 PR tree-optimization/79352
38296 * gimple-fold.c (get_range_strlen): Add argument.
38297 (get_range_strlen): Change return type to bool.
38298 (get_maxval_strlen): Pass in a dummy argument.
38299 * gimple-fold.h (get_range_strlen): Change return type to bool.
38300 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
38301 * tree.h (array_at_struct_end_p): Add argument.
38302 * tree.c (array_at_struct_end_p): Handle it.
38304 2017-02-03 Martin Liska <mliska@suse.cz>
38307 * multiple_target.c (create_dispatcher_calls): Redirect edge
38308 from a caller of a dispatcher.
38309 (expand_target_clones): Make the clones local.
38310 (ipa_target_clone): Do both target clones and resolvers.
38311 (ipa_dispatcher_calls): Remove the pass.
38312 (pass_dispatcher_calls::gate): Likewise.
38313 (make_pass_dispatcher_calls): Likewise.
38314 * passes.def (pass_target_clone): Put as very first IPA early
38317 2017-02-03 Martin Liska <mliska@suse.cz>
38319 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
38320 in case of a function with ifunc attribute.
38322 2017-02-03 Martin Liska <mliska@suse.cz>
38324 * cgraph.c (cgraph_node::dump): Dump function version info.
38325 * symtab.c (symtab_node::dump_base): Add missing new line.
38327 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38329 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
38330 (ifcombine_ifandif): Use it.
38332 2017-02-03 Martin Liska <mliska@suse.cz>
38334 * doc/invoke.texi: Document default value for
38335 use-after-scope-direct-emission-threshold.
38337 2017-02-03 Martin Liska <mliska@suse.cz>
38339 PR tree-optimization/79339
38340 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
38341 (format_floating): Likewise.
38343 2017-02-03 Martin Liska <mliska@suse.cz>
38346 * ipa-prop.c (ipa_node_params_t::insert): Remove current
38348 (ipa_node_params_t::remove): Likewise.
38349 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
38350 initialization from removed ipa_node_params_t::insert.
38351 (ipa_node_params::~ipa_node_params): Move from removed
38352 ipa_node_params_t::release.
38353 * symbol-summary.h (symbol_summary::m_released): New member.
38354 Do not release a summary twice. Do not allow to call finalizer
38355 for types of a summary that live in GGC memory.
38357 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38359 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
38362 2017-02-02 Martin Sebor <msebor@redhat.com>
38364 PR middle-end/79275
38365 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
38366 (format_string): Tighten up the range of output for non-constant
38367 strings and correct the expected range for wide non-constant strings.
38369 2017-02-02 Martin Sebor <msebor@redhat.com>
38371 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
38373 PR middle-end/32003
38374 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
38376 (-fdump-tree-@var): Add to index and document how to come up
38377 with pass-specific option and dump file names.
38378 (-fdump-passes): Clarify where to look for output.
38380 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38382 PR middle-end/77445
38383 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
38384 statistics of the analyzed path; allow threading for speed when
38385 any of BBs along the path are optimized for speed.
38387 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
38389 PR middle-end/78468
38390 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
38391 settings of the virtual registers.
38394 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
38396 * explow.c (get_dynamic_stack_size): Take known alignment of stack
38397 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
38400 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38402 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
38403 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
38405 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38407 * config/s390/s390.md: Add missing comments with the expanded
38409 * config/s390/vector.md: Likewise.
38410 * config/s390/vx-builtins.md: Likewise.
38412 2017-02-02 Jakub Jelinek <jakub@redhat.com>
38415 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
38416 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
38417 conditions on a single line.
38419 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38421 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38422 __S390_VX__ to __VX__.
38424 2017-02-01 Andrew Pinski <apinski@cavium.com>
38426 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
38427 stmt_info to record_stmt_cost.
38428 (vect_get_known_peeling_cost): Pass stmt_info if known to
38430 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
38431 cpu_vector_cost field into
38432 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
38433 field into vec_int_stmt_cost and vec_fp_stmt_cost.
38434 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
38435 splitting of scalar_stmt_cost and vec_stmt_cost.
38436 (thunderx_vector_cost): Likewise.
38437 (cortexa57_vector_cost): LIkewise.
38438 (exynosm1_vector_cost): Likewise.
38439 (xgene1_vector_cost): Likewise.
38440 (thunderx2t99_vector_cost): Improve after the splitting of the two
38442 (aarch64_builtin_vectorization_cost): Update for the splitting of
38443 scalar_stmt_cost and vec_stmt_cost.
38445 2017-02-01 Torvald Riegel <triegel@redhat.com>
38446 Richard Henderson <rth@redhat.com>
38448 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
38449 conditional on existance of a fast atomic load.
38450 * optabs-query.c (can_atomic_load_p): New function.
38451 * optabs-query.h (can_atomic_load_p): Declare it.
38452 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
38453 no fast atomic load is available for the particular size of access.
38454 (expand_atomic_compare_and_swap): Likewise.
38455 (expand_atomic_load): Likewise.
38456 (expand_atomic_store): Likewise.
38457 (expand_atomic_fetch_op): Likewise.
38458 * testsuite/lib/target-supports.exp
38459 (check_effective_target_sync_int_128): Remove x86 because it provides
38460 no fast atomic load.
38461 (check_effective_target_sync_int_128_runtime): Likewise.
38463 2017-02-01 Richard Biener <rguenther@suse.de>
38465 * graphite.c: Include tree-vectorizer.h for find_loop_location.
38466 (graphite_transform_loops): Provide opt-info for optimized nests.
38467 * tree-parloop.c (parallelize_loops): Provide opt-info for
38468 parallelized loops.
38470 2017-02-01 Richard Biener <rguenther@suse.de>
38472 PR middle-end/79315
38473 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
38474 was not set before.
38476 2017-02-01 Richard Biener <rguenther@suse.de>
38478 PR tree-optimization/71824
38479 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38480 Verify the loops are valid in the merged SESE region.
38481 (scop_detection::can_represent_loop_1): Check analyzing the
38482 evolution of the number of iterations in the region succeeds.
38484 2017-01-31 Ian Lance Taylor <iant@golang.org>
38486 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
38487 REG_ARGS_SIZE note to 32-bit push insns and call insn.
38489 2017-01-31 David Malcolm <dmalcolm@redhat.com>
38491 PR preprocessor/79210
38492 * input.c (get_substring_ranges_for_loc): Replace line_width
38493 assertion with error-handling.
38495 2017-01-31 Richard Biener <rguenther@suse.de>
38497 PR tree-optimization/77318
38498 * graphite-sese-to-poly.c (extract_affine): Fix assert.
38499 (create_pw_aff_from_tree): Take loop parameter.
38500 (add_condition_to_pbb): Pass loop of the condition to
38501 create_pw_aff_from_tree.
38503 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38505 * config/s390/s390.c (s390_asan_shadow_offset): New function.
38506 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
38508 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
38512 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
38514 (convert_int_to_float128): Likewise.
38515 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
38516 (convert_int_to_float128): Likewise.
38517 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
38518 (UNSPEC_IEEE128_CONVERT): Likewise.
38519 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
38520 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
38521 Use local variables for IBM extended format.
38522 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
38523 (fix_trunc<mode>si2_fprs): Likewise.
38524 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
38525 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
38526 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
38527 to know that we can now have integers of all sizes in vector
38529 (fix<uns>_<mode>di2_hw): Likewise.
38530 (float<uns>_<mode>si2_hw): Likewise.
38531 (fix_<mode>si2_hw): Likewise.
38532 (fixuns_<mode>si2_hw): Likewise.
38533 (float<uns>_<mode>di2_hw): Likewise.
38534 (float_<mode>di2_hw): Likewise.
38535 (float_<mode>si2_hw): Likewise.
38536 (floatuns_<mode>di2_hw): Likewise.
38537 (floatuns_<mode>si2_hw): Likewise.
38538 (xscvqp<su>wz_<mode>): Delete, no longer used.
38539 (xscvqp<su>dz_<mode>): Likewise.
38540 (xscv<su>dqp_<mode>): Likewise.
38541 (ieee128_mfvsrd_64bit): Likewise.
38542 (ieee128_mfvsrd_32bit): Likewise.
38543 (ieee128_mfvsrwz): Likewise.
38544 (ieee128_mtvsrw): Likewise.
38545 (ieee128_mtvsrd_64bit): Likewise.
38546 (ieee128_mtvsrd_32bit): Likewise.
38548 2017-01-31 Martin Liska <mliska@suse.cz>
38551 * ipa-prop.c (ipa_free_all_node_params): Call release method
38552 instead of ~sumbol_summary to not to trigger double times
38555 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
38557 PR tree-optimization/71691
38558 * bitmap.h (class auto_bitmap): New.
38559 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
38560 is_maybe_undefined instead of ssa_undefined_value_p.
38562 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38564 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38565 __S390_ARCH_LEVEL__ to __ARCH__.
38567 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38569 PR tree-optimization/79267
38570 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38571 if should_remove_lhs_p is true.
38573 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38576 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38577 (add_alignment_attribute): New.
38578 (base_type_die): Add alignment attribute.
38579 (subrange_type_die): Likewise.
38580 (modified_type_die): Likewise.
38581 (gen_array_type_die): Likewise.
38582 (gen_descr_array_type_die: Likewise.
38583 (gen_enumeration_type_die): Likewise.
38584 (gen_subprogram_die): Likewise.
38585 (gen_variable_die): Likewise.
38586 (gen_field_die): Likewise.
38587 (gen_ptr_to_mbr_type_die): Likewise.
38588 (gen_struct_or_union_type_die): Likewise.
38589 (gen_subroutine_type_die): Likewise.
38590 (gen_typedef_die): Likewise.
38591 (base_type_cmp): Compare alignment attribute.
38593 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38596 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38597 (setb_unsigned) New pattern for setb with CCUNS.
38598 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38599 subfc./subfe sequence to avoid overflow problems. Generate a
38600 shorter sequence with cmpld/setb for power9.
38601 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38602 for generating subfc. instruction.
38603 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38604 now uses this instruction.
38606 2017-01-30 Ian Lance Taylor <iant@google.com>
38609 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38610 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38612 2017-01-30 Martin Sebor <msebor@redhat.com>
38614 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38615 Move constant to the right of a relational operator.
38616 (get_mpfr_format_length, format_character, format_string): Ditto.
38617 (should_warn_p, maybe_warn): Same.
38619 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38621 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38624 * asan.c (get_translation_unit_decl): Remove function.
38625 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38627 2017-01-30 Martin Liska <mliska@suse.cz>
38629 PR gcov-profile/79259
38630 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38631 -fprofile-generate.
38633 2017-01-30 Martin Liska <mliska@suse.cz>
38636 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38637 Initialize variables with NULL value.
38639 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38642 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38644 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38646 2017-01-30 Richard Biener <rguenther@suse.de>
38648 PR tree-optimization/79276
38649 * tree-vrp.c (process_assert_insertions): Properly adjust common
38650 when removing a duplicate.
38652 2017-01-30 Richard Biener <rguenther@suse.de>
38654 PR tree-optimization/79256
38655 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38656 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38658 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38660 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38663 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38664 ("*r<noxa>sbg_<mode>_sll_bitmask")
38665 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38666 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38667 Use contiguous_bitmask_nowrap_operand.
38669 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38672 * config/rs6000/altivec.h (vec_xl): Revise #define.
38673 (vec_xst): Likewise.
38675 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38677 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38679 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38681 PR rtl-optimization/79194
38682 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38683 traps before call to bypass_conditional_jumps.
38685 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38687 PR tree-optimization/71374
38688 * lra-constraints.c (check_conflict_input_operands): New.
38689 (match_reload): Use it.
38691 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38694 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38695 account to calculate conflict_set.
38697 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38699 PR rtl-optimization/78559
38700 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38701 other_insn in combine.
38703 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38705 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38706 uint16_type_node for BT_UINT16.
38708 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38710 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38711 "RTL Tests" to menu.
38712 (GIMPLE Tests): New node.
38713 (RTL Tests): New node.
38715 2017-01-27 Richard Biener <rguenther@suse.de>
38717 PR tree-optimization/79245
38718 * tree-loop-distribution.c (distribute_loop): Apply cost
38719 modeling also to detected patterns.
38721 2017-01-27 Richard Biener <rguenther@suse.de>
38723 PR tree-optimization/71433
38724 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38725 (compare_assert_loc): New function.
38726 (process_assert_insertions): Sort and optimize assert locations
38727 to remove duplicates and push down identical assertions on
38728 edges to their destination block.
38730 2017-01-27 Richard Biener <rguenther@suse.de>
38732 PR tree-optimization/79244
38733 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38734 out SSA names even if abnormal.
38736 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38738 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38739 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38740 instead of MPFR_RNDN.
38742 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38745 * arm.c (arm_option_override): Don't call build_target_option_node
38746 until after doing all option overrides.
38747 (arm_valid_target_attribute_tree): Likewise.
38749 2017-01-27 Martin Liska <mliska@suse.cz>
38751 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38752 for {cd}tors and C++ {cd}tors.
38754 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38756 * config/s390/s390.md ("*setmem_long_and")
38757 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38759 2017-01-26 Martin Sebor <msebor@redhat.com>
38761 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38764 2017-01-26 Martin Sebor <msebor@redhat.com>
38766 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38767 HAVE_DFmode before using XFmode or DFmode.
38768 (parse_directive): Avoid using the z length modifier to avoid
38769 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38771 PR middle-end/78703
38772 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38773 to accept adjustment as an array.
38774 (get_int_range): New function.
38775 (struct directive): Make width and prec arrays.
38776 (directive::set_width, directive::set_precision): Call get_int_range.
38777 (format_integer, format_floating): Handle width and precision ranges.
38778 (format_string, parse_directive): Same.
38780 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38783 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38784 comdat_type_p set, just clone them, but keep the children in the
38788 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38789 which have direct callers with -fvar-tracking-assignments enabled
38791 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38792 inside of type units.
38794 2017-01-26 Martin Sebor <msebor@redhat.com>
38796 PR middle-end/78703
38797 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38799 (struct format_result): Replace number_chars, number_chars_min,
38800 and number_chars_max with a single member of struct result_range.
38802 (format_result::operator+=): Adjust.
38803 (struct fmtresult): Remove bounded. Handle likely and unlikely
38805 (fmtresult::adjust_for_width_or_precision): New function.
38806 (fmtresult:type_max_digits): New function.
38807 (bytes_remaining): Handle likely and unlikely counters.
38808 (min_bytes_remaining): Remove.
38809 (format_percent): Simplify.
38810 (format_integer, format_floating): Set likely and unlikely counters.
38811 (get_string_length, format_character, format_string): Same.
38812 (format_plain, should_warn_p): New function.
38813 (maybe_warn): Call should_warn_p. Update diagnostic messages
38814 and handle those for all directives, including plain strings.
38815 (format_directive): Handle likely and unlikely counters.
38816 Remove unnecessary quoting from diagnostics. Add an informational
38818 (add_bytes): Remove.
38819 (pass_sprintf_length::compute_format_length): Simplify.
38820 (try_substitute_return_value): Handle likely and unlikely counters.
38822 2017-01-26 Carl Love <cel@us.ibm.com>
38824 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
38825 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
38827 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
38830 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
38831 endianess for subregs into account.
38832 * lra-constraints.c (lra_constraints): Do risky transformations
38833 always on the first iteration.
38834 * lra-lives.c (check_pseudos_live_through_calls): Add arg
38835 last_call_used_reg_set.
38836 (process_bb_lives): Define and use last_call_used_reg_set.
38837 * lra.c (lra): Always continue after lra_constraints on the first
38840 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
38842 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
38844 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
38846 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38848 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
38849 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
38850 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
38851 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
38852 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
38853 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
38854 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
38855 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
38856 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
38858 2017-01-26 Marek Polacek <polacek@redhat.com>
38861 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
38862 for the third operand.
38864 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38866 PR middle-end/79236
38867 * omp-low.c (struct omp_context): Add simt_stmt field.
38868 (scan_omp_for): Return omp_context *.
38869 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
38870 context to the _simt_ SIMD stmt.
38871 (lower_omp_for): For combined SIMD with sibling _simt_
38872 SIMD, make sure to use the same decls in _looptemp_
38873 clauses as in the sibling.
38875 2017-01-26 David Sherwood <david.sherwood@arm.com>
38877 PR middle-end/79212
38878 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
38881 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38884 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
38885 emit fld b; fld a; if possible.
38887 * brig-builtins.def: Update copyright years.
38888 * config/arm/arm_acle_builtins.def: Update copyright years.
38890 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
38893 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
38894 constraint instead of o for the stxsd instruction.
38896 2017-01-25 Carl Love <cel@us.ibm.com>
38898 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
38899 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
38901 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
38903 * doc/invoke.texi (C++ Dialect Options): Fix typo.
38905 2017-01-25 Richard Biener <rguenther@suse.de>
38907 PR tree-optimization/69264
38908 * target.def (vector_alignment_reachable): Improve documentation.
38909 * doc/tm.texi: Regenerate.
38910 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
38912 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
38913 earlier changes with respect to TYPE_USER_ALIGN.
38914 (vector_alignment_reachable_p): Likewise. Improve dumping.
38916 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38919 * config/arm/arm.md (xordi3): Force constant operand into a register
38922 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38924 * doc/invoke.texi (-fstore-merging): Correct default optimization
38925 levels at which it is enabled.
38926 (-O): Move -fstore-merging from list to...
38929 2017-01-25 Richard Biener <rguenther@suse.de>
38932 * omp-expand.c: Include debug.h.
38933 (expand_omp_taskreg): Make sure to generate early debug before
38934 outlining anything from a function.
38935 (expand_omp_target): Likewise.
38936 (grid_expand_target_grid_body): Likewise.
38938 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
38941 * asan.c (get_translation_unit_decl): New function.
38942 (asan_add_global): Extract modules file name from globals
38943 TRANSLATION_UNIT_DECL name.
38945 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
38948 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
38949 for long calls with APCS frame and VFP.
38951 2017-01-24 David Malcolm <dmalcolm@redhat.com>
38953 * cfg.c (original_copy_tables_initialized_p): New function.
38954 * cfg.h (original_copy_tables_initialized_p): New decl.
38955 * cfgrtl.c (relink_block_chain): Guard the call to
38956 free_original_copy_tables with a call to
38957 original_copy_tables_initialized_p.
38958 * cgraph.h (symtab_node::native_rtl_p): New decl.
38959 * cgraphunit.c (symtab_node::native_rtl_p): New function.
38960 (symtab_node::needed_p): Don't assert for early assembly output
38961 for __RTL functions.
38962 (cgraph_node::finalize_function): Set "force_output" for __RTL
38964 (cgraph_node::analyze): Bail out early for __RTL functions.
38965 (analyze_functions): Update assertion to support __RTL functions.
38966 (cgraph_node::expand): Bail out early for __RTL functions.
38967 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
38969 * function.h (struct function): Update comment for field
38971 * gimple-expr.c: Include "tree-pass.h".
38972 (gimple_has_body_p): Return false for __RTL functions.
38973 * Makefile.in (OBJS): Add run-rtl-passes.o.
38974 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
38976 (gcc::pass_manager::get_clean_slate): New accessor.
38977 * passes.c: Include "insn-addr.h".
38978 (should_skip_pass_p): Add logging. Update logic for running
38979 "expand" to be compatible with both __GIMPLE and __RTL. Guard
38980 property-provider override so it is only done for gimple passes.
38982 (skip_pass): New function.
38983 (execute_one_pass): Call skip_pass when skipping passes.
38984 * read-md.c (md_reader::read_char): Support filtering
38985 the input to a subset of line numbers.
38986 (md_reader::md_reader): Initialize fields
38987 m_first_line and m_last_line.
38988 (md_reader::read_file_fragment): New function.
38989 * read-md.h (md_reader::read_file_fragment): New decl.
38990 (md_reader::m_first_line): New field.
38991 (md_reader::m_last_line): New field.
38992 * read-rtl-function.c (function_reader::create_function): Only
38993 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
38994 curr_properties. Set DECL_INITIAL to a dummy block.
38995 (read_rtl_function_body_from_file_range): New function.
38996 * read-rtl-function.h (read_rtl_function_body_from_file_range):
38998 * run-rtl-passes.c: New file.
38999 * run-rtl-passes.h: New file.
39001 2017-01-24 Jeff Law <law@redhat.com>
39003 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39006 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39008 PR tree-optimization/79159
39009 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39010 (record_nonwrapping_iv): Improve boundary using above function if no
39011 value range information.
39013 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39014 Martin Jambor <mjambor@suse.cz>
39016 * brig-builtins.def: New file.
39017 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39018 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39019 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39020 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39021 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39022 * builtin-types.def (BT_INT8): New.
39023 (BT_INT16): Likewise.
39024 (BT_UINT8): Likewise.
39025 (BT_UINT16): Likewise.
39026 (BT_FN_ULONG): Likewise.
39027 (BT_FN_UINT_INT): Likewise.
39028 (BT_FN_UINT_ULONG): Likewise.
39029 (BT_FN_UINT_LONG): Likewise.
39030 (BT_FN_UINT_PTR): Likewise.
39031 (BT_FN_ULONG_PTR): Likewise.
39032 (BT_FN_INT8_FLOAT): Likewise.
39033 (BT_FN_INT16_FLOAT): Likewise.
39034 (BT_FN_UINT32_FLOAT): Likewise.
39035 (BT_FN_UINT16_FLOAT): Likewise.
39036 (BT_FN_UINT8_FLOAT): Likewise.
39037 (BT_FN_UINT64_FLOAT): Likewise.
39038 (BT_FN_UINT16_UINT32): Likewise.
39039 (BT_FN_UINT32_UINT16): Likewise.
39040 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39041 (BT_FN_INT_PTR_INT): Likewise.
39042 (BT_FN_UINT_PTR_UINT): Likewise.
39043 (BT_FN_LONG_PTR_LONG): Likewise.
39044 (BT_FN_ULONG_PTR_ULONG): Likewise.
39045 (BT_FN_VOID_UINT64_UINT64): Likewise.
39046 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39047 (BT_FN_INT8_INT8_INT8): Likewise.
39048 (BT_FN_INT16_INT16_INT16): Likewise.
39049 (BT_FN_INT_INT_INT): Likewise.
39050 (BT_FN_UINT_FLOAT_UINT): Likewise.
39051 (BT_FN_FLOAT_UINT_UINT): Likewise.
39052 (BT_FN_ULONG_UINT_UINT): Likewise.
39053 (BT_FN_ULONG_UINT_PTR): Likewise.
39054 (BT_FN_ULONG_ULONG_ULONG): Likewise.
39055 (BT_FN_UINT_UINT_UINT): Likewise.
39056 (BT_FN_VOID_UINT_PTR): Likewise.
39057 (BT_FN_UINT_UINT_PTR: Likewise.
39058 (BT_FN_UINT32_UINT64_PTR): Likewise.
39059 (BT_FN_INT_INT_UINT_UINT): Likewise.
39060 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
39061 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
39062 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
39063 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
39064 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
39065 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
39066 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
39067 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
39068 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
39069 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
39070 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
39071 * doc/frontends.texi: List BRIG FE.
39072 * doc/install.texi (Testing): Add BRIG tesring requirements.
39073 * doc/invoke.texi (Overall Options): Mention BRIG.
39074 * doc/standards.texi (Standards): Doucment BRIG HSA version.
39076 2017-01-24 Richard Biener <rguenther@suse.de>
39078 PR translation/79208
39079 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
39081 2017-01-24 Martin Jambor <mjambor@suse.cz>
39084 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
39085 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
39086 and known_contexts.
39088 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
39090 PR middle-end/79123
39091 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
39092 casts from signed to unsigned really don't have a range.
39094 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
39096 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
39097 GMP_RNDx for compatiblity.
39099 2017-01-24 Martin Liska <mliska@suse.cz>
39102 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
39103 that would prevent us to call alloca with -1 as argument.
39105 2017-01-24 Jakub Jelinek <jakub@redhat.com>
39107 * dwarf2out.c (output_compilation_unit_header, output_file_names):
39108 Avoid -Wformat-security warning.
39110 2017-01-23 Andrew Pinski <apinski@cavium.com>
39112 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
39115 2017-01-23 Martin Sebor <msebor@redhat.com>
39117 PR middle-end/78703
39118 * gimple-ssa-sprintf.c (warn_level): New global.
39119 (format_integer): Use it here and throughout the rest of the file.
39120 Use the same switch to compute sign as base.
39121 (maybe_warn): New function.
39122 (format_directive): Factor out warnings into maybe_warn.
39123 Add debugging output. Use warn_level.
39124 (add_bytes): Use warn_level.
39125 (pass_sprintf_length::compute_format_length): Add debugging output.
39126 (try_substitute_return_value): Same.
39127 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
39129 PR middle-end/78703
39130 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
39131 (struct fmtresult, format_integer, format_floating): Adjust.
39132 (fmtresult::fmtresult): Set max correctly in two argument ctor.
39133 (get_string_length, format_string,format_directive): Same.
39134 (pass_sprintf_length::compute_format_length): Same.
39135 (try_substitute_return_value): Simplify slightly.
39137 PR middle-end/78703
39138 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
39139 (fmtresult::operator+=): Outlined.
39140 (struct fmtresult): Add ctors.
39141 (struct conversion_spec): Rename...
39142 (struct directive): ...to this. Add and remove data members.
39143 (directive::set_width, directive::set_precision): New functions.
39144 (format_percent): Use fmtresult ctor.
39145 (get_width_and_precision): Remove.
39146 (format_integer): Make naming changes. Avoid computing width and
39148 (format_floating): Same. Adjust indentation.
39149 (format_character, format_none): New functions.
39150 (format_string): Moved character handling to format_character.
39151 (format_directive): Remove arguments, change return type.
39152 (parse_directive): New function.
39153 (pass_sprintf_length::compute_format_length): Move directive
39154 parsing to parse_directive.
39156 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39158 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
39159 (assign_assembler_name_if_needed): ... this.
39160 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
39161 (assign_assembler_name_if_needed): ... this.
39162 (free_lang_data_in_cgraph): Adjust callers.
39163 * cgraphunit.c (cgraph_node::analyze): Likewise.
39164 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
39167 2017-01-23 Richard Biener <rguenther@suse.de>
39169 PR tree-optimization/79088
39170 PR tree-optimization/79188
39171 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
39172 resetting loop bounds after last path deletion. Reset loop
39173 bounds of the target loop, make code match the comments.
39174 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
39175 Make sure loops need no fixups.
39177 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
39179 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
39180 exponent support with double type for first argument.
39181 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
39182 type returned by __builtin_vec_extract_sig,
39183 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
39184 functions from "vector int" to "vector unsigned int" or from
39185 "vector long long int" to "vector unsigned long long int".
39186 Changed type returned by __builtin_vec_extract_exp,
39187 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
39188 functions from "vector int" to "vector unsigned int" or from
39189 "vector long long int" to "vector unsigned long long int".
39190 Changed return type of __builtin_vec_test_data_class,
39191 __builtin_vec_test_data_class_sp, and
39192 __builtin_vec_test_data_class_dp from "vector int" to
39193 "vector bool int" or from "vector long long int" to "vector bool
39194 long long int" and changed second argument type from "unsigned
39195 int" to "int". Added new overloaded function forms "vector float
39196 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
39197 "vector float __builtin_vec_insert_exp_sp (vector float, vector
39198 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
39199 double, vector unsigned long long int)" and "vector double
39200 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
39201 long int)". Changed return type of
39202 __builtin_scalar_test_data_class and
39203 __builtin_scalar_test_data_class_sp and
39204 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
39205 int" and changed second argument from "unsigned int" to "int".
39206 Changed type returned by __builtin_scalar_test_neg,
39207 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
39208 from "int" to "bool int". Added new overloaded function form
39209 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
39210 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
39211 exponent double-precision with floating point first argument.
39212 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
39213 documentation of scalar_test_data_class, scalar_test_neg,
39214 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
39215 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
39216 vec_test_data_class built-in functions to reflect refinements in
39217 their type signatures.
39219 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
39221 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
39223 (aarch64_elf_asm_destructor): Likewise.
39225 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
39227 PR rtl-optimization/78634
39228 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
39229 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
39230 * ifcvt.c (noce_try_cmove): Add missing cost check.
39232 PR rtl-optimization/71724
39233 * combine.c (if_then_else_cond): Look for situations where it is
39234 beneficial to undo the work of one of the recursive calls.
39236 2017-01-23 Bin Cheng <bin.cheng@arm.com>
39238 PR tree-optimization/70754
39239 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
39240 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
39241 combined stmt before it if not NULL.
39242 (combine_chains): Process refs reversely and compute dominance point
39245 2017-01-23 Martin Liska <mliska@suse.cz>
39247 PR tree-optimization/79196
39248 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
39249 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
39251 (strlen_optimize_stmt): Call the renamed function.
39253 2017-01-23 Michael Matz <matz@suse.de>
39255 PR tree-optimization/78384
39256 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
39258 2017-01-23 Richard Biener <rguenther@suse.de>
39260 PR tree-optimization/79186
39261 * tree-vrp.c (register_new_assert_for): Make sure we've seen
39262 both incoming edges before moving an assert.
39264 2017-01-23 Martin Jambor <mjambor@suse.cz>
39266 * ipa-prop.c (load_from_param_1): Removed.
39267 (load_from_unmodified_param): Bits from load_from_param_1 put back
39269 (load_from_param): Removed.
39270 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
39271 with stmt. Reverted back to use of load_from_unmodified_param.
39273 2017-01-23 Martin Jambor <mjambor@suse.cz>
39276 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
39277 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
39278 field a pointer to garbage collected vector, mark lattices and
39279 ipcp_orig_node with GTY((skip)).
39280 (ipa_get_param_count): Adjust to descriptors being a pointer.
39281 (ipa_get_param): Likewise.
39282 (ipa_get_type): Likewise.
39283 (ipa_get_param_move_cost): Likewise.
39284 (ipa_set_param_used): Likewise.
39285 (ipa_get_controlled_uses): Likewise.
39286 (ipa_set_controlled_uses): Likewise.
39287 (ipa_is_param_used): Likewise.
39288 (ipa_node_params_t): Move into garbage collector. New methods insert
39290 (ipa_node_params_sum): Annotate wth GTY(()).
39291 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
39293 (ipa_load_from_parm_agg): Adjust declaration.
39294 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
39295 * ipa-profile.c (ipa_profile): Likewise.
39296 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
39297 (ipa_populate_param_decls): Make descriptors parameter garbage
39299 (ipa_dump_param): Adjust to descriptors being a pointer.
39300 (ipa_alloc_node_params): Likewise.
39301 (ipa_initialize_node_params): Likewise.
39302 (load_from_param_1): Make descriptors parameter garbage collected.
39303 (load_from_unmodified_param): Likewise.
39304 (load_from_param): Likewise.
39305 (ipa_load_from_parm_agg): Likewise.
39306 (ipa_node_params::~ipa_node_params): Removed.
39307 (ipa_free_all_node_params): Remove call to delete operator.
39308 (ipa_node_params_t::insert): New.
39309 (ipa_node_params_t::remove): Likewise.
39310 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
39311 copy known_csts and known_contexts vectors.
39312 (ipa_read_node_info): Adjust to descriptors being a pointer.
39313 (ipcp_modif_dom_walker): Make m_descriptors field garbage
39315 (ipcp_transform_function): Make descriptors variable garbage
39318 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
39320 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
39321 * config/i386/avx512dqintrin.h: Ditto.
39322 * config/i386/avx512fintrin.h: Ditto.
39323 * config/i386/i386.c: Handle new builtins.
39324 * config/i386/i386-builtin.def: Add new builtins.
39325 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
39326 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
39328 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39329 Martin Liska <mliska@suse.cz>
39331 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
39332 * asan.c (asan_expand_poison_ifn): Support stores and use
39333 appropriate ASAN report function.
39334 * internal-fn.c (expand_ASAN_POISON_USE): New function.
39335 * internal-fn.def (ASAN_POISON_USE): Declare.
39336 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
39337 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
39338 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
39339 ASAN_POISON calls w/o LHS.
39340 * tree-ssa.c (execute_update_addresses_taken): Create clobber
39341 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
39342 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
39343 * gimplify.c (asan_poison_variables): Add attribute
39344 use_after_scope_memory to variables that really needs to live
39346 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
39347 having the attribute.
39349 2017-01-23 Martin Liska <mliska@suse.cz>
39351 * asan.c (create_asan_shadow_var): New function.
39352 (asan_expand_poison_ifn): Likewise.
39353 * asan.h (asan_expand_poison_ifn): New declaration.
39354 * internal-fn.c (expand_ASAN_POISON): Likewise.
39355 * internal-fn.def (ASAN_POISON): New builtin.
39356 * sanopt.c (pass_sanopt::execute): Expand
39357 asan_expand_poison_ifn.
39358 * tree-inline.c (copy_decl_for_dup_finish): Make function
39360 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
39361 * tree-ssa.c (is_asan_mark_p): New function.
39362 (execute_update_addresses_taken): Rewrite local variables
39363 (identified just by use-after-scope as addressable) into SSA.
39365 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39367 * doc/install.texi (Specific): opensource.apple.com uses https
39368 now. Remove trailing slash.
39370 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39372 * README.Portability: Remove note on an Irix compatibility issue.
39374 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
39376 * gcov.c (INCLUDE_ALGORITHM): Define.
39377 (INCLUDE_VECTOR): Define.
39378 No longer include <vector> and <algorithm> directly.
39380 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
39382 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
39384 * doc/invoke.texi (Code Gen Options): Ditto.
39386 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
39389 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
39391 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
39393 rtl-optimization/79125
39394 * cprop.c (local_cprop_pass): Handle cases where we make an
39395 unconditional trap.
39397 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
39401 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
39402 read from, for big endian.
39404 2017-01-20 Jiong Wang <jiong.wang@arm.com>
39406 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
39407 register pauth builtins for LP64 only.
39409 2017-01-20 Marek Polacek <polacek@redhat.com>
39412 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
39415 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39417 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
39419 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
39420 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
39421 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
39423 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39426 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
39427 in big-endian mode when they are not a single duplicated value.
39429 2017-01-20 Richard Biener <rguenther@suse.de>
39431 * BASE-VER: Bump to 7.0.1.
39433 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39435 * omp-low.c (omplow_simd_context): New struct. Use it...
39436 (lower_rec_simd_input_clauses): ...here and...
39437 (lower_rec_input_clauses): ...here to hold common data. Adjust all
39438 references to idx, lane, max_vf, is_simt.
39440 2017-01-20 Graham Markall <graham.markall@embecosm.com>
39442 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
39445 2017-01-20 Martin Jambor <mjambor@suse.cz>
39447 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
39448 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
39450 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
39451 (GTFILES): Rename hsa.c to hsa-common.c.
39452 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
39453 * hsa-dump.c: Likewise.
39454 * hsa-gen.c: Likewise.
39455 * hsa-regalloc.c: Likewise.
39456 * ipa-hsa.c: Likewise.
39457 * omp-expand.c: Likewise.
39458 * omp-low.c: Likewise.
39459 * toplev.c: Likewise.
39461 2017-01-20 Marek Polacek <polacek@redhat.com>
39464 * doc/invoke.texi: Document -Wduplicated-branches.
39465 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
39466 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
39467 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
39468 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
39469 return 0 only when not OEP_LEXICOGRAPHIC.
39470 (fold_build_cleanup_point_expr): Use the expression
39471 location when building CLEANUP_POINT_EXPR.
39472 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
39473 * tree.c (add_expr): Handle error_mark_node.
39475 2017-01-20 Martin Liska <mliska@suse.cz>
39478 * tree-profile.c (init_ic_make_global_vars): Do not call
39480 (gimple_init_gcov_profiler): Likewise.
39482 2017-01-20 Martin Liska <mliska@suse.cz>
39485 * cgraph.h (maybe_create_reference): Remove argument and
39487 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
39489 * ipa-cp.c (create_specialized_node): Likewise.
39490 * symtab.c (symtab_node::maybe_create_reference): Handle
39491 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
39493 2017-01-20 Martin Liska <mliska@suse.cz>
39495 * read-rtl-function.c (function_reader::create_function): Use
39496 build_decl instread of build_decl_stat.
39498 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
39500 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
39501 * config/i386/avx512dqintrin.h: Ditto.
39502 * config/i386/avx512fintrin.h: Ditto.
39503 * config/i386/i386-builtin-types.def: Add new types.
39504 * config/i386/i386.c: Handle new types.
39505 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
39506 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
39507 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
39508 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
39509 (__builtin_ia32_kshiftridi): New.
39510 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
39512 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
39516 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
39517 define to rs6000_init_stack_protect_guard.
39518 (rs6000_init_stack_protect_guard): New function.
39520 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39521 Yunqiang Su <yunqiang.su@imgtec.com>
39523 * config.gcc (supported_defaults): Add madd4.
39524 (with_madd4): Add validation.
39525 (all_defaults): Add madd4.
39526 * config/mips/mips.opt (mmadd4): New option.
39527 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39529 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39531 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
39532 (ISA_HAS_FUSED_MADD4): Likewise.
39533 * doc/invoke.texi (-mmadd4): Document the new option.
39534 * doc/install.texi (--with-madd4): Document the new option.
39536 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39538 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
39539 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
39540 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
39541 (aarch64_init_pauth_hint_builtins): New.
39542 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
39543 (aarch64_expand_builtin): Expand new builtins.
39545 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39547 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
39548 * combine-stack-adj.c (no_unhandled_cfa): Handle
39549 REG_CFA_TOGGLE_RA_MANGLE.
39550 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
39551 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
39552 info for return address signing.
39553 (aarch64_expand_epilogue): Likewise.
39555 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39557 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
39558 * config/aarch64/aarch64-protos.h
39559 (aarch64_return_address_signing_enabled): New declaration.
39560 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
39562 (aarch64_expand_prologue): Sign return address before it's pushed onto
39564 (aarch64_expand_epilogue): Authenticate return address fetched from
39566 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39567 (aarch64_attributes): New function attributes for "sign-return-address".
39568 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39569 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39570 ("*do_return"): Generate combined instructions according to key index.
39571 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39572 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39574 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39575 * config/aarch64/aarch64.opt (msign-return-address=): New.
39576 * doc/extend.texi (AArch64 Function Attributes): Documents
39577 "sign-return-address=".
39578 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39580 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39582 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39583 overall option summary.
39585 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39587 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39588 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39589 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39590 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39592 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39594 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39595 -mpower9-minmax by default for -mcpu=power9.
39596 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39597 128-bit floating point.
39599 2017-01-20 Alan Modra <amodra@gmail.com>
39601 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39602 optimizing for size.
39604 2017-01-20 Alan Modra <amodra@gmail.com>
39607 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39608 for strcmp and strncmp from corresponding builtin decl.
39610 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39612 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39613 instead of i386/rtems-64.h.
39614 * config/i386/rtems-64.h: Remove.
39616 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39620 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39622 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39624 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39626 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39627 Change int to HOST_WIDE_INT.
39628 * config/aarch64/aarch64-protos.h
39629 (aarch64_simd_gen_const_vector_dup): Likewise.
39630 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39632 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39634 * langhooks-def.h (lhd_type_for_size): New decl.
39635 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39636 * langhooks.c (lhd_type_for_size): New function, taken from
39639 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39641 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39642 define_bypass for CR latency.
39643 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39644 (power9-alu2): Add define_bypass for CR latency.
39646 (power9-mul): Update insn latency.
39647 (power9-mul-compare): Update insn latency, bypass latency and remove
39650 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39652 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39654 * config/aarch64/aarch64.md
39655 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39656 aarch64_nopcrelative_literal_loads.
39657 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39659 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39661 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39662 TARGET_LOONGSON_3A.
39663 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39665 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39668 * config.gcc (supported_defaults): Add lxc1-sxc1.
39669 (with_lxc1_sxc1): Add validation.
39670 (all_defaults): Add lxc1-sxc1.
39671 * config/mips/mips.opt (mlxc1-sxc1): New option.
39672 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39674 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39675 __mips_no_lxc1_sxc1.
39676 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39677 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39678 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39680 2017-01-19 Richard Biener <rguenther@suse.de>
39682 PR tree-optimization/72488
39683 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39684 sure to restore SSA info.
39685 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39687 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39689 PR rtl-optimization/79121
39690 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39691 of the inner type when shifting an extended value.
39693 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39696 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39697 interposable aliases.
39699 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39702 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39703 Use the evmergelohi instruction.
39704 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39705 (mov_sitf_e500_subreg8_2_be): Likewise.
39706 (mov_sitf_e500_subreg12_2_le): Likewise.
39707 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39708 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39709 (mov_sitf_e500_subreg8_2_le): Likewise.
39710 (mov_sitf_e500_subreg12_2_be): Likewise.
39712 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39714 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39715 attribute from vecsimple to vecperm.
39716 (altivec_vbpermq2): Likewise.
39718 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39721 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39723 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39724 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39725 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39726 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39727 case where N arg is SIZE_MAX.
39728 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39729 (cmpstrsi): Add pattern.
39731 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39733 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39734 __builtin_vec_revb builtins.
39735 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39736 built-in functions to support generation of the ISA 3.0 XXBR<x>
39737 vector byte reverse instructions.
39738 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39739 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39740 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39741 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39742 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39743 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39744 (P9V_BUILTIN_VEC_REVB): Likewise.
39745 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39746 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39747 (p9_xxbrq_v16qi): Likewise.
39748 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39749 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39750 (p9_xxbrh_v8hi): Likewise.
39751 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39752 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39753 vec_revb built-in functions.
39755 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39757 PR rtl-optimization/78952
39758 * config/i386/i386.md (any_extract): New code iterator.
39759 (*insvqi_2): Use any_extract for source operand.
39760 (*insvqi_3): Use any_shiftrt for source operand.
39762 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39764 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39766 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39768 2017-01-18 Matthias Klose <doko@ubuntu.com>
39770 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39772 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39774 * config/rs6000/altivec.h (vec_bperm): Change #define.
39775 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39776 (altivec_vbpermq2): New define_insn.
39777 (altivec_vbpermd): Likewise.
39778 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39779 function interface.
39780 (VBPERMD): Likewise.
39781 (VBPERM): New polymorphic function interface.
39782 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39783 Add entries for P9V_BUILTIN_VEC_VBPERM.
39784 * doc/extend.texi: Add interfaces for vec_bperm.
39786 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39788 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39789 first letter of error messages.
39790 (s390_resolve_overloaded_builtin): Likewise.
39791 * config/s390/s390.c (s390_expand_builtin): Likewise.
39792 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39793 (s390_valid_target_attribute_inner_p): Likewise.
39794 * config/s390/s390.md ("tabort"): Likewise.
39796 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39798 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39799 (ISA_AVOID_DIV_HILO): New macro.
39800 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39801 (ISA_HAS_DDIV): Likewise.
39803 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39805 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39807 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39809 * doc/invoke.texi (fabi-version): Spelling fix.
39811 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39814 * doc/invoke.texi (fabi-version): Mention mangling fix for
39817 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39820 * doc/invoke.texi (fabi-version): Document discriminator mangling.
39822 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
39825 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
39826 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
39828 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
39830 (stack_protect_test): Ditto.
39831 * config/rs6000/rs6000.opt (mstack-protector-guard=,
39832 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
39834 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
39835 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
39836 -mstack-protector-guard-offset=.
39837 (RS/6000 and PowerPC Options): Ditto.
39839 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39841 * config/i386/i386.h (MASK_CLASS_P): New define.
39842 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
39843 there are no registers from different register sets also when
39844 mask registers are used. Update function comment.
39845 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
39846 to (*k/*r) and (*k/*km) alternatives.
39848 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
39850 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
39851 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
39852 (EH_RETURN_HANDLER_RTX): New define.
39853 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
39854 Force frame pointer in EH return functions.
39855 (aarch64_expand_epilogue): Add barrier for eh_return.
39856 (aarch64_final_eh_return_addr): Remove.
39857 (aarch64_eh_return_handler_rtx): New function.
39858 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
39860 (aarch64_eh_return_handler_rtx): New prototype.
39862 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39864 * config/rs6000/altivec.h (vec_rlmi): New #define.
39865 (vec_vrlnm): Likewise.
39866 (vec_rlnm): Likewise.
39867 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
39868 (UNSPEC_VRLNM): Likewise.
39869 (VIlong): New mode iterator.
39870 (altivec_vrl<VI_char>mi): New define_insn.
39871 (altivec_vrl<VI_char>nm): Likewise.
39872 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
39874 (VRLDNM): Likewise.
39875 (RLNM): New polymorphic function entry.
39876 (VRLWMI): New monomorphic function entry.
39877 (VRLDMI): Likewise.
39878 (RLMI): New polymorphic function entry.
39879 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
39880 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
39881 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
39884 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39887 * dwarf2out.c (field_byte_offset): Restore the
39888 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
39889 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
39890 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
39893 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39896 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
39898 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39901 * dwarf2out.c (add_data_member_location_attribute): For constant
39902 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
39903 instead of DW_AT_data_member_location, DW_AT_bit_offset and
39904 DW_AT_byte_size attributes.
39906 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39908 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
39909 after forcing to constant memory when the code model is medium.
39911 2017-01-17 Julia Koval <julia.koval@intel.com>
39914 * config/i386/avx512fintrin.h
39915 (_mm512_i32gather_ps): Change __addr type to void const*.
39916 (_mm512_mask_i32gather_ps): Ditto.
39917 (_mm512_i32gather_pd): Ditto.
39918 (_mm512_mask_i32gather_pd): Ditto.
39919 (_mm512_i64gather_ps): Ditto.
39920 (_mm512_mask_i64gather_ps): Ditto.
39921 (_mm512_i64gather_pd): Ditto.
39922 (_mm512_mask_i64gather_pd): Ditto.
39923 (_mm512_i32gather_epi32): Ditto.
39924 (_mm512_mask_i32gather_epi32): Ditto.
39925 (_mm512_i32gather_epi64): Ditto.
39926 (_mm512_mask_i32gather_epi64): Ditto.
39927 (_mm512_i64gather_epi32): Ditto.
39928 (_mm512_mask_i64gather_epi32): Ditto.
39929 (_mm512_i64gather_epi64): Ditto.
39930 (_mm512_mask_i64gather_epi64): Ditto.
39931 (_mm512_i32scatter_ps): Change __addr type to void*.
39932 (_mm512_mask_i32scatter_ps): Ditto.
39933 (_mm512_i32scatter_pd): Ditto.
39934 (_mm512_mask_i32scatter_pd): Ditto.
39935 (_mm512_i64scatter_ps): Ditto.
39936 (_mm512_mask_i64scatter_ps): Ditto.
39937 (_mm512_i64scatter_pd): Ditto.
39938 (_mm512_mask_i64scatter_pd): Ditto.
39939 (_mm512_i32scatter_epi32): Ditto.
39940 (_mm512_mask_i32scatter_epi32): Ditto.
39941 (_mm512_i32scatter_epi64): Ditto.
39942 (_mm512_mask_i32scatter_epi64): Ditto.
39943 (_mm512_i64scatter_epi32): Ditto.
39944 (_mm512_mask_i64scatter_epi32): Ditto.
39945 (_mm512_i64scatter_epi64): Ditto.
39946 (_mm512_mask_i64scatter_epi64): Ditto.
39947 * config/i386/avx512pfintrin.h
39948 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
39949 (_mm512_mask_prefetch_i32gather_ps): Ditto.
39950 (_mm512_mask_prefetch_i64gather_pd): Ditto.
39951 (_mm512_mask_prefetch_i64gather_ps): Ditto.
39952 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
39953 (_mm512_prefetch_i32scatter_ps): Ditto.
39954 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
39955 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
39956 (_mm512_prefetch_i64scatter_pd): Ditto.
39957 (_mm512_prefetch_i64scatter_ps): Ditto.
39958 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
39959 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
39960 * config/i386/avx512vlintrin.h
39961 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
39962 (_mm_mmask_i32gather_ps): Ditto.
39963 (_mm256_mmask_i32gather_pd): Ditto.
39964 (_mm_mmask_i32gather_pd): Ditto.
39965 (_mm256_mmask_i64gather_ps): Ditto.
39966 (_mm_mmask_i64gather_ps): Ditto.
39967 (_mm256_mmask_i64gather_pd): Ditto.
39968 (_mm_mmask_i64gather_pd): Ditto.
39969 (_mm256_mmask_i32gather_epi32): Ditto.
39970 (_mm_mmask_i32gather_epi32): Ditto.
39971 (_mm256_mmask_i32gather_epi64): Ditto.
39972 (_mm_mmask_i32gather_epi64): Ditto.
39973 (_mm256_mmask_i64gather_epi32): Ditto.
39974 (_mm_mmask_i64gather_epi32): Ditto.
39975 (_mm256_mmask_i64gather_epi64): Ditto.
39976 (_mm_mmask_i64gather_epi64): Ditto.
39977 (_mm256_i32scatter_ps): Change __addr type to void*.
39978 (_mm256_mask_i32scatter_ps): Ditto.
39979 (_mm_i32scatter_ps): Ditto.
39980 (_mm_mask_i32scatter_ps): Ditto.
39981 (_mm256_i32scatter_pd): Ditto.
39982 (_mm256_mask_i32scatter_pd): Ditto.
39983 (_mm_i32scatter_pd): Ditto.
39984 (_mm_mask_i32scatter_pd): Ditto.
39985 (_mm256_i64scatter_ps): Ditto.
39986 (_mm256_mask_i64scatter_ps): Ditto.
39987 (_mm_i64scatter_ps): Ditto.
39988 (_mm_mask_i64scatter_ps): Ditto.
39989 (_mm256_i64scatter_pd): Ditto.
39990 (_mm256_mask_i64scatter_pd): Ditto.
39991 (_mm_i64scatter_pd): Ditto.
39992 (_mm_mask_i64scatter_pd): Ditto.
39993 (_mm256_i32scatter_epi32): Ditto.
39994 (_mm256_mask_i32scatter_epi32): Ditto.
39995 (_mm_i32scatter_epi32): Ditto.
39996 (_mm_mask_i32scatter_epi32): Ditto.
39997 (_mm256_i32scatter_epi64): Ditto.
39998 (_mm256_mask_i32scatter_epi64): Ditto.
39999 (_mm_i32scatter_epi64): Ditto.
40000 (_mm_mask_i32scatter_epi64): Ditto.
40001 (_mm256_i64scatter_epi32): Ditto.
40002 (_mm256_mask_i64scatter_epi32): Ditto.
40003 (_mm_i64scatter_epi32): Ditto.
40004 (_mm_mask_i64scatter_epi32): Ditto.
40005 (_mm256_i64scatter_epi64): Ditto.
40006 (_mm256_mask_i64scatter_epi64): Ditto.
40007 (_mm_i64scatter_epi64): Ditto.
40008 (_mm_mask_i64scatter_epi64): Ditto.
40009 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40010 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40011 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40012 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40013 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40014 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40015 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40016 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40017 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40018 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40019 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40020 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40021 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40022 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40023 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40024 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40025 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40026 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40027 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40028 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40029 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40030 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40031 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40032 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40033 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40034 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40035 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40036 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40037 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40038 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40039 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40040 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40041 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40042 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40043 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40044 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40045 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40046 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40047 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40048 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40049 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40050 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40051 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40052 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40053 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
40054 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
40055 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
40056 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
40057 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
40058 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
40059 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
40060 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
40061 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
40062 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
40063 definitions accordingly.
40065 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
40066 Kuan-Lin Chen <kuanlinchentw@gmail.com>
40069 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
40072 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
40075 * ira-conflicts.c (ira_build_conflicts): Update total conflict
40076 hard regs for inner regno.
40078 2017-01-17 Martin Liska <mliska@suse.cz>
40081 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
40082 assumption and add comment.
40084 2017-01-17 Nathan Sidwell <nathan@acm.org>
40086 * ipa-visibility.c (localize_node): New function, broken out of ...
40087 (function_and_variable_visibility): ... here. Call it.
40089 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40091 PR middle-end/77445
40092 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
40093 correctly set frequency of oudgoing edge.
40094 (duplicate_thread_path): Fix profile updating.
40096 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40099 * configure.ac: Add GCC_BASE_VER.
40100 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
40101 version from BASE-VER file.
40102 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
40103 (gcc.o): Depend on $(BASEVER).
40104 * common.opt (dumpfullversion): New option.
40105 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
40106 * doc/invoke.texi: Document -dumpfullversion.
40107 * doc/install.texi: Document --with-gcc-major-version-only.
40108 * configure: Regenerated.
40110 2017-01-17 Richard Biener <rguenther@suse.de>
40112 PR tree-optimization/71433
40113 * tree-vrp.c (register_new_assert_for): Merge same asserts
40114 on all incoming edges.
40115 (process_assert_insertions_for): Handle insertions at the
40118 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
40120 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
40121 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
40123 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
40126 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
40129 2017-01-17 Alan Modra <amodra@gmail.com>
40132 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
40133 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
40134 symbolic stack limit when pic.
40136 2017-01-16 Martin Sebor <msebor@redhat.com>
40138 PR tree-optimization/78608
40139 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
40141 2017-01-16 Jeff Law <law@redhat.com>
40144 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
40145 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
40146 for several include directories that may be relative to sysroot.
40147 * config/i386/x-mingw32 (gplus_includedir): Define.
40148 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
40149 (native_system_includedir): Likewise.
40150 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
40151 override if TARGET_SYSTEM_ROOT is defined.
40152 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
40154 PR tree-optimization/79090
40155 PR tree-optimization/33562
40156 PR tree-optimization/61912
40157 PR tree-optimization/77485
40158 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
40159 and computed trims into the dump file.
40161 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40163 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
40165 2017-01-16 Jakub Jelinek <jakub@redhat.com>
40168 * gimplify.c (gimplify_init_constructor): If want_value and
40169 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
40173 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
40174 sequence. Formatting fixes.
40175 (doloop_optimize): Formatting fixes.
40178 * gcc.c (debug_level_greater_than_spec_func): New function.
40179 (static_spec_functions): Add debug-level-gt spec function.
40180 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
40182 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40183 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40184 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
40185 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
40186 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
40187 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
40189 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
40191 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
40192 QImode fixups to general and mask registers only.
40194 2017-01-16 Carl Love <cel@us.ibm.com>
40196 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
40197 for built-in functions
40198 vector signed char vec_nabs (vector signed char)
40199 vector signed short vec_nabs (vector signed short)
40200 vector signed int vec_nabs (vector signed int)
40201 vector signed long long vec_nabs (vector signed long long)
40202 vector float vec_nabs (vector float)
40203 vector double vec_nabs (vector double)
40204 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
40206 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
40207 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
40208 * doc/extend.texi: Update the documentation file for the new built-in
40211 2017-01-16 Martin Sebor <msebor@redhat.com>
40213 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
40216 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40218 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
40219 UNSPEC_VSX__XXSPLTD to require special splat handling.
40221 2017-01-16 David Malcolm <dmalcolm@redhat.com>
40224 * system.h: Poison strndup.
40226 2017-01-16 Alan Modra <amodra@gmail.com>
40229 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
40232 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
40234 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
40236 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
40238 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
40239 call recog here. Assert that INSN_CODE (insn) is non-negative.
40241 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40244 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
40246 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
40247 in the currently scheduled RTL fragment.
40249 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40251 PR rtl-optimization/78751
40252 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
40255 2017-01-14 Jeff Law <law@redhat.com>
40257 PR tree-optimization/79090
40258 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
40259 variable length stores.
40260 (compute_trims): Delete dead assignment to *trim_tail.
40261 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
40264 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
40266 PR rtl-optimization/78626
40267 PR rtl-optimization/78727
40268 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
40269 of a block, and split such blocks after everything else is finished.
40271 2017-01-14 Alan Modra <amodra@gmail.com>
40274 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
40275 target legitimate_combined_insn.
40276 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
40277 (rs6000_legitimate_combined_insn): New function.
40278 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
40280 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
40281 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
40282 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
40284 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
40286 * doc/frontends.texi (G++ and GCC): Remove references to Java.
40288 2017-01-13 Jeff Law <law@redhat.com>
40290 PR tree-optimization/33562
40291 PR tree-optimization/61912
40292 PR tree-optimization/77485
40293 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
40295 (delete_dead_assignment): Likewise.
40296 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
40297 statement to delete_dead_call and delete_dead_assignment.
40299 2017-01-13 David Malcolm <dmalcolm@redhat.com>
40302 * substring-locations.c (format_warning_va): Strengthen case 1 so
40303 that both endpoints of the substring must be within the format
40304 range for just the substring to be printed.
40306 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
40308 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
40309 * config/i386/i386.c (ix86_target_string): Add missing options
40310 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
40311 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
40312 flags_other and ix86_target_other to flags2_other. Display unknown
40314 (ix86_valid_target_attribute_inner_p): Add missing options and
40315 reorder options by implied ISAs, as in ix86_target_string.
40317 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40319 * hash-table.h (hash_table::too_empty_p): New function.
40320 (hash_table::expand): Use it.
40321 (hash_table::traverse): Likewise.
40322 (hash_table::empty_slot): Use sizeof (value_type) instead of
40323 sizeof (PTR) to convert bytes to elements. Shrink the table
40324 if the current size is excessive for the current number of
40327 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40329 * ira-costs.c (record_reg_classes): Break from the inner loop
40330 early once alt_fail is known to be true. Update outer loop
40331 handling accordingly.
40333 2017-01-13 Jeff Law <law@redhat.com>
40335 * tree-ssa-dse.c (decrement_count): New function.
40336 (increment_start_addr, maybe_trim_memstar_call): Likewise.
40337 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
40338 when we know the partially dead statement is a mem* function.
40340 PR tree-optimization/61912
40341 PR tree-optimization/77485
40342 * tree-ssa-dse.c: Include expr.h.
40343 (maybe_trim_constructor_store): New function.
40344 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
40346 PR tree-optimization/33562
40347 PR tree-optimization/61912
40348 PR tree-optimization/77485
40349 * doc/invoke.texi: Document new dse-max-object-size param.
40350 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
40351 * tree-ssa-dse.c: Include params.h.
40352 (dse_store_status): New enum.
40353 (initialize_ao_ref_for_dse): New, partially extracted from
40355 (valid_ao_ref_for_dse, normalize_ref): New.
40356 (setup_live_bytes_from_ref, compute_trims): Likewise.
40357 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
40358 (maybe_trim_partially_dead_store): Likewise.
40359 (maybe_trim_complex_store): Likewise.
40360 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
40361 Track what bytes live from the original store. Return tri-state
40362 for dead, partially dead or live.
40363 (dse_dom_walker): Add constructor, destructor and new private members.
40364 (delete_dead_call, delete_dead_assignment): New extracted from
40366 (dse_optimize_stmt): Make a member of dse_dom_walker.
40367 Use initialize_ao_ref_for_dse.
40369 PR tree-optimization/33562
40370 PR tree-optimization/61912
40371 PR tree-optimization/77485
40372 * sbitmap.h (bitmap_count_bits): Prototype.
40373 (bitmap_clear_range, bitmap_set_range): Likewise.
40374 * sbitmap.c (bitmap_clear_range): New function.
40375 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
40377 2017-01-13 Martin Liska <mliska@suse.cz>
40380 * function.c (set_cfun): Add new argument force.
40381 * function.h (set_cfun): Likewise.
40382 * ipa-inline-transform.c (inline_call): Use the function when
40383 strict alising from is dropped for function we inline to.
40385 2017-01-13 Richard Biener <rguenther@suse.de>
40387 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
40388 for dumping GIMPLE INTEGER_CSTs.
40390 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
40392 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
40393 to 201112L since C++17.
40395 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
40398 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
40399 if -fsanitize=kernel-address is present.
40401 2017-01-13 Richard Biener <rguenther@suse.de>
40403 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
40404 as _Literal ( type ) number in case usual suffixes do not
40405 preserve all information.
40407 2017-01-13 Richard Biener <rguenther@suse.de>
40409 PR tree-optimization/77283
40410 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
40411 and ssa-iterators.h.
40412 (is_feasible_trace): Implement a cost model based on joiner
40415 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
40418 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
40419 char or short to __float128/_Float128 directly.
40421 2017-01-12 Martin Sebor <msebor@redhat.com>
40423 to -Wformat-overflow.
40424 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
40425 (min_bytes_remaining): Same.
40426 (get_string_length): Same.
40427 (format_string): Same.
40428 (format_directive): Same.
40430 (pass_sprintf_length::handle_gimple_call): Same.
40432 2017-01-12 Jakub Jelinek <jakub@redhat.com>
40434 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
40435 info.nowrite calls with no lhs that can't throw. Return bool
40436 whether gsi_remove has been called or not.
40437 (pass_sprintf_length::handle_gimple_call): Return bool whether
40438 try_substitute_return_value called gsi_remove. Formatting fix.
40439 (pass_sprintf_length::execute): Don't use gsi_remove if
40440 handle_gimple_call returned true.
40443 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
40444 be removed due to side-effects, don't remove following barrier nor
40445 turn the successor edge into fallthru edge.
40447 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40450 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
40451 element-reversing loads and stores as not swappable.
40453 2017-01-12 Nathan Sidwell <nathan@acm.org>
40454 Nicolai Stange <nicstange@gmail.com>
40456 * combine.c (try_combine): Don't ignore result of overlap checking
40457 loop. Combine overlap & asm check into single loop.
40459 2017-01-12 Richard Biener <rguenther@suse.de>
40461 * tree-pretty-print.c (dump_generic_node): Provide -gimple
40462 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
40464 2017-01-12 Richard Biener <rguenther@suse.de>
40466 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
40467 and TS_TARGET_OPTION directly derive from TS_BASE.
40468 * tree-core.h (tree_optimization_option): Derive from tree_base.
40469 (tree_target_option): Likewise.
40471 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
40473 * config/i386/i386.c (memory_address_length): Increase len
40474 only when rip_relative_addr_p returns false.
40476 2017-01-11 Julia Koval <julia.koval@intel.com>
40478 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
40479 (OPTION_MASK_ISA_SGX_SET): New.
40480 (ix86_handle_option): Handle OPT_msgx.
40481 * config.gcc: Added sgxintrin.h.
40482 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
40483 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
40484 * config/i386/i386.c (ix86_target_string): Add -msgx.
40486 (ix86_option_override_internal): Handle new options.
40487 (ix86_valid_target_attribute_inner_p): Add sgx.
40488 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
40489 * config/i386/i386.opt: Add msgx.
40490 * config/i386/sgxintrin.h: New file.
40491 * config/i386/x86intrin.h: Add sgxintrin.h.
40493 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40496 * fold-const.c (maybe_nonzero_address): Return 1 for function
40498 (tree_single_nonzero_warnv_p): Don't handle function local objects
40502 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
40505 2017-01-11 David Malcolm <dmalcolm@redhat.com>
40508 * opts.c: Include "spellcheck.h"
40509 (struct string_fragment): New struct.
40510 (struct edit_distance_traits<const string_fragment &>): New
40512 (get_closest_sanitizer_option): New function.
40513 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
40515 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40517 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
40519 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
40520 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
40521 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
40522 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
40523 for initial die_offset if dwarf_split_debug_info.
40524 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
40525 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
40527 (output_skeleton_debug_sections): Formatting fix. Use
40528 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
40529 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
40531 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
40533 * config/arm/cortex-a53.md: Add bypasses for
40534 cortex_a53_r2f_cvt.
40535 (cortex_a53_r2f): Only use for transfers.
40536 (cortex_a53_f2r): Likewise.
40537 (cortex_a53_r2f_cvt): Add reservation for conversions.
40538 (cortex_a53_f2r_cvt): Likewise.
40540 2017-01-11 Tamar Christina <tamar.christina@arm.com>
40542 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
40543 to all inlined functions, change static to extern.
40545 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
40548 * config/arm/arm.c (legitimize_pic_address): Handle reference to
40550 (arm_assemble_integer): Likewise.
40552 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
40554 * config.gcc: Use new awk script to check CPU, FPU and architecture
40555 parameters for --with-... options.
40556 * config/arm/parsecpu.awk: New file
40557 * config/arm/arm-cpus.in: New file.
40558 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
40560 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
40562 * config/arm/t-arm: Update dependency rules.
40563 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
40564 of processing .def files.
40565 * config/arm/genopt.sh: Deleted.
40566 * config/arm/gentune.sh: Deleted.
40567 * config/arm/arm-cores.def: Deleted.
40568 * config/arm/arm-arches.def: Deleted.
40569 * config/arm/arm-fpus.def: Deleted.
40570 * config/arm/arm-tune.md: Regenerated.
40571 * config/arm/arm-tables.opt: Regenerated.
40572 * config/arm/arm-cpu.h: New generated file.
40573 * config/arm/arm-cpu-data.h: New generated file.
40574 * config/arm/arm-cpu-cdata.h: New generated file.
40576 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40579 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40581 (input_varpool_node): Unpack dynamically_initialized bit.
40583 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40585 PR rtl-optimization/79032
40586 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40587 the alignment of the adjusted memory reference against that of MODE,
40588 instead of the alignment of the original memory reference.
40590 2017-01-11 Martin Jambor <mjambor@suse.cz>
40592 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40594 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40595 decorated functions.
40597 2017-01-11 Richard Biener <rguenther@suse.de>
40599 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40600 set range/nonnull info for PHI results. Do not set it on
40601 stmts marked for removal.
40603 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40605 * expr.c (store_field): In the bitfield case, fetch the return value
40606 from the registers before applying a single big-endian adjustment.
40607 Always do a final load for a BLKmode value not larger than a word.
40609 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40612 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40613 that we correctly handle column numbers greater than
40614 LINE_MAP_MAX_COLUMN_NUMBER.
40616 2017-01-10 Martin Sebor <msebor@redhat.com>
40618 PR middle-end/78245
40619 * gimple-ssa-sprintf.c (get_destination_size): Call
40620 {init,fini}object_sizes.
40621 * tree-object-size.c (addr_object_size): Adjust.
40622 (pass_through_call): Adjust.
40623 (pass_object_sizes::execute): Adjust.
40624 * tree-object-size.h (fini_object_sizes): Declare.
40626 2017-01-10 Martin Sebor <msebor@redhat.com>
40628 PR tree-optimization/78775
40629 * builtins.c (get_size_range): Move...
40630 * calls.c: ...to here.
40631 (alloc_max_size): Accept zero argument.
40632 (operand_signed_p): Remove.
40633 (maybe_warn_alloc_args_overflow): Call get_size_range.
40634 * calls.h (get_size_range): Declare.
40636 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40638 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40639 from TI's devices.csv file as of September 2016.
40640 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40642 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40644 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40645 * doc/invoke.texi: Likewise.
40646 * doc/md.texi: Likewise.
40647 * doc/objc.texi: Likewise.
40649 2017-01-10 Joshua Conner <joshconner@google.com>
40651 * config/arm/fuchsia-elf.h: New file.
40652 * config/fuchsia.h: New file.
40653 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40654 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40656 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40658 2016-01-10 Richard Biener <rguenther@suse.de>
40660 PR tree-optimization/79034
40661 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40662 Propagate out degenerate PHIs in the joiner.
40664 2017-01-10 Martin Liska <mliska@suse.cz>
40666 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40667 (sort_congruence_classes_by_decl_uid): Likewise.
40668 (sort_congruence_class_groups_by_decl_uid): Likewise.
40669 (sem_item_optimizer::merge_classes): Sort class, groups in these
40670 classes and members in the groups by DECL_UID of declarations.
40671 This would make merge operations stable.
40673 2017-01-10 Martin Liska <mliska@suse.cz>
40675 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40676 usage of m_classes_vec.
40677 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40678 (sem_item_optimizer::get_group_by_hash): Likewise.
40679 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40680 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40681 (sem_item_optimizer::verify_classes): Likewise.
40682 (sem_item_optimizer::process_cong_reduction): Likewise.
40683 (sem_item_optimizer::dump_cong_classes): Likewise.
40684 (sem_item_optimizer::merge_classes): Likewise.
40685 * ipa-icf.h (congruence_class_hash): Rename from
40686 congruence_class_group_hash. Remove declaration of m_classes_vec.
40688 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40690 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40691 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40692 * config.gcc: Add avx512vpopcntdqintrin.h.
40693 * config/i386/avx512vpopcntdqintrin.h: New.
40694 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40695 * config/i386/i386-builtin-types.def: Add new types.
40696 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40697 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40698 __builtin_ia32_vpopcountq_v8di_mask): New.
40699 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40700 __AVX512VPOPCNTDQ__.
40701 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40702 (PTA_AVX512VPOPCNTDQ): Define.
40703 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40704 TARGET_AVX512VPOPCNTDQ_P): Define.
40705 * config/i386/i386.opt: Add mavx512vpopcntdq.
40706 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40707 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40709 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40711 PR middle-end/77484
40712 * predict.def (PRED_CALL): Set to 67.
40714 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40716 * expr.c (store_field): In the bitfield case, if the value comes from
40717 a function call and is of an aggregate type returned in registers, do
40718 not modify the field mode; extract the value in all cases if the mode
40719 is BLKmode and the size is not larger than a word.
40721 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40724 * config/i386/cpuid.h: Fix undefined behavior.
40726 2017-01-04 Jeff Law <law@redhat.com>
40728 PR tree-optimization/79007
40729 PR tree-optimization/67955
40730 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40731 conservative for pt.null when flag_non_call_exceptions is on.
40733 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40735 PR translation/79019
40736 PR translation/79020
40737 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40738 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40739 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40741 * config/avr/avr.opt (maccumulate-args): Likewise.
40742 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40743 * common.opt (freport-bug): Likewise.
40744 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40745 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40746 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40747 translatable string.
40748 * config/i386/i386.c (function_value_32): Likewise.
40749 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40750 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40752 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40753 * common/config/msp430/msp430-common.c (msp430_handle_option):
40755 * symtab.c (symtab_node::verify_base): Likewise.
40756 * opts.c (set_debug_level): Likewise.
40757 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40758 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40759 missing whitespace to translatable strings.
40760 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40761 * config/sh/superh.h: Likewise.
40762 * config/i386/xopintrin.h: Likewise.
40763 * config/i386/znver1.md: Likewise.
40764 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40765 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40766 * double-int.h (struct double_int): Likewise.
40767 * double-int.c (div_and_round_double): Likewise.
40768 * wide-int.cc: Likewise.
40769 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40770 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40771 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40772 (crossjumps_occurred): ... this.
40773 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40776 PR tree-optimization/78899
40777 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40778 returning bool return struct loop *, NULL for failure and the new
40780 (versionable_outer_loop_p): Don't version outer loop if it has
40781 dont_vectorized bit set.
40782 (tree_if_conversion): When versioning outer loop, ensure
40783 tree_if_conversion is performed also on the inner loop of the
40784 non-vectorizable outer loop copy.
40785 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40786 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40787 prevent vectorization of it.
40788 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40789 the outer loop vectorization of the non-scalar version is attempted
40790 before vectorization of the inner loop in scalar version. If
40791 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40792 vectorization of its inner loop.
40793 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40794 has 2 inner loops, rename also on edges from bb whose single pred
40795 is outer_loop->header. Fix typo in function comment.
40797 2017-01-09 Martin Sebor <msebor@redhat.com>
40800 * asan.c (asan_emit_stack_protection): Increase local buffer size
40801 to avoid snprintf truncation warning.
40803 2017-01-09 Andrew Pinski <apinski@cavium.com>
40805 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40806 to reference thunderx2t99 for the tuning structure
40807 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40809 (thunderx2t99_extra_costs): This.
40810 * config/aarch64/aarch64-tune.md: Regenerate.
40811 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
40812 (vulcan_addrcost_table): This.
40813 (vulcan_regmove_cost): Rename to ...
40814 (thunderx2t99_regmove_cost): This.
40815 (vulcan_vector_cost): Rename to ...
40816 (thunderx2t99_vector_cost): this.
40817 (vulcan_branch_cost): Rename to ...
40818 (thunderx2t99_branch_cost): This.
40819 (vulcan_tunings): Rename to ...
40820 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
40821 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
40823 2017-01-09 Martin Jambor <mjambor@suse.cz>
40827 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
40828 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
40829 (propagate_vr_accross_jump_function): Use the above function for all
40830 value range computations for pass-through jump functions and type
40831 converasion from explicit value range values.
40832 (ipcp_propagate_stage): Do not attempt to deduce types of formal
40833 parameters from TYPE_ARG_TYPES.
40834 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
40835 (ipa_write_node_info): Stream type of the actual argument.
40836 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
40838 2017-01-09 Martin Liska <mliska@suse.cz>
40841 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
40842 (lookup_compiler): Do not show error message with have_E.
40844 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40846 PR tree-optimization/78938
40847 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
40848 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
40849 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
40850 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
40853 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40855 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
40858 2017-01-09 Richard Biener <rguenther@suse.de>
40860 PR tree-optimization/78997
40861 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
40862 name condition properly.
40864 2017-01-09 Richard Biener <rguenther@suse.de>
40867 * dwarf2out.c (is_cxx): New overload with context.
40868 (is_naming_typedef_decl): Use it.
40870 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40872 * invoke.texi (Option Summary): Correct spacing in option lists
40873 and add line breaks to fix over-long lines.
40875 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40877 PR middle-end/17660
40879 * extend.texi (Common Variable Attributes): Add xref to GCC
40880 Internals manual to explain mode attribute keywords.
40882 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40885 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
40886 and Preprocessor Options.
40887 (Options for Linking): Document -pthread here....
40888 (RS/6000 and PowerPC Options): ...not here.
40889 (Solaris 2 Options): ...or here.
40890 * doc/cppopts.texi: Document -pthread.
40892 2017-01-08 Martin Sebor <msebor@redhat.com>
40894 PR middle-end/77708
40895 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
40896 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
40897 New member functions.
40898 (format_directive): Used them.
40900 (pass_sprintf_length::handle_gimple_call): Same.
40901 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
40902 to avoid truncation for any argument.
40903 (extract_affine_mul): Same.
40904 * tree.c (get_file_function_name): Same.
40906 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40908 PR middle-end/77484
40909 * predict.def (PRED_INDIR_CALL): Set to 86.
40911 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40913 PR preprocessor/54124
40914 * doc/cppopts.texi: Reformat -d subtable to list the full name
40915 of the options. Add cross-reference to the docs for the general
40916 compiler -d options.
40917 * doc/invoke.texi (Developer Options): Add cross-reference to the
40918 preprocessor-specific -d option documentation.
40920 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40922 PR preprocessor/13498
40923 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
40924 redudant material, and reflect new command-line options.
40925 (System Headers): Likewise.
40927 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40929 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
40930 -isystem, and -idirafter. Copy-edit.
40931 * doc/cppopts.texi: Copy-edit. Remove contradiction about
40932 default for -ftrack-macro-expansion. Delete obsolete and
40933 badly-formatted implementation details about -fdebug-cpp output.
40934 * doc/cppwarnopts.texi: Copy-edit.
40936 2017-01-07 David Malcolm <dmalcolm@redhat.com>
40939 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40940 that the transition from a max line width >= 1<<10 to narrower
40941 lines works correctly.
40943 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
40945 * doc/options.texi (PerFunction): New.
40946 * opt-functions.awk (switch_flags): Map both Optimization and
40947 PerFunction to CL_OPTIMIZATION.
40948 * opth-gen.awk: Test for PerFunction flag along with
40950 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
40951 it only when the latter is present. Skip those that don't in
40952 the hash function generator.
40953 * common.opt (fvar-tracking): Mark as PerFunction instead of
40955 (fvar-tracking-assignments): Likewise.
40956 (fvar-tracking-assignments-toggle): Likewise.
40957 (fvar-tracking-uninit): Likewise.
40959 2017-01-07 Jakub Jelinek <jakub@redhat.com>
40961 PR translation/79018
40962 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
40965 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
40968 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
40969 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
40970 TARGET_LONG_JUMP_TABLE_OFFSETS.
40971 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
40972 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
40973 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
40974 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
40975 * config/m68k/m68k.md (tablejump expander): Likewise.
40976 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
40977 TARGET_LONG_JUMP_TABLE_OFFSETS.
40978 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
40979 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
40981 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
40982 David Holsgrove <david.holsgrove@xilinx.com>
40984 * common/config/microblaze/microblaze-common.c
40985 (TARGET_EXCEPT_UNWIND_INFO): Remove.
40986 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
40988 * config/microblaze/microblaze.c (microblaze_must_save_register)
40989 (microblaze_expand_epilogue, microblaze_return_addr): Handle
40991 (microblaze_eh_return): New function.
40992 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
40993 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
40994 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
40995 * config/microblaze/microblaze.md (eh_return): New pattern.
40997 2017-01-06 Jakub Jelinek <jakub@redhat.com>
40999 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41000 GCC_DIAGNOSTIC_STRINGIFY): Define.
41002 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41004 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41006 * config/arm/arm.md (<mcrr>): New.
41008 * config/arm/arm.c (arm_arch5te): New.
41009 (arm_option_override): Set arm_arch5te.
41010 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41012 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41013 (arm_mcrr_qualifiers): ... this. New.
41014 (MRRC_QUALIFIERS): Define to...
41015 (arm_mrrc_qualifiers): ... this. New.
41016 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41018 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41019 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41020 (MRRCI, mrrc, MRRC): New.
41021 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41022 VUNSPEC_MRRC2): New.
41024 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41026 * config/arm/arm.md (<mcr>): New.
41028 * config/arm/arm.c (arm_coproc_builtin_available): Add
41029 support for mcr, mrc, mcr2 and mrc2.
41030 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41031 (arm_mcr_qualifiers): ... this. New.
41032 (MRC_QUALIFIERS): Define to ...
41033 (arm_mrc_qualifiers): ... this. New.
41034 (MCR_QUALIFIERS): Define to ...
41035 (arm_mcr_qualifiers): ... this. New.
41036 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41038 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41039 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41040 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41041 VUNSPEC_MRC2): New.
41043 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41045 * config/arm/arm.md (*ldc): New.
41049 * config/arm/arm.c (arm_coproc_builtin_available): Add
41050 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41051 (arm_coproc_ldc_stc_legitimate_address): New.
41052 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41053 'qualifier_const_pointer'.
41054 (LDC_QUALIFIERS): Define to...
41055 (arm_ldc_qualifiers): ... this. New.
41056 (STC_QUALIFIERS): Define to...
41057 (arm_stc_qualifiers): ... this. New.
41058 * config/arm/arm-protos.h
41059 (arm_coproc_ldc_stc_legitimate_address): New.
41060 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
41061 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
41062 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
41063 stc2, stcl, stc2l): New.
41064 * config/arm/constraints.md (Uz): New.
41065 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
41066 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
41067 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
41068 VUNSPEC_STC2L): New.
41070 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41072 * config/arm/arm.md (<cdp>): New.
41073 * config/arm/arm.c (neon_const_bounds): Rename this ...
41074 (arm_const_bounds): ... this.
41075 (arm_coproc_builtin_available): New.
41076 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
41077 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
41078 (CDP_QUALIFIERS): Define to...
41079 (arm_cdp_qualifiers): ... this. New.
41081 (arm_expand_builtin_args): Add case for 6 arguments.
41082 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
41083 (arm_const_bounds): ... this.
41084 (arm_coproc_builtin_available): New.
41085 * config/arm/arm_acle.h (__arm_cdp): New.
41087 * config/arm/arm_acle_builtins.def (cdp): New.
41089 * config/arm/iterators.md (CDPI,CDP,cdp): New.
41090 * config/arm/neon.md: Rename all 'neon_const_bounds' to
41091 'arm_const_bounds'.
41092 * config/arm/types.md (coproc): New.
41093 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
41094 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
41095 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
41096 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
41098 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41100 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
41101 (UBINOP_QUALIFIERS): New.
41103 (acle_builtin_data): New. Change comment.
41104 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
41105 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
41106 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
41107 arm_acle_builtins.def.
41108 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
41109 (arm_init_acle_builtins): New.
41110 (CRC32_BUILTIN): Remove.
41111 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
41112 crc32cb, crc32ch and crc32cw.
41113 (arm_init_crc32_builtins): Remove.
41114 (arm_init_builtins): Use arm_init_acle_builtins rather
41115 than arm_init_crc32_builtins.
41116 (arm_expand_acle_builtin): New.
41117 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
41118 * config/arm/arm_acle_builtins.def: New.
41120 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41122 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
41123 (arm_builtin_datum): ... this.
41124 (arm_init_neon_builtin): Rename to ...
41125 (arm_init_builtin): ... this. Add a new parameters PREFIX
41126 and USE_SIG_IN_NAME.
41127 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
41128 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
41129 'arm_builtin_datum'.
41130 (arm_init_vfp_builtins): Likewise.
41131 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
41132 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
41133 (arm_expand_neon_args): Rename to ...
41134 (arm_expand_builtin_args): ... this. Rename builtin_arg
41135 enum values and differentiate between ARG_BUILTIN_MEMORY
41136 and ARG_BUILTIN_NEON_MEMORY.
41137 (arm_expand_neon_builtin_1): Rename to ...
41138 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
41139 values, arm_expand_builtin_args and add bool parameter NEON.
41140 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
41141 (arm_expand_vfp_builtin): Likewise.
41142 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
41144 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41146 PR middle-end/77484
41147 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
41148 * predict.c (tree_estimate_probability_bb): Reverse direction of
41149 polymorphic call predictor.
41151 2017-01-06 David Malcolm <dmalcolm@redhat.com>
41153 * passes.c (execute_one_pass): Split out pass-skipping logic into...
41154 (determine_pass_name_match): ...this new function and...
41155 (should_skip_pass_p): ...this new function.
41157 2017-01-06 Nathan Sidwell <nathan@acm.org>
41159 * ipa-visibility.c (function_and_variable_visibility): Reformat
41160 comments and long lines. Remove extrneous if.
41161 * symtab.c (symtab_node::make_decl_local): Fix code format.
41162 (symtab_node::set_section_for_node): Fix comment typo.
41164 2017-01-06 Martin Liska <mliska@suse.cz>
41167 * lra-constraints.c: Rename invariant to lra_invariant.
41168 * predict.c (set_even_probabilities): Initialize e to NULL.
41170 2017-01-05 Martin Sebor <msebor@redhat.com>
41172 PR tree-optimization/78910
41173 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
41174 (format_integer): Correct off-by-one error in the handling
41175 of precision with negative numbers in signed conversions..
41177 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
41179 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
41181 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41183 PR tree-optimization/71016
41184 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
41185 factor_out_conditional_conversion. Formatting fix.
41186 (factor_out_conditional_conversion): Add cond_stmt argument.
41187 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
41188 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
41191 2017-01-05 David Malcolm <dmalcolm@redhat.com>
41193 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
41194 read-rtl-function.o, and selftest-rtl.o.
41195 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
41196 (selftest::aarch64_test_loading_full_dump): New function.
41197 (selftest::aarch64_run_selftests): New function.
41198 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
41199 selftest::aarch64_run_selftests.
41200 * config/i386/i386.c
41201 (selftest::ix86_test_loading_dump_fragment_1): New function.
41202 (selftest::ix86_test_loading_call_insn): New function.
41203 (selftest::ix86_test_loading_full_dump): New function.
41204 (selftest::ix86_test_loading_unspec): New function.
41205 (selftest::ix86_run_selftests): Call the new functions.
41206 * emit-rtl.c (maybe_set_max_label_num): New function.
41207 * emit-rtl.h (maybe_set_max_label_num): New decl.
41208 * function.c (instantiate_decls): Guard call to
41209 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
41210 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
41212 * gensupport.c (gen_reader::gen_reader): Pass "false"
41213 for new "compact" param of rtx_reader.
41214 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
41215 rather than an empty string for NULL strings.
41216 * read-md.c: Potentially include config.h rather than bconfig.h.
41217 Wrap include of errors.h with #ifdef GENERATOR_FILE.
41218 (have_error): New global, copied from errors.c.
41219 (md_reader::read_name): Rename to...
41220 (md_reader::read_name_1): ...this, adding "out_loc" param,
41221 and converting "missing name or number" to returning false, rather
41223 (md_reader::read_name): Reimplement in terms of read_name_1.
41224 (md_reader::read_name_or_nil): New function.
41225 (md_reader::read_string): Handle "(nil)" by returning NULL.
41226 (md_reader::md_reader): Add new param "compact".
41227 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
41228 (md_reader::read_file): New method.
41229 * read-md.h (md_reader::md_reader): Add new param "compact".
41230 (md_reader::read_file): New method.
41231 (md_reader::is_compact): New accessor.
41232 (md_reader::read_name): Convert return type from void to file_location.
41233 (md_reader::read_name_or_nil): New decl.
41234 (md_reader::read_name_1): New decl.
41235 (md_reader::m_compact): New field.
41236 (noop_reader::noop_reader): Pass "false" for new "compact" param
41238 (rtx_reader::rtx_reader): Add new "compact" param.
41239 (rtx_reader::read_rtx_operand): Make virtual and convert return
41240 type from void to rtx.
41241 (rtx_reader::read_until): New decl.
41242 (rtx_reader::handle_any_trailing_information): New virtual function.
41243 (rtx_reader::postprocess): New virtual function.
41244 (rtx_reader::finalize_string): New virtual function.
41245 (rtx_reader::m_in_call_function_usage): New field.
41246 (rtx_reader::m_reuse_rtx_by_id): New field.
41247 * read-rtl-function.c: New file.
41248 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
41249 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
41250 (selftest::verify_three_block_rtl_cfg): New decl.
41251 * read-rtl-function.h: New file.
41252 * read-rtl.c: Potentially include config.h rather than bconfig.h.
41253 For host, include function.h, memmodel.h, and emit-rtl.h.
41254 (one_time_initialization): New function.
41255 (struct compact_insn_name): New struct.
41256 (compact_insn_names): New array.
41257 (find_code): Handle insn codes in compact dumps.
41258 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
41259 (bind_subst_iter_and_attr): Likewise.
41260 (add_condition_to_string): Likewise.
41261 (add_condition_to_rtx): Likewise.
41262 (apply_attribute_uses): Likewise.
41263 (add_current_iterators): Likewise.
41264 (apply_iterators): Likewise.
41265 (initialize_iterators): Guard usage of apply_subst_iterator with
41266 #ifdef GENERATOR_FILE.
41267 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
41268 (md_reader::read_mapping): Likewise.
41269 (add_define_attr_for_define_subst): Likewise.
41270 (add_define_subst_attr): Likewise.
41271 (read_subst_mapping): Likewise.
41272 (check_code_iterator): Likewise.
41273 (rtx_reader::read_rtx): Likewise. Move one-time initialization
41275 (one_time_initialization): New function.
41276 (rtx_reader::read_until): New method.
41277 (read_flags): New function.
41278 (parse_reg_note_name): New function.
41279 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
41280 Handle reuse_rtx ids.
41281 Wrap iterator lookup within #ifdef GENERATOR_FILE.
41282 Add parsing support for RTL dumps, mirroring the special-cases in
41283 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
41284 values, and calling handle_any_trailing_information.
41285 (rtx_reader::read_rtx_operand): Convert return type from void
41286 to rtx, returning return_rtx. Handle case 'e'. Call
41287 finalize_string on XSTR and XTMPL fields.
41288 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
41289 "(nil)" values were omitted. Call the postprocess vfunc on the
41291 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
41292 class ctor. Initialize m_in_call_function_usage. Call
41293 one_time_initialization.
41294 * rtl-tests.c (selftest::test_uncond_jump): Call
41295 set_new_first_and_last_insn.
41296 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
41297 * selftest-rtl.c: New file.
41298 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
41299 (selftest::get_insn_by_uid): New decl.
41300 * selftest-run-tests.c (selftest::run_tests): Call
41301 read_rtl_function_c_tests.
41302 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
41303 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
41306 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
41308 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
41309 operands in a special way. Assert that pos+len <= mode precision.
41311 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41313 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
41314 3 argument Alias with unlimited for the negative form.
41315 (fno-vect-cost-model): Removed.
41317 2017-01-05 Martin Liska <mliska@suse.cz>
41319 * hsa-gen.c (gen_hsa_divmod): New function.
41320 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
41322 2017-01-05 Martin Liska <mliska@suse.cz>
41325 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
41328 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41330 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
41331 small constant length operands.
41333 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41335 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
41336 between loop iterations.
41338 2017-01-05 Martin Liska <mliska@suse.cz>
41341 * gimplify.c (gimplify_decl_expr): Compare to
41342 asan_poisoned_variables instread of checking flags.
41343 (gimplify_target_expr): Likewise.
41344 (gimplify_expr): Likewise.
41345 (gimplify_function_tree): Conditionally initialize
41346 asan_poisoned_variables.
41348 2017-01-04 Jeff Law <law@redhat.com>
41350 PR tree-optimizatin/78812
41351 * rtl.h (contains_mem_rtx_p): Prototype.
41352 * ifcvt.c (containts_mem_rtx_p): Move from here to...
41353 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
41354 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
41355 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
41356 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
41358 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41360 * input.c (assert_char_at_range): Default-initialize actual_range.
41362 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41364 * df-scan.c (df_ref_create_structure): Make regno unsigned,
41365 to match the caller.
41367 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41369 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
41370 insns after final jump in test to emit dummy move.
41372 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41374 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
41375 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
41377 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41379 * multiple_target.c (create_dispatcher_calls): Init e_next.
41380 * tree-ssa-loop-split.c (split_loop): Init border.
41381 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
41384 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
41389 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
41390 (altivec_register_operand): Do not return true if the operand
41391 contains a SUBREG mixing SImode and SFmode.
41392 (vsx_register_operand): Likewise.
41393 (vsx_reg_sfsubreg_ok): New predicate.
41394 (vfloat_operand): Do not return true if the operand contains a
41395 SUBREG mixing SImode and SFmode.
41396 (vint_operand): Likewise.
41397 (vlogical_operand): Likewise.
41398 (gpc_reg_operand): Likewise.
41399 (int_reg_operand): Likewise.
41400 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
41401 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
41402 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
41404 (rs6000_emit_move_si_sf_subreg): New helper function.
41405 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
41406 fixup SUBREGs involving SImode and SFmode.
41407 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
41408 numbers for the new peephole2 optimization.
41409 (peephole2 for SFmode unions): New peephole2 to optimize cases in
41410 the GLIBC math library that do AND/IOR/XOR operations on single
41411 precision floating point.
41412 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
41413 target macros to say whether we need to avoid SUBREGs mixing
41415 (TARGET_ALLOW_SF_SUBREG): Likewise.
41416 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
41417 (UNSPEC_SI_FROM_SF): Likewise.
41418 (iorxor): Change spacing.
41419 (and_ior_xor): New iterator for AND, IOR, and XOR.
41420 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
41421 (movdi_from_sf_zero_ext): Likewise.
41422 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
41423 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
41424 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
41425 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
41426 (fms<mode>4): Likewise.
41427 (fnma<mode>4): Likewise.
41428 (fnms<mode>4): Likewise.
41429 (nfma<mode>4): Likewise.
41430 (nfms<mode>4): Likewise.
41432 2017-01-04 Marek Polacek <polacek@redhat.com>
41435 * doc/invoke.texi: Document -Wpointer-compare.
41437 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41439 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
41442 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
41443 descriptions for -gdwarf-5 and emit them as uleb128 instead of
41446 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
41449 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
41450 documentation of the powerpc_popcntb_ok attribute.
41451 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
41452 code to issue warning messages if a requested CPU configuration is
41453 not supported by the binary (assembler and loader) toolchain.
41454 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
41455 made to define a built-in function that has been disabled.
41456 (paired_init_builtins): Add assertion to prevent ICE if attempt is
41457 made to define a built-in function that has been disabled.
41458 (altivec_init_builtins): Add comment explaining why definition
41459 of the DST built-in functions is not preceded by an assertion
41460 check. Add assertions to prevent ICE if attempts are made to
41461 define an altivec predicate or an abs* built-in function that has
41463 (htm_init_builtins): Add comment explaining why definition of the
41464 htm built-in functions is not preceded by an assertion check.
41466 2017-01-04 Jeff Law <law@redhat.com>
41468 PR tree-optimizatin/67955
41469 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
41470 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
41471 the points-to solution does not include pt_null. Use DECL_PT_UID
41474 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
41476 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
41477 Use gen_int_mode instead of gen_lopwart for const_int operands.
41479 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41481 PR tree-optimization/71563
41482 * match.pd: Simplify X << Y into X if Y is known to be 0 or
41483 out of range value - has low bits known to be zero.
41485 2017-01-04 Alan Modra <amodra@gmail.com>
41487 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
41488 * configure: Regenerate.
41489 * config.in: Regenerate.
41491 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41494 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
41495 a substring of the message, but strcmp with the whole message. Ifdef
41496 ENABLE_NLS, translate the message first using dgettext.
41498 2017-01-03 Jeff Law <law@redhat.com>
41500 PR tree-optimizatin/78856
41501 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
41502 (mark_threaded_blocks): Remove code to truncate thread paths that
41503 cross multiple loop headers. Instead invalidate the cached loop
41504 iteration information and handle case of a thread path walking
41505 into an irreducible region.
41507 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
41510 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
41511 assertions. Add support for doing the signbit if the IEEE 128-bit
41512 floating point value is in a GPR.
41513 * config/rs6000/rs6000.md (Fsignbit): Delete.
41514 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
41515 Update the length attribute if the value is in a GPR.
41516 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
41517 the sign or zero extension instruction, since the value is always 0/1.
41518 (signbit<mode>2_dm2): Delete using <Fsignbit>.
41521 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
41522 extracting SImode to a GPR register so that we can generate a
41523 store, limit the vector to be in a traditional Altivec register
41524 for the vextuwrx instruction.
41526 2017-01-03 Ian Lance Taylor <iant@google.com>
41528 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
41530 2017-01-03 Martin Sebor <msebor@redhat.com>
41532 PR tree-optimization/78696
41533 * gimple-ssa-sprintf.c (format_floating): Correct handling of
41534 precision. Use MPFR for %f for greater fidelity. Correct handling
41536 (pass_sprintf_length::compute_format_length): Set width and precision
41537 specified by asrerisk to void_node for vararg functions.
41538 (try_substitute_return_value): Adjust dump output.
41540 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
41542 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
41544 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
41546 * doc/invoke.texi (SPARC options): Document -mlra as the default.
41547 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
41548 -mlra/-mno-lra was passed to the compiler.
41550 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
41552 PR rtl-optimization/65618
41553 * emit-rtl.c (try_split): Move initialization of "before" and
41554 "after" to just before the call to emit_insn_after_setloc.
41556 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
41558 * doc/md.texi (Standard Names): Remove reference to Java frontend.
41560 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
41562 * dwarf2out.c (gen_enumeration_type_die): When
41563 -gno-strict-dwarf, add a DW_AT_encoding attribute.
41565 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41567 PR tree-optimization/78965
41568 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41569 Change first argument from const call_info & to call_info &. For %n
41570 set info.nowrite to false.
41572 PR middle-end/78901
41573 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41574 possibly throwing calls.
41576 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41577 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41578 and fns handling, rather than in a separate case SSA_NAME.
41580 2017-01-02 Jeff Law <law@redhat.com>
41582 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41583 fixes for first_period and second_period variables.
41585 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41588 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41589 (*insvqi_1): New insn pattern.
41590 (*insvqi_1_mem_rex64): Ditto.
41591 (*insvqi_2): Ditto.
41592 (*insvqi_3): Rename from *insvqi.
41594 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41596 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41598 * doc/cfg.texi (Edges): Remove reference to Java.
41599 (Maintaining the CFG): Ditto.
41601 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41603 PR middle-end/77674
41604 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41605 transparent aliases.
41607 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41609 PR middle-end/77484
41610 * predict.def (PRED_CALL): Update hitrate.
41611 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41612 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41613 into direct/indirect/polymorphic variants.
41615 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41617 Update copyright years.
41619 * gcc.c (process_command): Update copyright notice dates.
41620 * gcov-dump.c (print_version): Ditto.
41621 * gcov.c (print_version): Ditto.
41622 * gcov-tool.c (print_version): Ditto.
41623 * gengtype.c (create_file): Ditto.
41624 * doc/cpp.texi: Bump @copying's copyright year.
41625 * doc/cppinternals.texi: Ditto.
41626 * doc/gcc.texi: Ditto.
41627 * doc/gccint.texi: Ditto.
41628 * doc/gcov.texi: Ditto.
41629 * doc/install.texi: Ditto.
41630 * doc/invoke.texi: Ditto.
41632 Copyright (C) 2017 Free Software Foundation, Inc.
41634 Copying and distribution of this file, with or without modification,
41635 are permitted in any medium without royalty provided the copyright
41636 notice and this notice are preserved.