1 2017-12-27 Tom de Vries <tom@codesourcery.com>
3 * config/nvptx/nvptx.c (nvptx_option_override): Disable
6 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
8 PR rtl-optimization/83513
9 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
10 before priority comparison.
12 2017-12-25 Jakub Jelinek <jakub@redhat.com>
15 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
16 ix86_isa_flags2 to ix86_isa_flags.
17 * config/i386/i386-c.c (ix86_target_macros_internal): Test
18 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
19 isa_flags rather than isa_flags2.
20 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
21 and -mavx512bitalg from isa2_opts to isa_opts.
22 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
23 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
24 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
26 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
27 OPTION_MASK_ISA_AVX512F ored with another option similarly to
28 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
29 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
30 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
31 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
32 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
33 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
34 builtins from bdesc_args2 to bdesc_args section.
35 (__builtin_ia32_compressstoreuqi512_mask,
36 __builtin_ia32_compressstoreuhi512_mask,
37 __builtin_ia32_compressstoreuqi256_mask,
38 __builtin_ia32_expandloadqi512_mask,
39 __builtin_ia32_expandloadqi512_maskz,
40 __builtin_ia32_expandloadhi512_mask,
41 __builtin_ia32_expandloadhi512_maskz,
42 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
43 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
44 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
45 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
46 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
47 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
48 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
49 __builtin_ia32_vpshldv_v32hi_maskz,
50 __builtin_ia32_vpopcountb_v64qi_mask,
51 __builtin_ia32_vpopcountw_v32hi_mask,
52 __builtin_ia32_vpshufbitqmb512_mask,
53 __builtin_ia32_vpshufbitqmb256_mask): Add
54 " | OPTION_MASK_ISA_AVX512BW".
55 (__builtin_ia32_expandloadqi256_mask,
56 __builtin_ia32_expandloadqi256_maskz,
57 __builtin_ia32_vpopcountb_v32qi_mask): Add
58 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
59 (__builtin_ia32_expandloadhi256_mask,
60 __builtin_ia32_expandloadhi256_maskz,
61 __builtin_ia32_expandloadqi128_mask,
62 __builtin_ia32_expandloadqi128_maskz,
63 __builtin_ia32_expandloadhi128_mask,
64 __builtin_ia32_expandloadhi128_maskz,
65 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
66 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
67 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
68 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
69 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
70 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
71 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
72 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
73 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
74 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
75 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
76 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
77 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
78 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
79 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
80 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
81 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
82 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
83 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
84 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
85 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
86 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
87 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
88 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
89 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
90 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
91 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
92 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
93 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
94 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
95 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
96 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
97 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
98 __builtin_ia32_vpopcountw_v8hi_mask): Add
99 " | OPTION_MASK_ISA_AVX512VL".
100 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
101 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
102 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
103 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
104 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
105 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
106 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
107 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
108 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
109 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
110 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
111 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
112 _mm_mask_bitshuffle_epi64_mask): Likewise.
113 * common/config/i386/i386-common.c
114 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
115 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
116 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
117 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
118 OPTION_MASK_ISA_AVX512BITALG_UNSET.
119 (OPTION_MASK_ISA2_AVX512F_UNSET,
120 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
121 (ix86_handle_option): For -mno-general-regs-only, clear from
122 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
123 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
124 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
125 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
126 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
127 for moving from ix86_isa_flags2 to ix86_isa_flags.
129 2017-12-23 Jakub Jelinek <jakub@redhat.com>
132 * fold-const.c (struct contains_label_data): New type.
133 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
134 inside of a SWITCH_BODY seen during the walk.
135 (contains_label_p): Use walk_tree instead of
136 walk_tree_without_duplicates, prepare data for contains_label_1 and
139 2017-12-22 Jakub Jelinek <jakub@redhat.com>
142 * tree-iterator.c (alloc_stmt_list): Start with cleared
143 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
147 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
148 enable avx512f explicitly in #pragma GCC target.
149 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
150 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
151 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
152 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
153 __builtin_ia32_vpdpbusds_v8si_mask,
154 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
155 __builtin_ia32_vpdpbusds_v4si_mask,
156 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
157 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
158 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
159 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
160 __builtin_ia32_vpdpwssds_v8si_mask,
161 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
162 __builtin_ia32_vpdpwssds_v4si_mask,
163 __builtin_ia32_vpdpwssds_v4si_maskz): Use
164 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
165 just OPTION_MASK_ISA_AVX512VNNI.
167 2017-12-22 Martin Jambor <mjambor@suse.cz>
170 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
173 2017-12-22 Julia Koval <julia.koval@intel.com>
174 Sebastian Peryt <sebastian.peryt@intel.com>
176 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
177 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
178 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
179 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
180 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
181 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
182 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
183 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
184 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
185 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
186 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
187 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
188 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
189 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
190 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
191 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
192 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
193 _mm256_maskz_popcnt_epi64): New intrinsics.
194 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
195 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
196 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
197 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
198 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
199 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
200 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
201 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
202 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
203 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
204 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
205 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
206 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
207 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
208 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
209 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
210 __builtin_ia32_vpshufbitqmb256_mask,
211 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
212 * config/i386/i386-c.c (__AVX512BITALG__): New.
213 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
214 (ix86_valid_target_attribute_inner_p): Ditto.
215 (ix86_expand_args_builtin): Handle new types.
216 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
217 * config/i386/i386.opt: Add -mavx512bitalg.
218 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
219 avx512bitalgintrin.h.
220 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
221 (vpopcount<mode><mask_name>): Add more types.
222 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
223 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
225 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
227 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
228 Or in OPTION_MASK_ISA_AVX512F_SET.
229 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
230 OPTION_MASK_ISA_AVX512VNNI_UNSET.
231 (ix86_handle_option): Adjust for
232 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
233 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
235 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
236 isa_flag instead of isa_flag2.
237 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
238 isa_opts2 to isa_opts.
239 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
242 2017-12-22 Mike Stump <mikestump@comcast.net>
243 Eric Botcazou <ebotcazou@adacore.com>
245 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
247 2017-12-21 Jakub Jelinek <jakub@redhat.com>
250 * config/i386/i386.c (ix86_function_arg_boundary): Return
251 PARM_BOUNDARY for TYPE_EMPTY_P types.
254 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
255 if navail is >= dir.len.
257 2017-12-21 Steve Ellcey <sellcey@cavium.com>
259 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
262 2017-12-21 Jakub Jelinek <jakub@redhat.com>
264 PR rtl-optimization/80747
265 PR rtl-optimization/83512
266 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
267 succ edge from ENTRY, copy partition from e->dest to the newly
269 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
270 ENTRY, use BB_PARTITION of its successor block as current_partition.
271 Don't copy partition when splitting succ edge from ENTRY.
273 PR tree-optimization/83523
274 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
275 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
276 (convert_mult_to_fma): Likewise.
278 PR tree-optimization/83521
279 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
280 gimple_build_assign without code on result of
281 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
284 2017-12-21 Andrew Pinski <apinski@cavium.com>
285 Steve Ellcey <sellcey@cavium.com>
287 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
288 multi-arch for ilp32.
290 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
293 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
294 constraints to operand 2.
295 (*ashl<mode>3_mask_1): Ditto.
296 (*<shift_insn><mode>3_mask): Ditto.
297 (*<shift_insn><mode>3_mask_1): Ditto.
298 (*<rotate_insn><mode>3_mask): Ditto.
299 (*<rotate_insn><mode>3_mask_1): Ditto.
301 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
303 * reorg.c (make_return_insns): Reemit each insn with its own location.
305 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
307 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
308 generation for cases where splatting a value is not useful.
309 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
310 across a vec_duplicate and a paradoxical subreg forming a vector
311 mode to a vec_concat.
313 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
315 * combine.c (simplify_set): Do not transform subregs to zero_extends
316 if the destination is not a scalar int mode.
318 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
321 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
322 shared zero if the input has overflowed.
324 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
326 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
327 support for Cortex-A55 and Cortex-A75.
329 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
331 * common/config/arm/arm-common.c (compare_opt_names): Add function
332 comment. Use strcmp instead of manual loop.
334 2017-12-21 Martin Liska <mliska@suse.cz>
336 PR gcov-profile/83509
337 * gcov-dump.c (dump_gcov_file): Do not read info about
338 support_unexecuted_blocks for gcda files.
340 2017-12-21 Jakub Jelinek <jakub@redhat.com>
342 PR rtl-optimization/82973
343 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
344 (valid_for_const_vector_p): ... this.
345 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
346 (valid_for_const_vector_p): ... this. Adjust function comment.
347 (gen_vec_duplicate): Adjust caller.
348 * optabs.c (expand_vector_broadcast): Likewise.
349 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
350 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
352 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
353 instead of GET_CODE == CONST_FIXED.
354 (simplify_subreg): Use CONST_FIXED_P macro instead of
355 GET_CODE == CONST_FIXED.
358 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
359 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
360 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
361 (ix86_option_override_internal): Adjust for
362 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
363 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
364 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
365 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
366 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
367 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
368 def_builtin2 for CET builtins.
369 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
370 up formatting in IX86_BUILTIN_RDPID code.
371 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
372 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
373 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
374 ix86_isa_flags2 to ix86_isa_flags.
375 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
377 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
378 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
379 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
381 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
382 Or in OPTION_MASK_ISA_AVX512F_SET.
383 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
384 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
385 (ix86_handle_option): Adjust for
386 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
387 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
389 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
390 Alan Hayward <alan.hayward@arm.com>
391 David Sherwood <david.sherwood@arm.com>
393 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
394 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
395 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
396 Track polynomial offsets.
398 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
399 Alan Hayward <alan.hayward@arm.com>
400 David Sherwood <david.sherwood@arm.com>
402 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
403 Treat drb->init as a poly_int. Fail if its misalignment wrt
404 vector_alignment isn't known.
406 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
407 Alan Hayward <alan.hayward@arm.com>
408 David Sherwood <david.sherwood@arm.com>
410 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
412 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
413 (vect_loop_versioning): Take the loop versioning threshold as a
415 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
416 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
417 versioning_threshold.
418 (vect_analyze_loop_2): Compute the loop versioning threshold
419 whenever loop versioning is needed, and store it in the new
420 field rather than combining it with the cost model threshold.
421 (vect_transform_loop): Update call to vect_loop_versioning.
422 Try to combine the loop versioning and cost thresholds here.
424 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
425 Alan Hayward <alan.hayward@arm.com>
426 David Sherwood <david.sherwood@arm.com>
428 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
429 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
430 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
431 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
432 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
433 (group_compare_offset): Update accordingly.
434 (split_small_address_groups_p): Likewise.
435 (record_use): Take addr_offset as a poly_uint64 rather than
436 an unsigned HOST_WIDE_INT.
437 (strip_offset): Return the offset as a poly_uint64 rather than
438 an unsigned HOST_WIDE_INT.
439 (record_group_use, split_address_groups): Track polynomial offsets.
440 (add_iv_candidate_for_use): Likewise.
441 (addr_offset_valid_p): Take the offset as a poly_int64 rather
442 than a HOST_WIDE_INT.
443 (strip_offset_1): Return the offset as a poly_int64 rather than
446 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
447 Alan Hayward <alan.hayward@arm.com>
448 David Sherwood <david.sherwood@arm.com>
450 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
451 rather than a HOST_WIDE_INT.
452 * tree.c (get_binfo_at_offset): Likewise.
454 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
455 Alan Hayward <alan.hayward@arm.com>
456 David Sherwood <david.sherwood@arm.com>
458 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
459 rather than a HOST_WIDE_INT.
460 * tree-sra.c (build_ref_for_offset): Likewise.
462 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
463 Alan Hayward <alan.hayward@arm.com>
464 David Sherwood <david.sherwood@arm.com>
466 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
468 * tree.c (mem_ref_offset): Likewise.
469 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
470 * builtins.c (get_object_alignment_2): Likewise.
471 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
472 * gimple-fold.c (get_base_constructor): Likewise.
473 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
474 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
476 * ipa-polymorphic-call.c
477 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
478 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
479 (get_ancestor_addr_info): Likewise.
480 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
481 * match.pd: Likewise.
482 * tree-data-ref.c (dr_analyze_innermost): Likewise.
483 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
484 * tree-eh.c (tree_could_trap_p): Likewise.
485 * tree-object-size.c (addr_object_size): Likewise.
486 * tree-ssa-address.c (copy_ref_info): Likewise.
487 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
488 (indirect_refs_may_alias_p): Likewise.
489 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
490 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
491 (non_rewritable_mem_ref_base): Likewise.
492 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
493 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
494 * varasm.c (decode_addr_const): Likewise.
496 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
497 Alan Hayward <alan.hayward@arm.com>
498 David Sherwood <david.sherwood@arm.com>
500 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
501 offsets for MEM_REFs.
503 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
504 Alan Hayward <alan.hayward@arm.com>
505 David Sherwood <david.sherwood@arm.com>
507 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
508 as a poly_uint64 rather than an unsigned int.
509 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
511 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
512 Alan Hayward <alan.hayward@arm.com>
513 David Sherwood <david.sherwood@arm.com>
515 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
517 (decode_addr_const): Update accordingly.
519 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
520 Alan Hayward <alan.hayward@arm.com>
521 David Sherwood <david.sherwood@arm.com>
523 * tree.h (bit_field_size, bit_field_offset): New functions.
524 * hsa-gen.c (gen_hsa_addr): Use them.
525 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
526 (simplify_vector_constructor): Likewise.
527 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
528 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
529 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
530 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
533 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
534 Alan Hayward <alan.hayward@arm.com>
535 David Sherwood <david.sherwood@arm.com>
537 * expr.h (emit_group_load, emit_group_load_into_temps)
538 (emit_group_store): Take the size as a poly_int64 rather than an int.
539 * expr.c (emit_group_load_1, emit_group_load): Likewise.
540 (emit_group_load_into_temp, emit_group_store): Likewise.
542 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
543 Alan Hayward <alan.hayward@arm.com>
544 David Sherwood <david.sherwood@arm.com>
546 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
547 an unsigned int to a poly_uint64.
548 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
549 sizes as poly_uint64s rather than unsigned ints.
550 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
553 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
554 Alan Hayward <alan.hayward@arm.com>
555 David Sherwood <david.sherwood@arm.com>
557 * lra-constraints.c (emit_inc): Change inc_amount from an int
560 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
561 Alan Hayward <alan.hayward@arm.com>
562 David Sherwood <david.sherwood@arm.com>
564 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
566 (add_stack_var, stack_var_cmp, partition_stack_vars)
567 (dump_stack_var_partition): Update accordingly.
568 (alloc_stack_frame_space): Take the size as a poly_int64 rather
569 than a HOST_WIDE_INT.
570 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
571 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
572 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
573 rather than a HOST_WIDE_INT.
575 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
576 Alan Hayward <alan.hayward@arm.com>
577 David Sherwood <david.sherwood@arm.com>
579 * target.def (return_pops_args): Treat both the input and output
580 sizes as poly_int64s rather than HOST_WIDE_INTS.
581 * targhooks.h (default_return_pops_args): Update accordingly.
582 * targhooks.c (default_return_pops_args): Likewise.
583 * doc/tm.texi: Regenerate.
584 * emit-rtl.h (incoming_args): Change pops_args, size and
585 outgoing_args_size from int to poly_int64_pod.
586 * function.h (expr_status): Change x_pending_stack_adjust and
587 x_stack_pointer_delta from int to poly_int64.
588 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
589 (ARGS_SIZE_RTX): Update accordingly.
590 * calls.c (highest_outgoing_arg_in_use): Change from int to
592 (stack_usage_watermark, stored_args_watermark): New variables.
593 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
594 (emit_call_1): Change the stack_size and rounded_stack_size
595 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
597 (save_fixed_argument_area): Check stack_usage_watermark.
598 (initialize_argument_information): Change old_pending_adj from
599 a HOST_WIDE_INT * to a poly_int64_pod *.
600 (compute_argument_block_size): Return the size as a poly_int64
602 (finalize_must_preallocate): Track polynomial argument sizes.
603 (compute_argument_addresses): Likewise.
604 (internal_arg_pointer_based_exp): Track polynomial offsets.
605 (mem_overlaps_already_clobbered_arg_p): Rename to...
606 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
607 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
608 Check stored_args_used_watermark.
609 (load_register_parameters): Update accordingly.
610 (check_sibcall_argument_overlap_1): Likewise.
611 (combine_pending_stack_adjustment_and_call): Take the unadjusted
612 args size as a poly_int64 rather than an int. Return a bool
613 indicating whether the optimization was possible and return
614 the new adjustment by reference.
615 (check_sibcall_argument_overlap): Track polynomail argument sizes.
616 Update stored_args_watermark.
617 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
618 (expand_call): Likewise. Maintain stack_usage_watermark and
619 stored_args_watermark. Update calls to
620 combine_pending_stack_adjustment_and_call.
621 (emit_library_call_value_1): Handle polynomial argument sizes.
622 Call stack_region_maybe_used_p and mark_stack_region_used.
623 Maintain stack_usage_watermark.
624 (store_one_arg): Likewise. Update call to
625 mem_overlaps_already_clobbered_arg_p.
626 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
628 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
629 * config/microblaze/microblaze.c (microblaze_function_prologue):
631 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
632 TARGET_RETURN_POPS_ARGS interface.
633 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
635 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
636 * config/i386/i386.c (ix86_return_pops_args): Update for new
637 TARGET_RETURN_POPS_ARGS interface.
638 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
639 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
640 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
641 TARGET_RETURN_POPS_ARGS interface.
642 * config/vax/vax.c (vax_return_pops_args): Likewise.
643 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
644 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
645 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
646 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
647 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
648 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
649 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
650 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
651 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
652 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
653 and x_stack_pointer_delta from int to poly_int64.
654 * dojump.c (do_pending_stack_adjust): Update accordingly.
655 * explow.c (allocate_dynamic_stack_space): Handle polynomial
656 stack_pointer_deltas.
657 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
658 (pad_to_arg_alignment): Track polynomial offsets.
659 (assign_parm_find_stack_rtl): Likewise.
660 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
661 * toplev.c (output_stack_usage): Update reference to
662 current_function_pushed_stack_size.
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 * function.c (in_arg_offset, var_offset, dynamic_offset)
669 (out_arg_offset, cfa_offset): Change from int to poly_int64.
670 (instantiate_new_reg): Return the new offset as a poly_int64_pod
671 rather than a HOST_WIDE_INT.
672 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
673 (instantiate_virtual_regs_in_insn): Likewise.
675 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
676 Alan Hayward <alan.hayward@arm.com>
677 David Sherwood <david.sherwood@arm.com>
679 * rtl.h (get_args_size, add_args_size_note): New functions.
680 (find_args_size_adjust): Return a poly_int64 rather than a
682 (fixup_args_size_notes): Likewise. Make the same change to the
683 end_args_size parameter.
684 * rtlanal.c (get_args_size, add_args_size_note): New functions.
685 * builtins.c (expand_builtin_trap): Use add_args_size_note.
686 * calls.c (emit_call_1): Likewise.
687 * explow.c (adjust_stack_1): Likewise.
688 * cfgcleanup.c (old_insns_match_p): Update use of
689 find_args_size_adjust.
690 * combine.c (distribute_notes): Track polynomial arg sizes.
691 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
692 end_true_args_size, beg_delay_args_size and end_delay_args_size
693 from HOST_WIDE_INT to poly_int64.
694 (add_cfi_args_size): Take the args_size as a poly_int64 rather
695 than a HOST_WIDE_INT.
696 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
697 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
698 polynomial arg sizes.
699 * emit-rtl.c (try_split): Use get_args_size.
700 * recog.c (peep2_attempt): Likewise.
701 * reload1.c (reload_as_needed): Likewise.
702 * expr.c (find_args_size_adjust): Return the adjustment as a
703 poly_int64 rather than a HOST_WIDE_INT.
704 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
705 to a poly_int64 and change the return type in the same way.
706 (emit_single_push_insn): Track polynomial arg sizes.
708 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
709 Alan Hayward <alan.hayward@arm.com>
710 David Sherwood <david.sherwood@arm.com>
712 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
713 from HOST_WIDE_INT to poly_int64.
714 * expr.c (push_block, emit_push_insn): Likewise.
716 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
717 Alan Hayward <alan.hayward@arm.com>
718 David Sherwood <david.sherwood@arm.com>
720 * function.h (frame_space): Change start and length from HOST_WIDE_INT
722 (get_frame_size): Return the size as a poly_int64 rather than a
724 (frame_offset_overflow): Take the offset as a poly_int64 rather
725 than a HOST_WIDE_INT.
726 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
727 (assign_stack_temp): Likewise for the size.
728 * function.c (get_frame_size): Return a poly_int64 rather than
730 (frame_offset_overflow): Take the offset as a poly_int64 rather
731 than a HOST_WIDE_INT.
732 (try_fit_stack_local): Take the start, length and size as poly_int64s
733 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
734 rather than a HOST_WIDE_INT.
735 (add_frame_space): Take the start and end as poly_int64s rather than
737 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
738 (assign_stack_temp): Likewise for the size.
739 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
741 (find_temp_slot_from_address): Handle polynomial offsets.
742 (combine_temp_slots): Likewise.
743 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
745 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
746 poly_int64 rather than a HOST_WIDE_INT.
747 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
748 than a HOST_WIDE_INT.
749 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
750 polynomial frame offsets.
751 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
752 as a poly_int64 rather than an int.
753 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
754 * config/v850/v850-protos.h (compute_frame_size): Likewise.
755 * config/v850/v850.c (compute_frame_size): Likewise.
756 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
757 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
758 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
759 * config/pa/pa.c (pa_compute_frame_size): Likewise.
760 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
761 rather than a HOST_WIDE_INT.
762 * explow.c (get_dynamic_stack_base): Likewise.
763 * final.c (final_start_function): Use the constant lower bound
764 of the frame size for -Wframe-larger-than.
765 * ira.c (do_reload): Adjust for new get_frame_size return type.
766 * lra.c (lra): Likewise.
767 * reload1.c (reload): Likewise.
768 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
769 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
770 * rtlanal.c (get_initial_register_offset): Return the offset as
771 a poly_int64 rather than a HOST_WIDE_INT.
773 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
774 Alan Hayward <alan.hayward@arm.com>
775 David Sherwood <david.sherwood@arm.com>
777 * reload1.c (elim_table): Change initial_offset, offset and
778 previous_offset from HOST_WIDE_INT to poly_int64_pod.
779 (offsets_at): Change the target array's element type from
780 HOST_WIDE_INT to poly_int64_pod.
781 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
782 (elimination_costs_in_insn, update_eliminable_offsets)
783 (verify_initial_elim_offsets, set_offsets_for_label)
784 (init_eliminable_invariants): Update after above changes.
786 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
787 Alan Hayward <alan.hayward@arm.com>
788 David Sherwood <david.sherwood@arm.com>
790 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
791 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
792 (decomposition): Change start and end from HOST_WIDE_INT
794 (decompose, immune_p): Update accordingly.
795 (find_inc_amount): Return a poly_int64 rather than an int.
796 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
799 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
800 Alan Hayward <alan.hayward@arm.com>
801 David Sherwood <david.sherwood@arm.com>
803 * tree.h (get_inner_reference): Return the bitsize and bitpos
804 as poly_int64_pods rather than HOST_WIDE_INT.
805 * fold-const.h (ptr_difference_const): Return the pointer difference
806 as a poly_int64_pod rather than a HOST_WIDE_INT.
807 * expr.c (get_inner_reference): Return the bitsize and bitpos
808 as poly_int64_pods rather than HOST_WIDE_INT.
809 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
811 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
812 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
813 (optimize_bit_field_compare): Update call to get_inner_reference.
814 (decode_field_reference): Likewise.
815 (fold_unary_loc): Track polynomial offsets and sizes.
816 (split_address_to_core_and_offset): Return the bitpos as a
817 poly_int64_pod rather than a HOST_WIDE_INT.
818 (ptr_difference_const): Likewise for the pointer difference.
819 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
820 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
821 * dbxout.c (dbxout_expand_expr): Likewise.
822 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
823 (loc_list_from_tree_1, fortran_common): Likewise.
824 * gimple-laddress.c (pass_laddress::execute): Likewise.
825 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
826 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
827 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
828 * tree-affine.c (tree_to_aff_combination): Likewise.
829 (get_inner_reference_aff): Likewise.
830 * tree-data-ref.c (split_constant_offset_1): Likewise.
831 (dr_analyze_innermost): Likewise.
832 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
833 * tree-sra.c (ipa_sra_check_caller): Likewise.
834 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
835 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
836 (instrument_bool_enum_load, instrument_object_size): Likewise.
837 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
838 to get_inner_reference.
839 * hsa-gen.c (gen_hsa_addr): Likewise.
840 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
841 * tsan.c (instrument_expr): Likewise.
842 * match.pd: Update call to ptr_difference_const.
844 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
845 Alan Hayward <alan.hayward@arm.com>
846 David Sherwood <david.sherwood@arm.com>
848 * fold-const.c (fold_comparison): Track sizes and offsets as
849 poly_int64s rather than HOST_WIDE_INTs when folding address
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 * expr.h (get_bit_range): Return the bitstart and bitend as
857 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
858 as a poly_int64 rather than a HOST_WIDE_INT.
859 * expr.c (get_bit_range): Likewise.
860 (expand_assignment): Update call accordingly.
861 * fold-const.c (optimize_bit_field_compare): Likewise.
863 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
864 Alan Hayward <alan.hayward@arm.com>
865 David Sherwood <david.sherwood@arm.com>
867 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
868 (aarch64_legitimate_address_p): Use it instead of an rtx code,
869 as an optional final parameter.
870 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
871 (aarch64_legitimate_address_p): Likewise.
872 (aarch64_print_address_internal): Take an aarch64_addr_query_type
873 instead of an rtx code.
874 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
875 (aarch64_legitimate_address_hook_p): Likewise.
876 (aarch64_print_ldpstp_address): Likewise.
877 (aarch64_print_operand_address): Likewise.
878 (aarch64_address_cost): Likewise.
879 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
880 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
881 (aarch64_mem_pair_lanes_operand): Likewise.
883 2017-12-20 Richard Biener <rguenther@suse.de>
885 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
886 update_stmt_if_modified.
888 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
890 PR tree-optimization/83491
891 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
892 before walking uses. Improve coding style and comments.
894 2017-12-20 Tom de Vries <tom@codesourcery.com>
896 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
898 2017-12-20 Jakub Jelinek <jakub@redhat.com>
901 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
903 (pass_ipa_free_fn_summary): Add small_p private data member,
904 initialize to false in the ctor.
905 (pass_ipa_free_fn_summary::clone,
906 pass_ipa_free_fn_summary::set_pass_param,
907 pass_ipa_free_fn_summary::gate): New methods.
908 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
909 | TODO_dump_symtab if small_p.
910 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
911 entry and add another instance of the pass with false parm after
913 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
914 ipa_free_fn_summary here.
916 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
918 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
920 2017-12-20 Martin Sebor <msebor@redhat.com>
923 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
924 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
925 (expand_builtin_strncmp): Same.
927 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
930 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
931 allowed debug stmts before labels.
932 (expand_gimple_basic_block): Likewise.
933 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
934 * gimple-iterator.h (gsi_after_labels): Likewise.
935 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
936 rename reused variable, and simplify using gsi_move_before.
937 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
938 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
939 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
940 (gimple_verify_flow_info, gimple_block_label): Likewise.
941 (make_blocks): Move debug markers after adjacent labels.
942 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
943 allowed debug insns outside blocks.
944 * df-scan.c (df_insn_delete): Likewise.
945 * lra-constraints.c (update_ebb_live_info): Likewise.
946 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
947 (vt_initialize, delete_vta_debug_insns): Likewise.
948 (reemit_marker_as_note): Drop BB parm. Adjust callers.
950 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
951 Alan Hayward <alan.hayward@arm.com>
952 David Sherwood <david.sherwood@arm.com>
954 * poly-int-types.h (round_down_to_byte_boundary): New macro.
955 (round_up_to_byte_boundary): Likewise.
956 * expr.h (get_bit_range): Add temporary shim.
957 * gimple-ssa-store-merging.c (store_operand_info): Change the
958 bitsize, bitpos, bitregion_start and bitregion_end fields from
959 unsigned HOST_WIDE_INT to poly_uint64.
960 (merged_store_group): Likewise load_align_base.
961 (compatible_load_p, compatible_load_p): Update accordingly.
962 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
963 (split_group, imm_store_chain_info::output_merged_store): Likewise.
964 (mem_valid_for_store_merging): Return the bitsize, bitpos,
965 bitregion_start and bitregion_end as poly_uint64s rather than
966 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
967 (handled_load): Take the bitsize, bitpos,
968 bitregion_start and bitregion_end as poly_uint64s rather than
969 unsigned HOST_WIDE_INTs.
970 (pass_store_merging::process_store): Update call to
971 mem_valid_for_store_merging.
973 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
974 Alan Hayward <alan.hayward@arm.com>
975 David Sherwood <david.sherwood@arm.com>
977 * builtins.c (get_object_alignment_2): Track polynomial offsets
978 and sizes. Update the alignment handling.
980 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
981 Alan Hayward <alan.hayward@arm.com>
982 David Sherwood <david.sherwood@arm.com>
984 * tree.h (get_inner_reference): Add a version that returns the
985 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
986 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
987 the case in which bitpos is not associated with the first byte.
989 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
990 Alan Hayward <alan.hayward@arm.com>
991 David Sherwood <david.sherwood@arm.com>
993 * tree-affine.h (get_inner_reference_aff): Return the size as a
995 * tree-affine.c (get_inner_reference_aff): Likewise.
996 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
997 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
999 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1000 Alan Hayward <alan.hayward@arm.com>
1001 David Sherwood <david.sherwood@arm.com>
1003 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1004 HOST_WIDE_INT rather than a poly_int64.
1006 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1007 Alan Hayward <alan.hayward@arm.com>
1008 David Sherwood <david.sherwood@arm.com>
1010 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1011 HOST_WIDE_INT to poly_int64_pod.
1012 (perform_symbolic_merge): Update accordingly.
1013 (bswap_replace): Likewise.
1015 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1016 Alan Hayward <alan.hayward@arm.com>
1017 David Sherwood <david.sherwood@arm.com>
1019 * tree-affine.h (aff_tree::offset): Change from widest_int
1021 (wide_int_ext_for_comb): Delete.
1022 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1023 constants as poly_widest_int rather than widest_int.
1024 (aff_combination_constant_multiple_p): Return the multiplier
1025 as a poly_widest_int.
1026 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1028 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1029 version static and add an overload for poly_widest_int.
1030 (aff_combination_const, aff_combination_add_cst)
1031 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1032 the constants as poly_widest_int rather than widest_int.
1033 (tree_to_aff_combination): Generalize INTEGER_CST case to
1035 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1036 (aff_combination_add_product, aff_combination_mult): Handle
1038 (aff_combination_constant_multiple_p): Return the multiplier
1039 as a poly_widest_int.
1040 * tree-predcom.c (determine_offset): Return the offset as a
1042 (split_data_refs_to_components, suitable_component_p): Update
1044 (valid_initializer_p): Update call to
1045 aff_combination_constant_multiple_p.
1046 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1047 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1048 as a poly_int64 rather than a HOST_WIDE_INT.
1049 (get_address_cost): Handle polynomial offsets.
1050 (iv_elimination_compare_lt): Likewise.
1051 (rewrite_use_nonlinear_expr): Likewise.
1053 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1054 Alan Hayward <alan.hayward@arm.com>
1055 David Sherwood <david.sherwood@arm.com>
1057 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1058 as a poly_int64_pod rather than a HOST_WIDE_INT.
1059 (get_addr_base_and_unit_offset): Likewise.
1060 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1061 (get_addr_base_and_unit_offset): Likewise.
1062 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1063 to poly_int64 in example.
1064 * fold-const.c (fold_binary_loc): Update call to
1065 get_addr_base_and_unit_offset.
1066 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1067 (maybe_canonicalize_mem_ref_addr): Likewise.
1068 (gimple_fold_stmt_to_constant_1): Likewise.
1069 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1071 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
1072 * match.pd: Likewise.
1073 * omp-low.c (lower_omp_target): Likewise.
1074 * tree-sra.c (build_ref_for_offset): Likewise.
1075 (build_debug_ref_for_model): Likewise.
1076 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
1077 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
1078 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
1079 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1080 (constant_pointer_difference): Likewise.
1081 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
1082 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
1083 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
1084 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1085 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
1086 (set_ssa_val_to): Likewise.
1087 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
1088 (maybe_diag_stxncpy_trunc): Likewise.
1089 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1090 * tree.c (build_simple_mem_ref_loc): Likewise.
1091 (array_at_struct_end_p): Likewise.
1093 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1094 Alan Hayward <alan.hayward@arm.com>
1095 David Sherwood <david.sherwood@arm.com>
1097 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
1098 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1099 (get_ref_base_and_extent_hwi): Declare.
1100 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
1101 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1102 (get_ref_base_and_extent_hwi): New function.
1103 * cfgexpand.c (expand_debug_expr): Update call to
1104 get_ref_base_and_extent.
1105 * dwarf2out.c (add_var_loc_to_decl): Likewise.
1106 * gimple-fold.c (get_base_constructor): Return the offset as a
1107 poly_int64_pod rather than a HOST_WIDE_INT.
1108 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
1109 * ipa-polymorphic-call.c
1110 (ipa_polymorphic_call_context::set_by_invariant)
1111 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
1112 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
1113 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
1114 rather than get_ref_base_and_extent.
1115 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
1116 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
1117 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
1119 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
1120 call to get_ref_base_and_extent.
1121 * tree-sra.c (create_access, get_access_for_expr): Likewise.
1122 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
1123 (stmt_kills_ref_p): Likewise.
1124 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1125 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
1127 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
1128 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
1129 when calling native_encode_expr.
1130 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
1131 call to get_ref_base_and_extent.
1132 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
1133 get_ref_base_and_extent.
1134 * var-tracking.c (track_expr_p): Likewise.
1136 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1137 Alan Hayward <alan.hayward@arm.com>
1138 David Sherwood <david.sherwood@arm.com>
1140 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
1141 HOST_WIDE_INT to poly_int64_pod.
1142 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
1143 polynomail parameter offsets.
1145 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1146 Alan Hayward <alan.hayward@arm.com>
1147 David Sherwood <david.sherwood@arm.com>
1149 * gengtype.c (main): Handle poly_int64_pod.
1150 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
1151 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
1152 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
1153 from HOST_WIDE_INT to poly_int64_pod.
1154 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
1155 (copy_cfa): New function.
1156 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
1157 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
1158 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
1159 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
1160 (initial_return_save): Treat offsets as poly_ints.
1161 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
1162 offset is nonconstant.
1163 (reg_save): Take the offset as a poly_int64. Fall back to
1164 DW_CFA_expression for nonconstant offsets.
1165 (queue_reg_save): Take the offset as a poly_int64.
1166 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
1168 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1169 Alan Hayward <alan.hayward@arm.com>
1170 David Sherwood <david.sherwood@arm.com>
1172 * rtl.h (operand_subword, operand_subword_force): Take the offset
1173 as a poly_uint64 an unsigned int.
1174 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
1176 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1177 Alan Hayward <alan.hayward@arm.com>
1178 David Sherwood <david.sherwood@arm.com>
1180 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
1181 'p' format code. Use INT_LIST rather than SUBREG as the example of
1182 a code with an XINT and an XEXP. Remove the implication that
1183 accessing an rtx field using XINT is expected to work.
1184 * rtl.def (SUBREG): Change format from "ei" to "ep".
1185 * rtl.h (rtunion::rt_subreg): New field.
1186 (XCSUBREG): New macro.
1187 (SUBREG_BYTE): Use it.
1188 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
1189 Update constructor accordingly.
1190 (subreg_shape::operator ==): Update accordingly.
1191 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
1192 than an unsigned int.
1193 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
1194 a poly_uint64 rather than an unsigned int.
1195 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
1196 than an unsigned int.
1197 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
1198 (subreg_size_highpart_offset): Return a poly_uint64 rather than
1199 an unsigned int. Take the sizes as poly_uint64s.
1200 (subreg_offset_from_lsb): Return a poly_uint64 rather than
1201 an unsigned int. Take the shift as a poly_uint64 rather than
1203 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
1204 as a poly_uint64 rather than an unsigned int.
1205 (simplify_subreg_regno): Likewise.
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 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
1211 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
1212 poly_uint64 rather than an unsigned int.
1213 * rtl.c (rtx_format): Describe 'p' in comment.
1214 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
1215 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
1216 offset as a poly_uint64 rather than an unsigned int.
1217 (byte_lowpart_offset): Return the memory offset as a poly_int64
1219 (subreg_memory_offset): Likewise. Take the subreg offset as a
1220 poly_uint64 rather than an unsigned int.
1221 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
1222 mode sizes as poly_uint64s rather than unsigned ints. Return a
1223 poly_uint64 rather than an unsigned int.
1224 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
1225 (copy_insn_1): Handle 'p'.
1226 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
1227 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
1228 an unsigned int. Return the shift in the same way.
1229 (subreg_lsb): Return the shift as a poly_uint64 rather than an
1231 (subreg_size_offset_from_lsb): Take the sizes and shift as
1232 poly_uint64s rather than unsigned ints. Return the offset as
1234 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
1235 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
1237 * rtlhash.c (add_rtx): Handle 'p'.
1238 * genemit.c (gen_exp): Likewise.
1239 * gengenrtl.c (type_from_format, gendef): Likewise.
1240 * gensupport.c (subst_pattern_match, get_alternatives_number)
1241 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
1242 (subst_dup): Likewise.
1243 * gengtype.c (adjust_field_rtx_def): Likewise.
1244 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
1245 (match_pattern_2): Likewise.
1246 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
1247 (rtx_test::subreg_field): New function.
1248 (operator ==, safe_to_hoist_p, transition_parameter_type)
1249 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
1250 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
1251 * genpeep.c (match_rtx): Likewise.
1252 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
1253 (rtx_writer::print_rtx_operand): Handle 'p'.
1254 (print_value): Handle SUBREG.
1255 * read-rtl.c (apply_int_iterator): Likewise.
1256 (rtx_reader::read_rtx_operand): Handle 'p'.
1257 * alias.c (rtx_equal_for_memref_p): Likewise.
1258 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
1259 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
1261 * calls.c (expand_call): Likewise.
1262 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
1263 (make_extraction, gen_lowpart_for_combine): Likewise.
1264 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
1266 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
1267 rather than an unsigned int. Treat subreg offsets as poly_ints.
1268 (exp_equiv_p): Handle 'p'.
1269 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
1270 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
1271 * dse.c (find_shift_sequence): Likewise.
1272 * dwarf2out.c (rtl_for_decl_location): Likewise.
1273 * expmed.c (extract_low_bits): Likewise.
1274 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
1275 (expand_expr_real_2): Likewise.
1276 * final.c (alter_subreg): Likewise.
1277 (leaf_renumber_regs_insn): Handle 'p'.
1278 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
1279 Treat subreg offsets as poly_ints.
1280 * fwprop.c (forward_propagate_and_simplify): Likewise.
1281 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
1282 * ira.c (get_subreg_tracking_sizes): Likewise.
1283 * ira-conflicts.c (go_through_subreg): Likewise.
1284 * ira-lives.c (process_single_reg_class_operands): Likewise.
1285 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
1286 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
1287 as a poly_uint64 rather than an unsigned int.
1288 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
1289 subreg offsets as poly_ints.
1290 * lra-constraints.c (operands_match_p): Handle 'p'.
1291 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
1292 * lra-spills.c (assign_mem_slot): Likewise.
1293 * postreload.c (move2add_valid_value_p): Likewise.
1294 * recog.c (general_operand, indirect_operand): Likewise.
1295 * regcprop.c (copy_value, maybe_mode_change): Likewise.
1296 (copyprop_hardreg_forward_1): Likewise.
1297 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
1298 (record_subregs_of_mode): Likewise.
1299 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
1300 * reload.c (operands_match_p): Handle 'p'.
1301 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
1302 * reload1.c (alter_reg, choose_reload_regs): Likewise.
1303 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
1304 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
1305 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
1306 (simplify_const_poly_int_tests<N>::run): Likewise.
1307 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
1308 a poly_uint64 rather than an unsigned int.
1309 * valtrack.c (debug_lowpart_subreg): Likewise.
1310 * var-tracking.c (var_lowpart): Likewise.
1311 (loc_cmp): Handle 'p'.
1313 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1314 Alan Hayward <alan.hayward@arm.com>
1315 David Sherwood <david.sherwood@arm.com>
1317 * ira.c (get_subreg_tracking_sizes): New function.
1318 (init_live_subregs): Take an integer size rather than a register.
1319 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
1320 to init_live_subregs.
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 * expr.c (store_constructor_field): Change bitsize from a
1327 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
1328 HOST_WIDE_INT to a poly_int64.
1329 (store_constructor): Change size from a HOST_WIDE_INT to
1331 (store_field): Likewise bitsize and bitpos.
1333 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1334 Alan Hayward <alan.hayward@arm.com>
1335 David Sherwood <david.sherwood@arm.com>
1337 * expmed.h (store_bit_field): Change bitregion_start and
1338 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
1339 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
1340 (store_bit_field_1, store_integral_bit_field, store_bit_field)
1341 (store_fixed_bit_field, store_split_bit_field): Likewise.
1342 * expr.c (store_constructor_field, store_field): Likewise.
1343 (optimize_bitfield_assignment_op): Likewise. Make the same change
1344 to bitsize and bitpos.
1345 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
1346 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
1347 same change in the constructor arguments.
1348 (get_best_mode): Change bitregion_start and bitregion_end from
1349 unsigned HOST_WIDE_INT to poly_uint64.
1350 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1351 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
1353 (bit_field_mode_iterator::next_mode): Update for new types
1354 of m_bitregion_start and m_bitregion_end.
1355 (get_best_mode): Change bitregion_start and bitregion_end from
1356 unsigned HOST_WIDE_INT to poly_uint64.
1358 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1359 Alan Hayward <alan.hayward@arm.com>
1360 David Sherwood <david.sherwood@arm.com>
1362 * rtl.h (simplify_gen_subreg): Add a temporary overload that
1363 accepts poly_uint64 offsets.
1364 * expmed.h (extract_bit_field): Take bitsize and bitnum as
1365 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1366 * expmed.c (lowpart_bit_field_p): Likewise.
1367 (extract_bit_field_as_subreg): New function, split out from...
1368 (extract_bit_field_1): ...here. Take bitsize and bitnum as
1369 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
1370 extractions, check that BITSIZE matches the size of the extracted
1371 value and that BITNUM is an exact multiple of that size.
1372 If all else fails, try forcing the value into memory if
1373 BITNUM is variable, and adjusting the address so that the
1374 offset is constant. Split the part that can only handle constant
1375 bitsize and bitnum out into...
1376 (extract_integral_bit_field): ...this new function.
1377 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
1378 rather than unsigned HOST_WIDE_INTs.
1380 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1381 Alan Hayward <alan.hayward@arm.com>
1382 David Sherwood <david.sherwood@arm.com>
1384 * expmed.h (store_bit_field): Take bitsize and bitnum as
1385 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1386 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
1387 that returns the byte size.
1388 (store_bit_field_1): Take bitsize and bitnum as
1389 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
1390 to simple_mem_bitfield_p. Split the part that can only handle
1391 constant bitsize and bitnum out into...
1392 (store_integral_bit_field): ...this new function.
1393 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
1394 than unsigned HOST_WIDE_INTs.
1395 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
1397 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1398 Alan Hayward <alan.hayward@arm.com>
1399 David Sherwood <david.sherwood@arm.com>
1401 * lra-int.h (lra_reg): Change offset from int to poly_int64.
1402 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
1404 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1405 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
1406 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
1407 offset as a poly_int64 rather than an int.
1408 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
1409 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1410 * lra-constraints.c (equiv_address_substitution): Track offsets
1412 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
1413 (curr_insn_transform): Handle the new form of sp_offset.
1414 * lra-eliminations.c (lra_elim_table): Change previous_offset
1415 and offset from HOST_WIDE_INT to poly_int64.
1416 (print_elim_table, update_reg_eliminate): Update accordingly.
1417 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
1418 (get_elimination): Update accordingly.
1419 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
1420 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1421 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
1422 poly_int64 offsets generally.
1423 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
1424 (mark_not_eliminable, init_elimination): Update accordingly.
1425 (remove_reg_equal_offset_note): Return a bool and pass the new
1426 offset back by pointer as a poly_int64.
1427 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
1428 rather than a HOST_WIDE_INT.
1429 (do_remat): Track offsets poly_int64s.
1430 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
1432 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1433 Alan Hayward <alan.hayward@arm.com>
1434 David Sherwood <david.sherwood@arm.com>
1436 * rtl.h (mem_attrs): Add a default constructor. Change size and
1437 offset from HOST_WIDE_INT to poly_int64.
1438 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
1439 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
1440 (widen_memory_access): Take the sizes and offsets as poly_int64s
1441 rather than HOST_WIDE_INTs.
1442 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
1443 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
1445 (adjust_offset_for_component_ref): Change the offset from a
1446 HOST_WIDE_INT to a poly_int64.
1447 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
1448 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
1449 * dce.c (find_call_stack_args): Likewise.
1450 * dse.c (record_store): Likewise.
1451 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
1452 * print-rtl.c (rtx_writer::print_rtx): Likewise.
1453 * read-rtl-function.c (test_loading_mem): Likewise.
1454 * rtlanal.c (may_trap_p_1): Likewise.
1455 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1456 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
1457 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
1458 (mem_attrs::mem_attrs): New function.
1459 (set_mem_attributes_minus_bitpos): Change bitpos from a
1460 HOST_WIDE_INT to poly_int64.
1461 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
1462 (clear_mem_offset, clear_mem_size, change_address)
1463 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
1464 initialize mem_attrs.
1465 (set_mem_offset, set_mem_size, adjust_address_1)
1466 (adjust_automodify_address_1, offset_address, widen_memory_access):
1467 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
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 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
1474 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
1475 rather than 0 to represent an unknown size. Assert that the size
1476 is known when the mode isn't BLKmode.
1477 (may_trap_p_1): Use -1 for unknown sizes.
1478 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
1480 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1481 Alan Hayward <alan.hayward@arm.com>
1482 David Sherwood <david.sherwood@arm.com>
1484 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
1485 to poly_int64. Update commentary for positions_needed.large.
1486 (read_info_type): Change offset and width from HOST_WIDE_INT
1488 (set_usage_bits): Likewise.
1489 (canon_address): Return the offset as a poly_int64 rather than
1490 a HOST_WIDE_INT. Use strip_offset_and_add.
1491 (set_all_positions_unneeded, any_positions_needed_p): Use
1492 positions_needed.large to track stores with non-constant widths.
1493 (all_positions_needed_p): Likewise. Take the offset and width
1494 as poly_int64s rather than ints. Assert that rhs is nonnull.
1495 (record_store): Cope with non-constant offsets and widths.
1496 Nullify the rhs of an earlier store if we can't tell which bytes
1498 (find_shift_sequence): Take the access_size and shift as poly_int64s
1500 (get_stored_val): Take the read_offset and read_width as poly_int64s
1501 rather than HOST_WIDE_INTs.
1502 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
1503 non-constant offsets and widths.
1505 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1506 Alan Hayward <alan.hayward@arm.com>
1507 David Sherwood <david.sherwood@arm.com>
1509 * inchash.h (inchash::hash::add_poly_int): New function.
1510 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1511 Use poly_int64 rather than HOST_WIDE_INT.
1512 (ao_ref::max_size_known_p): New function.
1513 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1514 rather than HOST_WIDE_INT.
1515 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
1516 to temporaries until its interface is adjusted to match.
1517 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
1518 (aliasing_component_refs_p, decl_refs_may_alias_p)
1519 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
1520 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
1521 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
1523 * alias.c (ao_ref_from_mem): Likewise.
1524 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1525 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
1526 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
1527 (maybe_trim_complex_store, maybe_trim_constructor_store)
1528 (live_bytes_read, dse_classify_store): Likewise.
1529 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
1530 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
1531 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
1532 (vn_reference_lookup_3): Likewise.
1533 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
1535 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1536 Alan Hayward <alan.hayward@arm.com>
1537 David Sherwood <david.sherwood@arm.com>
1539 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
1540 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
1541 instead of ranges_overlap_p.
1543 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1544 Alan Hayward <alan.hayward@arm.com>
1545 David Sherwood <david.sherwood@arm.com>
1547 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
1548 sizes as poly_int64s rather than HOST_WIDE_INTs.
1550 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1551 Alan Hayward <alan.hayward@arm.com>
1552 David Sherwood <david.sherwood@arm.com>
1554 * gimple-fold.h (fold_ctor_reference): Take the offset and size
1555 as poly_uint64 rather than unsigned HOST_WIDE_INT.
1556 * gimple-fold.c (fold_ctor_reference): Likewise.
1558 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1559 Alan Hayward <alan.hayward@arm.com>
1560 David Sherwood <david.sherwood@arm.com>
1562 * target.def (dwarf_poly_indeterminate_value): New hook.
1563 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
1564 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
1565 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
1566 * doc/tm.texi: Regenerate.
1567 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
1568 offset as a poly_int64.
1569 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
1570 offset as a poly_int64.
1571 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
1572 Take the offset as a poly_int64.
1573 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1574 (frame_pointer_fb_offset): Change to a poly_int64.
1575 (int_loc_descriptor): Take the offset as a poly_int64. Use
1576 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1577 (based_loc_descr): Take the offset as a poly_int64.
1578 Use strip_offset_and_add to handle (plus X (const)).
1579 Use new_reg_loc_descr instead of an open-coded version of the
1580 previous implementation.
1581 (mem_loc_descriptor): Handle CONST_POLY_INT.
1582 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1583 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1585 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1586 Alan Hayward <alan.hayward@arm.com>
1587 David Sherwood <david.sherwood@arm.com>
1589 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1590 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1591 * inchash.h (inchash::hash::add_poly_hwi): New function.
1592 * gengtype.c (main): Register poly_int64.
1593 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1594 offset as a poly_int.
1595 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1596 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1597 offset as a poly_int64.
1598 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1599 * print-rtl.c (print_poly_int): New function.
1600 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1602 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1604 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1605 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1606 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1607 Enforce track_offset_p for parts of a PARALLEL.
1608 (vt_add_function_parameter): Use const_offset for the final
1609 offset to track. Use get_tracked_reg_offset for the parts
1612 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1613 Alan Hayward <alan.hayward@arm.com>
1614 David Sherwood <david.sherwood@arm.com>
1616 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1617 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1618 * doc/tm.texi: Regenerate.
1619 * hooks.h (hook_bool_uint_uint_true): Delete.
1620 (hook_bool_puint64_puint64_true): Declare.
1621 * hooks.c (hook_bool_uint_uint_true): Delete.
1622 (hook_bool_puint64_puint64_true): New function.
1623 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1624 instead of unsigned ints.
1625 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1626 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1628 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1629 Alan Hayward <alan.hayward@arm.com>
1630 David Sherwood <david.sherwood@arm.com>
1632 * optabs.h (expand_operand): Add an int_value field.
1633 (create_expand_operand): Add an int_value parameter and use it
1634 to initialize the new expand_operand field.
1635 (create_integer_operand): Replace with a declaration of a function
1636 that accepts poly_int64s. Move the implementation to...
1637 * optabs.c (create_integer_operand): ...here.
1638 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1639 the mode preserves the value of int_value, instead of calling
1640 const_int_operand on the rtx. Use gen_int_mode to generate
1643 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1644 Alan Hayward <alan.hayward@arm.com>
1645 David Sherwood <david.sherwood@arm.com>
1647 * dumpfile.h (dump_dec): Declare.
1648 * dumpfile.c (dump_dec): New function.
1649 * pretty-print.h (pp_wide_integer): Turn into a function and
1650 declare a poly_int version.
1651 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1653 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1654 Alan Hayward <alan.hayward@arm.com>
1655 David Sherwood <david.sherwood@arm.com>
1657 * doc/generic.texi (POLY_INT_CST): Document.
1658 * tree.def (POLY_INT_CST): New tree code.
1659 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1660 * tree-core.h (tree_poly_int_cst): New struct.
1661 (tree_node): Add a poly_int_cst field.
1662 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1663 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1664 instead of a wide_int_ref.
1665 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1667 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1668 instead of an unsigned HOST_WIDE_INT.
1669 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1670 (ptrdiff_tree_p): Declare.
1671 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1672 extern inline implementations if the target doesn't use POLY_INT_CST.
1673 (poly_int_tree_p): New function.
1674 (wi::unextended_tree): New class.
1675 (wi::int_traits <unextended_tree>): New override.
1676 (wi::extended_tree): Add a default constructor.
1677 (wi::extended_tree::get_tree): New function.
1678 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1679 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1680 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1681 (wi::tree_to_poly_wide_ref): New typedefs.
1682 (wi::ints_for): Provide overloads for extended_tree and
1684 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1685 (wi::to_wide): New functions.
1686 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1687 * tree.c (poly_int_cst_hasher): New struct.
1688 (poly_int_cst_hash_table): New variable.
1689 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1690 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1692 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1693 (init_ttree): Initialize poly_int_cst_hash_table.
1694 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1695 a poly_int64 instead of a HOST_WIDE_INT.
1696 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1697 instead of an unsigned HOST_WIDE_INT.
1698 (wide_int_to_tree): Rename to...
1699 (wide_int_to_tree_1): ...this.
1700 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1701 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1702 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1703 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1704 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1705 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1707 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1708 (streamer_read_tree_body): Likewise.
1709 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1710 (streamer_write_tree_body): Likewise.
1711 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1712 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1713 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1714 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1715 * gimple-expr.h (is_gimple_constant): Likewise.
1716 * gimplify.c (maybe_with_size_expr): Likewise.
1717 * print-tree.c (print_node): Likewise.
1718 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1719 * tree-pretty-print.c (dump_generic_node): Likewise.
1720 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1721 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1722 * tree-vrp.c (compare_values_warnv): Likewise.
1723 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1724 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1725 (force_expr_to_var_cost): Likewise.
1726 * tree-ssa-loop.c (for_each_index): Likewise.
1727 * fold-const.h (build_invariant_address, size_int_kind): Take a
1728 poly_int64 instead of a HOST_WIDE_INT.
1729 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1730 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1732 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1733 (int_const_binop_2): New function, split out from...
1734 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1735 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1736 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1737 * gimple-ssa-strength-reduction.c (slsr_process_add)
1738 (slsr_process_mul): Check for INTEGER_CSTs before using them
1740 * stor-layout.c (bits_from_bytes): New function.
1741 (bit_from_pos): Use it.
1742 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1743 by BITS_PER_UNIT to get the TYPE_SIZE.
1744 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1745 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1747 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1748 Alan Hayward <alan.hayward@arm.com>
1749 David Sherwood <david.sherwood@arm.com>
1751 * doc/rtl.texi (const_poly_int): Document. Also document the
1752 rtl sharing behavior.
1753 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1754 * rtl.h (const_poly_int_def): New struct.
1755 (rtx_def::u): Add a cpi field.
1756 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1757 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1758 (wi::rtx_to_poly_wide_ref): New typedef
1759 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1760 (poly_int_rtx_p): New functions.
1761 (trunc_int_for_mode): Declare a poly_int64 version.
1762 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1763 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1765 (strip_offset): Declare.
1766 (strip_offset_and_add): New function.
1767 * rtl.def (CONST_POLY_INT): New rtx code.
1768 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1769 (shared_const_p): Use poly_int_rtx_p.
1770 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1772 (gen_int_shift_amount): Likewise.
1773 * emit-rtl.c (const_poly_int_hasher): New class.
1774 (const_poly_int_htab): New variable.
1775 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1776 (const_poly_int_hasher::hash): New function.
1777 (const_poly_int_hasher::equal): Likewise.
1778 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1779 (immed_wide_int_const): Rename to...
1780 (immed_wide_int_const_1): ...this and make static.
1781 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1782 instead of a wide_int_ref.
1783 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1784 (gen_lowpart_common): Handle CONST_POLY_INT.
1785 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1786 * cselib.c (cselib_hash_rtx): Likewise.
1787 * dwarf2out.c (const_ok_for_output_1): Likewise.
1788 * expr.c (convert_modes): Likewise.
1789 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1790 * rtlhash.c (add_rtx): Likewise.
1791 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1792 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1793 Handle existing CONST_POLY_INT rtxes.
1794 * expmed.h (expand_shift): Take a poly_int64 instead of a
1796 * expmed.c (expand_shift): Likewise.
1797 * rtlanal.c (strip_offset): New function.
1798 (commutative_operand_precedence): Give CONST_POLY_INT the same
1799 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1801 * rtl-tests.c (const_poly_int_tests): New struct.
1802 (rtl_tests_c_tests): Use it.
1803 * simplify-rtx.c (simplify_const_unary_operation): Handle
1805 (simplify_const_binary_operation): Likewise.
1806 (simplify_binary_operation_1): Fold additions of symbolic constants
1807 and CONST_POLY_INTs.
1808 (simplify_subreg): Handle extensions and truncations of
1810 (simplify_const_poly_int_tests): New struct.
1811 (simplify_rtx_c_tests): Use it.
1812 * wide-int.h (storage_ref): Add default constructor.
1813 (wide_int_ref_storage): Likewise.
1814 (trailing_wide_ints): Use GTY((user)).
1815 (trailing_wide_ints::operator[]): Add a const version.
1816 (trailing_wide_ints::get_precision): New function.
1817 (trailing_wide_ints::extra_size): Likewise.
1819 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1820 Alan Hayward <alan.hayward@arm.com>
1821 David Sherwood <david.sherwood@arm.com>
1823 * emit-rtl.h (gen_int_shift_amount): Declare.
1824 * emit-rtl.c (gen_int_shift_amount): New function.
1825 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
1827 * calls.c (shift_return_value): Likewise.
1828 * cse.c (fold_rtx): Likewise.
1829 * dse.c (find_shift_sequence): Likewise.
1830 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
1831 (expand_shift, expand_smod_pow2): Likewise.
1832 * lower-subreg.c (shift_cost): Likewise.
1833 * optabs.c (expand_superword_shift, expand_doubleword_mult)
1834 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
1835 (expand_vec_perm_var): Likewise.
1836 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1837 (simplify_binary_operation_1): Likewise.
1838 * combine.c (try_combine, find_split_point, force_int_to_mode)
1839 (simplify_shift_const_1, simplify_shift_const): Likewise.
1840 (change_zero_ext): Likewise. Use simplify_gen_binary.
1842 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1844 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
1846 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1848 * doc/invoke.texi (ARM Options): Document accepted extension options
1849 for -march=armv8.3-a.
1851 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
1854 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
1855 or --with-float=softfp, set the default CPU to arm10e.
1857 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
1859 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
1860 * config/visium/predicates.md (const_shift_operand): Likewise.
1861 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
1862 (visium_legitimize_reload_address): Likewise.
1864 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1867 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
1869 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1872 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
1874 2017-12-20 Julia Koval <julia.koval@intel.com>
1876 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
1877 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
1878 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
1879 * config.gcc: Include vpclmulqdqintrin.h.
1880 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
1881 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
1883 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
1884 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
1885 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
1886 * config/i386/i386.c (isa2_opts): Add -mcx16.
1887 (isa_opts): Add -mpclmulqdq, remove -mcx16.
1888 (ix86_option_override_internal): Move mcx16 to flags2.
1889 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
1890 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
1891 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
1892 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
1893 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
1894 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
1895 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
1896 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
1897 * doc/invoke.texi: Add -mvpclmulqdq.
1899 2017-12-20 Tom de Vries <tom@codesourcery.com>
1902 * config/i386/i386.c (ix86_static_chain): Move
1903 DECL_STATIC_CHAIN test ...
1904 * calls.c (rtx_for_static_chain): ... here. New function.
1905 * calls.h (rtx_for_static_chain): Declare.
1906 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
1907 instead of targetm.calls.static_chain.
1908 * df-scan.c (df_get_entry_block_def_set): Same.
1910 2017-12-19 Tom de Vries <tom@codesourcery.com>
1912 PR tree-optimization/83493
1913 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
1916 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1918 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
1919 inform with hardcoded english plural handling.
1921 2017-12-18 Jeff Law <law@redhat.com>
1923 PR tree-optimization/83477
1924 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
1925 a non-virtual PHI, always push a new range.
1927 2017-12-19 Martin Sebor <msebor@redhat.com>
1930 * builtins.c (compute_objsize): Handle non-constant offsets.
1932 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1934 PR tree-optimization/83444
1935 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
1936 character load case, if get_stridx on MEM_REF's operand doesn't
1937 look usable, retry with get_addr_stridx.
1939 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
1942 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
1943 (variable_tracking_main_1): Keep markers even when VTA fails.
1946 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
1947 even if there are markers before it.
1948 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
1950 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1952 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
1953 typical order conditions.
1954 * sel-sched.c (extract_new_fences_from): Likewise.
1955 * config/visium/constraints.md (J, K, L): Likewise.
1956 * config/visium/predicates.md (const_shift_operand): Likewise.
1957 * config/visium/visium.c (visium_legitimize_address,
1958 visium_legitimize_reload_address): Likewise.
1959 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
1960 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
1961 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
1962 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
1963 SET_DUMP_DETAIL): Likewise.
1964 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
1965 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
1966 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
1967 avr_set_core_architecture, avr_set_current_function,
1968 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
1969 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
1970 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
1971 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
1972 avr_map_decompose, avr_fold_builtin): Likewise.
1973 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
1974 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
1975 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
1976 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
1977 * config/m32c/m32c.c (m32c_conditional_register_usage,
1978 m32c_address_cost): Likewise.
1979 * config/m32c/predicates.md (shiftcount_operand,
1980 longshiftcount_operand): Likewise.
1981 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
1982 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
1983 can_use_cdx_ldstw): Likewise.
1984 * config/nios2/nios2.h (CDX_REG_P): Likewise.
1985 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
1987 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
1988 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
1989 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
1990 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
1991 * config/vax/vax.c (adjacent_operands_p): Likewise.
1992 * config/ft32/constraints.md (L, b, KA): Likewise.
1993 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
1995 * cfgexpand.c (expand_stack_alignment): Likewise.
1996 * gcse.c (insert_expr_in_table): Likewise.
1997 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
1998 * cgraphunit.c (cgraph_node::expand): Likewise.
1999 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2000 * emit-rtl.c (add_insn): Likewise.
2001 * input.c (dump_location_info): Likewise.
2002 * passes.c (NEXT_PASS): Likewise.
2003 * read-rtl-function.c (parse_note_insn_name,
2004 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2006 * sched-rgn.c (sched_rgn_init): Likewise.
2007 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2008 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2009 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2011 * explow.c (eliminate_constant_term): Likewise.
2012 * final.c (leaf_renumber_regs_insn): Likewise.
2013 * cfgrtl.c (print_rtl_with_bb): Likewise.
2014 * genhooks.c (emit_init_macros): Likewise.
2015 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2016 * tree-data-ref.c (conflict_fn): Likewise.
2017 * selftest.c (assert_streq): Likewise.
2018 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2019 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2020 fold_binary_loc, multiple_of_p): Likewise.
2021 * reload.c (push_reload, find_equiv_reg): Likewise.
2022 * et-forest.c (et_nca, et_below): Likewise.
2023 * dbxout.c (dbxout_symbol_location): Likewise.
2024 * reorg.c (relax_delay_slots): Likewise.
2025 * dojump.c (do_compare_rtx_and_jump): Likewise.
2026 * gengtype-parse.c (type): Likewise.
2027 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2028 simplify_const_relational_operation): Likewise.
2029 * reload1.c (do_output_reload): Likewise.
2030 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2031 * gengtype.c (type_for_name): Likewise.
2032 * gimple-ssa-sprintf.c (format_directive): Likewise.
2034 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2037 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2038 accessing it. Adjust comment.
2040 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2043 * predict.c (zero_one_minusone): New function.
2044 (apply_return_prediction): Avoid return prediction for functions
2045 returning only -1, 0 and 1 values, unless they only return -1 and 0
2048 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2050 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2051 fall-through warning.
2052 (arc_compute_frame_size): Remove unused variables.
2053 (arc_print_operand): Fix fprintif format.
2054 (arc_can_follow_jump): Clean fall-through warning.
2056 2017-12-19 Marek Polacek <polacek@redhat.com>
2059 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2062 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2064 * config/arc/arc.c (overriderregs): New variable.
2065 (arc_override_options): Track fixed/call saved/call options.
2066 (arc_conditional_register_usage): Check against overrideregs
2067 variable whenever we change register properties.
2069 2017-12-19 Nathan Sidwell <nathan@acm.org>
2071 * opts.c (finish_options): Don't prefix dump_base_name if it
2072 already contains directories.
2074 2017-12-19 Martin Liska <mliska@suse.cz>
2076 PR rtl-optimization/82675
2077 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
2078 more element in sbitmap.
2080 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
2082 * gimplify.c (gimplify_expr): Use error_operand_p.
2084 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
2087 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
2090 2017-12-19 Daniel Cederman <cederman@gaisler.com>
2092 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
2093 the jump is to a label.
2095 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2097 PR tree-optimization/83444
2098 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
2103 * ipa-inline.c (flatten_remove_node_hook): New function.
2104 (ipa_inline): Keep only nodes with flatten attribute at the end of
2105 the array in the order from ipa_reverse_postorder, only walk that
2106 portion of array for flattening, if there is more than one such
2107 node, temporarily register a removal hook and ignore removed nodes.
2109 PR tree-optimization/80631
2110 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
2111 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
2112 IFN_REDUC_MAX or IFN_REDUC_MIN.
2114 2017-12-18 Jeff Law <law@redhat.com>
2116 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
2117 record symbolic equivalences from backedges in the CFG.
2120 2017-11-19 Jeff Law <law@redhat.com>
2122 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2123 of degenerates resulting from ignoring an edge.
2125 2017-12-18 Martin Sebor <msebor@redhat.com>
2128 PR tree-optimization/78450
2129 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
2130 (handle_builtin_strlen): Call it.
2132 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
2134 PR rtl-optimization/83424
2135 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
2137 2017-12-18 Marek Polacek <polacek@redhat.com>
2140 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
2141 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
2144 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2146 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
2147 (umaddsidi4, umaddsidi_split): Likewise.
2149 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2151 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
2154 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2157 * config/s390/s390.c (s390_option_override): Avoid strncpy.
2159 2017-12-18 Richard Biener <rguenther@suse.de>
2161 PR tree-optimization/81877
2162 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
2163 (outermost_indep_loop): Adjust.
2164 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
2165 (can_sm_ref_p): Adjust.
2167 2017-12-18 Richard Biener <rguenther@suse.de>
2170 * tree.c (array_at_struct_end_p): Return true if the underlying
2171 object has space for at least one element in excess of what
2172 the array domain specifies.
2174 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2176 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
2177 example to avoid overfull hbox.
2178 * doc/invoke.texi (Option Summary): Add missing @gol.
2179 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
2182 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2183 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2185 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
2186 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
2187 (AArch64 Options): Document -mverbose-cost-dump.
2188 (ARM Options): Likewise, plus -mflip-thumb.
2190 2017-12-17 Martin Sebor <msebor@redhat.com>
2193 * gimple-ssa-warn-restrict.c
2194 (builtin_memref::offset_out_of_bounds): Correct the handling of
2197 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2199 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
2201 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
2203 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
2205 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
2207 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
2210 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
2212 * doc/invoke.texi: Fix some typos.
2214 2017-12-16 Martin Sebor <msebor@redhat.com>
2216 PR tree-optimization/78918
2217 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
2218 * builtins.c (check_sizes): Rename...
2219 (check_access): ...to this. Rename function arguments for clarity.
2220 (check_memop_sizes): Adjust names.
2221 (expand_builtin_memchr, expand_builtin_memcpy): Same.
2222 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
2223 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
2224 (check_strncat_sizes, expand_builtin_strncat): Same.
2225 (expand_builtin_strncpy, expand_builtin_memset): Same.
2226 (expand_builtin_bzero, expand_builtin_memcmp): Same.
2227 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
2228 (maybe_emit_sprintf_chk_warning): Same.
2229 (expand_builtin_strcpy): Adjust.
2230 (expand_builtin_stpcpy): Same.
2231 (expand_builtin_with_bounds): Detect out-of-bounds accesses
2232 in pointer-checking forms of memcpy, memmove, and mempcpy.
2233 (gcall_to_tree_minimal, max_object_size): Define new functions.
2234 * builtins.h (max_object_size): Declare.
2235 * calls.c (alloc_max_size): Call max_object_size instead of
2236 hardcoding ssizetype limit.
2237 (get_size_range): Handle new argument.
2238 * calls.h (get_size_range): Add a new argument.
2239 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
2240 * doc/invoke.texi (-Wrestrict): Adjust, add example.
2241 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
2243 (gimple_fold_builtin_memory_chk): Same.
2244 (gimple_fold_builtin_stxcpy_chk): New function.
2245 * gimple-ssa-warn-restrict.c: New source.
2246 * gimple-ssa-warn-restrict.h: New header.
2247 * gimple.c (gimple_build_call_from_tree): Propagate location.
2248 * passes.def (pass_warn_restrict): Add new pass.
2249 * tree-pass.h (make_pass_warn_restrict): Declare.
2250 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
2252 (handle_builtin_strcat): Same.
2253 (strlen_optimize_stmt): Rename...
2254 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
2255 stpncpy, strncpy, and their checking forms.
2257 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
2259 PR rtl-optimization/82849
2260 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
2261 and get_max_loop_iterations_int.
2263 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2264 Alan Hayward <alan.hayward@arm.com>
2265 David Sherwood <david.sherwood@arm.com>
2267 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
2268 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
2269 size as a poly_uint64.
2270 (mode_for_vector, mode_for_int_vector): Take the number of vector
2271 elements as a poly_uint64.
2272 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
2273 size as a poly_uint64.
2274 (mode_for_vector, mode_for_int_vector): Take the number of vector
2275 elements as a poly_uint64.
2277 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2278 Alan Hayward <alan.hayward@arm.com>
2279 David Sherwood <david.sherwood@arm.com>
2281 * machmode.h (MACRO_MODE): New macro.
2282 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
2283 * targhooks.c (default_libcall_value, default_secondary_reload)
2284 (default_memory_move_cost, default_register_move_cost)
2285 (default_class_max_nregs): Likewise.
2287 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2288 Alan Hayward <alan.hayward@arm.com>
2289 David Sherwood <david.sherwood@arm.com>
2291 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
2292 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
2294 * genautomata.c (main): Likewise.
2295 * genconditions.c (write_header): Likewise.
2296 * genemit.c (main): Likewise.
2297 * genextract.c (print_header): Likewise.
2298 * genopinit.c (main): Likewise.
2299 * genoutput.c (output_prologue): Likewise.
2300 * genpeep.c (main): Likewise.
2301 * genpreds.c (write_insn_preds_c): Likewise.
2302 * genrecog.c (writer_header): Likewise.
2303 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
2304 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
2305 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
2306 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
2307 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
2308 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
2309 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
2310 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
2311 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
2312 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
2313 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
2314 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
2315 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
2316 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
2317 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
2318 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
2319 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
2320 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
2321 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
2322 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
2323 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
2324 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
2325 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
2326 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
2327 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
2328 * config/darwin.c (IN_TARGET_CODE): Likewise.
2329 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
2330 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
2331 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
2332 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
2333 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
2334 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
2335 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
2336 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
2337 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
2338 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
2339 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
2340 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
2341 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
2342 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
2343 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
2344 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
2345 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
2346 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
2347 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
2348 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
2349 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
2350 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
2351 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
2352 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
2353 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
2354 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
2355 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
2356 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
2357 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
2358 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
2359 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
2360 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
2361 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
2362 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
2363 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
2364 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
2365 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
2366 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
2367 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
2368 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
2369 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
2370 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
2371 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
2372 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
2373 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
2374 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
2375 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
2376 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
2377 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
2378 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
2379 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
2380 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
2381 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
2382 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
2383 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
2384 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
2385 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
2386 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
2387 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
2388 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
2389 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2390 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
2391 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
2392 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
2393 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
2394 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
2395 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
2396 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
2397 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
2398 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
2399 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
2400 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2401 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
2402 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
2403 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
2404 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
2405 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
2406 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
2407 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
2408 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
2409 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
2410 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
2411 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
2412 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
2413 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
2414 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
2415 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
2416 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
2417 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
2418 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
2419 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
2420 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
2421 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
2422 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
2423 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
2424 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
2425 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
2426 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
2427 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
2428 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
2429 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
2430 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
2431 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
2432 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
2433 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
2434 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
2436 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2437 Alan Hayward <alan.hayward@arm.com>
2438 David Sherwood <david.sherwood@arm.com>
2440 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
2443 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2444 Alan Hayward <alan.hayward@arm.com>
2445 David Sherwood <david.sherwood@arm.com>
2447 * doc/generic.texi (VEC_SERIES_EXPR): Document.
2448 * doc/md.texi (vec_series@var{m}): Document.
2449 * tree.def (VEC_SERIES_EXPR): New tree code.
2450 * tree.h (build_vec_series): Declare.
2451 * tree.c (build_vec_series): New function.
2452 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
2453 * tree-pretty-print.c (dump_generic_node): Likewise.
2454 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
2455 * tree-inline.c (estimate_operator_cost): Likewise.
2456 * expr.c (expand_expr_real_2): Likewise.
2457 * optabs-tree.c (optab_for_tree_code): Likewise.
2458 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
2459 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
2460 * expmed.c (make_tree): Handle VEC_SERIES.
2461 * optabs.def (vec_series_optab): New optab.
2462 * optabs.h (expand_vec_series_expr): Declare.
2463 * optabs.c (expand_vec_series_expr): New function.
2464 * tree-vect-generic.c (expand_vector_operations_1): Check that
2465 the operands also have vector type.
2467 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2468 Alan Hayward <alan.hawyard@arm.com>
2469 David Sherwood <david.sherwood@arm.com>
2471 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
2472 (VEC_COND_EXPR): Add missing @tindex.
2473 * doc/md.texi (vec_duplicate@var{m}): Document.
2474 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
2475 * tree.c (build_vector_from_val): Add stubbed-out handling of
2476 variable-length vectors, using VEC_DUPLICATE_EXPR.
2477 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
2478 * cfgexpand.c (expand_debug_expr): Likewise.
2479 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2480 * tree-inline.c (estimate_operator_cost): Likewise.
2481 * tree-pretty-print.c (dump_generic_node): Likewise.
2482 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
2483 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
2484 (test_vec_duplicate_folding): New function.
2485 (fold_const_c_tests): Call it.
2486 * optabs.def (vec_duplicate_optab): New optab.
2487 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
2488 * optabs.h (expand_vector_broadcast): Declare.
2489 * optabs.c (expand_vector_broadcast): Make non-static. Try using
2490 vec_duplicate_optab.
2491 * expr.c (store_constructor): Try using vec_duplicate_optab for
2493 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
2495 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
2498 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
2499 div/mod latencies a bit.
2501 2017-12-15 Jeff Law <law@redhat.com>
2503 PR tree-optimization/36550
2504 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
2505 (mark_threaded_blocks): Rewrite code to avoid block copying when
2506 optimizing for size. Don't pessimize blocks which will be
2507 copied, but all the statements will be dead.
2509 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
2511 PR tree-optimization/81165
2512 * tree-ssa-threadupdate.c (uses_in_bb): New.
2513 (estimate_threading_killed_stmts): New.
2514 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
2515 * tree-ssa-threadedge.c
2516 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
2519 2017-12-15 Jeff Law <law@redhat.com>
2521 PR tree-optimization/83410
2522 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
2523 threads when parallelizing loops.
2525 2017-12-15 Jakub Jelinek <jakub@redhat.com>
2527 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
2529 * config/alpha/alpha.c (vms_attribute_table): Swap
2530 affects_type_identity and handler fields, adjust comments.
2531 * config/mips/mips.c (mips_attribute_table): Likewise.
2532 * config/visium/visium.c (visium_attribute_table): Likewise.
2533 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
2534 * config/microblaze/microblaze.c (microblaze_attribute_table):
2536 * config/spu/spu.c (spu_attribute_table): Likewise.
2537 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
2538 * config/arc/arc.c (arc_attribute_table): Likewise.
2539 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
2540 * config/v850/v850.c (v850_handle_interrupt_attribute,
2541 v850_handle_data_area_attribute): Formatting fixes.
2542 (v850_attribute_table): Swap affects_type_identity and handler
2543 fields, adjust comments.
2544 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
2545 * config/arm/arm.c (arm_attribute_table): Likewise.
2546 * config/avr/avr.c (avr_attribute_table): Likewise.
2547 * config/s390/s390.c (s390_attribute_table): Likewise.
2548 * config/sh/sh.c (sh_attribute_table): Likewise.
2549 * config/i386/i386.c (ix86_handle_cconv_attribute,
2550 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
2551 (ix86_attribute_table): Swap affects_type_identity and handler
2552 fields, adjust comments.
2553 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2554 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
2555 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
2556 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
2557 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
2558 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2559 * config/rx/rx.c (rx_attribute_table): Likewise.
2560 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
2561 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
2562 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
2563 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
2564 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2565 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
2566 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
2567 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
2568 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2569 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2570 * doc/plugins.texi (user_attr): Likewise. Add NULL for
2572 * attribs.c (empty_attribute_table): Swap affects_type_identity and
2574 (register_scoped_attributes, decl_attributes): Formatting fixes.
2576 PR tree-optimization/83269
2577 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2578 subtraction in arg0's type if type is signed and arg0 is unsigned.
2582 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2583 convert? on @0 instead of convert. Check type of @1, not @0.
2584 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2585 convert? on @0 instead of convert. Check type of @1, not @0.
2586 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2587 only optimize if either both @1 and @2 types are narrower
2588 precision, or both are wider or equal precision, and in the former
2589 case only if both have undefined overflow.
2591 2017-12-15 Richard Biener <rguenther@suse.de>
2594 * internal-fn.def (IFN_NOP): Add.
2595 * internal-fn.c (expand_NOP): Do nothing.
2596 * lto-streamer-in.c (input_function): Instead of removing
2597 sanitizer calls replace them with IFN_NOP calls.
2599 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2600 Alan Hayward <alan.hayward@arm.com>
2601 David Sherwood <david.sherwood@arm.com>
2603 * dse.c (store_info, read_info_type): Replace begin and end with
2605 (print_range): New function.
2606 (set_all_positions_unneeded, any_positions_needed_p)
2607 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2609 (record_store): Likewise. Optimize the case in which all positions
2611 (get_stored_val): Replace read_begin and read_end with read_offset
2613 (replace_read): Update call accordingly.
2615 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2617 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2618 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2619 (loop_cand::loop_cand): Initialize above members.
2620 (loop_cand::supported_operations): Delete.
2621 (loop_cand::can_interchange_p): Inline above function.
2622 (loop_cand::classify_simple_reduction): Record number of constant
2623 initialized simple reductions.
2624 (should_interchange_loops): New parameters. Check stmt cost of loops
2626 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2627 Update call to should_interchange_loops.
2628 (should_interchange_loop_nest): Update call to
2629 should_interchange_loops.
2631 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2634 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2635 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2636 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2638 2017-12-15 Julia Koval <julia.koval@intel.com>
2640 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2641 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
2643 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2644 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2645 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2647 2017-12-15 Julia Koval <julia.koval@intel.com>
2649 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2650 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2651 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2652 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2653 _mm_aesenc_epi128): New intrinsics.
2655 2017-12-15 Julia Koval <julia.koval@intel.com>
2657 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2658 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
2660 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2661 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2662 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2664 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2666 * gimple-ssa-strength-reduction.c (analyze_increments):
2667 Distinguish replacement costs for constant strides from those for
2670 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2672 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2674 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2676 * doc/invoke.texi: Document -Wcast-function-type.
2677 * recog.h (stored_funcptr): Change signature.
2678 * tree-dump.c (dump_node): Avoid warning.
2679 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2681 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2684 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2687 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2689 PR tree-optimization/83312
2690 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2691 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2692 "val" param, and to cope with arbitrary basic blocks.
2693 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2694 handle NULL_TREE for "val", dropping "bb" param.
2695 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2696 drop "bb" param. Handle NULL_TREE for "val".
2697 (find_case_label_for_value): Make "switch_stmt" param const.
2698 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2700 (vrp_prop::check_all_array_refs): Reimplement as...
2701 (check_array_bounds_dom_walker::before_dom_children): ...this new
2702 vfunc. Replace linear search through BB block list, excluding
2703 those with non-executable in-edges via dominator walk.
2705 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2707 * config/arm/arm.opt (mverbose-cost-dump): New option.
2708 * config/arm/arm.c (arm_rtx_costs): Use it.
2710 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2713 * reload1.c (emit_input_reload_insns): Skip debug markers.
2715 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2717 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2721 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2722 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2723 * config/sh/sh.c: Skip debug insns besides notes.
2724 * config/sh/sh.md: Likewise.
2725 * config/sh/sh_treg_combine.cc: Likewise.
2726 * config/sh/sync.md: Likewise.
2728 2017-12-14 Tom de Vries <tom@codesourcery.com>
2730 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2731 Add item for weak_undefined.
2733 2017-12-14 Richard Biener <rguenther@suse.de>
2735 PR tree-optimization/67842
2736 * sese.h (bb_in_region): Remove #if 0'ed code.
2738 2017-12-14 Richard Biener <rguenther@suse.de>
2740 PR tree-optimization/83326
2741 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2742 may_be_zero parameter and handle it by not marking the first
2743 peeled copy as not exiting the loop.
2744 (try_peel_loop): Likewise.
2745 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2746 to handle the case of constant or zero iterations and perform
2747 loop header copying on-the-fly.
2749 2017-12-14 Richard Biener <rguenther@suse.de>
2751 PR tree-optimization/83418
2752 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2753 Instead of asserting we don't get unfolded comparisons deal with them.
2755 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2758 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2759 basic blocks. Assert debug bind insns don't appear outside of bbs,
2760 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2763 PR tree-optimization/83198
2764 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2765 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2766 value if arg is a REAL_CST with incompatible type.
2768 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2769 Bin Cheng <bin.cheng@arm.com>
2772 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2774 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2777 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2779 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2780 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2782 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2783 Alan Hayward <alan.hayward@arm.com>
2784 David Sherwood <david.sherwood@arm.com>
2786 * poly-int.h: New file.
2787 * poly-int-types.h: Likewise.
2788 * coretypes.h: Include them.
2789 (POLY_INT_CONVERSION): Define.
2790 * target.def (estimated_poly_value): New hook.
2791 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2792 * doc/tm.texi: Regenerate.
2793 * doc/poly-int.texi: New file.
2794 * doc/gccint.texi: Include it.
2795 * doc/rtl.texi: Describe restrictions on subreg modes.
2796 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2797 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2798 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2799 * targhooks.h (default_estimated_poly_value): Declare.
2800 * targhooks.c (default_estimated_poly_value): New function.
2801 * target.h (estimated_poly_value): Likewise.
2802 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2803 (wi::unary_traits): Delete.
2804 (wi::binary_traits::signed_shift_result_type): Define for
2805 offset_int << HOST_WIDE_INT, etc.
2806 (generic_wide_int::operator <<=): Define for all types and use
2807 wi::lshift instead of <<.
2808 (wi::hwi_with_prec): Add a default constructor.
2809 (wi::ints_for): New class.
2810 (operator <<): Define for all wide-int types.
2811 (operator /): New function.
2812 (operator %): Likewise.
2813 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2814 (ASSERT_MAYBE_NE_AT): New macros.
2816 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2817 Dominik Vogt <vogt@linux.vnet.ibm.com>
2820 * emit-rtl.c (init_emit): Remove ??? comment.
2821 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2822 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
2823 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
2824 alignment of 3 virtual registers to BITS_PER_WORD.
2826 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
2828 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
2830 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
2831 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
2833 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2836 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
2838 2017-12-13 Jakub Jelinek <jakub@redhat.com>
2840 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
2841 with the exception of debug begin stmt markers appear before
2845 * final.c (rest_of_handle_final): Call variable_tracking_main only
2846 if !flag_var_tracking.
2848 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2849 Jakub Jelinek <jakub@redhat.com>
2853 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
2854 labels debug stmts that can only appear after labels.
2856 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
2858 PR rtl-optimization/82398
2859 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
2860 EXPR_USEFULNESS in priority comparison.
2862 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
2864 PR rtl-optimization/83393
2865 * combine.c (move_deaths): If reg_stat points to a too new insn in
2866 last_death, do not use it: find the proper insn instead.
2868 2017-12-12 Jeff Law <law@redhat.com>
2870 PR tree-optimization/83298
2871 PR tree-optimization/83362
2872 PR tree-optimization/83383
2873 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
2874 push_value_range a public interface. Add new argument to
2875 record_ranges_from_stmt.
2876 * gimple-ssa-evrp-analyze.c
2877 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
2878 Update comments. Handle recording temporary equivalences.
2879 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
2880 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
2881 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
2882 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
2883 gimple-ssa-evrp-analyze.h.
2884 (record_temporary_equivalences_from_phis): Add new argument. When
2885 the PHI arg is an SSA_NAME, set the result's range to the range
2887 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
2888 from statements too.
2889 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
2890 Pass it down to children as needed.
2891 (thread_outgoing_edges): Likewise.
2892 (thread_across_edge): Likewise. Push/pop range state as needed.
2893 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
2895 2017-12-12 Julia Koval <julia.koval@intel.com>
2897 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
2898 (PTA_CANNONLAKE): Remove PTA_CLWB.
2900 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2903 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
2904 (vcondv2div2df): Likewise.
2905 (vconduv2dfv2di): Likewise.
2907 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2909 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
2910 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
2911 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
2912 ATTR_MATHFN_FPROUNDING_ERRNO.
2914 2017-12-12 Richard Biener <rguenther@suse.de>
2916 PR tree-optimization/83385
2917 * graphite-scop-detection.c (get_order, order): Remove.
2918 (bb_to_rpo): New global.
2920 (build_scops): Sort pbbs in RPO order.
2922 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
2924 * combine.c (simplify_set): Do not transform subregs to zero_extends
2925 if the destination mode is a vector mode.
2927 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2929 PR tree-optimization/83359
2930 * tree-cfg.h (fold_loop_internal_call): Declare.
2931 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
2932 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
2933 (find_loop_dist_alias): New function.
2934 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
2935 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
2938 PR tree-optimization/80631
2939 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
2940 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
2941 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
2942 hardcoding zero as the value if COND_EXPR is never true. For
2943 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
2944 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
2945 hardcoding MAX_EXPR as the reduction operation.
2946 (is_nonwrapping_integer_induction): Allow negative step.
2947 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
2948 vect_create_epilog_for_reduction, if no value is suitable, don't
2949 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
2951 2017-12-12 Richard Biener <rguenther@suse.de>
2953 PR tree-optimization/81889
2954 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
2955 range info from the non-wrapping IV instead of just the range
2958 2017-12-12 Julia Koval <julia.koval@intel.com>
2960 * config.gcc: Add vaesintrin.h.
2961 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2963 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2964 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2966 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2967 * config/i386/immintrin.h: Include vaesintrin.h.
2968 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2969 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2970 _mm_aesdec_epi128): New intrinsics.
2972 2017-12-12 Julia Koval <julia.koval@intel.com>
2974 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
2975 OPTION_MASK_ISA_VAES_UNSET): New.
2976 (ix86_handle_option): Handle -mvaes.
2977 * config/i386/cpuid.h: Define bit_VAES.
2978 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
2979 * config/i386/i386-c.c (__VAES__): New.
2980 * config/i386/i386.c (ix86_target_string): Add -mvaes.
2981 (ix86_valid_target_attribute_inner_p): Ditto.
2982 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
2983 * config/i386/i386.opt: Add -mvaes.
2984 * doc/invoke.texi: Ditto.
2986 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
2988 * debug.h (gcc_debug_hooks): Add inline_entry.
2989 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
2990 * debug.c (do_nothing_debug_hooks): Likewise.
2991 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2992 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
2993 (dwarf2_lineno_debug_hooks): Likewise.
2995 * common.opt (gstatement-frontiers): New, setting
2996 debug_nonbind_markers_p.
2997 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
2998 * toplev.c (process_options): Autodetect value for debug statement
3000 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3001 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3003 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3004 markers. Integrate source bind into debug stmt expand loop.
3005 (pass_expand::execute): Check debug marker limit. Avoid deep
3006 TER and expand debug locations for debug bind insns only.
3007 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3009 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3010 * final.c (reemit_insn_block_notes): Take current block from
3011 nonbind markers. Declare note where it's first set.
3012 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3013 begin stmt markers if enabled.
3014 (notice_source_line): Handle nonbind markers. Fail if their
3015 location is unknown or that of builtins.
3016 (rest_of_handle_final): Convert begin stmt markers to notes if
3017 var-tracking didn't run.
3018 (rest_of_clean_state): Skip begin stmt markers.
3019 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3021 * function.c (allocate_struct_function): Set begin_stmt_markers.
3022 * function.h (struct function): Add debug_marker_count counter
3023 and debug_nonbind_markers flag.
3024 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3025 * gimple-low.c (lower_function_body): Adjust
3026 debug_nonbind_markers.
3027 (lower_stmt): Drop or skip gimple debug stmts.
3028 (lower_try_catch): Skip debug stmts.
3029 * gimple.c (gimple_build_debug_begin_stmt): New.
3030 (gimple_copy): Increment debug_marker_count if copying one.
3031 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3032 * gimplify.c (rexpr_location): New.
3033 (rexpr_has_location): New.
3034 (warn_switch_unreachable_r): Handle gimple debug stmts.
3035 (shortcut_cond_r): Call expr_location.
3037 (find_goto_label): New.
3038 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3040 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3042 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3043 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3044 (LANG_HOOKS_INITIALIZER): ... this.
3045 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3046 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3048 (update_ebb_live_info): Skip debug insn markers.
3049 * lra.c (debug_insn_static_data): Rename to...
3050 (debug_bind_static_data): ... this.
3051 (debug_marker_static_data): New.
3052 (lra_set_insn_recog_data): Select one of the above depending
3054 (lra_update_isn_regno_info): Don't assume debug insns have
3056 (push_insns): Skip debug insns.
3057 * lto-streamer-in.c (input_function): Drop debug stmts
3058 depending on active options. Adjust debug_nonbind_markers.
3059 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3060 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3061 begin stmt marker notes.
3062 (print_insn): Likewise.
3063 * recog.c (extract_insn): Recognize rtl for debug markers.
3064 * rtl.def (DEBUG_MARKER): New.
3065 * tree-inline.c: Include params.h.
3066 (remap_gimple_stmt): Handle nonbind markers.
3067 (maybe_move_debug_stmts_to_successors): Likewise.
3068 (copy_debug_stmt): Likewise.
3069 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
3070 markers regardless of no side effects.
3071 (tsi_link_before): Don't update container's side effects when adding
3072 a begin stmt marker.
3073 (tsi_link_after): Likewise.
3074 (expr_first): Skip begin stmt markers.
3075 (expr_last): Likewise.
3076 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
3077 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3078 Disregard nonbind markers.
3079 * tree.c (make_node_stat): Don't set side effects for begin stmt
3081 (build1_stat): Likewise.
3082 * tree.def (DEBUG_BEGIN_STMT): New.
3083 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
3084 * var-tracking.c (delete_debug_insns): Renamed to...
3085 (delete_vta_debug_insns): ... this.
3086 (reemit_marker_as_note): New.
3087 (vt_initialize): Reemit markers.
3088 (delete_vta_debug_insns): Likewise.
3089 (vt_debug_insns_local): Reemit or delete markers.
3090 (variable_tracking_main_1): Likewise.
3091 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
3092 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
3093 (gimple_debug_nonbind_marker_p): New.
3094 (gimple_build_debug_bind): Adjust.
3095 (gimple_build_debug_begin_stmt): New.
3096 * doc/invoke.texi (max-debug-marker-count): New param.
3097 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
3098 (debug_parameter_ref, debug_marker): New.
3099 (NOTE_INSN_BEGIN_STMT): New.
3100 (DEBUG_INSN): Describe begin stmt markers.
3102 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
3103 without debug insns, we wouldn't, but clean up debug insns
3104 after a control flow insn nevertheless.
3106 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
3107 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
3108 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
3109 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
3110 (rtl_tidy_fallthru_edge): Likewise.
3111 (rtl_verify_fallthru): Likewise.
3112 (rtl_verify_bb_layout): Likewise.
3113 (skip_insns_after_block): Likewise.
3114 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
3115 * dwarf2out.c: Include print-rtl.h.
3116 (dwarf2out_next_real_insn): New.
3117 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
3118 Dump debug binds in asm comments.
3119 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
3120 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
3121 callers to use gsi_start_nondebug_bb instead.
3122 (gsi_after_labels): Skip gimple debug stmts.
3123 (gsi_start_nondebug): New.
3124 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
3125 (proper_loop_form_for_interchange): Adjust.
3126 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
3127 * gimple.h (gimple_seq_last_nondebug_stmt): New.
3128 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
3129 (collect_fallthrough_labels): Likewise.
3130 (should_warn_for_implicit_fallthrough): Likewise.
3131 (warn_implicit_fallthrough_r): Likewise.
3132 (expand_FALLTHROUGH_r): Likewise.
3133 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
3134 (graphite_copy_stmts_from_block): Skip nonbind markers.
3135 * haifa-sched.c (sched_extend_bb): Skip debug insns.
3136 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
3137 * jump.c (clean_barriers): Skip debug insns.
3138 * omp-expand.c (expand_parallel_call): Skip debug insns.
3139 (expand_task_call): Likewise.
3140 (remove_exit_barrier): Likewise.
3141 (expand_omp_taskreg): Likewise.
3142 (expand_omp_for_init_counts): Likewise.
3143 (expand_omp_for_generic): Likewise.
3144 (expand_omp_for_static_nochunk): Likewise.
3145 (expand_omp_for_static_chunk): Likewise.
3146 (expand_omp_simd): Likewise.
3147 (expand_omp_taskloop_for_outer): Likewise.
3148 (expand_omp_taskloop_for_inner): Likewise.
3149 (expand_oacc_for): Likewise.
3150 (expand_omp_sections): Likewise.
3151 (expand_omp_single): Likewise.
3152 (expand_omp_synch): Likewise.
3153 (expand_omp_atomic_load): Likewise.
3154 (expand_omp_atomic_store): Likewise.
3155 (expand_omp_atomic_fetch_op): Likewise.
3156 (expand_omp_atomic_pipeline): Likewise.
3157 (expand_omp_atomic_mutex): Likewise.
3158 (expand_omp_target): Likewise.
3159 (grid_expand_omp_for_loop): Likewise.
3160 (grid_expand_target_grid_body): Likewise.
3161 (build_omp_regions_1): Likewise.
3162 * omp-low.c (check_combined_parallel): Skip debug stmts.
3163 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
3164 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
3165 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
3167 * tree-cfg.c (make_blobs_1): Skip debug stmts.
3168 (make_edges): Likewise.
3169 (cleanup_dead_labels): Likewise.
3170 (gimple_can_merge_blocks_p): Likewise.
3171 (stmt_starts_bb_p): Likewise.
3172 (gimple_block_label): Likewise.
3173 (gimple_redirect_edge_and_branch): Likewise.
3174 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
3176 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
3178 * tree-pretty-print (print_declaration): Omit initializer in slim
3180 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
3182 (eliminate_unnecessary_stmts): Stabilize block removal order.
3183 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
3184 * var-tracking.c (get_first_insn): New.
3185 (vt_emit_notes): Call it.
3186 (vt_initialize): Walk any insns before the first BB.
3187 (delete_debug_insns): Likewise.
3189 * gimple.h (enum gimple_debug_subcode): Add
3190 GIMPLE_DEBUG_BEGIN_STMT.
3191 (gimple_debug_begin_stmt_p): New.
3192 (gimple_debug_nonbind_marker_p): New.
3193 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
3194 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
3195 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
3196 * insn-notes.def (BEGIN_STMT): New.
3197 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
3198 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
3199 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
3200 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
3201 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
3202 (INSN_DEBUG_MARKER_KIND): New.
3203 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
3204 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
3205 (INSN_VAR_LOCATION_PTR): New.
3206 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
3208 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
3209 deps only if debug bind insns are enabled.
3210 (pass_expand::execute): Avoid deep TER and expand
3211 debug locations for debug bind insns only.
3212 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
3213 debug stmts special handling down to debug bind stmts.
3214 * combine.c (try_combine): Narrow debug insns special handling
3215 down to debug bind insns.
3216 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
3217 Narrow debug insns preexisting special handling down to debug
3219 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
3220 handling down to debug bind insns.
3221 * function.c (instantiate_virtual_regs): Skip debug markers,
3222 adjust handling of debug binds.
3223 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
3224 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
3225 * haifa-sched.c (schedule_insn): Narrow special handling of debug
3226 insns to debug bind insns.
3227 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
3228 special handling of debug stmts to debug bind stmts.
3229 * ipa-split.c (split_function): Likewise.
3230 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
3231 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
3233 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
3234 * regrename.c (build_def_use): Likewise.
3235 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
3236 (pass_cprop_hardreg): Narrow special casing of debug insns to
3238 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
3239 * reload1.c (reload): Likewise.
3240 * sese.c (sese_insert_phis_for_liveouts): Narrow special
3241 casing of debug stmts to debug bind stmts.
3242 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
3243 * ssa-iterators.h (num_imm_uses): Likewise.
3244 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
3245 debug stmts to debug bind stmts.
3246 * tree-inline.c (tree_function_versioning): Narrow special casing
3247 of debug stmts to debug bind stmts.
3248 * tree-loop-distribution.c (generate_loops_for_partition):
3249 Narrow special casing of debug stmts to debug bind stmts.
3250 * tree-sra.c (analyze_access_subtree): Narrow special casing
3251 of debug stmts to debug bind stmts.
3252 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
3253 stmts to debug bind stmts.
3254 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
3255 casing of debug stmts to debug bind stmts.
3256 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
3257 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
3258 casing of debug stmts to debug bind stmts.
3259 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3261 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
3262 debug stmts to debug bind stmts.
3263 (gimple_replace_ssa_lhs): Likewise.
3264 (insert_debug_temp_for_var_def): Likewise.
3265 (insert_debug_temps_for_defs): Likewise.
3266 (reset_debug_uses): Likewise.
3267 * tree-ssanames.c (release_ssa_name_fn): Likewise.
3268 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
3269 (adjust_debug_stmts): Likewise.
3270 (adjust_phi_and_debug_stmts): Likewise.
3271 (vect_do_peeling): Likewise.
3272 * tree-vect-loop.c (vect_transform_loop): Likewise.
3273 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
3274 * var-tracking.c (adjust_mems): Narrow special casing of debug
3275 insns to debug bind insns.
3276 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
3277 (compute_bb_dataflow, vt_find_locations): Likewise.
3278 (vt_expand_loc, emit_notes_for_changes): Likewise.
3279 (vt_init_cfa_base): Likewise.
3280 (vt_emit_notes): Likewise.
3281 (vt_initialize): Likewise.
3282 (vt_finalize): Likewise.
3284 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
3285 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
3286 (next_nonnote_nondebug_insn_bb): New.
3287 (prev_nonnote_nondebug_insn_bb): New.
3288 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
3289 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
3290 (prev_nonnote_nondebug_insn_bb): Declare.
3291 (next_nonnote_nondebug_insn_bb): Declare.
3292 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
3293 * cfgrtl.c (get_last_bb_insn): Likewise.
3294 * lra.c (push_insns): Likewise.
3296 2017-12-11 David Malcolm <dmalcolm@redhat.com>
3299 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
3301 * selftest.c (selftest::test_runner::test_runner): New ctor.
3302 (selftest::test_runner::~test_runner): New dtor.
3303 * selftest.h (class selftest::test_runner): New class.
3305 2017-12-11 Carl Love <cel@us.ibm.com>
3307 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
3308 vec_extract_fp32_from_shortl]): Add #defines.
3309 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
3310 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
3311 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
3312 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
3313 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
3314 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
3315 * doc/extend.texi: Add documentation for the added builtins.
3317 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
3319 PR rtl-optimization/80693
3320 PR rtl-optimization/81019
3321 PR rtl-optimization/81020
3322 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
3323 are not mentioned in i3. Place the REG_UNUSED note on i2,
3324 possibly modified to REG_DEAD, if it did not originate in i3.
3326 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3328 * recog.c (store_data_bypass_p_1): New function.
3329 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
3330 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
3332 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
3334 PR rtl-optimization/83361
3335 * ifcvt.c (if_convert): Call fixup_partitions.
3337 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
3339 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3340 early folding of splat_u{8,16,32}.
3342 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3344 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
3345 output_operand_lossage first argument with capital letter.
3346 (aarch64_override_options): Don't start error and sorry first argument
3347 with capital letter.
3349 2017-12-11 Andi Kleen <ak@linux.intel.com>
3351 PR gcov-profile/83355
3352 * auto-profile.c (string_table::get_index_by_decl): Don't
3353 recurse when abstract origin points to itself.
3355 2017-12-11 Bin Cheng <bin.cheng@arm.com>
3357 PR tree-optimization/83320
3358 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
3359 (prune_datarefs_not_in_loop): Ditto.
3361 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
3363 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
3365 2017-12-10 Jakub Jelinek <jakub@redhat.com>
3367 PR tree-optimization/83337
3368 * gimple-loop-interchange.cc (compute_access_stride): Handle
3369 bitfield DRs properly.
3371 2017-12-09 Jakub Jelinek <jakub@redhat.com>
3373 PR tree-optimization/83338
3374 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
3375 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
3376 vector of unsigned integers to vector of signed integers.
3378 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
3380 PR rtl-optimization/83317
3381 * lra-constraints.c (process_address_1): Add insn code check.
3383 2017-12-08 Michael Matz <matz@suse.de>
3385 Fix PR tree-optimization/83323
3386 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
3387 head-controlled loops and loop BBs.
3388 * common.opt (funroll-and-jam): Remove, instead ...
3389 (floop-unroll-and-jam): ... reuse this option.
3390 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
3391 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
3392 (-floop-unroll-and-jam): ... this option.
3394 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3396 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
3397 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
3398 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
3400 2017-12-08 Richard Biener <rguenther@suse.de>
3403 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
3404 handle accesses outside of zero-sized vars.
3406 2017-12-08 Martin Jambor <mjambor@suse.cz>
3408 PR tree-optimization/83141
3409 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
3410 test for MEM_REFs implicitely changing types with padding. Remove
3412 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
3414 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3416 * config/arc/arc.c (arc_attribute_table): Add exclusions to
3418 * config/avr/avr.c (avr_attribute_table): Likewise.
3419 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3420 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3421 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3422 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
3423 of struct attribute_spec.
3424 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3426 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
3429 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
3431 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
3434 * profile-count.c (profile_count::from_gcov_type): Move from
3435 profile-count.h; handle overflow.
3436 * profile-count.h (profile_count::from_gcov_type): Move offline.
3438 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
3440 PR rtl-optimization/83304
3441 * combine.c (move_deaths): If we do not know where a register died,
3444 2017-12-08 Richard Biener <rguenther@suse.de>
3446 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
3447 Provide -fopt-info-loop feedback when we interchange in a nest.
3449 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3451 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
3452 for armv6 ARM CPU IDs.
3454 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3456 * common/config/arm/arm-common.c: Include <algorithm>.
3457 (INCLUDE_VECTOR): Define.
3458 (compare_opt_names): New function.
3459 (arm_rewrite_selected_arch): Only strip out extensions that can be
3460 expressed through -mfpu. Sort the remaining extensions
3463 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3465 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
3466 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
3467 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
3468 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
3470 2017-06-08 Tristan Gingold <gindold@adacore.com>
3473 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
3474 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
3476 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3478 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
3479 of struct attribute_spec.
3481 2017-12-08 Julia Koval <julia.koval@intel.com>
3483 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
3484 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
3485 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
3486 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
3487 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
3488 _mm_maskz_dpwssds_epi32): Ditto.
3490 2017-12-08 Richard Biener <rguenther@suse.de>
3492 PR tree-optimization/81303
3493 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
3494 conditions try to create a comparison vector type matching
3495 the data vector type.
3496 (vectorizable_condition): Adjust.
3497 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3498 Leave invariant conditions alone in case we can vectorize those.
3500 2017-12-08 Julia Koval <julia.koval@intel.com>
3502 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
3503 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
3504 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
3505 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
3506 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
3508 2017-12-08 Julia Koval <julia.koval@intel.com>
3510 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
3511 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
3512 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
3513 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
3514 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
3515 _mm_maskz_dpbusds_epi32): New intrinsics.
3517 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3519 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
3521 (ld<bh>io_signed): Likewise.
3522 (st<bhw>io): Likewise.
3523 * config/nios2/predicates.md (ldstio_memory_operand): Allow
3524 SMALL_INT12 constant integer operand.
3526 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3528 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
3530 (nios2_large_constant_allowed): ...this. Adjust uses.
3531 (nios2_plus_symbolic_constant_p): Rename to...
3532 (nios2_plus_large_constant_p): ...this. Adjust uses.
3533 (nios2_legitimate_address_p): Correct CONST_INT handling.
3534 (nios2_symbolic_memory_operand_p): Rename to...
3535 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
3536 (nios2_large_constant_p): Check for large constant integers too.
3537 (nios2_split_large_constant): Handle constant integers.
3538 (nios2_split_symbolic_memory_operand): Rename to...
3539 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
3540 (nios2_legitimize_constant_address): Handle constant integers.
3541 (r0rel_constant_p): Handle small constant integers.
3542 (nios2_print_operand_address): Handle r0-relative integer addresses.
3543 * config/nios2/nios2-protos.h: Adjust for renamed functions.
3544 * config/nios2/nios2.md: Adjust for renamed functions.
3546 2017-12-07 Andrew Waterman <andrew@sifive.com>
3548 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
3549 (TARGET_HAVE_SRODATA_SECTION): New define.
3550 (riscv_select_section): New function.
3552 2017-12-08 Joseph Myers <joseph@codesourcery.com>
3553 Alexander Monakov <amonakov@ispras.ru>
3554 Jakub Jelinek <jakub@redhat.com>
3557 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
3559 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3561 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
3562 a bool success value. Don't call output_operand_lossage here.
3563 (aarch64_print_ldpstp_address): Return a bool success value.
3564 (aarch64_print_operand_address): Call output_addr_const if
3565 aarch64_print_address_internal fails.
3566 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
3567 'y'; call output_operand_lossage instead. Call output_operand_lossage
3568 if aarch64_print_ldpstp_address fails.
3570 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3572 * tree-vector-builder.h
3573 (tree_vector_builder::binary_encoded_nelts): Declare.
3574 * tree-vector-builder.c
3575 (tree_vector_builder::binary_encoded_nelts): New function.
3576 * fold-const.c (negate_expr_p): Likewise.
3577 (operand_equal_p, fold_checksum_tree): Likewise.
3578 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3579 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3580 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3581 (uniform_vector_p): Likewise.
3582 * varasm.c (const_hash_1, compare_constant): Likewise.
3583 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3584 (valid_lattice_transition): Operate directly on the VECTOR_CST
3586 * ipa-icf.c: Include tree-vector-builder.h.
3587 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3588 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3590 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3592 * tree.c (build_vector): Delete.
3593 * tree.h (build_vector): Make static and move into the self-testing
3596 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3598 * vector-builder.h (vector_builder::derived): New const overload.
3599 (vector_builder::elt): New function.
3600 * tree-vector-builder.h (tree_vector_builder::type): New function.
3601 (tree_vector_builder::apply_step): Declare.
3602 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3604 * gimple-fold.h (tree_vector_builder): Declare.
3605 (gimple_build_vector): Take a tree_vector_builder instead of a
3606 type and vector of elements.
3607 * gimple-fold.c (gimple_build_vector): Likewise.
3608 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3610 (get_initial_defs_for_reduction): Likewise.
3611 (vectorizable_induction): Likewise.
3613 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3615 * tree-vector-builder.h
3616 (tree_vector_builder::new_binary_operation): Declare.
3617 * tree-vector-builder.c
3618 (tree_vector_builder::new_binary_operation): New function.
3619 * fold-const.c (fold_relational_const): Use it.
3620 (const_binop): Likewise. Check that both input vectors have
3621 the same number of elements, thus excluding things like WIDEN_SUM.
3622 Check whether it is possible to operate directly on the encodings
3625 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3627 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3628 new_unary_operation, operating only on the encoded elements.
3629 (const_unop): Likewise.
3630 (exact_inverse): Likewise.
3631 (distributes_over_addition_p): New function.
3632 (const_binop): Use tree_vector_builder and new_unary_operation
3633 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3634 on the encoded elements unless the encoding is strided and the
3635 operation does not distribute over addition.
3636 (fold_convert_const): Use tree_vector_builder and
3637 new_unary_operation. Operate only on the encoded elements
3638 for truncating integer conversions, or for non-stepped encodings.
3640 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3642 * config/sparc/sparc.c: Include tree-vector-builder.h.
3643 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3644 * expmed.c: Include tree-vector-builder.h.
3645 (make_tree): Use tree_vector_builder instead of build_vector.
3646 * fold-const.c: Include tree-vector-builder.h.
3647 (const_binop): Use tree_vector_builder instead of build_vector.
3648 (const_unop): Likewise.
3649 (native_interpret_vector): Likewise.
3650 (fold_vec_perm): Likewise.
3651 (fold_ternary_loc): Likewise.
3652 * gimple-fold.c: Include tree-vector-builder.h.
3653 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3655 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3656 (simplify_vector_constructor): Use tree_vector_builder instead
3658 * tree-vect-generic.c: Include tree-vector-builder.h.
3659 (add_rshift): Use tree_vector_builder instead of build_vector.
3660 (expand_vector_divmod): Likewise.
3661 (optimize_vector_constructor): Likewise.
3662 * tree-vect-loop.c: Include tree-vector-builder.h.
3663 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3664 of build_vector. Explicitly use a stepped encoding for
3666 * tree-vect-slp.c: Include tree-vector-builder.h.
3667 (vect_get_constant_vectors): Use tree_vector_builder instead
3669 (vect_transform_slp_perm_load): Likewise.
3670 (vect_schedule_slp_instance): Likewise.
3671 * tree-vect-stmts.c: Include tree-vector-builder.h.
3672 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3673 (vect_gen_perm_mask_any): Likewise.
3674 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3675 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3677 (build_vector_from_val): Likewise. Explicitly use a duplicate
3680 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3682 * doc/generic.texi (VECTOR_CST): Describe new representation of
3684 * vector-builder.h: New file.
3685 * tree-vector-builder.h: Likewise.
3686 * tree-vector-builder.c: Likewise.
3687 * Makefile.in (OBJS): Add tree-vector-builder.o.
3688 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3689 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3690 (tree_vector): Change the number of elements to
3691 vector_cst_encoded_nelts.
3692 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3693 (VECTOR_CST_ELTS): Delete.
3694 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3695 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3696 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3697 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3698 (VECTOR_CST_ENCODED_ELT): Likewise.
3699 (vector_cst_encoded_nelts): New function.
3700 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3701 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3702 (vector_cst_int_elt, vector_cst_elt): Declare.
3703 * tree.c: Include tree-vector-builder.h.
3704 (tree_code_size): Abort if passed VECTOR_CST.
3705 (tree_size): Update for new VECTOR_CST layout.
3706 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3707 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3708 (build_vector): Use tree_vector_builder.
3709 (vector_cst_int_elt, vector_cst_elt): New functions.
3710 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3711 encoded elements and then create the vector in the canonical form.
3712 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3713 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3714 (tree_c_tests): Call test_vector_cst_patterns.
3715 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3717 (hash_tree): Likewise.
3718 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3719 (streamer_write_tree_header): Likewise.
3720 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3721 (streamer_alloc_tree): Likewise. Update call to make_vector.
3722 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3724 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3726 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3727 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3728 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3730 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3731 Richard Biener <rguenther@suse.de>
3733 PR tree-optimization/81303
3734 * Makefile.in (gimple-loop-interchange.o): New object file.
3735 * common.opt (floop-interchange): Reuse the option from graphite.
3736 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3737 -floop-interchange and mention it for -O3.
3738 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3739 * gimple-loop-interchange.cc: New file.
3740 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3741 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3742 * passes.def (pass_linterchange): New pass.
3743 * timevar.def (TV_LINTERCHANGE): New time var.
3744 * tree-pass.h (make_pass_linterchange): New declaration.
3745 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3746 interchange. Record IV before/after increment in new parameters.
3747 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3748 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3750 (check_reduction_path): ...New function here.
3751 * tree-vectorizer.h (check_reduction_path): New declaration.
3753 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3756 PR rtl-optimization/80818
3757 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3758 always early clobbered.
3759 * lra-lives.c (process_bb_lives): Check input hard regs for early
3760 clobbered non-operand hard reg.
3762 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3765 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3766 types_compatible_p, just that TYPE_MODE is the same.
3768 2017-12-07 Martin Sebor <msebor@redhat.com>
3771 * attribs.c (empty_attribute_table): Initialize new member of
3772 struct attribute_spec.
3773 (decl_attributes): Add argument. Handle mutually exclusive
3774 combinations of attributes.
3775 (selftests::test_attribute_exclusions): New function.
3776 (selftests::attribute_c_tests): Ditto.
3777 * attribs.h (decl_attributes): Add default argument.
3778 * selftest.h (attribute_c_tests): Declare.
3779 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3780 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3782 * doc/extend.texi (Common Function Attributes): Update const and pure.
3783 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3784 of struct attribute_spec.
3785 * config/arc/arc.c (arc_attribute_table): Same.
3786 * config/arm/arm.c (arm_attribute_table): Same.
3787 * config/avr/avr.c ( avr_attribute_table): Same.
3788 * config/bfin/bfin.c (bfin_attribute_table): Same.
3789 * config/cr16/cr16.c (cr16_attribute_table): Same.
3790 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3791 * config/h8300/h8300.c (h8300_attribute_table): Same.
3792 * config/i386/i386.c (ix86_attribute_table): Same.
3793 * config/ia64/ia64.c (ia64_attribute_table): Same.
3794 * config/m32c/m32c.c (m32c_attribute_table): Same.
3795 * config/m32r/m32r.c (m32r_attribute_table): Same.
3796 * config/m68k/m68k.c (m68k_attribute_table): Same.
3797 * config/mcore/mcore.c (mcore_attribute_table): Same.
3798 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3799 * config/mips/mips.c (mips_attribute_table): Same.
3800 * config/msp430/msp430.c (msp430_attribute_table): Same.
3801 * config/nds32/nds32.c (nds32_attribute_table): Same.
3802 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3803 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3804 * config/rl78/rl78.c (rl78__attribute_table): Same.
3805 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3806 * onfig/rx/rx.c (rx_attribute_table): Same.
3807 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3808 * config/sh/sh.c (sh_attribute_table): Same.
3809 * config/sparc/sparc.c (sparc_attribute_table): Same.
3810 * config/spu/spu.c (spu_attribute_table): Same.
3811 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3812 * config/v850/v850.c (v850_attribute_table): Same.
3813 * config/visium/visium.c (visium_attribute_table): Same.
3815 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3818 * config/arm/arm.c (INCLUDE_STRING): Define.
3819 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3820 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
3823 2017-12-07 Michael Matz <matz@suse.de>
3825 Add unroll and jam pass
3827 * gimple-loop-jam.c: New file.
3828 * Makefile.in (OBJS): Add gimple-loop-jam.o.
3829 * common.opt (funroll-and-jam): New option.
3830 * opts.c (default_options_table): Add unroll-and-jam at -O3.
3831 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
3832 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
3833 * passes.def: Add pass_loop_jam.
3834 * timevar.def (TV_LOOP_JAM): Add.
3835 * tree-pass.h (make_pass_loop_jam): Declare.
3836 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
3837 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
3838 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
3839 to flow_loop_tree_node_add.
3840 (duplicate_subloops, copy_loops_to): Append to sibling list.
3841 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
3842 * doc/invoke.texi (-funroll-and-jam): Document new option.
3843 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
3845 2017-12-07 Richard Biener <rguenther@suse.de>
3847 PR tree-optimization/83296
3848 PR tree-optimization/67769
3849 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
3850 flow sensitive info in an unrelated BB.
3851 (value_replacement): Use reset_flow_sensitive_info.
3852 (minmax_replacement): Reset flow sensitive info on the def
3853 we move. Do not reset flow sensitive info in the whole BB
3854 we move the stmt to.
3855 (abs_replacement): Likewise.
3857 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3860 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3861 rs6000_cpu to the given -mcpu=, or to the default processor.
3863 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3865 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
3866 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
3867 instead of rs6000_cpu_attr.
3868 (rs6000_adjust_cost): Ditto.
3869 (is_microcoded_insn): Ditto.
3870 (rs6000_adjust_priority): Ditto.
3871 (rs6000_issue_rate): Ditto.
3872 (rs6000_use_sched_lookahead): Ditto.
3873 (rs6000_use_sched_lookahead_guard): Ditto.
3874 (rs6000_sched_reorder): Ditto.
3875 (force_new_group): Ditto.
3876 * config/rs6000/rs6000.md (cpu attribute): Ditto.
3877 (group_ending_nop): Ditto.
3879 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3881 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
3882 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
3883 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
3884 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
3886 (rs6000_reassociation_width): Ditto.
3887 (rs6000_emit_epilogue): Ditto.
3888 (rs6000_adjust_cost): Ditto.
3889 (is_microcoded_insn): Ditto.
3890 (is_cracked_insn): Ditto.
3891 (rs6000_adjust_priority): Ditto.
3892 (rs6000_sched_reorder): Ditto.
3893 (rs6000_sched_reorder2): Ditto.
3894 (insn_must_be_first_in_group): Ditto.
3895 (insn_must_be_last_in_group): Ditto.
3896 (rs6000_register_move_cost): Ditto.
3897 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
3900 2017-12-07 Julia Koval <julia.koval@intel.com>
3902 * config.gcc: Add vaesintrin.h.
3903 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3905 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3906 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3908 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3909 * config/i386/immintrin.h: Include vaesintrin.h.
3910 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3911 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3912 _mm_aesdec_epi128): New intrinsics.
3914 2017-12-06 David Malcolm <dmalcolm@redhat.com>
3916 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
3917 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
3918 c-family/c-spellcheck.cc.
3919 (best_macro_match::best_macro_match): Likewise.
3921 (struct edit_distance_traits<cpp_hashnode *>): Move to
3922 c-family/c-spellcheck.h.
3923 (class best_macro_match): Likewise.
3925 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3927 PR tree-optimization/83293
3928 * gimple-ssa-strength-reduction.c (insert_initializers): Use
3929 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
3930 might insert into empty bb.
3933 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
3934 simplify for plus with :c added, and pointer_plus without that.
3935 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
3936 with undefined overflow and the conversion is not widening,
3937 perform negation in utype and only convert to type afterwards.
3938 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
3939 simplify for plus with :c added, and pointer_plus without that.
3940 If type is integral with undefined overflow and the conversion is
3941 not widening, perform minus in utype and only convert to type
3942 afterwards. Move the last pointer_diff_expr simplify into the
3945 2017-12-06 Martin Sebor <msebor@redhat.com>
3947 PR tree-optimization/82646
3948 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
3949 strncpy, not maxlen.
3951 2017-12-06 Martin Sebor <msebor@redhat.com>
3953 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
3956 PR tree-optimization/83075
3957 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
3958 strncat/strncpy don't change length of source string.
3960 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
3963 2017-11-29 Martin Aberg <maberg@gaisler.com>
3965 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3966 to prevent b2bst errata sequence.
3967 (sqrtdf2_fix): Likewise.
3969 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3971 PR tree-optimization/81945
3972 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
3973 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
3974 to dest_cfun has orig_loop_num set, either remap it to the new
3975 loop number if the loop got moved too, or clear it.
3977 2017-12-05 Steve Ellcey <sellcey@cavium.com>
3979 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
3981 (thunderx2t99_nothing): New insn reservation.
3982 (thunderx2t99_mrs): New insn reservation.
3983 (thunderx2t99_multiple): New insn reservation.
3984 (thunderx2t99_alu_basi): Add bfx to reservation.
3985 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
3987 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3990 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
3992 2017-12-05 Bin Cheng <bin.cheng@arm.com>
3994 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
3995 tree-ssa-pre.c::remove_dead_inserted_code.
3996 * tree-ssa-dce.h: New file.
3997 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
3998 (remove_dead_inserted_code): Move and rename to function
3999 tree-ssa-dce.c::simple_dce_from_worklist.
4000 (pass_pre::execute): Update use.
4002 2017-12-05 Richard Biener <rguenther@suse.de>
4004 PR tree-optimization/83277
4005 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4006 to code-gen liveout vars.
4008 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4010 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4011 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4012 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4015 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4017 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4019 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4020 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4023 2017-12-05 Richard Biener <rguenther@suse.de>
4025 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4026 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4027 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4028 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4030 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4032 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4034 2017-12-05 Martin Liska <mliska@suse.cz>
4035 Jakub Jelinek <jakub@redhat.com>
4037 * doc/invoke.texi: Document the options.
4038 * flag-types.h (enum sanitize_code): Add
4039 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4040 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4041 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4042 * opts.c: Define new sanitizer options.
4043 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4044 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4046 2017-12-05 Julia Koval <julia.koval@intel.com>
4048 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4049 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4050 (ix86_handle_option): Handle -mavx512vnni.
4051 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4052 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4053 * config/i386/i386-c (__AVX512VNNI__): New.
4054 * config/i386/i386.c (ix86_target_string): Handle new option.
4055 (ix86_valid_target_attribute_inner_p): Handle new option.
4056 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4057 * config/i386/i386.opt (mavx512vnni): New option.
4059 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4062 * athlon.md: Disable for generic.
4063 * haswell.md: Enable for generic.
4064 * i386.c (ix86_sched_init_global): Add core hooks for generic.
4065 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
4067 (ix86_adjust_cost): Move generic to haswell path.
4069 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
4071 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
4072 instead of MEM_P in a couple more places. Fix formatting issues.
4074 2017-12-04 Jim Wilson <jimw@sifive.com>
4076 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
4077 instead of GP_REG_LAST-1.
4078 (riscv_adjust_libcall_cfi_prologue): Likewise.
4079 (riscv_adjust_libcall_cri_epilogue): Likewise.
4080 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
4083 2017-12-04 Luis Machado <luis.machado@linaro.org>
4085 * ipa-pure-const.c (check_decl): Add missing newline.
4086 (state_from_flags): Likewise.
4088 2017-12-04 Jeff Law <law@redhat.com>
4090 PR tree-optimizatin/78496
4091 * gimple-ssa-evrp-analyze.h
4092 (evrp_range_analyzer::get_vr_values): Simplify.
4093 * gimple-ssa-evrp-analyze.c: Corresponding changes.
4094 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
4095 and gimple-ssa-evrp-analyze.h.
4096 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
4097 (simplify_stmt_for_jump_threading): Copy a blob of code from
4098 tree-vrp.c to use ranges to simplify statements.
4099 (dom_opt_dom_walker::before_dom_children): Call
4100 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
4101 (dom_opt_dom_walker::after_dom_children): Similarly for
4102 evrp_range_analyzer::leave.
4103 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
4106 * gimple-ssa-evrp-analyze.c
4107 (evrp_range_analyzer::extract_range_from_stmt): Always use
4108 vr_values::update_value_range so preexisting range info is
4109 medged with new range info, even if the new range is VR_VARYING.
4111 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4113 * combine.c: Adjust comment.
4114 (use_crosses_set_p): Delete.
4115 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
4116 (try_combine): Ditto.
4118 2017-12-04 Richard Biener <rguenther@suse.de>
4120 PR tree-optimization/83255
4121 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
4122 Re-add zero-iteration check.
4124 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4126 PR rtl-optimization/83245
4127 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
4128 hard registers as earlyclobber, also if not in an asm.
4130 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4134 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4137 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4138 rs6000_cpu based on cpu_index, not tune_index.
4140 2017-12-04 Richard Biener <rguenther@suse.de>
4142 PR tree-optimization/83238
4143 * graphite-scop-detection.c (scop_detection::merge_sese): Make
4144 code match comment, rejecting invalid SESE regions.
4146 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
4148 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
4149 require base operand is a REG_POINTER prior to reload on targets
4150 with non-equivalent space registers.
4152 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4154 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
4155 (update_specialized_profile): Fix updating of counts.
4156 (perhaps_add_new_callers): Likewise.
4158 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4161 * x86-tune.def: Remove obsolette FIXMEs.
4162 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
4163 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
4164 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
4166 (X86_TUNE_PAD_RETURNS): Disable for generic.
4168 2017-12-02 Jakub Jelinek <jakub@redhat.com>
4170 PR tree-optimization/83170
4171 PR tree-optimization/83241
4172 * gimple-ssa-store-merging.c
4173 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
4174 gimple_vuse (ins_stmt) in case it has changed.
4175 (imm_store_chain_info::output_merged_store): Likewise.
4177 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
4181 * tree-cfg.c (pass_warn_function_return::execute): Handle
4182 __builtin_ubsan_handle_missing_return like __builtin_unreachable
4183 with BUILTINS_LOCATION.
4187 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
4188 is BLKmode for vector field with vector raw mode, use TYPE_MODE
4189 instead of DECL_MODE.
4191 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
4192 last argument to rtx pointer.
4193 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
4194 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
4195 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
4196 depending on the chosen ISAs.
4197 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
4198 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
4199 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
4201 * config/i386/sse.md (mov<mode>_internal): Likewise.
4202 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
4204 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4206 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
4208 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4210 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
4211 parameters from prototype.
4212 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
4213 parameters. Don't print a comment.
4214 (emit_fusion_gpr_load): Adjust.
4215 (emit_fusion_load_store): Adjust.
4216 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
4217 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
4218 comment on the second line.
4220 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4223 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4224 rs6000_cpu based on cpu_index, not tune_index.
4226 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4228 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
4229 which_alternative instead of which_alternative + 1.
4230 (output_asm_insn): Print an extra tab if the template is short.
4232 2017-12-01 Jim Wilson <jimw@sifive.com>
4234 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
4236 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
4237 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
4239 * doc/tm.texi: Regenerate.
4241 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
4244 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
4245 whether we can allocate pseudos before trying to fix an address.
4246 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
4247 memory address is indexed or indirect.
4248 (floatuns_<mode>si2_hw2): Likewise.
4250 2017-12-01 Jason Merrill <jason@redhat.com>
4252 * Makefile.in (TAGS): Add c-family/*.cc.
4254 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
4256 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
4257 (call_value_insn): Likewise.
4258 (sibcall_insn): Likewise.
4259 (sibcall_value_insn): Likewise.
4260 (movsi_aarch64): Likewise.
4261 (movdi_aarch64): Likewise.
4262 (add_losym_): Likewise.
4263 (ldr_got_small_): Likewise.
4264 (ldr_got_small_sidi): Likewise.
4265 (ldr_got_small_28k_): Likewise.
4266 (ldr_got_small_28k_sidi): Likewise.
4267 * config/aarch64/aarch64.c (aarch64_print_address_internal):
4268 Move output_addr_const to symbolic case. Add error check.
4270 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4272 * config/s390/predicates.md (plus16_Q_operand): New predicate.
4273 * config/s390/s390.md: Disable MVC merging peephole if it would
4274 disable operand forwarding.
4275 (new peephole2): Split MVCs if it would turn them into up to 2
4278 2017-12-01 Richard Biener <rguenther@suse.de>
4280 PR tree-optimization/83232
4281 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
4282 detection of same access. Instead of breaking the group here
4283 do not consider the duplicate. Add comment explaining real fix.
4285 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
4287 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
4289 2017-12-01 Sudakshina Das <sudi.das@arm.com>
4291 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
4293 2017-12-01 Jakub Jelinek <jakub@redhat.com>
4295 * function.h (struct function): Remove cilk_frame_decl,
4296 is_cilk_function and calls_cilk_spawn fields.
4297 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
4299 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
4300 cilk_elemental field.
4301 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
4302 * target.def: Adjust comment.
4303 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
4304 Don't test cilk_elemental.
4306 PR tree-optimization/83233
4307 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
4308 bswap_stat name for the struct.
4311 * tree.h (SWITCH_BREAK_LABEL_P): Define.
4312 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
4313 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
4314 SWITCH_BREAK_LABEL_P set on the label.
4315 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
4316 added for default case if it was missing and not all cases covered.
4317 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
4318 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
4320 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
4321 to avoid -Wimplicit-fallthrough warning.
4322 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
4323 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
4325 PR tree-optimization/83221
4326 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
4328 (init_reassoc): Formatting fix.
4331 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
4332 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
4336 * tree-cfg.c: Include asan.h.
4337 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
4339 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4341 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
4343 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4345 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
4346 vpcmpeqd instruction.
4348 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4350 * config/i386/i386.c (standard_sse_constant_opcode): Fix
4351 registers type for 128bit mode.
4353 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4355 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
4357 * gengtype-state.c (read_state_pair): Likewise.
4358 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
4359 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
4360 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
4361 * ifcvt.c (noce_try_store_flag_constants): Likewise.
4362 * tree-ssa-ccp.c (ccp_finalize): Likewise.
4363 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
4364 * builtins.c (fold_builtin_3): Likewise.
4365 * graphite-scop-detection.c
4366 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
4367 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
4370 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
4373 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
4374 parameter. Return true if ignore_decl_rtl_set_p is true and other
4375 conditions are satisfied.
4376 * asan.h (asan_protect_global): Add new parameter.
4377 * varasm.c (categorize_decl_for_section): Pass true as second parameter
4378 to asan_protect_global calls.
4380 2017-11-30 Jim Wilson <jimw@sifive.com>
4382 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
4383 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
4384 -mstrict-align, add info on default value. Delete redundant lines for
4385 -mabi. Add missing -mexplicit-relocs docs.
4387 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4389 * config/arc/arc.md (trap): New pattern.
4391 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4393 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
4394 instruction to end into a delay slot.
4395 * config/arc/arc.md (cond_delay_insn): Check if the instruction
4396 can be placed into a delay slot against reg_note.
4398 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4400 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
4401 labels number of usages.
4403 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4405 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
4407 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
4409 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4412 * internal-fn.c (expand_mul_overflow): Optimize unsigned
4413 multiplication by power of 2 constant into two shifts + comparison.
4415 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
4418 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
4420 2017-11-30 Richard Biener <rguenther@suse.de>
4422 PR tree-optimization/83202
4423 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
4424 allow_peel argument and guard peeling.
4425 (canonicalize_loop_induction_variables): Likewise.
4426 (canonicalize_induction_variables): Pass false.
4427 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
4428 peeling from cunrolli.
4430 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4432 * combine.c (try_combine): Print a message to dump file whenever
4433 I0, I1, or I2 cannot be combined into I3.
4435 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4437 PR rtl-optimization/83156
4438 PR rtl-optimization/82621
4439 * combine.c (try_combine): Don't split an I2 if one of the dests is
4440 set again before I3. Allow unused dests.
4442 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4444 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
4446 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
4448 PR rtl-optimization/80818
4449 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
4450 recursively. Use insn code for clobber.
4451 (lra_set_insn_recog_data): Pass the new arg to
4452 collect_non_operand_hard_regs.
4453 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
4455 (lra_update_insn_regno_info): Pass insn to
4456 add_regs_to_insn_regno_info.
4458 2017-11-29 Jim Wilson <jimw@sifive.com>
4459 Andrew Waterman <andrew@sifive.com>
4461 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
4462 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
4463 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
4464 use SINGLE_SHIFT_COST.
4465 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
4466 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
4468 2017-11-29 Julia Koval <julia.koval@intel.com>
4470 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
4471 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
4472 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
4473 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
4474 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
4475 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
4476 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
4477 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
4478 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
4479 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
4480 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
4481 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
4482 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
4483 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
4484 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
4485 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
4486 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
4487 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
4488 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
4489 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
4490 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
4491 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
4492 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
4493 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
4494 __builtin_ia32_vpshldv_v2di_mask,
4495 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
4496 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
4497 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
4499 2017-11-29 Julia Koval <julia.koval@intel.com>
4501 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
4502 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
4503 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
4504 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
4505 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
4506 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
4507 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
4508 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
4509 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
4510 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
4511 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
4512 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
4513 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
4514 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
4515 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
4516 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
4517 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
4518 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
4519 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4520 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
4521 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
4523 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4525 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
4526 movsi_pic_gotdata_op instruction as a load for the UT699 errata
4529 2017-11-29 Martin Aberg <maberg@gaisler.com>
4531 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4532 to prevent b2bst errata sequence.
4533 (sqrtdf2_fix): Likewise.
4535 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4537 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
4538 (div_sqrt_insn_p): New function.
4539 (sparc_do_work_around_errata): Insert NOP instructions to
4540 prevent sequences that could trigger the TN-0013 errata for
4541 certain LEON3 processors.
4542 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
4543 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
4544 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
4545 (in_branch_delay): Prevent div and sqrt in delay slot if
4547 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
4549 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4551 * config/sparc/sparc.c (atomic_insn_p): New function.
4552 (sparc_do_work_around_errata): Insert NOP instructions to
4553 prevent sequences that could trigger the TN-0010 errata for
4555 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
4556 instruction referable in atomic_insns_p.
4558 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4560 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
4561 (atomic_compare_and_swap_leon3_1): Likewise.
4564 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4566 * config/sparc/sparc.c (fpop_insn_p): New function.
4567 (sparc_do_work_around_errata): Insert NOP instructions to
4568 prevent sequences that could trigger the TN-0012 errata for
4570 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
4571 * config/sparc/sparc.md (fix_gr712rc): New attribute.
4572 (in_branch_annul_delay): Prevent floating-point instructions
4573 in delay slot of annulled integer branch.
4575 2017-11-29 Richard Biener <rguenther@suse.de>
4577 PR tree-optimization/83202
4578 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4580 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4581 nodes vectorized to the same stmts multiple times.
4582 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4583 (vect_analyze_slp_instance): Adjust.
4584 (scalar_stmts_to_slp_tree_map_t): New typedef.
4585 (vect_schedule_slp_instance): Add a map recording the SLP node
4586 representing the vectorized stmts for a set of scalar stmts.
4587 Avoid code-generating redundancies.
4588 (vect_schedule_slp): Allocate map and pass it down.
4590 2017-11-29 Nathan Sidwell <nathan@acm.org>
4593 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4596 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4598 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4599 cases for printing LDP/STP memory addresses.
4600 (aarch64_print_address_internal): Renamed from
4601 aarch64_print_operand_address, added parameter, add Pmode check.
4602 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4603 (aarch64_print_operand_address): Indirect to
4604 aarch64_print_address_internal.
4605 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4606 'y' operand output specifier.
4608 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4611 * tree.c (build_simple_mem_ref_loc): Handle
4612 get_addr_base_and_unit_offset returning a MEM_REF.
4615 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4618 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4621 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4623 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4625 (layout::move_to_column): Likewise.
4627 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4630 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4631 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4633 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4634 Martin Jambor <mjambor@suse.cz>
4637 * tree.h (expr_type_first_operand_type_p): Declare
4638 * tree.c (expr_type_first_operand_type_p): New function.
4639 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4640 (ipa_value_from_jfunc): Adjust declaration.
4641 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4642 Use it as result type for arithmetics, unless it is NULL in which case
4643 be more conservative.
4644 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4645 ipa_get_jf_pass_through_result.
4646 (propagate_vals_across_pass_through): Likewise.
4647 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4648 is to propagate_vals_across_pass_through.
4649 (propagate_constants_across_call): Pass PARM_TYPE to
4650 propagate_scalar_across_jump_function.
4651 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4652 ipa_value_from_jfunc.
4653 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4654 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4655 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4656 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4657 target_type, pass it to ipa_value_from_jfunc.
4658 (update_indirect_edges_after_inlining): Pass parameter type to
4659 try_make_edge_direct_simple_call.
4661 2017-11-28 Jeff Law <law@redhat.com>
4663 * gimple-ssa-evrp-analyze.c
4664 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4665 refine ranges if scev_initialized_p returns true.
4666 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4668 2017-11-28 Julia Koval <julia.koval@intel.com>
4670 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4671 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4672 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4673 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4674 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4675 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4676 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4677 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4678 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4679 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4680 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4681 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4682 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4683 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4684 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4685 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4686 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4687 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4688 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4689 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4690 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4691 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4693 2017-11-28 Julia Koval <julia.koval@intel.com>
4695 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4696 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4697 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4698 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4699 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4700 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4701 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4702 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4703 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4704 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4705 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4706 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4707 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4708 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4709 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4710 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4711 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4712 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4713 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4714 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4715 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4716 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4717 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4718 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4719 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4720 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4721 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4722 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4723 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4724 __builtin_ia32_vpshld_v2di_mask): New builtins.
4725 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4726 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4728 2017-11-28 Richard Biener <rguenther@suse.de>
4730 PR tree-optimization/80776
4731 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4733 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4735 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4736 If the incoming edge is an effective fallthru because the other
4737 edge only reaches a __builtin_unreachable () then record ranges
4738 derived from the controlling condition in SSA info.
4739 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4740 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4742 2017-11-28 Olivier Hainque <hainque@adacore.com>
4744 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4745 on mingw build hosts.
4747 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4750 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4752 2017-11-28 Richard Biener <rguenther@suse.de>
4755 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4756 copies generated from memcpy use a character array as reference
4759 2017-11-28 Julia Koval <julia.koval@intel.com>
4760 Sebastian Peryt <sebastian.peryt@intel.com>
4762 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4763 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4764 cilk-common.o, cilk.h, cilk-common.c): Remove.
4766 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4767 * builtins.c (is_builtin_name): Remove cilkplus condition.
4768 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4769 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4770 cilk-builtins.def, cilkplus.def): Remove.
4771 * cif-code.def (CILK_SPAWN): Remove.
4772 * cilk-builtins.def: Delete.
4773 * cilk-common.c: Ditto.
4775 * cilkplus.def: Ditto.
4776 * config/darwin.h (fcilkplus): Delete.
4777 * cppbuiltin.c: Ditto.
4778 * doc/extend.texi: Remove cilkplus doc.
4779 * doc/generic.texi: Ditto.
4780 * doc/invoke.texi: Ditto.
4781 * doc/passes.texi: Ditto.
4782 * gcc.c (fcilkplus): Remove.
4783 * gengtype.c (cilk.h): Remove.
4784 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4786 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4788 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4790 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4791 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4792 cilkplus conditions.
4793 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4794 inline_read_section): Ditto.
4795 * ipa-inline-analysis.c (cilk.h): Remove.
4796 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4797 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4798 append_linker_options): Remove condition for fcilkplus.
4799 * lto/lto-lang.c (cilk.h): Remove.
4800 (lto_init): Remove condition for fcilkplus.
4801 * omp-expand.c (expand_cilk_for_call): Delete.
4802 (expand_omp_taskreg, expand_omp_for_static_chunk,
4803 expand_omp_for): Remove cilkplus
4805 (expand_cilk_for): Delete.
4806 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4807 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4808 execute_lower_omp, diagnose_sb_0): Ditto.
4809 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4810 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4811 * tree-nested.c: Ditto.
4812 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4813 (dump_generic_node): Ditto.
4814 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4815 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4817 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4819 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4821 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
4822 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
4823 for vector mode and !TARGET_SIMD.
4825 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4827 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
4829 * tree.h (SWITCH_LABELS): Remove.
4830 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
4831 assert SWITCH_BODY is non-NULL.
4832 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
4834 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
4836 PR tree-optimization/80788
4837 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
4838 has TREE_OVERFLOW set, call drop_tree_overflow.
4840 2017-11-28 Richard Biener <rguenther@suse.de>
4842 PR tree-optimization/83158
4843 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
4845 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
4848 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
4849 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
4851 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
4853 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
4854 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
4856 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4859 * varasm.c (compare_constant): Take the mode of the constants into
4860 account when comparing floating point constants.
4862 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
4864 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
4865 from explicit instantiation of debug_helper.
4866 * vec.h (DEFINE_DEBUG_VEC): Ditto.
4868 2017-11-27 Richard Biener <rguenther@suse.de>
4870 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
4873 2017-11-27 Richard Biener <rguenther@suse.de>
4875 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
4876 (type_hash_canon): Also clear int_cst_hash_table entry for
4878 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
4880 2017-11-27 Tamar Christina <tamar.christina@arm.com>
4882 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4883 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4884 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4886 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4888 * hash-map.h (gt_cleare_cache): Avoid UB.
4890 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
4892 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
4893 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
4894 (decide_unroll_runtime_iterations): Likewise.
4895 (decide_unroll_stupid): Likewise.
4897 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4900 * config/i386/i386.md: Add a loop with incssp.
4902 2017-11-27 Martin Jambor <mjambor@suse.cz>
4904 PR tree-optimization/81248
4905 * tree-sra.c (splice_param_accesses): Remove size check.
4906 (decide_one_param_reduction): Fix size check.
4907 * gimple-pretty-print.c (dump_profile): Silence warning.
4908 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
4910 2017-11-27 Jakub Jelinek <jakub@redhat.com>
4913 * dbxout.c (lastlineno): New variable.
4914 (dbx_debug_hooks): Use dbxout_switch_text_section as
4915 switch_text_section debug hook.
4916 (dbxout_function_end): Switch to current_function_section
4917 rather than function_section. If crtl->has_bb_partition,
4918 output just one N_FUN, depending on in_cold_section_p.
4919 (dbxout_source_line): Remember last lineno in lastlineno.
4920 (dbxout_switch_text_section): New function.
4921 (dbxout_function_decl): Adjust dbxout_block caller.
4922 (dbx_block_with_cold_children): New function.
4923 (dbxout_block): Return true if any LBRAC/RBRAC have been
4924 emitted. Use dbx_block_with_cold_children at depth == 0
4925 in second partition. Add PARENT_BLOCKNUM argument, pass
4926 it optionally adjusted to children. Output LBRAC/RBRAC
4927 around recursive call only if the block is in the current
4928 partition, if not and anything was output, emit empty
4930 * final.c (final_scan_insn): Compute cold_function_name
4931 before calling switch_text_section debug hook. Call
4932 that hook even if dwarf2out_do_frame if not emitting
4936 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
4937 TREE_READONLY decls.
4939 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4941 PR rtl-optimization/82488
4942 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
4944 2017-11-26 Julia Koval <julia.koval@intel.com>
4946 * config/i386/i386.c (processor_target_table): Add skylake_cost for
4948 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
4951 2017-11-26 Julia Koval <julia.koval@intel.com>
4953 * config/i386/driver-i386.c (host_detect_local_cpu):
4954 Detect skylake-avx512.
4956 2017-11-26 Julia Koval <julia.koval@intel.com>
4958 * config.gcc: Add -march=cannonlake.
4959 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
4960 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
4961 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
4962 (PTA_CANNONLAKE): New.
4963 (processor_target_table): Add cannonlake.
4964 (ix86_option_override_internal): Ditto.
4965 (fold_builtin_cpu): Ditto.
4966 (get_builtin_code_for_version): Handle cannonlake.
4967 (M_INTEL_COREI7_CANNONLAKE): New.
4968 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
4969 * doc/invoke.texi: Add -march=cannonlake.
4971 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
4973 * plugin.c (add_new_plugin): Use platform-specific library extensions.
4974 (try_init_one_plugin): Alternative implementation for MinGW.
4975 * Makefile.in (plugin_implib): New.
4976 (gengtype-lex.c): Fix broken AIX workaround.
4977 * configure: Regenerate.
4978 * doc/plugins.texi: Document support for MinGW.
4980 2017-11-25 Jakub Jelinek <jakub@redhat.com>
4982 PR rtl-optimization/81553
4983 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
4984 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
4985 is a shift where C1 has different mode than the whole shift, use C1's
4986 mode for MULT rather than the shift's mode.
4989 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
4990 builtins not enabled in the currently selected ISA.
4992 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
4994 PR tree-optimization/71026
4995 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
4996 (insert_reciprocals): Change to insert reciprocals before a division
4997 by a square and to insert the square of a reciprocal.
4998 (execute_cse_reciprocals_1): Change to consider division by a square.
4999 (register_division_in): Add importance parameter.
5001 2017-11-24 Richard Biener <rguenther@suse.de>
5003 PR tree-optimization/82402
5004 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5005 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5007 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5009 * match.pd (0-ptr): New transformation.
5011 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5014 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5015 when skipping checking.
5017 2017-11-24 Richard Biener <rguenther@suse.de>
5019 PR tree-optimization/83128
5020 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5021 (vn_reference_lookup_3): Likewise.
5023 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5026 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5027 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5030 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5031 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5032 check for the latter with gimple_call_builtin_p. Do not handle
5033 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5035 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5037 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5040 2017-11-23 Julia Koval <julia.koval@intel.com>
5042 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5043 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5044 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5045 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5046 _mm512_maskz_expandloadu_epi16): New intrinsics.
5047 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5048 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5049 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5050 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5051 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5052 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5053 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5054 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5055 _mm256_maskz_expandloadu_epi8): New intrinsics.
5056 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5057 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5058 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5059 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5060 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5061 * config/i386/sse.md (VI248_VLBW): New iterator.
5062 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5064 2017-11-23 Julia Koval <julia.koval@intel.com>
5066 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
5067 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
5068 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
5069 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
5070 _mm512_mask_compressstoreu_epi16): New.
5071 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
5072 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
5073 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
5074 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
5075 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
5076 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
5077 _mm256_mask_compressstoreu_epi8): New.
5078 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
5079 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
5080 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
5081 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
5082 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
5083 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
5084 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
5085 __builtin_ia32_compresshi128_mask,
5086 __builtin_ia32_compressstoreuqi512_mask,
5087 __builtin_ia32_compressstoreuhi512_mask,
5088 __builtin_ia32_compressstoreuqi256_mask,
5089 __builtin_ia32_compressstoreuqi128_mask,
5090 __builtin_ia32_compressstoreuhi256_mask,
5091 __builtin_ia32_compressstoreuhi128_mask): New builtins.
5092 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
5094 (ix86_expand_special_args_builtin): Handle new types.
5095 (s4fma_expand): Handle new builtin array.
5096 * config/i386/immintrin.h: Include new headers.
5097 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
5098 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
5100 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5102 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
5103 fixes. Declare temp and g variables at the top in order to avoid
5104 {} in most of the cases.
5106 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
5108 * match.pd (ptr-0): New transformation.
5110 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5112 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
5113 (struct addr_mode_cost_table): New.
5114 (struct tune_params): Add field addr_mode_costs.
5115 * config/arm/arm.c (generic_addr_mode_costs): New.
5116 (arm_slowmul_tune): Initialise addr_mode_costs field.
5117 (arm_fastmul_tune): Likewise.
5118 (arm_strongarm_tune): Likewise.
5119 (arm_xscale_tune): Likewise.
5120 (arm_9e_tune): Likewise.
5121 (arm_marvell_pj4_tune): Likewise.
5122 (arm_v6t2_tune): Likewise.
5123 (arm_cortex_tune): Likewise.
5124 (arm_cortex_a8_tune): Likewise.
5125 (arm_cortex_a7_tune): Likewise.
5126 (arm_cortex_a15_tune): Likewise.
5127 (arm_cortex_a35_tune): Likewise.
5128 (arm_cortex_a53_tune): Likewise.
5129 (arm_cortex_a57_tune): Likewise.
5130 (arm_exynosm1_tune): Likewise.
5131 (arm_xgene1_tune): Likewise.
5132 (arm_cortex_a5_tune): Likewise.
5133 (arm_cortex_a9_tune): Likewise.
5134 (arm_cortex_a12_tune): Likewise.
5135 (arm_cortex_a73_tune): Likewise.
5136 (arm_v7m_tune): Likewise.
5137 (arm_cortex_m7_tune): Likewise.
5138 (arm_v6m_tune): Likewise.
5139 (arm_fa726te_tune): Likewise.
5140 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
5142 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5144 * config/arm/arm.c (arm_mem_costs): New function.
5145 (arm_rtx_costs_internal): Use arm_mem_costs.
5147 2017-11-23 Mark Wielaard <mark@klomp.org>
5149 * dwarf2out.c (init_sections_and_labels): Use generation to create
5150 unique ranges_section_label and ranges_base_label. Return generation.
5151 (output_rnglists): Add generation argument. Use generation to create
5152 unique ranges labels.
5153 (dwarf2out_finish): Get generation from init_sections_and_labels
5154 and pass generation to output_rnglists.
5156 2017-11-23 Mike Stump <mikestump@comcast.net>
5157 Eric Botcazou <ebotcazou@adacore.com>
5159 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
5160 * cfgloop.h (struct loop): Add unroll field.
5161 * function.h (struct function): Add has_unroll bitfield.
5162 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
5163 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
5164 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
5165 (pass_rtl_unroll_loops::gate): Likewise.
5166 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
5167 for which loop->unroll==1.
5168 (decide_unroll_constant_iterations): Use note for consistency and
5169 take loop->unroll into account. Return early if loop->unroll is set.
5170 Fix thinko in existing test.
5171 (decide_unroll_runtime_iterations): Use note for consistency and
5172 take loop->unroll into account.
5173 (decide_unroll_stupid): Likewise.
5174 * lto-streamer-in.c (input_cfg): Read loop->unroll.
5175 * lto-streamer-out.c (output_cfg): Write loop->unroll.
5176 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
5178 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
5179 (print_loop): Print loop->unroll if set.
5180 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
5181 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
5182 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
5184 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
5185 loop->unroll is set and smaller than the trip count. Otherwise bypass
5186 entirely the heuristics if loop->unroll is set. Remove dead note.
5187 Fix off-by-one bug in other note.
5188 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
5189 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
5191 (tree_unroll_loops_completely): Make static.
5192 (pass_complete_unroll::execute): Use correct type for variable.
5193 (pass_complete_unrolli::execute): Fix formatting.
5194 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
5196 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
5198 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
5199 enable when TARGET_PREFER_AVX128 is set.
5201 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5203 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
5205 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
5206 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
5208 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5210 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
5211 frequencies. Fix estimation of aggregate parameters.
5213 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5215 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
5218 2017-11-23 Tom de Vries <tom@codesourcery.com>
5220 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
5222 2017-11-23 Tom de Vries <tom@codesourcery.com>
5224 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
5227 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
5230 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
5231 sibcall_value_pcrel_fdpic): Use local variable instead of
5233 (calli_tbr_rel): Add missing operand 2.
5234 (call_valuei_tbr_rel): Add missing operand 3.
5236 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5239 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
5240 bitpos/bitsize covering the whole destination, use store_expr only if
5241 the complex mode is the same. Otherwise, use expand_normal and if
5242 it returns CONCAT, subreg each part separately instead of trying to
5243 subreg the whole result.
5245 2017-11-23 Richard Biener <rguenther@suse.de>
5247 PR tree-optimization/23094
5248 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
5249 come here from walking over backedges in the first iteration.
5250 (vn_reference_lookup_3): Skip clobbers that store the same value.
5252 2017-11-23 Richard Biener <rguenther@suse.de>
5254 PR tree-optimization/81403
5255 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
5256 a block we need a leader relative to.
5257 (phi_translate_1): For nary processing require a leader from
5258 get_representative_for given we run expression simplification
5259 using match-and-simplify. Remove previous fix.
5261 2017-11-22 Jeff Law <law@redhat.com>
5263 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
5264 Use new method allocate_value_range rather than accessing the
5265 vrp_value_range_pool data member directly.
5266 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
5267 to use extract_range_from_stmt method to avoid need for
5268 extract_range_from_assignment method.
5269 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
5270 method rather than setting values_propgated data member directly.
5271 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
5272 and values propagated data members and extract_range_from_assignment
5273 method. Reorder private data members to conform to standards.
5274 Add new methods set_lattice_propagation_complete and
5275 allocate_value_range.
5277 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
5279 PR rtl-optimization/83030
5280 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
5281 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
5282 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
5283 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
5284 CROSSING_JUMP_P flag is already set before setting it.
5285 * cfgrtl.c (fixup_partition_crossing): Likewise.
5286 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
5289 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5291 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
5292 Handle the case where both arguments are using gen_const_vec_series.
5294 2017-11-22 David Malcolm <dmalcolm@redhat.com>
5297 * pretty-print.c (pp_format): Move quoting implementation to
5298 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
5299 to pass address of "quote" local.
5300 (pp_begin_quote): New function.
5301 (pp_end_quote): New function.
5302 * pretty-print.h (printer_fn): Convert penultimate param from bool
5304 (pp_begin_quote): New decl.
5305 (pp_end_quote): New decl.
5306 * tree-diagnostic.c (default_tree_printer): Convert penultimate
5307 param from bool to bool *.
5308 * tree-diagnostic.h (default_tree_printer): Likewise.
5310 2017-11-22 Jeff Law <law@redhat.com>
5312 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
5313 Initialize vr_values.
5314 (evrp_range_analyzer::try_find_new_range): Call methods attached to
5315 vr_values via vr_values class instance rather than delegators.
5316 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
5317 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5318 (evrp_range_analyzer::push_value_range): Likewise.
5319 (evrp_range_analyzer::pop_value_range): Likewise.
5320 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
5321 most delegators. Those remaining are exposed as public interfaces.
5322 Make vr_values a pointer and private.
5323 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
5325 (evrp_range_analyzer::get_vr_value): New method.
5326 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
5327 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
5328 (class evrp_dom_walker): Attach evrp_folder class, initialize
5329 it in the ctor. Remove temporary delegators.
5330 (evrp_dom_walker::before_dom_children): Call methods in attached
5331 evrp_range_analyzer class via class instance pointer. Use
5332 free value_range_constant_singleton to remove need for
5333 op_with_constant_singleton_value delegator method. Do not
5334 create a vrp_prop class instance for every call! Narrow
5335 scope of a couple variables.
5336 (evrp_dom_walker::cleanup): Call methods in attached
5337 evrp_range_analyzer class via class instance pointer.
5338 * vr-values.h (class vr_values): Privatize many methods and
5341 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5343 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
5344 * cfgexpand.c (expand_debug_expr): Remove handling for them.
5345 * expr.c (expand_expr_real_2): Likewise.
5346 * fold-const.c (const_unop): Likewise.
5347 * optabs-tree.c (optab_for_tree_code): Likewise.
5348 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
5349 * tree-inline.c (estimate_operator_cost): Likewise.
5350 * tree-pretty-print.c (dump_generic_node): Likewise.
5351 (op_code_prio): Likewise.
5352 (op_symbol_code): Likewise.
5353 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
5354 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
5355 * internal-fn.c (direct_internal_fn_optab): New function.
5356 (direct_internal_fn_array, direct_internal_fn_supported_p
5357 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
5358 * fold-const-call.c (fold_const_reduction): New function.
5359 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
5361 * tree-vect-loop.c: Include internal-fn.h.
5362 (reduction_code_for_scalar_code): Rename to...
5363 (reduction_fn_for_scalar_code): ...this and return an internal
5365 (vect_model_reduction_cost): Take an internal_fn rather than
5367 (vect_create_epilog_for_reduction): Likewise. Build calls rather
5369 (vectorizable_reduction): Use internal functions rather than tree
5370 codes for the reduction operation. Update calls to the functions
5372 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
5373 Use calls to internal functions rather than REDUC tree codes.
5374 * config/aarch64/aarch64-simd.md: Update comment accordingly.
5376 2017-11-22 Olivier Hainque <hainque@adacore.com>
5378 * config/vxworks.c (vxworks_override_options): Pick default
5379 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
5380 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
5381 DWARF_GNAT_ENCODINGS_DEFAULT.
5382 * config/vxworksae.h: Likewise.
5384 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
5386 PR tree-optimization/83104
5387 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
5390 2017-11-22 Marek Polacek <polacek@redhat.com>
5391 H.J. Lu <hongjiu.lu@intel.com>
5392 Jason Merrill <jason@redhat.com>
5397 * calls.c (initialize_argument_information): Call
5398 warn_parameter_passing_abi target hook.
5399 (store_one_arg): Use 0 for empty record size. Don't push 0 size
5400 argument onto stack.
5401 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
5402 * common.opt: Update -fabi-version description.
5403 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
5404 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
5406 (ix86_is_empty_record): New function.
5407 (ix86_warn_parameter_passing_abi): New function.
5408 (TARGET_EMPTY_RECORD_P): Redefine.
5409 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
5410 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
5411 * doc/tm.texi: Regenerated.
5412 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
5413 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
5414 * dwarf2out.c (get_ultimate_context): Move to tree.c.
5415 * explow.c (hard_function_value): Call arg_int_size_in_bytes
5416 instead of int_size_in_bytes.
5417 * expr.c (copy_blkmode_to_reg): Likewise.
5418 * function.c (aggregate_value_p): Return 0 for empty types.
5419 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
5421 (locate_and_pad_parm): Call arg size_in_bytes instead
5423 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
5424 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
5425 * target.def (empty_record_p, warn_parameter_passing_abi): New target
5427 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
5428 (std_gimplify_va_arg_expr): Skip empty records. Call
5429 arg_size_in_bytes instead size_in_bytes.
5430 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
5431 * tree-core.h (tree_type_common): Add empty_flag.
5432 (tree_decl_common): Update comments.
5433 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
5435 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5436 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
5438 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5439 * tree.c (default_is_empty_type): New function.
5440 (default_is_empty_record): New function.
5441 (arg_int_size_in_bytes): New function.
5442 (arg_size_in_bytes): New function.
5443 (get_ultimate_context): New function.
5444 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
5445 TRANSLATION_UNIT_WARN_EMPTY_P.
5446 (default_is_empty_record, arg_int_size_in_bytes,
5447 arg_size_in_bytes, get_ultimate_context): Declare.
5449 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5451 * config/arm/arm.c (cmse_clear_registers): New function.
5452 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
5453 code by call to cmse_clear_registers.
5454 (cmse_nonsecure_entry_clear_before_return): Likewise.
5456 2017-11-22 Tamar Christina <tamar.christina@arm.com>
5458 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
5459 (vdot_s32, vdotq_s32): New.
5460 (vdot_lane_u32, vdotq_lane_u32): New.
5461 (vdot_lane_s32, vdotq_lane_s32): New.
5464 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5467 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
5468 for unsigned values with fewer HWIs than the precision.
5469 (test_overflow): New function.
5470 (wide_int_cc_tests): Call it.
5472 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5474 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
5475 has the same mode class as Pmode.
5477 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5479 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
5480 padding_bits_to_clear_ptr.
5481 (cmse_nonsecure_entry_clear_before_return): Likewise.
5483 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5485 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
5486 auto_sbitap instead of integer bitfield to control register needing
5489 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5491 PR tree-optimization/83044
5492 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
5493 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
5494 that need the upper bound. Subtract offset from
5495 get_addr_base_and_unit_offset only if positive and subtract it
5496 before division by eltsize rather than after it.
5499 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
5500 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
5501 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
5505 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
5506 expand_binop with *mul_widen_optab, make sure at least one of the
5507 operands doesn't have VOIDmode.
5510 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
5512 PR rtl-optimization/82044
5513 PR tree-optimization/82042
5514 * dse.c (record_store): Check for overflow.
5515 (check_mem_read_rtx): Properly check for overflow if width == -1, call
5516 add_wild_read instead of clear_rhs_from_active_local_stores on
5517 overflow and log it into dump_file.
5519 2017-11-22 Richard Biener <rguenther@suse.de>
5521 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
5522 fake edges to exit when looking for a place to insert.
5523 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
5525 (insert_into_preds_of_block): Commit edge insertion immediately,
5526 assert that doesn't require new BBs.
5527 (fini_pre): Release expressions.
5528 (pass_pre::execute): Shuffle things around a bit, if the fn
5529 is too large do not compute AVAIL either as this is really the
5532 2017-11-22 Richard Biener <rguenther@suse.de>
5534 PR tree-optimization/83089
5535 * tree-if-conv.c (pass_if_conversion::execute): If anything
5536 changed reset SCEV and free the number of iteration estimates.
5538 2017-11-21 Martin Sebor <msebor@redhat.com>
5540 PR tree-optimization/82945
5541 * calls.h (warn_nonstring_bound): Remove unused function.
5543 2017-11-21 Martin Sebor <msebor@redhat.com>
5545 PR tree-optimization/82945
5546 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
5547 * calls.h (maybe_warn_nonstring_arg): Declare new function.
5548 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
5550 (initialize_argument_information): Call maybe_warn_nonstring_arg.
5551 * calls.h (get_attr_nonstring_decl): Declare new function.
5552 * doc/extend.texi (attribute nonstring): Update.
5553 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
5554 get_attr_nonstring_decl and handle it.
5555 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
5556 detection of nul-termination.
5557 (strlen_to_stridx): Change to a pointer.
5558 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
5559 (pass_strlen::execute): Same.
5561 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
5563 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
5564 for the new option -mprefer-vector-width=[none|128|256|512].
5565 * config/i386/i386.c (ix86_target_string): remove old style options
5566 -mprefer-avx256 and make -mprefer-avx128 as alias.
5567 (ix86_option_override_internal): Apply defaults for the
5568 -mprefer-vector-width=[128|256] option.
5569 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
5570 Implement macros to work with -mprefer-vector-width=.
5571 * config/i386/i386.opt: Implemented option
5572 -mprefer-vector-width=[none|128|256|512].
5573 * doc/invoke.texi: Documentation for
5574 -mprefer-vector-width=[none|128|256|512].
5576 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5578 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5580 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5582 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5584 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5586 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5588 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5589 * expr.c (expand_expr_real_2): Likewise.
5590 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5591 fold_binary_loc): Likewise.
5592 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5593 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5594 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5595 MINUS_EXPR transformations.
5596 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5597 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5598 * tree-inline.c (estimate_operator_cost): Likewise.
5599 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5600 op_symbol_code): Likewise.
5601 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5602 * vr-values.c (extract_range_from_binary_expr): Likewise.
5603 * varasm.c (initializer_constant_valid_p_1): Likewise.
5604 * tree.def: New tree code POINTER_DIFF_EXPR.
5606 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5608 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5609 integer suffix to movbe mnemonic.
5610 (*bswaphi2_movbe): Ditto.
5611 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5613 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5616 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5617 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5618 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5620 2017-11-21 Martin Liska <mliska@suse.cz>
5622 * tree-inline.c (expand_call_inline): Remove not needed
5625 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5626 Jakub Jelinek <jakub@redhat.com>
5629 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5630 Remove static keyword from f variable.
5632 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5634 PR tree-optimization/83086
5635 * gimple-ssa-store-merging.c
5636 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5637 rather than n.base_addr.
5639 2017-11-21 Martin Liska <mliska@suse.cz>
5641 PR rtl-optimization/82044
5642 PR tree-optimization/82042
5643 * dse.c (check_mem_read_rtx): Check for overflow.
5645 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5647 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5650 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5652 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5653 and its corresponding call arg location note.
5655 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5656 Andrew Burgess <andrew.burgess@embecosm.com>
5658 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5660 (arc_return_slot_offset): Likewise.
5661 (arc_eh_return_address_location): New declaration.
5662 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5663 (MUST_SAVE_REGISTER): Add exception handler case.
5664 (MUST_SAVE_RETURN_ADDR): Likewise.
5665 (arc_frame_pointer_required): Likewise.
5666 (arc_frame_pointer_needed): New function.
5667 (arc_compute_frame_size): Changed.
5668 (arc_expand_prologue): Likewise.
5669 (arc_expand_epilogue): Likewise.
5670 (arc_initial_elimination_offset): Likewise.
5671 (arc_return_slot_offset): Delete.
5672 (arc_eh_return_address_location): New function.
5673 (arc_builtin_setjmp_frame_value): Likewise.
5674 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5675 (EH_RETURN_STACKADJ_RTX): Define.
5676 (EH_RETURN_HANDLER_RTX): Likewise.
5677 * config/arc/arc.md (eh_return): Delete.
5679 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5681 * print-tree.h (debug_vec_tree): Remove prototype.
5682 * gdbinit.in (pvt): Remove macro.
5684 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5686 PR tree-optimization/83047
5687 * gimple-ssa-store-merging.c
5688 (imm_store_chain_info::output_merged_store): If the loads with the
5689 same vuse are in different basic blocks, for load_gsi pick a load
5690 location that is dominated by the other loads.
5693 * config/i386/i386.c (ix86_memmodel_check): Start
5694 -Winvalid-memory-model diagnostics with lowercase letter.
5697 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5698 set in_cold_section_p to the partition containing loc_list->first.
5699 When seeing loc_list->last_before_switch node, update secname and
5700 perform range_across_switch second partition handling only after that.
5703 * run-rtl-passes.c: Include debug.h.
5704 (run_rtl_passes): Call debug_hooks->assembly_start.
5705 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5709 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5710 OPTAB_DIRECT in calls to expand_simple_binop.
5712 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5715 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5717 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5720 * spellcheck.h (best_match::blithely_get_best_candidate): New
5723 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5725 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5726 with lower case letter.
5728 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5730 * config/i386/i386.md (bswaphi2): New expander.
5731 (*bswaphi2_movbe): New insn pattern.
5732 (bswaphi -> rorhi pepehole2): New peephole pattern.
5734 2017-11-20 Jeff Law <law@redhat.com>
5736 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5737 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5738 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5739 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5740 gimple-ssa-evrp-analyze.h.
5742 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5743 set BB_VISITED here.
5744 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5746 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5748 * tree-predcom.c: Add general comment on Store-Store chains.
5749 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5751 (get_chain_last_ref_at): Rename into...
5752 (get_chain_last_write_at): ...this.
5753 (get_chain_last_write_before_load): New function.
5754 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5755 CT_STORE_STORE when write reference is added.
5756 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5757 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5758 (initialize_root_vars_store_elim_1): Ditto.
5759 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5760 definition is created.
5761 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5762 chain by replacing it with dominant stored value.
5764 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5766 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5768 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5770 * vr-values.c (extract_range_from_binary_expr): Use a full range
5773 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5775 * config/arm/arm.md (R4_REGNUM): Define constant.
5776 (nonsecure_call_internal): Remove r4 clobber.
5777 (nonsecure_call_value_internal): Likewise.
5778 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5779 clobber and resequence match_operands.
5780 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5781 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5782 (nonsecure_call_value_reg_thumb2): Likewise.
5784 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5786 PR tree-optimization/78821
5787 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5788 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5789 to the address of the base rather than the base itself.
5790 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5791 (find_bswap_or_nop_finalize): New function.
5792 (find_bswap_or_nop): Use it.
5793 (bswap_replace): Return a tree rather than bool, change first
5794 argument from gimple * to gimple_stmt_iterator, allow inserting
5795 into an empty sequence, allow ins_stmt to be NULL - then emit
5796 all stmts into gsi. Fix up MEM_REF address gimplification.
5797 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5798 (struct store_immediate_info): Add N and INS_STMT non-static
5800 (store_immediate_info::store_immediate_info): Initialize them
5801 from newly added ctor args.
5802 (merged_store_group::apply_stores): Formatting fixes. Sort by
5804 (stmts_may_clobber_ref_p): For stores call also
5805 refs_anti_dependent_p.
5806 (gather_bswap_load_refs): New function.
5807 (imm_store_chain_info::try_coalesce_bswap): New method.
5808 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5809 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5810 (imm_store_chain_info::output_merged_store): Fail if number of
5811 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5812 and NOP_EXPR rhs_code.
5813 (pass_store_merging::process_store): Compute n and ins_stmt, if
5814 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5815 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5817 (pass_store_merging::execute): Calculate dominators.
5819 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5820 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5821 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
5822 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5823 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
5824 class pass_optimize_bswap, bswap_replace,
5825 pass_optimize_bswap::execute): Moved to ...
5826 * gimple-ssa-store-merging.c: ... this file.
5827 Include optabs-tree.h.
5828 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
5829 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5830 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
5831 anonymous namespace, remove static keywords.
5832 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
5833 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
5835 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
5838 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
5840 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
5842 * vec.h (debug_helper): New function.
5843 (DEFINE_DEBUG_VEC): New macro.
5844 * hash-set.h (debug_helper): New function.
5845 (DEFINE_DEBUG_HASH_SET): New macro.
5846 * cfg.c (debug_slim (edge)): New function.
5847 Call DEFINE_DEBUG_VEC for edges.
5848 Call DEFINE_DEBUG_HASH_SET for edges.
5849 * cfghooks.c (debug_slim (basic_block)): New function.
5850 Call DEFINE_DEBUG_VEC for basic blocks.
5851 Call DEFINE_DEBUG_HASH_SET for basic blocks.
5852 * print-tree.c (debug_slim): New function to handle trees.
5853 Call DEFINE_DEBUG_VEC for trees.
5854 Call DEFINE_DEBUG_HASH_SET for trees.
5855 (debug (vec<tree, va_gc>) &): Remove.
5856 (debug (<vec<tree, va_gc>) *): Remove.
5857 * print-rtl.c (debug_slim): New function to handle const_rtx.
5858 Call DEFINE_DEBUG_VEC for rtx_def.
5859 Call DEFINE_DEBUG_VEC for rtx_insn.
5860 Call DEFINE_DEBUG_HASH_SET for rtx_def.
5861 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
5862 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
5863 (debug (vec<rtx_insn *> *ptr): Remove.
5864 (debug_insn_vector): Remove.
5865 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
5867 2017-11-20 Tom de Vries <tom@codesourcery.com>
5869 PR rtl-optimization/82020
5870 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
5871 IF_THEN_ELSE condition.
5873 2017-11-19 Jeff Law <law@redhat.com>
5875 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
5876 of degenerates resulting from ignoring an edge.
5878 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5881 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
5883 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5886 * profile-count.c (profile_count::to_sreal_scale): Fix return value
5887 for uninitialied counts.
5889 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5892 * tree-inline.c (estimate_num_insns): Set to 1 at least.
5894 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5897 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
5899 2017-11-19 Tom de Vries <tom@codesourcery.com>
5901 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
5904 2017-11-19 Tom de Vries <tom@codesourcery.com>
5906 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
5909 2017-11-19 Tom de Vries <tom@codesourcery.com>
5911 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5914 2017-11-19 Tom de Vries <tom@codesourcery.com>
5916 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
5918 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
5920 2017-11-19 Tom de Vries <tom@codesourcery.com>
5922 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
5924 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
5925 semicolon after MCORE_EXPORT_NAME call.
5927 2017-11-19 Tom de Vries <tom@codesourcery.com>
5930 * vmsdbgout.c (vmsdbgout_early_finish): New function.
5931 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
5933 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
5935 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
5936 (cgraph_edge::clone): Cleanup updating of profile.
5937 * ipa-cp.c (update_profiling_info): Likewise.
5938 * ipa-inline-transform.c (inline_transform): Likewise.
5939 * ipa-inline.c (inline_small_functions): Add missing space to dump.
5940 * ipa-split.c (execute_split_functions): Do not split when function
5942 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
5943 * profile-count.c (profile_count::dump): Add global0.
5944 (profile_count::to_cgraph_frequency): Do not ICE when entry is
5946 (profile_count::to_sreal_scale): Likewise.
5947 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
5948 (profile_count::combine_with_ipa_count): New function.
5949 * profile-count.h (profile_guessed_global0adjusted): New.
5950 (profile_count::adjusted_zero): New.
5951 (profile_count::global0adjusted): New.
5952 (profile_count::combine_with_ipa_count): New.
5953 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
5954 correct profile of return block of split functions.
5955 (copy_cfg_body): Remove unused profile_count.
5956 (copy_body): Likewise.
5957 (expand_call_inline): Update.
5958 (tree_function_versioning): Update.
5960 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
5962 * hash-set.h (hash_set::empty): New.
5963 * tree-ssa-threadbackward.h: Delete.
5964 * tree-ssa-threadbackward.c (class thread_jumps): New.
5965 Move max_threaded_paths into class.
5966 (fsm_find_thread_path): Remove arguments that are now in class.
5967 (profitable_jump_thread_path): Rename to...
5968 (thread_jumps::profitable_jump_thread_path): ...this.
5969 (convert_and_register_jump_thread_path): Rename to...
5970 (thread_jumps::convert_and_register_current_path): ...this.
5971 (check_subpath_and_update_thread_path): Rename to...
5972 (thread_jumps::check_subpath_and_update_thread_path): ...this.
5973 (register_jump_thread_path_if_profitable): Rename to...
5974 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
5975 (handle_phi): Rename to...
5976 (thread_jumps::handle_phi): ...this.
5977 (handle_assignment): Rename to...
5978 (thread_jumps::handle_assignment): ...this.
5979 (fsm_find_control_statement_thread_paths): Rename to...
5980 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
5981 (find_jump_threads_backwards): Rename to...
5982 (thread_jumps::find_jump_threads_backwards): ...this.
5983 Initialize path local data.
5984 (pass_thread_jumps::execute): Call find_jump_threads_backwards
5985 from within thread_jumps class.
5986 (pass_early_thread_jumps::execute): Same.
5988 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5990 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
5992 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5994 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
5995 to canonical location.
5997 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6000 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6002 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6004 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6006 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6007 rcrt1.o%s/grcrt1.o%s for -static-pie.
6009 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6011 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6012 ix86_shift_rotate_cost): Break out from ...
6013 (ix86_rtx_costs): ... here.
6014 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6017 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6019 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6020 when entry block was promoted unlikely.
6021 (estimate_bb_frequencies): Increase frequency scale.
6022 * profile-count.h (profile_count): Export precision info.
6024 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6026 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6027 disturbin profile of entry block.
6029 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6031 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6034 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6036 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6037 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6039 2017-11-17 Jeff Law <law@redhat.com>
6041 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6042 from evrp_dom_walker class. Various methods moved into new class.
6043 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6044 (evrp_range_analyzer::enter): New method.
6045 (evrp_range_analyzer::leave): New method.
6046 (evrp_dom_walker): Remove delegators no longer needed by this class.
6047 Replace vr_values data member with evrp_range_analyzer
6049 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6050 method extracted from evrp_dom_walker::before_dom_children.
6051 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6052 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6054 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6055 Add private copy constructor and move assignment operators.
6056 Privatize methods and class data where trivially possible.
6057 (evrp_dom_walker::cleanup): New function, extracted from
6058 execute_early_vrp. Simplify access to class data.
6060 * vr-values.h (get_output_for_vrp): Prototype.
6061 * vr-values.c (get_output_for_vrp): New function extracted from
6062 vrp_visit_assignment_or_call and extract_range_from_stmt.
6063 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
6065 2017-11-17 Luis Machado <luis.machado@linaro.org>
6067 * config/aarch64/aarch64.c
6068 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
6069 (qdf24xx_tunings) <autoprefetcher_model>: Set to
6070 tune_params::AUTOPREFETCHER_WEAK.
6072 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6075 * config/arm/arm.c (arm_valid_target_attribute_rec):
6076 Parse "arch=" and "+<ext>".
6077 (arm_valid_target_attribute_tree): Re-init global options.
6078 (arm_option_override): Make non-static.
6079 (arm_options_perform_arch_sanity_checks): Make errors fatal.
6080 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
6081 (__ARM_FEATURE_CRC32): Support undef.
6082 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
6083 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
6085 2017-11-17 David Malcolm <dmalcolm@redhat.com>
6087 * gdbinit.in (break-on-diagnostic): New command.
6089 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6091 * config/i386/i386.c (ix86_expand_epilogue): Change simple
6092 return to indirect jump for EH return if control-flow
6093 protection is enabled. Change explicit 'false' argument in
6094 pro_epilogue_adjust_stack with a value of
6096 * config/i386/i386.md (simple_return_indirect_internal):
6097 Remove SImode restriction to support 64-bit.
6099 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6101 * combine.c (added_notes_insn): New.
6102 (try_combine): Handle added_notes_insn like added_links_insn.
6103 Rewrite return value code.
6104 (distribute_notes): Set added_notes_insn to the earliest insn we added
6107 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6109 PR rtl-optimization/82621
6110 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
6111 dest of one of those SETs is unused.
6113 2017-11-17 Richard Biener <rguenther@suse.de>
6116 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
6117 * tree-pretty-print.c (dump_generic_node): Handle
6118 annot_expr_parallel_kind.
6119 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
6120 * gimplify.c (gimple_boolify): Likewise.
6122 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6124 * config.gcc (extra_headers): Add cet.h for x86 targets.
6125 * config/i386/cet.h: New file.
6126 * doc/install.texi: Add --enable-cet/--disable-cet.
6128 2017-11-17 Richard Biener <rguenther@suse.de>
6130 PR tree-optimization/83017
6131 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
6132 (gen_parallel_loop): Properly count iterations.
6133 (parallelize_loops): Handle loop->can_be_parallel independent
6134 of flag_loop_parallelize_all. Make static profitability test match
6136 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
6137 * invoke.texi (parloops-min-per-thread): Document.
6139 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
6141 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
6142 upstreaming review comments.
6144 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6146 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
6147 with fast unaligned access.
6148 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
6150 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
6152 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
6153 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
6155 2017-11-17 Richard Biener <rguenther@suse.de>
6157 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
6158 folding of references.
6160 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
6163 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
6164 of replacing call to strncmp with corresponding call to strcmp when
6167 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
6169 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
6170 option prefer-avx256 for skylake-avx512 configuration.
6171 * config/i386/i386.c (ix86_option_override_internal): Ditto.
6172 (get_builtin_code_for_version): Ditto.
6174 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6175 Monk Chiang <sh.chiang04@gmail.com>
6177 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
6178 (FIXED_REGISTERS): Reserve more register numbers.
6179 (CALL_USED_REGISTERS): Likewise.
6180 (REG_ALLOC_ORDER): Likewise.
6181 (REG_CLASS_CONTENTS): Likewise.
6182 (REGISTER_NAMES): Likewise.
6184 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6185 Kito Cheng <kito.cheng@gmail.com>
6187 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
6189 * config/nds32/iterators.md: Add vector mode iterators and attributes.
6191 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
6193 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
6194 parameter list for vec_splats.
6196 2017-11-16 Joseph Myers <joseph@codesourcery.com>
6198 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
6200 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
6202 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
6205 * doc/install.texi: Document incompatibility of Perl >=5.6.26
6206 with the required version of automake 1.11.6.
6208 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
6210 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
6212 (DU_C2_3_power9): Correct reservation combinations.
6213 (FP_DIV_power9, VEC_DIV_power9): New.
6214 (power9-alu): Split out rotate/shift...
6215 (power9-rot): ...to here, correct dispatch resource.
6216 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
6218 (power9-fp): Correct latency.
6219 (power9-sdiv): Add div/sqrt resource.
6220 (power9-ddiv): Correct latency, add div/sqrt resource.
6221 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
6222 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
6224 (power9-qpdiv, power9-qpmul): Adjust resource usage.
6226 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
6228 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
6229 switch statement mapping KF built-ins to TF built-ins if we don't
6230 have the proper ISA 3.0 assembler support.
6232 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6234 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
6235 (lower_emutls_function_body): Do not compute it.
6237 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6239 * ipa-split.c (split_bb_info): Turn time to sreal.
6240 (split_point): Likewise.
6241 (dump_split_point): Likewise.
6242 (fine_split_points): Likewise.
6243 (execute_split_functions): Only zero split_bbs; turn time to sreals.
6245 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6247 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
6250 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6252 * predict.c (combine_predictions_for_bb): Preserve zero predicted
6254 (expensive_function_p): Remove useless assert.
6255 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
6257 2017-11-16 Martin Sebor <msebor@redhat.com>
6259 PR tree-optimization/82588
6260 PR tree-optimization/82583
6261 * tree-vrp.c (check_array_ref): Handle flexible array members,
6262 string literals, and inner indices.
6263 (search_for_addr_array): Add detail to diagnostics.
6265 2017-11-16 Nathan Sidwell <nathan@acm.org>
6269 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
6270 (SET_DECL_ASSEMBLER_NAME): Forward to
6271 overwrite_decl_assembler_name.
6272 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
6273 (overwrite_decl_assembler_name): Declare.
6274 * tree.c (overwrite_decl_assembler_name): New.
6275 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
6276 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
6277 (LANG_HOOKS_INITIALIZER): Add it.
6278 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
6279 * langhooks.c (lhd_set_decl_assembler_name): Use
6280 SET_DECL_ASSEMBLER_NAME.
6281 (lhd_overwrite_decl_assembler_name): Default implementation.
6283 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6284 Jackson Woodruff <jackson.woodruff@arm.com>
6286 PR tree-optimization/71026
6287 * match.pd: Canonicalize constant multiplies in division.
6289 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6291 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
6292 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
6293 Remove OPT_fomit_frame_pointer entry.
6294 * common/config/arc/arc-common.c: Likewise.
6295 * common/config/arm/arm-common.c: Likewise.
6296 * common/config/avr/avr-common.c: Likewise.
6297 * common/config/c6x/c6x-common.c: Likewise.
6298 * common/config/cr16/cr16-common.c: Likewise.
6299 * common/config/cris/cris-common.c: Likewise.
6300 * common/config/epiphany/epiphany-common.c: Likewise.
6301 * common/config/fr30/fr30-common.c: Likewise.
6302 * common/config/frv/frv-common.c: Likewise.
6303 * common/config/ia64/ia64-common.c: Likewise.
6304 * common/config/iq2000/iq2000-common.c: Likewise.
6305 * common/config/lm32/lm32-common.c: Likewise.
6306 * common/config/m32r/m32r-common.c: Likewise.
6307 * common/config/mcore/mcore-common.c: Likewise.
6308 * common/config/microblaze/microblaze-common.c: Likewise.
6309 * common/config/mips/mips-common.c: Likewise.
6310 * common/config/mmix/mmix-common.c: Likewise.
6311 * common/config/mn10300/mn10300-common.c: Likewise.
6312 * common/config/nios2/nios2-common.c: Likewise.
6313 * common/config/pa/pa-common.c: Likewise.
6314 * common/config/pdp11/pdp11-common.c: Likewise.
6315 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
6316 * common/config/riscv/riscv-common.c: Likewise.
6317 * common/config/rs6000/rs6000-common.c: Likewise.
6318 * common/config/rx/rx-common.c: Likewise.
6319 * common/config/s390/s390-common.c: Likewise.
6320 * common/config/sh/sh-common.c: Likewise.
6321 * common/config/sparc/sparc-common.c: Likewise.
6322 * common/config/tilegx/tilegx-common.c: Likewise.
6323 * common/config/tilepro/tilepro-common.c: Likewise.
6324 * common/config/v850/v850-common.c: Likewise.
6325 * common/config/visium/visium-common.c: Likewise.
6326 * common/config/xstormy16/xstormy16-common.c: Likewise.
6327 * common/config/xtensa/xtensa-common.c: Likewise.
6328 * invoke.texi (-fomit-frame-pointer): Update documentation.
6330 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6332 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
6334 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6336 * cfg.c (scale_bbs_frequencies_int,
6337 cale_bbs_frequencies_gcov_type): Remove.
6338 * cfg.h (scale_bbs_frequencies_int,
6339 cale_bbs_frequencies_gcov_type): Remove.
6341 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6343 * tree-ssa-loop-manip.c
6344 (scale_dominated_blocks_in_loop): Update to profile counts.
6345 (tree_transform_and_unroll_loop): Likewise.
6347 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6349 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
6350 scale_bbs_frequencies_int.
6352 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6354 * final.c (compute_alignments): Use counts rather than frequencies.
6356 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6358 * cfgloopanal.c: Include sreal.h
6359 (average_num_loop_insns): Use counts and sreal for accounting.
6361 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6363 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
6366 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6367 Kito Cheng <kito.cheng@gmail.com>
6369 * config/nds32/constraints.md: Provide more constraints.
6370 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
6371 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
6372 support constraints usage.
6374 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6376 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
6378 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6379 Kito Cheng <kito.cheng@gmail.com>
6381 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
6382 * config/nds32/nds32.opt: Refine the layout.
6383 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
6384 TARGET_EXT_STRING): Support new options.
6385 * config/nds32/nds32.h: Likewise.
6386 * config/nds32/nds32.md: Likewise.
6387 * config/nds32/nds32-predicates.c: Likewise.
6388 * config/nds32/constraints.md: Likewise.
6389 * common/config/nds32/nds32-common.c: Likewise.
6391 2017-11-16 Julia Koval <julia.koval@intel.com>
6394 * config/i386/gfniintrin.h: Add sse check.
6395 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
6397 2017-11-16 Julia Koval <julia.koval@intel.com>
6399 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
6400 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
6401 (ix86_handle_option): Handle -mavx512vbmi2.
6402 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
6403 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
6404 * config/i386/i386-c.c (__AVX512VBMI2__): New.
6405 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
6406 (ix86_valid_target_attribute_inner_p): Ditto.
6407 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
6408 * config/i386/i386.opt (mavx512vbmi2): New option.
6409 * doc/invoke.texi: Add new option.
6411 2017-11-16 Julia Koval <julia.koval@intel.com>
6413 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
6414 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
6415 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
6416 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
6417 _mm512_gf2p8mul_epi8): New intrinsics.
6418 * config/i386/i386-builtin-types.def
6419 (V64QI_FTYPE_V64QI_V64QI): New type.
6420 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
6421 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
6422 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
6423 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
6424 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
6425 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
6427 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
6429 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
6430 explicitly as a stream of bytes.
6432 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6434 * config/rs6000/altivec.h (vec_xst_be): New #define.
6435 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
6436 and externalize from *altivec_vperm_<mode>_internal.
6437 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
6439 (XL_BE_V8HI): Likewise.
6440 (XL_BE_V4SI): Likewise.
6441 (XL_BE_V4SI): Likewise.
6442 (XL_BE_V2DI): Likewise.
6443 (XL_BE_V4SF): Likewise.
6444 (XL_BE_V2DF): Likewise.
6445 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
6446 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
6447 all array entries with these keys: VSX_BUILTIN_VEC_XL,
6448 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
6449 VSX_BUILTIN_VEC_XST_BE.
6450 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
6451 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
6453 (altivec_init_builtins): Replace conditional calls to def_builtin
6454 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
6455 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
6456 with unconditional calls. Remove calls to def_builtin for
6457 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
6458 __builtin_vec_xst_be.
6459 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
6460 to define_expand, and add alternate RTL generation for P8.
6461 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
6462 vsx_ld_elemrev_v8hi.
6463 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
6464 add alternate RTL generation for P8.
6465 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
6466 vsx_ld_elemrev_v16qi.
6467 (vsx_st_elemrev_v8hi): Convert define_insn
6468 to define_expand, and add alternate RTL generation for P8.
6469 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
6470 vsx_st_elemrev_v8hi.
6471 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
6472 add alternate RTL generation for P8.
6473 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
6474 vsx_st_elemrev_v16qi.
6476 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
6479 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
6480 TARGET_AVX512ER check.
6481 (ix86_option_override_internal): Set MASK_VZEROUPPER if
6482 neither -mzeroupper nor -mno-zeroupper is used and
6483 TARGET_EMIT_VZEROUPPER is set.
6484 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
6485 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
6487 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6489 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6490 folding of vector compares.
6491 (fold_build_vec_cmp): New helper function.
6492 (fold_compare_helper): New helper function.
6493 (builtin_function_type): Add compare builtins to the list of functions
6494 having unsigned arguments. Cosmetic updates to comment indentation.
6495 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
6496 the not+eq combination.
6498 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6500 PR tree-optimization/82726
6501 PR tree-optimization/70754
6502 * tree-predcom.c (order_drefs_by_pos): New function.
6503 (combine_chains): Move code setting has_max_use_after to...
6504 (try_combine_chains): ...here. New parameter. Sort combined chains
6505 according to position information.
6506 (tree_predictive_commoning_loop): Update call to above function.
6507 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
6509 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6511 PR tree-optimization/82726
6513 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6515 PR tree-optimization/70754
6516 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6517 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6518 combined stmt before it if not NULL.
6519 (combine_chains): Process refs reversely and compute dominance point
6523 2017-02-23 Bin Cheng <bin.cheng@arm.com>
6525 PR tree-optimization/79663
6526 * tree-predcom.c (combine_chains): Process refs in reverse order
6527 only for ZERO length chains, and add explaining comment.
6529 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6531 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
6532 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
6533 * doc/invoke.texi (ARM Options): Add armv8.3-a.
6535 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6537 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
6539 2017-11-15 Martin Liska <mliska@suse.cz>
6541 * tree-cfg.c (pass_warn_function_return::execute):
6542 Compare warn_return_type for greater than zero.
6544 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
6548 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
6549 to return true on Xeon and not on Xeon Phi.
6550 (ix86_check_avx256_register): Changed to ...
6551 (ix86_check_avx_upper_register): ... this. Add extra check for
6552 VALID_AVX512F_REG_OR_XI_MODE.
6553 (ix86_avx_u128_mode_needed): Changed
6554 ix86_check_avx256_register to ix86_check_avx_upper_register.
6555 (ix86_check_avx256_stores): Changed to ...
6556 (ix86_check_avx_upper_stores): ... this. Changed
6557 ix86_check_avx256_register to ix86_check_avx_upper_register.
6558 (ix86_avx_u128_mode_after): Changed
6559 avx_reg256_found to avx_upper_reg_found. Changed
6560 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
6561 (ix86_avx_u128_mode_entry): Changed
6562 ix86_check_avx256_register to ix86_check_avx_upper_register.
6563 (ix86_avx_u128_mode_exit): Ditto.
6564 * config/i386/i386.h: (host_detect_local_cpu): New define.
6566 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6568 * config/arm/xgene1.md (xgene1): Split into automatons
6569 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
6570 (xgene1_f_load): Adjust reservations and/or types.
6571 (xgene1_f_store): Likewise.
6572 (xgene1_load_pair): Likewise.
6573 (xgene1_store_pair): Likewise.
6574 (xgene1_fp_load1): Likewise.
6575 (xgene1_load1): Likewise.
6576 (xgene1_store1): Likewise.
6577 (xgene1_move): Likewise.
6578 (xgene1_alu): Likewise.
6579 (xgene1_simd): Likewise.
6580 (xgene1_bfm): Likewise.
6581 (xgene1_neon_load1): Likewise.
6582 (xgene1_neon_store1): Likewise.
6583 (xgene1_neon_logic): Likewise.
6584 (xgene1_neon_st1): Likewise.
6585 (xgene1_neon_ld1r): Likewise.
6586 (xgene1_alu_cond): Added.
6587 (xgene1_shift_reg): Likwise.
6588 (xgene1_bfx): Likewise.
6589 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6591 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6594 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6596 (can_widen_mult_without_libcall): New function.
6597 (expand_mul_overflow): If only checking unsigned mul overflow,
6598 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6599 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6600 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6601 (expand_DIVMOD): Formatting fix.
6602 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6603 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6604 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6606 PR tree-optimization/82977
6607 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6608 constructed temporary to strlen_to_stridx.put.
6610 2017-11-15 Martin Liska <mliska@suse.cz>
6612 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6613 * configure: Regenerate.
6615 2017-11-15 Martin Liska <mliska@suse.cz>
6618 * config/sh/sh-mem.cc: Use proper probability for
6621 2017-11-14 Jeff Law <law@redhat.com>
6623 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6624 the red zone for stack_clash_protection_final_dynamic_probe targets
6625 when the total dynamic stack size is zero bytes.
6627 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6628 blocks is post order.
6630 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6632 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6633 * final.c (rest_of_clean_state): Set it for the
6634 -fcompare-debug dump.
6635 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6636 class when TDF_COMPARE_DEBUG is set.
6638 * dwarf2out.c (gen_producer_string): Discard
6641 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6644 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6645 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6646 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6647 (__TGMATH_REAL_2_3): Remove macros.
6648 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6649 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6651 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6653 (remquo): Define using __TGMATH_REAL_3.
6655 2017-11-14 Jeff Law <law@redhat.com>
6657 * vr-values.c: New file with contents extracted from tree-vrp.c.
6658 * Makefile.in (OBJS): Add vr-values.o
6659 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6660 (set_value_range, set_and_canonicalize_value_range): Likewise.
6661 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6662 (value_range_constant_singleton, symbolic_range_p): Likewise.
6663 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6664 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6665 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6666 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6667 (find_case_label_range, find_case_label_index): Likewise.
6668 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6669 (range_int_cst_singleton_p, value_inside_range): Likewise.
6670 (get_single_symbol): Likewise.
6671 (switch_update): Move structure definition here.
6672 (to_remove_edges, to_update_switch_stmts): Provide externs.
6673 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6674 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6675 (vrp_val_is_min, set_value_range): Likewise.
6676 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6677 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6678 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6679 (range_is_nonnull, range_int_cst_p): Likewwise.
6680 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6681 (get_single_symbol, operand_less_p): Likewise
6682 (compare_values_warnv, compare_values): Likewise.
6683 (value_inside_range, value_range_constant_singleton): Likewise.
6684 (zero_nonzero_bitgs_from_vr): Likewise.
6685 (extract_range_from_binary_expr_1): Likewise.
6686 (overflow_comparison_p): Likewise.
6687 (to_remove_edges, to_update_switch_stmts): Likewise.
6688 (find_case_label-index, find_case_label_range): Likewise.
6689 (switch_update, set_value_range_to_nonnegative): Remove.
6690 (set_value_range_to_truthvalue): Likewise.
6691 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6692 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6693 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6694 (find_case_label_ranges, test_for_singularity): Likewise.
6695 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6696 (x_vr_values): Move to its remaining use site.
6698 2017-11-10 Jeff Law <law@redhat.com>
6700 * vr-values.h (VR_INITIALIZER): Move #define here.
6701 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6702 * Makefile.in (OBJS): Add tree-evrp.o
6703 * tree-vrp.h (assert_info): Move structure definition here.
6704 (set_value_range_to_varying): Prototype.
6705 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6706 (infer_value_range, register_edge_assert_for): Likewise.
6707 (stmt_interesting_for_vrp): Likewise.
6708 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6709 (set_value_range_to_varying): No longer static.
6710 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6711 (infer_value_range, register_edge_assert_for): Likewise.
6713 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6715 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6716 generate the XXBRD instruction.
6718 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6719 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6720 as being compatible if -mabi=ieeelongdouble.
6721 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6722 to setup float128 built-ins with hardware support.
6723 (BU_FLOAT128_HW_2): Likewise.
6724 (BU_FLOAT128_HW_3): Likewise.
6725 (BU_FLOAT128_HW_VSX_1): Likewise.
6726 (BU_FLOAT128_HW_VSX_2): Likewise.
6727 (scalar_extract_expq): Change float128 built-in functions to
6728 accommodate having both KFmode and TFmode functions. Use the
6729 KFmode variant as the default.
6730 (scalar_extract_sigq): Likewise.
6731 (scalar_test_neg_qp): Likewise.
6732 (scalar_insert_exp_q): Likewise.
6733 (scalar_insert_exp_qp): Likewise.
6734 (scalar_test_data_class_qp): Likewise.
6735 (sqrtf128_round_to_odd): Delete processing the round to odd
6736 built-in functions as special built-in functions, and define them
6737 as float128 built-ins. Use the KFmode variant as the default.
6738 (truncf128_round_to_odd): Likewise.
6739 (addf128_round_to_odd): Likewise.
6740 (subf128_round_to_odd): Likewise.
6741 (mulf128_round_to_odd): Likewise.
6742 (divf128_round_to_odd): Likewise.
6743 (fmaf128_round_to_odd): Likewise.
6744 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6745 support for KFmode and TFmode xststdcqp calls.
6746 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6747 point, switch the built-in handlers for the get/set float128
6748 exponent, get float128 mantissa, float128 test built-ins, and the
6749 float128 round to odd built-in functions. Eliminate creating the
6750 float128 round to odd built-in functions as special built-ins.
6751 (rs6000_init_builtins): Eliminate special creation of the float128
6752 round to odd built-in functions.
6753 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6754 function insns to support both TFmode and KFmode variants.
6755 (xsxsigqp_<mode>): Likewise.
6756 (xsiexpqpf_<mode>): Likewise.
6757 (xsiexpqp_<mode>): Likewise.
6758 (xststdcqp_<mode>): Likewise.
6759 (xststdcnegqp_<mode>): Likewise.
6760 (xststdcqp_<mode>): Likewise.
6762 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6764 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6765 unused path_in_freq_ptr parameter.
6766 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6768 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6770 * ipa-inline.c (edge_badness): Dump sreal frequency.
6771 (compute_inlined_call_time): Match natural implementaiton ...
6772 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6773 forgotten division by CGRAPH_FREQ_BASE.
6775 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6777 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6778 Solaris 11. Update comment.
6779 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6781 * config/sol2.h (STARTFILE_SPEC): Likewise.
6782 * configure: Regenerate.
6784 2017-11-14 Carl Love <cel@us.ibm.com>
6786 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6787 le_ and be_ prefixes to swap* variables. Remove
6788 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6791 2017-11-14 Jason Merrill <jason@redhat.com>
6793 Support GTY((cache)) on hash_map.
6794 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6795 (ggc_cache_remove): Override it instead of ggc_mx.
6796 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6797 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6798 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6799 (simple_cache_map_traits): Override maybe_mx.
6800 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6801 (hash_map): Friend gt_cleare_cache.
6802 (gt_cleare_cache): New.
6803 * tree.h (tree_cache_traits): New hash_map traits class.
6804 (tree_cache_map): New typedef.
6806 2017-11-14 Richard Biener <rguenther@suse.de>
6808 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6809 paramter and handling.
6810 (cleanup_control_flow_bb): Likewise.
6811 (cleanup_control_flow_pre): New helper performing a DFS walk
6812 to call cleanup_control_flow_bb in PRE order.
6813 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6814 via cleanup_control_flow_pre.
6816 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6818 * config/aarch64/aarch64-simd.md
6819 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6820 (*aarch64_simd_bsl<mode>_alt): Likewise.
6821 (aarch64_simd_bsldi_internal): New.
6822 (aarch64_simd_bsldi_alt): Likewise.
6824 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6826 * tracer.c (better_p): Do not compare frequencies.
6827 * reg-stack.c (better_edge): Likewise.
6828 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
6831 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6833 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
6834 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
6835 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
6836 * lto-streamer-in.c (input_function): Use update_max_bb_count.
6837 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
6838 * predict.c (maybe_hot_frequency_p): Inline to ...
6839 (maybe_hot_count_p): ... here; rewrite to counts.
6840 (counts_to_freqs): Rename to ...
6841 (update_max_bb_count): ... this one.
6842 (expensive_function_p): Use counts.
6843 (estimate_bb_frequencies): Update.
6844 (rebuild_frequencies): Update.
6845 * predict.h (counts_to_freqs): Rename to ...
6846 (update_max_bb_count): ... this one.
6847 * profile.c (compute_branch_probabilities): Add debug info
6848 * tree-inline.c (expand_call_inline): Update debug info.
6849 (optimize_inline_calls): Use update_max_bb_count..
6850 (tree_function_versioning): Use update_max_bb_count..
6851 * value-prof.c (gimple_value_profile_transformations):
6852 Do not use update_max_bb_count.
6854 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6856 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
6857 always use frequencies.
6859 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6861 * bb-reorder.c: Remove frequencies from comments.
6862 (better_edge_p): Use profile counts.
6863 (find_traces): Dump profile counts.
6864 (rotate_loop): Use profile counts.
6865 (find_traces_1_round): Likewise.
6866 (connect_better_edge_p): Use counts instead of probabilities for
6868 (copy_bb_p): Drop early check for non-0 frequency.
6869 (sanitize_hot_paths): Update comments.
6871 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6873 * ipa-split.c (struct split_point): Add count.
6874 (consider_split): Do not compute incoming frequency; compute incoming
6875 count and store it to split_point.
6876 (split_function): Set count of the call to split part correctly.
6878 2017-11-13 Carl Love <cel@us.ibm.com>
6880 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
6882 2017-11-13 Tom Tromey <tom@tromey.com>
6884 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
6886 2017-11-13 Carl Love <cel@us.ibm.com>
6888 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
6889 Add support for builtins:
6890 unsigned int vec_first_{,miss}_match_{,or_eos}index,
6891 vector {un,}signed {char,int,short},
6892 vector {un,}signed {char,int,short}) arguments.
6893 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
6894 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
6895 Add BU_P9V_AV_2 expansions for the builtins.
6896 * config/rs6000/altivec.h (vec_first_match_index,
6897 vec_first_mismatch_index, vec_first_match_or_eos_index,
6898 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
6899 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
6900 new extern declaration.
6901 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
6902 * config/rs6000/vsx.md (first_match_index_<mode>,
6903 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
6904 first_mismatch_or_eos_index_<mode>): Add define expand.
6905 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
6906 * doc/extend.texi: Update the built-in documenation file for the new
6909 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
6911 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
6912 into the min/max operations for _Float<N> and _Float<N>X types.
6914 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6917 * dwarf2out.c (do_eh_frame): New static variable.
6918 (dwarf2out_begin_prologue): Set it.
6919 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
6921 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6923 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
6925 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
6927 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6930 * debug.h (dwarf2out_do_eh_frame): Declare.
6931 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
6932 (dwarf2out_do_frame): Use it.
6933 (dwarf2out_do_cfi_asm): Likewise.
6934 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
6935 (dwarf2out_assembly_start): Likewise.
6936 (dwarf2out_begin_prologue): Fix comment.
6937 * toplev.c (compile_file): Always call dwarf2out_frame_finish
6938 if the target needs either debug or unwind DWARF2 info.
6939 * lto-opts.c (lto_write_options): Do not save -fexceptions,
6940 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
6941 -ftrapping-math, -ftrapv and -fwrapv.
6943 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6945 * cgraph.c (cgraph_edge::sreal_frequency): New function.
6946 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
6947 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
6948 (estimate_edge_size_and_time): Likewise.
6949 (ipa_merge_fn_summary_after_inlining): Likewise.
6950 * ipa-inline.c (cgraph_freq_base_rec): Remove.
6951 (compute_uninlined_call_time): Use sreal_frequency.
6952 (compute_inlined_call_time): Likewise.
6953 (ipa_inline): Do not initialize cgraph_freq_base_rec.
6954 * profile-count.c: Include sreal.h.
6955 (profile_count::to_sreal_scale): New.
6956 * profile-count.h: Forward declare sreal.
6957 (profile_count::to_sreal_scale): Declare.
6959 2017-11-13 Nathan Sidwell <nathan@acm.org>
6961 * diagnostic.c (maybe_line_and_column): New.
6962 (diagnostic_get_location_text): Use it.
6963 (diagnostic_report_current_module): Likewise.
6964 (test_diagnostic_get_location_text): Add tests.
6966 2017-11-13 Luis Machado <luis.machado@linaro.org>
6968 * doc/md.texi (Specifying processor pipeline description): Fix
6969 incorrect latency for the div instruction example.
6971 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6973 PR tree-optimization/78821
6974 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
6975 that bit_not_p is the same.
6976 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
6977 (split_group): Count precisely bit_not_p bits in each statement.
6978 (invert_op): New function.
6979 (imm_store_chain_info::output_merged_store): Use invert_op to
6980 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
6981 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
6983 2017-11-13 Martin Liska <mliska@suse.cz>
6985 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
6986 (struct source_info): Likewise.
6987 (add_branch_counts): Likewise.
6988 (add_line_counts): Likewise.
6989 (function_summary): Likewise.
6990 (output_intermediate_line): Likewise.
6991 (generate_results): Likewise.
6993 2017-11-13 Martin Liska <mliska@suse.cz>
6995 * gcov.c (struct block_info): Remove typedef for block_t.
6996 (struct line_info): Likewise.
6997 (line_info::has_block): Likewise.
6998 (EXIT_BLOCK): Likewise.
6999 (unblock): Likewise.
7000 (circuit): Likewise.
7001 (get_cycles_count): Likewise.
7002 (process_file): Likewise.
7003 (read_graph_file): Likewise.
7004 (solve_flow_graph): Likewise.
7005 (find_exception_blocks): Likewise.
7006 (add_line_counts): Likewise.
7007 (accumulate_line_info): Likewise.
7008 (output_line_details): Likewise.
7010 2017-11-13 Martin Liska <mliska@suse.cz>
7012 * gcov.c (struct arc_info): Remove typedef for arc_t.
7013 (struct line_info): Likewise.
7014 (add_branch_counts): Likewise.
7015 (output_branch_count): Likewise.
7016 (function_info::~function_info): Likewise.
7017 (circuit): Likewise.
7018 (output_intermediate_line): Likewise.
7019 (read_graph_file): Likewise.
7020 (solve_flow_graph): Likewise.
7021 (find_exception_blocks): Likewise.
7022 (add_line_counts): Likewise.
7023 (accumulate_line_info): Likewise.
7024 (output_line_details): Likewise.
7025 (output_function_details): Likewise.
7027 2017-11-13 Martin Liska <mliska@suse.cz>
7029 * gcov.c (struct function_info): Remove typedef for function_t.
7030 (struct source_info): Likewise.
7031 (source_info::get_functions_at_location): Likewise.
7032 (solve_flow_graph): Likewise.
7033 (find_exception_blocks): Likewise.
7034 (add_line_counts): Likewise.
7035 (output_intermediate_file): Likewise.
7036 (process_file): Likewise.
7037 (generate_results): Likewise.
7038 (release_structures): Likewise.
7039 (read_graph_file): Likewise.
7040 (read_count_file): Likewise.
7041 (accumulate_line_counts): Likewise.
7042 (output_lines): Likewise.
7044 2017-11-13 Martin Liska <mliska@suse.cz>
7046 * gcov.c (function_info::function_info): Remove num_counts
7047 and add vector<gcov_type>.
7048 (function_info::~function_info): Use the vector.
7049 (process_file): Likewise.
7050 (read_graph_file): Likewise.
7051 (read_count_file): Likewise.
7052 (solve_flow_graph): Likewise.
7054 2017-11-13 Martin Liska <mliska@suse.cz>
7056 * gcov.c (function_info::is_artificial): New function.
7057 (process_file): Erase all artificial early.
7058 (generate_results): Skip as all artificial are already
7061 2017-11-13 Martin Liska <mliska@suse.cz>
7063 * gcov.c (read_graph_file): Store to global vector of functions.
7064 (read_count_file): Iterate the vector.
7065 (process_file): Likewise.
7066 (generate_results): Likewise.
7067 (release_structures): Likewise.
7069 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7071 PR tree-optimization/82954
7072 * gimple-ssa-store-merging.c
7073 (imm_store_chain_info::coalesce_immediate_stores): If
7074 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
7076 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
7078 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7079 Upddate call to ENDIAN_LANE_N.
7080 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
7081 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
7082 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
7083 and use aarch64_endian_lane_rtx.
7084 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
7086 2017-11-12 Tom de Vries <tom@codesourcery.com>
7088 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
7090 2017-11-12 Tom de Vries <tom@codesourcery.com>
7092 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
7094 2017-11-12 Tom de Vries <tom@codesourcery.com>
7096 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
7098 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7099 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7100 * defaults.h (ASM_OUTPUT_LABELREF): Same.
7102 2017-11-11 Martin Sebor <msebor@redhat.com>
7105 * doc/extend.texi (attribute nonstring): Remove spurious argument.
7108 * prefic.c (translate_name): Replace strncpy with memcpy to
7109 avoid -Wstringop-truncation.
7111 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7113 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
7115 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7117 * predict.c (maybe_hot_frequency_p): Do not use cfun.
7119 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7121 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
7124 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7126 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
7127 enable generating XXBRH if the value is in a vector register.
7128 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
7129 value is in a vector register.
7130 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
7131 register to register bswap64's instead of doing the GPR sequence
7132 used on previous machines.
7133 (bswapdi2_xxbrd): New insn.
7134 (bswapdi2_reg): Disallow on ISA 3.0.
7135 (register to register bswap64 splitter): Do not split the insn on
7136 ISA 3.0 systems that use XXBRD.
7138 2017-11-10 Martin Sebor <msebor@redhat.com>
7141 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
7143 (find_subframework_file): Same.
7145 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7147 * auto-profile.c (afdo_indirect_call): Drop frequency.
7148 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
7149 (cgraph_node::create_edge): Drop frequency argument.
7150 (cgraph_node::create_indirect_edge): Drop frequency argument.
7151 (cgraph_edge::make_speculative): Drop frequency arguments.
7152 (cgraph_edge::resolve_speculation): Do not update frequencies
7153 (cgraph_edge::dump_edge_flags): Do not dump frequency.
7154 (cgraph_node::dump): Check consistency in IPA mode.
7155 (cgraph_edge::maybe_hot_p): Use IPA counter.
7156 (cgraph_edge::verify_count_and_frequency): Rename to ...
7157 (cgraph_edge::verify_count): ... this one; drop frequency checking.
7158 (cgraph_node::verify_node): Update.
7159 * cgraph.h (struct cgraph_edge): Drop frequency.
7160 (cgraph_edge::frequency): New function.
7161 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
7163 (cgraph_edge::rebuild_edges): Likewise.
7164 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
7165 (duplicate_thunk_for_node): Do not pass frequency.
7166 (cgraph_node::create_clone): Scale only counts.
7167 (cgraph_node::create_virtual_clone): Do not pass frequency.
7168 (cgraph_node::create_edge_including_clones): Do not pass frequency.
7169 (cgraph_node::create_version_clone): Do not pass frequency.
7170 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
7171 (cgraph_node::expand_thunk): Do not pass frequency.
7172 (cgraph_node::create_wrapper): Do not pass frequency.
7173 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
7175 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
7176 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
7177 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
7178 (gather_caller_stats): Use frequency function.
7179 (ipcp_cloning_candidate_p): Use frequency function.
7180 (ipcp_propagate_stage): Use frequency function.
7181 (get_info_about_necessary_edges): Use frequency function.
7182 (update_profiling_info): Update only IPA profile.
7183 (update_specialized_profile): Use frequency functoin.
7184 (perhaps_add_new_callers): Update only IPA profile.
7185 * ipa-devirt.c (ipa_devirt): Use IPA profile.
7186 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
7187 (dump_ipa_call_summary): Use frequency function.
7188 (estimate_edge_size_and_time): Use frequency function.
7189 (ipa_merge_fn_summary_after_inlining): Use frequency function.
7190 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
7191 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
7192 (update_noncloned_counts): ... ths one; scale counts only.
7193 (clone_inlined_nodes): Do not scale frequency.
7194 (inline_call): Do not pass frequency.
7195 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
7196 (compute_inlined_call_time): Use IPA profile.
7197 (want_inline_small_function_p): Use IPA profile.
7198 (want_inline_self_recursive_call_p): Use IPA profile.
7199 (edge_badness): Use IPA profile.
7200 (lookup_recursive_calls): Use IPA profile.
7201 (recursive_inlining): Do not pass frequency.
7202 (resolve_noninline_speculation): Do not update frequency.
7203 (inline_small_functions): Collect max of IPA profile.
7204 (dump_overall_stats): Dump IPA porfile.
7205 (dump_inline_stats): Dump IPA porfile.
7206 (ipa_inline): Collect IPA stats.
7207 * ipa-inline.h (clone_inlined_nodes): Update prototype.
7208 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
7209 (ipa_propagate_frequency): Use frequency function.
7210 (ipa_profile): Cleanup.
7211 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
7212 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
7213 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
7214 (input_node): Do not stream frequency.
7215 (input_edge): Do not stream frequency.
7216 (merge_profile_summaries): Scale only IPA profiles.
7217 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
7218 * predict.c (drop_profile): Do not recompute frequency.
7219 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
7220 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
7221 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
7222 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
7223 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
7224 * tree-inline.c (copy_bb): Do not scale frequency.
7225 (expand_call_inline): Do not scale frequency.
7226 (tree_function_versioning): Do not scale frequency.
7227 * ubsan.c (ubsan_create_edge): Do not pass frequency.
7229 2017-11-10 Julia Koval <julia.koval@intel.com>
7231 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
7232 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
7233 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
7234 (_mm256_maskz_gf2p8affine_epi64_epi8)
7235 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
7236 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
7237 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
7238 (__builtin_ia32_vgf2p8affineqb_v32qi)
7239 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
7240 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
7242 2017-11-10 Tamar Christina <tamar.christina@arm.com>
7246 (arm_option_override): Refactor.
7247 (arm_option_reconfigure_globals): New.
7248 (arm_options_perform_arch_sanity_checks): New.
7249 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
7251 (arm_options_perform_arch_sanity_checks): Likewise
7253 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
7255 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
7256 (power9-qpmul): New.
7257 * rs6000/rs6000.md ("type" attr): Add qmul.
7258 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
7259 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
7260 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
7262 2017-11-10 Martin Sebor <msebor@redhat.com>
7265 * builtins.c (compute_objsize): Handle arrays that
7266 compute_builtin_object_size likes to fail for. Make extern.
7267 * builtins.h (compute_objsize): Declare.
7268 (check_strncpy_sizes): New function.
7269 (expand_builtin_strncpy): Call check_strncpy_sizes.
7270 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
7271 -Wstringop-truncation.
7272 (gimple_fold_builtin_strncat): Same.
7273 * gimple.c (gimple_build_call_from_tree): Set call location.
7274 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
7275 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
7276 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
7277 (handle_builtin_strlen): Use strlen_to_stridx.
7278 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
7280 Use strlen_to_stridx.
7281 (pass_strlen::execute): Release strlen_to_stridx.
7282 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
7283 (-Wstringop-truncation): Document new option.
7285 2017-11-10 Martin Liska <mliska@suse.cz>
7287 PR gcov-profile/82702
7288 * gcov.c (main): Handle intermediate files in a different
7290 (get_gcov_intermediate_filename): New function.
7291 (output_gcov_file): Remove support of intermediate files.
7292 (generate_results): Allocate intermediate file.
7293 (release_structures): Clean-up properly fn_end.
7294 (output_intermediate_file): Start iterating with line 1.
7296 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7298 PR tree-optimization/82929
7299 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7300 ops_swapped_p non-static data member.
7301 (store_immediate_info::store_immediate_info): Clear it.
7302 (imm_store_chain_info::coalesce_immediate_stores): If swapping
7303 ops set ops_swapped_p.
7304 (count_multiple_uses): Handle ops_swapped_p.
7306 2017-11-10 Martin Liska <mliska@suse.cz>
7308 * coverage.c (coverage_init): Stream information about
7309 support of has_unexecuted_blocks.
7310 * doc/gcov.texi: Document that.
7311 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
7312 * gcov.c (read_graph_file): Likewise.
7313 (output_line_beginning): Fix a small issue with
7316 2017-11-10 Bin Cheng <bin.cheng@arm.com>
7318 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
7319 reference of trivial component.
7321 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7324 * gimple-ssa-store-merging.c
7325 (pass_store_merging::terminate_all_aliasing_chains): For
7326 gimple_store_p stmts also call refs_output_dependent_p.
7328 PR rtl-optimization/82913
7329 * compare-elim.c (try_merge_compare): Punt if def_insn is not
7332 2017-11-09 Jeff Law <law@redhat.com>
7334 * vr-values.h: New file with vr_values class.
7335 * tree-vrp.c: Include vr-values.h
7336 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
7337 data objects into the vr_values class.
7338 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
7339 (get_value_range): Make it a member function within vr_values class.
7340 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
7341 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
7342 (op_with_constant_singleton_value_range): Likewise.
7343 (extract_range_for_var_from_comparison_expr): Likewise.
7344 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
7345 (extract_range_from_binary_expr): Likewise.
7346 (extract_range_from_unary_expr): Likewise.
7347 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
7348 (check_for_binary_op_overflow, extract_range_basic): Likewise.
7349 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
7350 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
7351 (compare_name_with_value, compare_names): Likewise.
7352 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
7353 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
7354 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
7355 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
7356 (extract_range_from_phi_node): Likewise.
7357 (simplify_truth_ops_using_ranges): Likewise.
7358 (simplify_div_or_mod_using_ranges): Likewise.
7359 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
7360 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
7361 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
7362 (simplify_float_conversion_using_ranges): Likewise.
7363 (simplify_internal_call_using_ranges): Likewise.
7364 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
7365 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
7366 poitner via x_vr_values for calls into gimple folder.
7367 (vrp_initialize_lattice): Make this the vr_values ctor.
7368 (vrp_free_lattice): Make this the vr_values dtor.
7369 (set_vr_value): New function.
7370 (class vrp_prop): Add vr_values data member. Add various member
7371 functions as well as member functions that delegate to vr_values.
7372 (check_array_ref): Make a member function within vrp_prop class.
7373 (search_for_addr_array, vrp_initialize): Likewise.
7374 (vrp_finalize): Likewise. Revamp to avoid direct access to
7375 vr_value, values_propagated, etc.
7376 (check_array_bounds): Extract vrp_prop class instance pointer from
7377 walk info structure. Use it to call member functions.
7378 (check_all_array_refs): Make a member function within vrp_prop class.
7379 Smuggle class instance pointer via walk info structure.
7380 (x_vr_values): New local static.
7381 (vrp_valueize): Use x_vr_values to get class instance.
7382 (vr_valueize_1): Likewise.
7383 (class vrp_folder): Add vr_values data member. Add various member
7384 functions as well as member functions that delegate to vr_values.
7385 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
7386 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
7387 class instance from vr_values. Use it to call member functions.
7388 (vrp_dom_walker): Add vr_values data member.
7389 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
7390 instance via x_vr_values.
7391 (identify_jump_threads): Accept vr_values as argument. Store
7392 it into the walker structure.
7393 (evrp_dom_walker): Add vr_values class data member. Add various
7395 (evrp_dom_walker::try_find_new_range): Use vr_values data
7396 member to access the memory allocator.
7397 (evrp_dom_walker::before_dom_children): Store vr_values class
7398 instance into the vrp_folder class.
7399 (evrp_dom_walker::push_value_range): Rework to avoid direct
7400 access to num_vr_values and vr_value.
7401 (evrp_dom_walker::pop_value_range): Likewise.
7402 (execute_early_vrp): Remove call to vrp_initialize_lattice.
7403 Use vr_values to get to dump_all_value_ranges member function.
7404 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
7405 and simplify_cond_using_ranges_2 via vrp_prop class instance.
7406 Pass vr_values class instance down to identify_jump_threads.
7407 Remove call to vrp_free_lattice.
7408 (debug_all_value_ranges): Remove.
7410 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
7411 (vrp_folder): Likewise.
7413 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
7414 Get it from the existing bitmap instead.
7415 (vrp_intersect_ranges_1): Likewise.
7417 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7419 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7421 (store_immediate_info::store_immediate_info): Add bitnotp argument,
7422 set bit_not_p to it.
7423 (imm_store_chain_info::coalesce_immediate_stores): Break group
7424 if bit_not_p is different.
7425 (count_multiple_uses, split_group,
7426 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
7427 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
7428 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
7429 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
7430 to store_immediate_info ctor.
7432 2017-11-09 Jim Wilson <jimw@sifive.com>
7434 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
7435 (scan_prog_file): Likewise.
7437 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7439 * bb-reorder.c (max_entry_frequency): Remove.
7440 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
7441 connect_traces, push_to_next_round_p): Remove prototypes.
7442 (find_traces_1_round): Use counts only.
7443 (push_to_next_round_p): Likewise.
7444 (find_traces): Likewise.
7445 (rotate_loop): Likewise.
7446 (find_traces_1_round): Likewise.
7447 (connect_traces): Likewise.
7448 (edge_order): Likewise.
7450 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
7452 * config/arm/arm.c (output_return_instruction): Add comments to
7453 indicate requirement for cmse_nonsecure_entry return to account
7454 for the size of clearing instruction output here.
7455 (thumb_exit): Likewise.
7456 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
7457 return in hardfloat mode.
7459 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
7461 * config/rs6000/rs6000.c (machine_function): Add a bool,
7462 "toc_is_wrapped_separately".
7463 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
7464 if it wasn't explicitly set or unset, we are optimizing for speed, and
7465 doing separate shrink-wrapping.
7466 (rs6000_get_separate_components): Enable the TOC component if
7467 saving the TOC register in the prologue.
7468 (rs6000_components_for_bb): Handle the TOC component.
7469 (rs6000_emit_prologue_components): Store the TOC register where needed.
7470 (rs6000_set_handled_components): Mark TOC as handled, if handled.
7471 (rs6000_emit_prologue): Don't save the TOC if that is already done.
7473 2017-11-09 Martin Jambor <mjambor@suse.cz>
7475 * ipa-param-manipulation.c: New file.
7476 * ipa-param-manipulation.h: Likewise.
7477 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
7478 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
7479 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
7480 (ipa_parm_adjustment): Likewise.
7481 (ipa_parm_adjustment_vec): Likewise.
7482 (ipa_get_vector_of_formal_parms): Moved declaration to
7483 ipa-param-manipulation.h.
7484 (ipa_get_vector_of_formal_parm_types): Likewise.
7485 (ipa_modify_formal_parameters): Likewise.
7486 (ipa_modify_call_arguments): Likewise.
7487 (ipa_combine_adjustments): Likewise.
7488 (ipa_dump_param_adjustments): Likewise.
7489 (ipa_modify_expr): Likewise.
7490 (ipa_get_adjustment_candidate): Likewise.
7491 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
7492 ipa-param-manipulation.c.
7493 (ipa_get_vector_of_formal_parm_types): Likewise.
7494 (ipa_modify_formal_parameters): Likewise.
7495 (ipa_modify_call_arguments): Likewise.
7496 (ipa_modify_expr): Likewise.
7497 (get_ssa_base_param): Likewise.
7498 (ipa_get_adjustment_candidate): Likewise.
7499 (index_in_adjustments_multiple_times_p): Likewise.
7500 (ipa_combine_adjustments): Likewise.
7501 (ipa_dump_param_adjustments): Likewise.
7502 * tree-sra.c: Also include ipa-param-manipulation.h
7503 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
7506 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7507 Alan Hayward <alan.hayward@arm.com>
7508 David Sherwood <david.sherwood@arm.com>
7510 * doc/sourcebuild.texi (vect_masked_store): Document.
7512 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7513 Alan Hayward <alan.hayward@arm.com>
7514 David Sherwood <david.sherwood@arm.com>
7516 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
7518 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7519 Alan Hayward <alan.hayward@arm.com>
7520 David Sherwood <david.sherwood@arm.com>
7522 * doc/sourcebuild.texi (vect_variable_length): Document.
7524 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7525 Alan Hayward <alan.hayward@arm.com>
7526 David Sherwood <david.sherwood@arm.com>
7528 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
7530 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7531 Alan Hayward <alan.hayward@arm.com>
7532 David Sherwood <david.sherwood@arm.com>
7534 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
7536 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7537 Alan Hayward <alan.hayward@arm.com>
7538 David Sherwood <david.sherwood@arm.com>
7540 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
7541 previously undocumented selectors.
7542 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
7544 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7546 * doc/rtl.texi (const_vector): Say that elements can be
7547 const_wide_ints too.
7548 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
7549 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
7550 (gen_vec_duplicate): Use it instead of CONSTANT_P.
7551 * optabs.c (expand_vector_broadcast): Likewise.
7553 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7554 Alan Hayward <alan.hayward@arm.com>
7555 David Sherwood <david.sherwood@arm.com>
7557 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
7558 scaled index even if the unscaled address was invalid.
7559 Don't increase the complexity of using a scale in that case.
7561 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7562 Alan Hayward <alan.hayward@arm.com>
7563 David Sherwood <david.sherwood@arm.com>
7565 * doc/rtl.texi: Rewrite the subreg rules so that they partition
7566 the inner register into REGMODE_NATURAL_SIZE bytes rather than
7567 UNITS_PER_WORD bytes.
7568 * emit-rtl.c (validate_subreg): Divide subregs into blocks
7569 based on REGMODE_NATURAL_SIZE of the inner mode.
7570 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
7571 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
7572 * expmed.c (lowpart_bit_field_p): Divide the value up into
7573 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
7574 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
7575 whether something is likely to occupy more than one register.
7577 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7580 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7581 profile_count::adjust_for_ipa_scaling.
7582 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7583 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7585 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7587 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7589 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7590 (split_group): Add total_orig and total_new arguments, estimate the
7591 number of statements related to the store group without store merging
7592 and with store merging.
7593 (imm_store_chain_info::output_merged_store): Adjust split_group
7594 callers, punt if estimated number of statements with store merging
7595 is not smaller than estimated number of statements without it.
7597 (handled_load): Remove has_single_use checks.
7598 (pass_store_merging::process_store): Likewise.
7600 2017-11-09 Richard Biener <rguenther@suse.de>
7602 PR tree-optimization/82902
7603 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7605 2017-11-09 Martin Liska <mliska@suse.cz>
7608 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7611 2017-11-09 Martin Liska <mliska@suse.cz>
7613 PR tree-optimization/82669
7614 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7616 2017-11-09 Martin Liska <mliska@suse.cz>
7618 PR gcov-profile/48463
7619 * coverage.c (coverage_begin_function): Output also end locus
7620 of a function and information whether the function is
7622 * gcov-dump.c (tag_function): Parse and print the information.
7623 * gcov.c (INCLUDE_MAP): Add include.
7624 (INCLUDE_SET): Likewise.
7625 (struct line_info): Move earlier in the source file because
7626 of vector<line_info> in function_info structure.
7627 (line_info::line_info): Likewise.
7628 (line_info::has_block): Likewise.
7629 (struct source_info): Add new member index.
7630 (source_info::get_functions_at_location): New function.
7631 (function_info::group_line_p): New function.
7632 (output_intermediate_line): New function.
7633 (output_intermediate_file): Use the mentioned function.
7634 (struct function_start): New.
7635 (struct function_start_pair_hash): Likewise.
7636 (process_file): Add code that identifies group functions.
7637 Assign lines either to global or function scope.
7638 (generate_results): Skip artificial functions.
7639 (find_source): Assign index for each source file.
7640 (read_graph_file): Read new flag artificial and end_line.
7641 (add_line_counts): Assign it either to global of function scope.
7642 (accumulate_line_counts): Isolate core of the function to
7643 accumulate_line_info and call it for both function and global
7645 (accumulate_line_info): New function.
7646 (output_line_beginning): Fix GNU coding style.
7647 (print_source_line): New function.
7648 (output_line_details): Likewise.
7649 (output_function_details): Likewise.
7650 (output_lines): Iterate both source (global) scope and function
7652 (struct function_line_start_cmp): New class.
7653 * doc/gcov.texi: Reflect changes in documentation.
7655 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7658 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7659 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7660 and similarly for not instead of neg.
7662 2017-11-08 Andi Kleen <ak@linux.intel.com>
7664 * config/i386/i386.opt: Add -mforce-indirect-call.
7665 * config/i386/predicates.md: Check for flag_force_indirect_call.
7666 * doc/invoke.texi: Document -mforce-indirect-call
7668 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7670 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7672 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7673 config/riscv/riscv.c (predict.h): New include.
7674 (riscv_slow_unaligned_access_p): No longer static.
7675 (riscv_block_move_straight): Add require.
7676 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7678 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7681 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7682 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7683 nonimmediate_operand predicate for operand 1 instead of
7686 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7688 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7690 * config/aarch64/constraints.md (Uml): New constraint.
7691 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7694 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7696 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7697 of two vec_duplicates into a vec_concat.
7699 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7701 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7702 Simplify vec_merge of vec_duplicate and vec_concat.
7703 * config/aarch64/constraints.md (Utq): New constraint.
7704 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7707 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7709 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7710 Simplify vec_merge of vec_duplicate and const_vector.
7711 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7713 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7714 mode iterator. Update predicate on operand 1 to
7715 handle non-const_vec constants. Delete constraints.
7716 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7718 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7720 PR tree-optimization/78821
7721 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7723 (store_operand_info::store_operand_info): Initialize it to false.
7724 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7725 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7726 store in the group, and if chain_info is non-NULL, to ignore altogether
7728 (compatible_load_p): Fail if bit_not_p does not match.
7729 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7730 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7731 (pass_store_merging::process_store): Adjust
7732 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7733 call terminate_all_aliasing_chains newly when adding a store into
7734 a chain with non-NULL chain_info.
7736 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7738 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7740 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7742 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7744 (aarch64_layout_frame): Initialise emit_frame_chain.
7745 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7746 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7748 2017-11-08 Martin Liska <mliska@suse.cz>
7750 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7751 of gimple_call_internal_p.
7753 2017-11-07 Tom de Vries <tom@codesourcery.com>
7755 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7757 2017-11-07 Tom de Vries <tom@codesourcery.com>
7759 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7762 2017-11-08 Martin Liska <mliska@suse.cz>
7765 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7767 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7769 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7771 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7774 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7775 (*cmp<mode>_ccz_1): New insn with $k alternative.
7778 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7779 RTX_COMM_COMPARE as commutative as well.
7780 (ix86_binary_operator_ok): Formatting fix.
7781 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7782 *<code><mode>3<mask_name><round_saeonly_name>,
7783 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7784 *<s>mul<mode>3_highpart<mask_name>,
7785 *vec_widen_umult_even_v16si<mask_name>,
7786 *vec_widen_umult_even_v8si<mask_name>,
7787 *vec_widen_umult_even_v4si<mask_name>,
7788 *vec_widen_smult_even_v16si<mask_name>,
7789 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7790 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7791 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7792 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7793 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7794 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7795 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7796 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7797 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7798 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7799 ix86_binary_operator_ok. Formatting fixes.
7800 (*<plusminus_insn><mode>3<mask_name><round_name>,
7801 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7804 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7806 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7807 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7808 eq and ne if TARGET_ISEL.
7809 (cmp): New code_iterator.
7810 (UNS, UNSU_, UNSIK): New code_attrs.
7811 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7812 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7814 ("eq<mode>3"): ... this.
7815 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7817 ("ne<mode>3"): ... this.
7819 2017-11-07 Julia Koval <julia.koval@intel.com>
7822 * common/config/i386/i386-common.c
7823 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
7824 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
7825 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
7826 * config/i386/i386.opt: Ditto.
7827 * config/i386/i386.c (ix86_target_string): Ditto.
7828 (ix86_option_override_internal): Ditto.
7829 (ix86_init_mpx_builtins): Move MPX to args2.
7830 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
7831 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7832 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7833 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7834 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7835 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7836 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
7838 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
7841 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
7842 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
7843 (zero-extendsidi peephole2): Remove peephole.
7845 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7848 * stor-layout.c: Include attribs.h.
7849 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
7850 explicit lookup of "aligned" attribute.
7852 2017-11-07 Andrew Waterman <andrew@sifive.com>
7854 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
7855 (riscv_expand_block_move): Likewise.
7856 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
7858 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
7859 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
7860 * config/riscv/riscv.c (riscv_block_move_straight): New function.
7861 (riscv_adjust_block_mem): Likewise.
7862 (riscv_block_move_loop): Likewise.
7863 (riscv_expand_block_move): Likewise.
7864 * config/riscv/riscv.md (movmemsi): New pattern.
7866 2017-11-07 Michael Clark <michaeljclark@mac.com>
7868 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
7869 (MUSL_DYNAMIC_LINKER): Likewise.
7871 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
7873 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
7874 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
7877 2017-11-07 Richard Biener <rguenther@suse.de>
7879 * match.pd: Fix build.
7881 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
7882 Jackson Woodruff <jackson.woodruff@arm.com>
7884 PR tree-optimization/71026
7885 * match.pd: Canonicalize negate in division.
7887 2017-11-07 Sudakshina Das <sudi.das@arm.com>
7890 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
7892 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7894 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
7895 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
7897 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7899 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
7900 non-scalar integral types.
7901 * match.pd (negate_expr_p): Handle MINUS_EXPR.
7902 (-(A-B), -(~A)): New transformations.
7904 2017-11-07 Tom de Vries <tom@codesourcery.com>
7906 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
7907 semicolon after "do {} while (0)".
7908 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7909 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7910 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7911 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7912 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7914 2017-11-07 Tom de Vries <tom@codesourcery.com>
7916 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
7917 after "do {} while (0)".
7918 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7919 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7920 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7921 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7922 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7924 2017-11-07 Tom de Vries <tom@codesourcery.com>
7927 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
7929 (arm_rtx_costs_internal): Add missing semicolon after
7930 HANDLE_NARROW_SHIFT_ARITH call.
7932 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7934 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7935 disable isel if it was not set explicitly.
7937 2017-11-06 James Bowman <james.bowman@ftdichip.com>
7939 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
7940 (add_type_attribute) likewise.
7942 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
7944 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
7945 of struct ix86_frame.
7946 (ix86_initial_elimination_offset): Likewise.
7947 (ix86_expand_split_stack_prologue): Likewise.
7949 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
7951 * tree-vrp.h (enum value_range_type): Update stale comment.
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_expand_vec_perm)
7958 (aarch64_expand_vec_perm_const): Take the number of units too.
7959 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
7960 (aarch64_expand_vec_perm_const): Likewise.
7961 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
7962 (vec_perm<mode>): Update accordingly.
7964 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7965 Alan Hayward <alan.hayward@arm.com>
7966 David Sherwood <david.sherwood@arm.com>
7968 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
7969 Take the number of units too.
7970 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
7971 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
7972 but check for a vector mode before rather than after the call.
7973 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7974 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
7975 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
7976 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
7977 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
7978 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
7979 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
7980 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
7981 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
7982 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
7983 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
7984 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
7985 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
7986 (aarch64_sqdmull2_n<mode>): Update accordingly.
7988 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7989 Alan Hayward <alan.hayward@arm.com>
7990 David Sherwood <david.sherwood@arm.com>
7992 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
7993 the number of units too.
7994 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
7995 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
7996 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
7997 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
7998 (vec_store_lanesxi<mode>): Update accordingly.
8000 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8001 Alan Hayward <alan.hayward@arm.com>
8002 David Sherwood <david.sherwood@arm.com>
8004 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8005 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8006 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8007 of units rather than the mode.
8008 * config/aarch64/iterators.md (nunits): New mode attribute.
8009 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8010 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8011 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8012 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8013 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8014 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8015 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8016 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8017 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8018 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8019 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8020 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8021 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8022 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8023 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8024 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8025 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8026 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8027 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8028 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8029 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8030 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8031 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8032 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8033 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8034 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8035 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8036 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8037 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8038 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8039 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8040 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8041 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8042 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8043 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8044 (aarch64_simd_vec_setv2di): Likewise.
8046 2017-11-06 Carl Love <cel@us.ibm.com>
8048 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8050 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8051 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8053 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8055 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8056 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8057 Add power 8 macro expansions.
8058 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8059 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8060 power 8 instructions. (VSX_XXBR): Add iterator.
8062 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8064 * config/arm/arm.md (predicable_short_it): Change default to "no",
8065 improve documentation, remove uses that are identical to the default.
8066 (enabled_for_depr_it): Rename to enabled_for_short_it.
8067 * gcc/config/arm/arm-fixed.md (predicable_short_it):
8068 Remove default uses.
8069 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
8070 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
8071 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
8072 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
8074 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
8077 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
8078 float128 helper macros, which are no longer used after deleting
8079 the old 'q' built-in functions, and moving the round to odd
8080 built-in functions to being special built-in functions.
8081 (BU_FLOAT128_2): Likewise.
8082 (BU_FLOAT128_1_HW): Likewise.
8083 (BU_FLOAT128_2_HW): Likewise.
8084 (BU_FLOAT128_3_HW): Likewise.
8085 (FABSQ): Delete old 'q' built-in functions.
8086 (COPYSIGNQ): Likewise.
8087 (SQRTF128_ODD): Move round to odd built-in functions to be
8088 special built-in functions, so that we can handle
8089 -mabi=ieeelongdouble.
8090 (TRUNCF128_ODD): Likewise.
8091 (ADDF128_ODD): Likewise.
8092 (SUBF128_ODD): Likewise.
8093 (MULF128_ODD): Likewise.
8094 (DIVF128_ODD): Likewise.
8095 (FMAF128_ODD): Likewise.
8096 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
8097 built-in names to 'f128'.
8098 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
8099 old 'q' built-in functions, as the machine independent code for
8100 'f128' built-in functions handles this.
8101 (rs6000_expand_builtin): Add expansion for float128 round to odd
8102 functions, keying off on -mabi=ieeelongdouble of whether to use
8103 the KFmode or TFmode variant.
8104 (rs6000_init_builtins): Initialize the _Float128 round to odd
8106 * doc/extend.texi (PowerPC Built-in Functions): Document the old
8107 _Float128 'q' built-in functions are now mapped into the new
8108 'f128' built-in functions.
8110 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
8112 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
8113 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
8115 2017-11-06 David Malcolm <dmalcolm@redhat.com>
8118 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
8119 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
8120 * toplev.c: Include "ipa-fnsummary.h".
8121 (toplev::finalize): Call ipa_fnsummary_c_finalize.
8123 2017-11-06 Jakub Jelinek <jakub@redhat.com>
8125 PR tree-optimization/82838
8126 * gimple-ssa-store-merging.c
8127 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
8128 on a separate gimple_seq which is then appended to seq.
8130 2017-11-06 Jeff Law <law@redhat.com>
8133 * config/i386/i386.c (PROBE_INTERVAL): Remove.
8134 (get_probe_interval): New functions.
8135 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
8136 (ix86_adjust_stack_and_probe): Likewise.
8137 (output_adjust_stack_and_probe): Likewise.
8138 (ix86_emit_probe_stack_range): Likewise.
8139 (ix86_expand_prologue): Likewise.
8141 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8143 PR tree-optimization/82816
8144 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
8145 if the modes of the two types are the same.
8146 (convert_plusminus_to_widen): Likewise.
8148 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8150 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
8152 (usadv16qi): New define_expand.
8153 (usadv8hi): New define_expand.
8155 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
8158 * ipa-inline-transform.c (update_noncloned_frequencies): Always
8160 (inline_transform): Likewise.
8161 * predict.c (counts_to_freqs): Remove useless conditional.
8162 * profile-count.h (profile_count::apply_scale): Move sanity check.
8163 * tree-inline.c (copy_bb): Always scale.
8164 (copy_cfg_body): Likewise.
8166 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
8169 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
8171 (*cmp_ite0): Add enabled_for_depr_it attribute.
8172 (*cmp_ite1): Likewise.
8174 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8176 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
8179 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8181 * tree-vrp.c (vrp_int_const_binop): Return true on success and
8182 return the value by pointer.
8183 (extract_range_from_multiplicative_op_1): Update accordingly.
8184 Return as soon as an operation fails.
8186 2017-11-05 Tom de Vries <tom@codesourcery.com>
8189 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
8190 (DEF_SANITIZER_BUILTIN): ... here.
8191 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
8192 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
8194 2017-11-05 Tom de Vries <tom@codesourcery.com>
8197 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
8199 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
8200 ASM_OUTPUT_BEFORE_CASE_LABEL call.
8201 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
8203 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8204 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8205 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8207 2017-11-05 Tom de Vries <tom@codesourcery.com>
8210 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
8213 2017-11-04 Michael Clark <michaeljclark@mac.com>
8215 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
8216 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
8218 (*addsi3_extended): Likewise.
8219 (*addsi3_extended2): Likewise.
8220 (<optab>si3): Likewise.
8221 (<optab>di3): Likewise.
8222 (<optab><mode>3): Likewise.
8223 (<*optabe>si3_internal): Likewise.
8224 (zero_extendqi<SUPERQI:mode>2): Likewise.
8225 (*add<mode>hi3): Likewise.
8226 (*xor<mode>hi3): Likewise.
8227 (<optab>di3): Likewise.
8228 (*<optab>si3_extend): Likewise.
8229 (*sge<u>_<X:mode><GPR:mode>): Likewise.
8230 (*slt<u>_<X:mode><GPR:mode>): Likewise.
8231 (*sle<u>_<X:mode><GPR:mode>): Likewise.
8233 2017-11-04 Andrew Waterman <andrew@sifive.com>
8235 * config/riscv/riscv.c (riscv_option_override): Conditionally set
8236 TARGET_STRICT_ALIGN based upon -mtune argument.
8238 2017-11-04 Andrew Waterman <andrew@sifive.com>
8240 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
8242 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
8244 * config/i386/i386.c (choose_basereg): Use optional scratch
8245 register and add assertion.
8246 (x86_emit_outlined_ms2sysv_save): Use scratch register when
8247 needed, and don't allocate stack.
8248 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
8249 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
8250 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
8252 2017-11-03 Jeff Law <law@redhat.com>
8254 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
8255 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
8256 to probe at the start of a noreturn function.
8258 2017-11-03 Jakub Jelinek <jakub@redhat.com>
8260 PR tree-optimization/78821
8261 * gimple-ssa-store-merging.c: Update the file comment.
8262 (MAX_STORE_ALIAS_CHECKS): Define.
8263 (struct store_operand_info): New type.
8264 (store_operand_info::store_operand_info): New constructor.
8265 (struct store_immediate_info): Add rhs_code and ops data members.
8266 (store_immediate_info::store_immediate_info): Add rhscode, op0r
8267 and op1r arguments to the ctor, initialize corresponding data members.
8268 (struct merged_store_group): Add load_align_base and load_align
8270 (merged_store_group::merged_store_group): Initialize them.
8271 (merged_store_group::do_merge): Update them.
8272 (merged_store_group::apply_stores): Pick the constant for
8273 encode_tree_to_bitpos from one of the two operands, or skip
8274 encode_tree_to_bitpos if neither operand is a constant.
8275 (class pass_store_merging): Add process_store method decl. Remove
8276 bool argument from terminate_all_aliasing_chains method decl.
8277 (pass_store_merging::terminate_all_aliasing_chains): Remove
8278 var_offset_p argument and corresponding handling.
8279 (stmts_may_clobber_ref_p): New function.
8280 (compatible_load_p): New function.
8281 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
8282 if there is overlap and rhs_code is not INTEGER_CST. For
8283 non-overlapping stores terminate group if rhs is not mergeable.
8284 (get_alias_type_for_stmts): Change first argument from
8285 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
8286 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
8287 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
8289 (get_location_for_stmts): Change first argument from
8290 auto_vec<gimple *> & to vec<gimple *> &.
8291 (struct split_store): Remove orig_stmts data member, add orig_stores.
8292 (split_store::split_store): Create orig_stores rather than orig_stmts.
8293 (find_constituent_stmts): Renamed to ...
8294 (find_constituent_stores): ... this. Change second argument from
8295 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
8296 to info structures rather than the statements.
8297 (split_group): Rename ALLOW_UNALIGNED argument to
8298 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
8299 it. Adjust find_constituent_stores caller.
8300 (imm_store_chain_info::output_merged_store): Handle rhs_code other
8301 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
8302 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
8303 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
8304 (mem_valid_for_store_merging): New function.
8305 (handled_load): New function.
8306 (pass_store_merging::process_store): New method.
8307 (pass_store_merging::execute): Use process_store method. Adjust
8308 terminate_all_aliasing_chains caller.
8310 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8312 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8313 Return true for more constants, symbols and label references.
8314 (aarch64_valid_floating_const): Remove unused function.
8316 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8319 * config/aarch64/aarch64.c (aarch64_layout_frame):
8320 Undo forcing of LR at bottom of frame.
8322 2017-11-03 Jeff Law <law@redhat.com>
8325 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
8326 for int_registers_saved.
8328 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
8329 extracted from tree-ssa-dom.c.
8330 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
8331 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
8332 (record_equivalences_from_incoming_edge): Add additional argument
8333 to single_pred_edge_ignoring_loop_edges call.
8334 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
8335 (uncprop_dom_walker::before_dom_children): Add additional argument
8336 to single_pred_edge_ignoring_loop_edges call.
8337 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
8338 single_pred_edge_ignoring_loop_edges rather than open coding.
8339 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
8341 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
8343 * match.pd (-(-A)): Rewrite.
8345 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8347 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
8348 (rs6000_emit_int_cmove): New declaration.
8349 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
8350 (rs6000_emit_sISEL): Delete.
8351 (rs6000_emit_int_cmove): Make non-static.
8352 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
8353 instead of rs6000_emit_sISEL.
8355 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
8357 * asan.c (create_cond_insert_point): Maintain profile.
8358 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
8360 * basic-block.h (struct basic_block_def): Remove frequency.
8361 (EDGE_FREQUENCY): Use to_frequency
8362 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
8364 (find_traces): Update to use to_frequency.
8365 (find_traces_1_round): Likewise; use only IPA counts.
8366 (bb_to_key): Likewise.
8367 (connect_traces): Use IPA counts only.
8368 (copy_bb_p): Update to use to_frequency.
8369 (fix_up_crossing_landing_pad): Likewise.
8370 (sanitize_hot_paths): Likewise.
8371 * bt-load.c (basic_block_freq): Likewise.
8372 * cfg.c (init_flow): Set count_max to uninitialized.
8373 (check_bb_profile): Remove frequencies; check counts.
8374 (dump_bb_info): Do not dump frequencies.
8375 (update_bb_profile_for_threading): Update counts only.
8376 (scale_bbs_frequencies_int): Likewise.
8377 (MAX_SAFE_MULTIPLIER): Remove.
8378 (scale_bbs_frequencies_gcov_type): Update counts only.
8379 (scale_bbs_frequencies_profile_count): Update counts only.
8380 (scale_bbs_frequencies): Update counts only.
8381 * cfg.h (struct control_flow_graph): Add count-max.
8382 (update_bb_profile_for_threading): Update prototype.
8383 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
8384 (find_many_sub_basic_blocks): Likewise.
8385 * cfgcleanup.c (try_forward_edges): Likewise.
8386 (try_crossjump_to_edge): Likewise.
8387 * cfgexpand.c (expand_gimple_cond): Likewise.
8388 (expand_gimple_tailcall): Likewise.
8389 (construct_init_block): Likewise.
8390 (construct_exit_block): Likewise.
8391 * cfghooks.c (verify_flow_info): Check consistency of counts.
8392 (dump_bb_for_graph): Do not dump frequencies.
8393 (split_block_1): Do not update frequencies.
8394 (split_edge): Do not update frequencies.
8395 (make_forwarder_block): Do not update frequencies.
8396 (duplicate_block): Do not update frequencies.
8397 (account_profile_record): Do not update frequencies.
8398 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
8399 for global heuristics.
8400 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
8401 (expected_loop_iterations_unbounded): Use counts only.
8402 * cfgloopmanip.c (scale_loop_profile): Simplify.
8403 (create_empty_loop_on_edge): Simplify
8405 (duplicate_loop_to_header_edge): Simplify
8406 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
8407 (update_br_prob_note): Take care of removing note when profile
8409 (relink_block_chain): Do not dump frequency.
8410 (rtl_account_profile_record): Use to_frequency.
8411 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
8412 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
8413 (cgraph_update_edges_for_call_stmt_node): Likewise.
8414 (cgraph_edge::verify_count_and_frequency): Update.
8415 (cgraph_node::verify_node): Temporarily disable frequency verification.
8416 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
8417 to_cgraph_frequency.
8418 (cgraph_edge::rebuild_edges): Convert to ipa counts.
8419 * cgraphunit.c (init_lowered_empty_function): Do not initialize
8421 (cgraph_node::expand_thunk): Update profile.
8422 * except.c (dw2_build_landing_pads): Do not update frequency.
8423 * final.c (compute_alignments): Use to_frequency.
8424 (dump_basic_block_info): Do not dump frequency.
8425 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
8426 (dump_gimple_bb_header): Do not dump frequency.
8427 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
8429 * gimple-streamer-in.c (input_bb): Do not stream frequency.
8430 * gimple-streamer-out.c (output_bb): Do not stream frequency.
8431 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
8432 (init_before_recovery): Do not update frequency.
8433 (sched_create_recovery_edges): Do not update frequency.
8434 * hsa-gen.c (convert_switch_statements): Do not update frequency.
8435 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
8436 (ipa_cp_c_finalize): Set max_count to uninitialized.
8437 * ipa-fnsummary.c (get_minimal_bb): Use counts.
8438 (param_change_prob): Use counts.
8439 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
8441 * ipa-split.c (consider_split): Use to_frequency.
8442 (split_function): Use to_frequency.
8443 * ira-build.c (loop_compare_func): Likewise.
8444 (mark_loops_for_removal): Likewise.
8445 (mark_all_loops_for_removal): Likewise.
8446 * loop-doloop.c (doloop_modify): Do not update frequency.
8447 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
8449 * lto-streamer-in.c (input_function): Update count_max.
8450 * omp-expand.c (expand_omp_taskreg): Update count_max.
8451 * omp-simd-clone.c (simd_clone_adjust): Update profile.
8452 * predict.c (maybe_hot_frequency_p): Use to_frequency.
8453 (maybe_hot_count_p): Use ipa counts only.
8454 (maybe_hot_bb_p): Simplify.
8455 (maybe_hot_edge_p): Simplify.
8456 (probably_never_executed): Do not take frequency argument.
8457 (probably_never_executed_bb_p): Do not pass frequency.
8458 (probably_never_executed_edge_p): Likewise.
8459 (combine_predictions_for_bb): Check that profile is nonzero.
8460 (propagate_freq): Do not set frequency.
8461 (drop_profile): Simplify.
8462 (counts_to_freqs): Simplify.
8463 (expensive_function_p): Use to_frequency.
8464 (propagate_unlikely_bbs_forward): Simplify.
8465 (determine_unlikely_bbs): Simplify.
8466 (estimate_bb_frequencies): Add hack to silence graphite issues.
8467 (compute_function_frequency): Use ipa counts.
8468 (pass_profile::execute): Update.
8469 (rebuild_frequencies): Use counts only.
8470 (force_edge_cold): Use counts only.
8471 * profile-count.c (profile_count::dump): Dump new count types.
8472 (profile_count::differs_from_p): Check compatiblity.
8473 (profile_count::to_frequency): New function.
8474 (profile_count::to_cgraph_frequency): New function.
8475 * profile-count.h (struct function): Declare.
8476 (enum profile_quality): Add profile_guessed_local and
8477 profile_guessed_global0.
8478 (class profile_proability): Decrease number of bits to 29;
8479 update from_reg_br_prob_note and to_reg_br_prob_note.
8480 (class profile_count: Update comment; decrease number of bits
8481 to 61. Check compatibility.
8482 (profile_count::compatible_p): New private member function.
8483 (profile_count::ipa_p): New member function.
8484 (profile_count::operator<): Handle global zero correctly.
8485 (profile_count::operator>): Handle global zero correctly.
8486 (profile_count::operator<=): Handle global zero correctly.
8487 (profile_count::operator>=): Handle global zero correctly.
8488 (profile_count::nonzero_p): New member function.
8489 (profile_count::force_nonzero): New member function.
8490 (profile_count::max): New member function.
8491 (profile_count::apply_scale): Handle IPA scalling.
8492 (profile_count::guessed_local): New member function.
8493 (profile_count::global0): New member function.
8494 (profile_count::ipa): New member function.
8495 (profile_count::to_frequency): Declare.
8496 (profile_count::to_cgraph_frequency): Declare.
8497 * profile.c (OVERLAP_BASE): Delete.
8498 (compute_frequency_overlap): Delete.
8499 (compute_branch_probabilities): Do not use compute_frequency_overlap.
8500 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
8501 * sched-ebb.c (rank): Use counts only.
8502 * shrink-wrap.c (handle_simple_exit): Use counts only.
8503 (try_shrink_wrapping): Use counts only.
8504 (place_prologue_for_one_component): Use counts only.
8505 * tracer.c (find_best_predecessor): Use to_frequency.
8506 (find_trace): Use to_frequency.
8507 (tail_duplicate): Use to_frequency.
8508 * trans-mem.c (expand_transaction): Do not update frequency.
8509 * tree-call-cdce.c: Do not update frequency.
8510 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
8511 (gimple_merge_blocks): Likewise.
8512 (gimple_split_edge): Likewise.
8513 (gimple_duplicate_sese_region): Likewise.
8514 (gimple_duplicate_sese_tail): Likewise.
8515 (move_sese_region_to_fn): Likewise.
8516 (gimple_account_profile_record): Likewise.
8517 (insert_cond_bb): Likewise.
8518 * tree-complex.c (expand_complex_div_wide): Likewise.
8519 * tree-eh.c (lower_resx): Update profile.
8520 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
8522 (initialize_cfun): Do not initialize frequencies
8523 (freqs_to_counts): Delete.
8524 (copy_cfg_body): Ignore count parameter.
8525 (copy_body): Update.
8526 (expand_call_inline): Update count_max.
8527 (optimize_inline_calls): Update count_max.
8528 (tree_function_versioning): Update count_max.
8529 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
8530 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
8532 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
8533 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
8534 (try_peel_loop): Likewise.
8535 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
8537 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
8538 (tree_transform_and_unroll_loop): Do not use frequencies
8539 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
8540 Use reliable prediction only.
8541 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
8542 * tree-ssa-sink.c (select_best_block): Use to_frequency.
8543 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
8544 probability scaling.
8545 * tree-ssa-threadupdate.c (create_block_for_threading): Do
8546 not update frequency
8547 (any_remaining_duplicated_blocks): Likewise.
8548 (update_profile): Likewise.
8549 (estimated_freqs_path): Delete.
8550 (freqs_to_counts_path): Delete.
8551 (clear_counts_path): Delete.
8552 (ssa_fix_duplicate_block_edges): Likewise.
8553 (duplicate_thread_path): Likewise.
8554 * tree-switch-conversion.c (gen_inbound_check): Use counts.
8555 * tree-tailcall.c (decrease_profile): Do not update frequency.
8556 (eliminate_tail_call): Likewise.
8557 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
8558 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
8559 (optimize_mask_stores): Likewise.
8560 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8561 * ubsan.c (ubsan_expand_null_ifn): Update profile.
8562 (ubsan_expand_ptr_ifn): Update profile.
8563 * value-prof.c (gimple_ic): Simplify.
8564 * value-prof.h (gimple_ic): Update prototype.
8565 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
8566 * ipa-inline.c (compute_uninlined_call_time): Be sure that
8568 (want_inline_self_recursive_call_p): Likewise.
8569 (resolve_noninline_speculation): Only cummulate defined counts.
8570 (inline_small_functions): Use nonzero_p.
8571 (ipa_inline): Do not access freed node.
8573 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8575 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8576 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8578 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8580 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8581 non-legitimate address.
8583 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8585 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8586 (*lt0_<mode>di, *lt0_<mode>si): New.
8588 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8590 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8591 TARGET_PAIRED_FLOAT.
8593 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8594 Jim Wilson <jim.wilson@linaro.org>
8596 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8597 * config/aarch64/aarch64-tune.md: Regenerated.
8598 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8599 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8601 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8603 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8604 (arc_expand_prologue): Restore blink for millicode.
8605 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8607 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8610 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8611 gen_vec_duplicate after forcing the scalar into a register.
8613 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8615 * combine (try_combine): Print the insns input to try_combine to the
8618 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8621 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8622 Remove second argument from aarch64_process_target_attr call.
8623 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8625 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8627 (aarch64_handle_attr_arch): Remove second argument.
8628 (aarch64_handle_attr_cpu): Ditto.
8629 (aarch64_handle_attr_tune): Ditto.
8630 (aarch64_handle_attr_isa_flags): Ditto.
8631 (aarch64_process_one_target_attr): Ditto.
8632 (aarch64_process_target_attr): Ditto.
8633 (aarch64_option_valid_attribute_p): Remove second argument.
8634 on aarch64_process_target_attr call.
8636 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8638 * diagnostic.c: Include "selftest-diagnostic.h".
8639 (selftest::assert_location_text): New function.
8640 (selftest::test_diagnostic_get_location_text): New function.
8641 (selftest::diagnostic_c_tests): Call it.
8643 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8645 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8646 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8647 (class selftest::test_diagnostic_context): Move to...
8648 * selftest-diagnostic.c: New file.
8649 * selftest-diagnostic.h: New file.
8651 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8653 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8654 offset range for FT32B.
8655 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8656 * config/ft32/ft32.md: Add TARGET_NOPM.
8657 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8658 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8660 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8662 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8664 2017-11-02 Jeff Law <law@redhat.com>
8666 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8667 virtual keyword on FINAL OVERRIDE members.
8669 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8670 virtuals together. Add virtual destructor.
8671 (substitute_and_fold_engine): Similarly.
8673 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8675 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8677 2017-11-02 Richard Biener <rguenther@suse.de>
8679 PR tree-optimization/82795
8680 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8682 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8684 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8686 (gcc_GAS_CHECK_FEATURE): Remove.
8687 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8688 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8689 * configure: Regenerate.
8691 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8693 * config/arc/arc.c (hwloop_optimize): Account for empty
8696 2017-11-02 Richard Biener <rguenther@suse.de>
8699 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8700 Truncate ARRAY_REF index and element size.
8702 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8704 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8706 2017-11-01 Jeff Law <law@redhat.com>
8708 * tree-ssa-ccp.c (ccp_folder): New class derived from
8709 substitute_and_fold_engine.
8710 (ccp_folder::get_value): New member function.
8711 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8712 (ccp_fold_stmt): Remove prototype.
8713 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8714 * tree-ssa-copy.c (copy_folder): New class derived from
8715 substitute_and_fold_engine.
8716 (copy_folder::get_value): Renamed from get_value.
8717 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8718 * tree-vrp.c (vrp_folder): New class derived from
8719 substitute_and_fold_engine.
8720 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8721 (vrp_folder::get_value): New member function.
8722 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8723 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8724 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8725 provide a class interface to folder/substitute routines.
8726 (ssa_prop_fold_stmt_fn): Remove typedef.
8727 (ssa_prop_get_value_fn): Likewise.
8728 (subsitute_and_fold): Remove prototype.
8729 (replace_uses_in): Likewise.
8730 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8731 Renamed from replace_uses_in. Call the virtual member function
8732 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8733 (substitute_and_fold_dom_walker): Remove initialization of
8734 data member entries for calbacks. Add substitute_and_fold_engine
8735 member and initialize it.
8736 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8737 member functions for get_value, replace_phi_args_in c
8738 replace_uses_in, and fold_stmt calls.
8739 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8740 substitute_and_fold. Remove assert. Update ctor call.
8742 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8743 (ssa_prop_visit_phi_fn): Likewise.
8744 (class ssa_propagation_engine): New class to provide an interface
8746 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8748 (ssa_prop_visit_phi): Likewise.
8749 (ssa_propagation_engine::simulate_stmt): Moved into class.
8750 Call visit_phi/visit_stmt from the class rather than via
8751 file scoped static variables.
8752 (ssa_propagation_engine::simulate_block): Moved into class.
8753 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8754 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8755 set file scoped statics for the visit_stmt/visit_phi callbacks.
8756 * tree-complex.c (complex_propagate): New class derived from
8757 ssa_propagation_engine.
8758 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8759 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8760 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8762 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8763 ssa_propagation_engine.
8764 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8765 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8766 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8767 * tree-ssa-copy.c (copy_prop): New class derived from
8768 ssa_propagation_engine.
8769 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8770 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8771 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8772 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8773 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8774 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8775 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8777 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8779 PR rtl-optimization/82778
8780 PR rtl-optimization/82597
8781 * compare-elim.c (struct comparison): Add in_a_setter field.
8782 (find_comparison_dom_walker::before_dom_children): Remove killed
8783 bitmap and df_simulate_find_defs call, instead walk the defs.
8784 Compute last_setter and initialize in_a_setter. Merge definitions
8785 with first initialization for a few variables.
8786 (try_validate_parallel): Use insn_invalid_p instead of
8787 recog_memoized. Return insn rather than just the pattern.
8788 (try_merge_compare): Fix up comment. Don't uselessly test if
8789 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8791 (execute_compare_elim_after_reload): Remove df_chain_add_problem
8794 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8795 Alan Hayward <alan.hayward@arm.com>
8796 David Sherwood <david.sherwood@arm.com>
8798 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
8799 aarch64_hard_regno_nregs to get the number of registers
8802 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8803 Alan Hayward <alan.hayward@arm.com>
8804 David Sherwood <david.sherwood@arm.com>
8806 * config/aarch64/constraints.md (Upl): Rename to...
8808 * config/aarch64/aarch64.md
8809 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
8812 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8813 Alan Hayward <alan.hayward@arm.com>
8814 David Sherwood <david.sherwood@arm.com>
8816 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
8817 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8820 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8821 Alan Hayward <alan.hayward@arm.com>
8822 David Sherwood <david.sherwood@arm.com>
8824 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
8825 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
8826 (aarch64_evpc_dup): Generate rtl direcly, rather than using
8828 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
8829 of a single element.
8830 * config/aarch64/iterators.md: Add a comment above the permute
8831 unspecs to say that they are generated directly by
8832 aarch64_expand_vec_perm_const.
8833 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
8835 2017-11-01 Nathan Sidwell <nathan@acm.org>
8837 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
8839 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8841 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
8842 and medany code models, and describe what they do.
8844 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8846 Revert accidental duplicate:
8848 * combine.c (can_change_dest_mode): Reject changes in
8849 REGMODE_NATURAL_SIZE.
8851 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
8853 PR rtl-optimization/64682
8854 PR rtl-optimization/69567
8855 PR rtl-optimization/69737
8856 PR rtl-optimization/82683
8857 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
8858 register mentioned in the note, drop the note, unless it came from I3,
8859 in which case it should go to I3 again.
8861 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8863 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
8864 and return false if not.
8865 (clear_bytes_written_by, live_bytes_read): Update accordingly.
8867 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8869 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
8870 range is known to be empty.
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 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
8877 and CONST_VECTOR_NUNITS instead of computing the number of units from
8878 the byte sizes of the vector and element.
8879 (simplify_binary_operation_1): Likewise.
8880 (simplify_const_binary_operation): Likewise.
8881 (simplify_ternary_operation): Likewise.
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 * var-tracking.c (INT_MEM_OFFSET): Replace with...
8888 (int_mem_offset): ...this new function.
8889 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
8890 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
8891 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
8894 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8895 Alan Hayward <alan.hayward@arm.com>
8896 David Sherwood <david.sherwood@arm.com>
8898 * lower-subreg.c (interesting_mode_p): New function.
8899 (compute_costs, find_decomposable_subregs, decompose_register)
8900 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
8901 (resolve_clobber, dump_choices): Use it.
8903 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8904 Alan Hayward <alan.hayward@arm.com>
8905 David Sherwood <david.sherwood@arm.com>
8907 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
8909 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8910 Alan Hayward <alan.hayward@arm.com>
8911 David Sherwood <david.sherwood@arm.com>
8913 * alias.c (find_base_value, find_base_term): Only process integer
8914 truncations. Check the precision rather than the size.
8916 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8917 Alan Hayward <alan.hayward@arm.com>
8918 David Sherwood <david.sherwood@arm.com>
8920 * machmode.h (is_narrower_int_mode): New function
8921 * optabs.c (expand_float, expand_fix): Use it.
8922 * dwarf2out.c (rotate_loc_descriptor): Likewise.
8924 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8925 Alan Hayward <alan.hayward@arm.com>
8926 David Sherwood <david.sherwood@arm.com>
8928 * rtl.h (narrower_subreg_mode): New function.
8929 * ira-color.c (update_costs_from_allocno): Use it.
8931 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8932 Alan Hayward <alan.hayward@arm.com>
8933 David Sherwood <david.sherwood@arm.com>
8935 * optabs-query.h (convert_optab_p): New function, split out from...
8936 (convert_optab_handler): ...here.
8937 (widening_optab_handler): Delete.
8938 (find_widening_optab_handler): Remove permit_non_widening parameter.
8939 (find_widening_optab_handler_and_mode): Likewise. Provide an
8940 override that operates on mode class wrappers.
8941 * optabs-query.c (widening_optab_handler): Delete.
8942 (find_widening_optab_handler_and_mode): Remove permit_non_widening
8943 parameter. Assert that the two modes are the same class and that
8944 the "from" mode is narrower than the "to" mode. Use
8945 convert_optab_handler instead of widening_optab_handler.
8946 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
8947 instead of widening_optab_handler.
8948 * expr.c (expand_expr_real_2): Update calls to
8949 find_widening_optab_handler.
8950 * optabs.c (expand_widen_pattern_expr): Likewise.
8951 (expand_binop_directly): Take the insn_code as a parameter.
8952 (expand_binop): Only call find_widening_optab_handler for
8953 conversion optabs; use optab_handler otherwise. Update calls
8954 to find_widening_optab_handler and expand_binop_directly.
8955 Use convert_optab_handler instead of widening_optab_handler.
8956 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
8957 find_widening_optab_handler and use scalar_mode rather than
8959 (convert_plusminus_to_widen): Likewise.
8961 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8962 Alan Hayward <alan.hayward@arm.com>
8963 David Sherwood <david.sherwood@arm.com>
8965 * machmode.h (fixed_size_mode): New class.
8966 * rtl.h (get_pool_mode): Return fixed_size_mode.
8967 * gengtype.c (main): Add fixed_size_mode.
8968 * target.def (get_raw_result_mode): Return a fixed_size_mode.
8969 (get_raw_arg_mode): Likewise.
8970 * doc/tm.texi: Regenerate.
8971 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
8972 * targhooks.c (default_get_reg_raw_mode): Likewise.
8973 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
8974 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
8975 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
8976 (msp430_get_raw_result_mode): Likewise.
8977 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
8978 * dbxout.c (dbxout_parms): Require fixed-size modes.
8979 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
8980 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8981 * omp-low.c (lower_oacc_reductions): Likewise.
8982 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
8983 (simplify_subreg): Update accordingly.
8984 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
8985 (force_const_mem): Update accordingly. Return NULL_RTX for modes
8986 that aren't fixed-size.
8987 (get_pool_mode): Return a fixed_size_mode.
8988 (output_constant_pool_2): Take a fixed_size_mode.
8990 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8991 Alan Hayward <alan.hayward@arm.com>
8992 David Sherwood <david.sherwood@arm.com>
8994 * doc/rtl.texi (vec_series): Document.
8995 (const): Say that the operand can be a vec_series.
8996 * rtl.def (VEC_SERIES): New rtx code.
8997 * rtl.h (const_vec_series_p_1): Declare.
8998 (const_vec_series_p): New function.
8999 * emit-rtl.h (gen_const_vec_series): Declare.
9000 (gen_vec_series): Likewise.
9001 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9002 (gen_vec_series): Likewise.
9003 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9004 * simplify-rtx.c (simplify_unary_operation): Handle negations
9006 (simplify_binary_operation_series): New function.
9007 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9008 (test_vector_ops_series): New function.
9009 (test_vector_ops): Call it.
9010 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9011 gen_const_vec_series.
9012 (altivec_lvsr): Likewise.
9013 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9015 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9016 Alan Hayward <alan.hayward@arm.com>
9017 David Sherwood <david.sherwood@arm.com>
9019 * doc/rtl.texi (const): Update description of address constants.
9020 Say that vector constants are allowed too.
9021 * common.md (E, F): Use CONSTANT_P instead of checking for
9023 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9024 checking for CONST_VECTOR.
9025 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9027 * expr.c (expand_expr_real_2): Check for vector modes instead
9028 of checking for CONST_VECTOR.
9029 * rtl.h (const_vec_p): New function.
9030 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9031 (unwrap_const_vec_duplicate): Handle them here too.
9033 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9034 David Malcolm <dmalcolm@redhat.com>
9035 Alan Hayward <alan.hayward@arm.com>
9036 David Sherwood <david.sherwood@arm.com>
9038 * rtl.h (vec_duplicate_p): New function.
9039 * selftest-rtl.c (assert_rtx_eq_at): New function.
9040 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9041 (assert_rtx_eq_at): Declare.
9042 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9043 * selftest-run-tests.c (selftest::run_tests): Call it.
9044 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9045 (simplify_unary_operation_1): Recursively handle vector duplicates.
9046 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9048 (simplify_subreg): Handle subregs of vector duplicates.
9049 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9050 (selftest::simplify_rtx_c_tests): New functions.
9052 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9053 Alan Hayward <alan.hayward@arm.com>
9054 David Sherwood <david.sherwood@arm.com>
9056 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9057 (gen_vec_duplicate): Likewise.
9058 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9060 (gen_const_vector): ...here.
9061 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9062 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9063 whose elements are all equal.
9064 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
9065 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
9066 (simplify_relational_operation): Likewise.
9067 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
9069 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
9070 (aarch64_expand_vector_init): Likewise.
9071 * config/arm/arm.c (neon_vdup_constant): Likewise.
9072 (neon_expand_vector_init): Likewise.
9073 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
9074 (arm_block_set_unaligned_vect): Likewise.
9075 (arm_block_set_aligned_vect): Likewise.
9076 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
9077 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
9078 (expand_vec_perm_even_odd_pack): Likewise.
9079 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
9080 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
9081 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
9082 gen_const_vec_duplicate.
9083 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
9084 * config/mips/mips.c (mips_gen_const_int_vector): Use
9085 gen_const_vec_duplicate.
9086 (mips_expand_vector_init): Use CONST0_RTX.
9087 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
9088 (define_split): Use gen_const_vec_duplicate.
9089 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
9090 (define_split): Use gen_const_vec_duplicate.
9091 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
9092 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
9093 * config/spu/spu.c (spu_const): Likewise.
9095 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9096 Alan Hayward <alan.hayward@arm.com>
9097 David Sherwood <david.sherwood@arm.com>
9099 * combine.c (can_change_dest_mode): Reject changes in
9100 REGMODE_NATURAL_SIZE.
9102 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
9104 * configure.ac (--enable-libssp): New.
9105 (gcc_cv_libc_provides_ssp): Check for explicit setting before
9106 trying to determine target-specific default. Adjust indentation.
9107 * configure: Regenerated.
9108 * doc/install.texi (Configuration): Expand --disable-libssp
9111 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
9113 config/i386/i386.c (ix86_expand_epilogue): Correct stack
9116 2017-10-31 Martin Jambor <mjambor@suse.cz>
9119 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
9121 2017-10-31 David Malcolm <dmalcolm@redhat.com>
9123 * auto-profile.c (autofdo_source_profile::read): Use
9124 UNKNOWN_LOCATION rather than 0.
9125 * diagnostic-core.h (warning_at_rich_loc): Rename to...
9126 (warning_at): ...this overload.
9127 (warning_at_rich_loc_n): Rename to...
9128 (warning_n): ...this overload.
9129 (error_at_rich_loc): Rename to...
9130 (error_at): ...this overload.
9131 (pedwarn_at_rich_loc): Rename to...
9132 (pedwarn): ...this overload.
9133 (permerror_at_rich_loc): Rename to...
9134 (permerror): ...this overload.
9135 (inform_at_rich_loc): Rename to...
9136 (inform): ...this overload.
9137 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
9138 (diagnostic_n_impl_richloc): Rename to...
9139 (diagnostic_n_impl): ...this rich_location *-based decl.
9140 (inform_at_rich_loc): Rename to...
9141 (inform): ...this, and add an assertion.
9142 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
9143 (warning_at_rich_loc): Rename to...
9144 (warning_at): ...this, and add an assertion.
9145 (warning_at_rich_loc_n): Rename to...
9146 (warning_n): ...this, and add an assertion.
9147 (warning_n): Update location_t-based implementation for removal of
9148 location_t-based diagnostic_n_impl.
9149 (pedwarn_at_rich_loc): Rename to...
9150 (pedwarn): ...this, and add an assertion.
9151 (permerror_at_rich_loc): Rename to...
9152 (permerror): ...this, and add an assertion.
9153 (error_n): Update for removal of location_t-based diagnostic_n_impl.
9154 (error_at_rich_loc): Rename to...
9155 (error_at): ...this, and add an assertion.
9156 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
9157 (driver::do_spec_on_infiles): Likewise.
9158 * substring-locations.c (format_warning_va): Update for renaming
9159 of inform_at_rich_loc.
9161 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
9163 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
9164 _Float<N>X built-in functions so that the variant without the
9165 "__builtin_" prefix is only enabled for the GNU C and Objective C
9166 languages when they are in non-strict ANSI/ISO mode.
9167 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
9168 * target.def (floatn_builtin_p): Add a target hook to control
9169 whether _Float<N> and _Float<N>X built-in functions without the
9170 "__builtin_" prefix are enabled, and return true for C and
9171 Objective C in the default hook. Include langhooks.h in
9173 * targhooks.h (default_floatn_builtin_p): Likewise.
9174 * targhooks.c (default_floatn_builtin_p): Likewise.
9175 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
9176 floatn_builtin_p target hook.
9177 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
9179 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
9180 Eric Botcazou <ebotcazou@adacore.com>
9182 PR rtl-optimization/81803
9183 * lra-constraints.c (curr_insn_transform): Also reload the whole
9184 register for a strict subreg no wider than a word if this is for
9185 a WORD_REGISTER_OPERATIONS target.
9187 2017-10-31 Jason Merrill <jason@redhat.com>
9189 * gdbinit.in: Skip over inlines from timevar.h.
9191 2017-10-31 Martin Liska <mliska@suse.cz>
9193 * doc/gcov.texi: Document new option.
9194 * gcov.c (print_usage): Likewise print it.
9195 (process_args): Support the argument.
9196 (format_count): New function.
9197 (format_gcov): Use the function.
9199 2017-10-31 Martin Liska <mliska@suse.cz>
9201 * gcov.c (struct name_map): do not use typedef.
9202 Define operator== and operator<.
9203 (name_search): Remove.
9204 (name_sort): Remove.
9205 (main): Do not allocate names.
9206 (process_file): Add vertical space.
9207 (generate_results): Use std::find.
9208 (release_structures): Do not release memory.
9209 (find_source): Use std::find.
9211 2017-10-31 Martin Liska <mliska@suse.cz>
9213 * gcov.c (struct line_info): Remove it's typedef.
9214 (line_info::line_info): Add proper ctor.
9215 (line_info::has_block): Do not use a typedef.
9216 (struct source_info): Do not use typedef.
9217 (circuit): Likewise.
9218 (get_cycles_count): Likewise.
9219 (output_intermediate_file): Iterate via vector iterator.
9220 (add_line_counts): Use std::vector methods.
9221 (accumulate_line_counts): Likewise.
9222 (output_lines): Likewise.
9224 2017-10-31 Martin Liska <mliska@suse.cz>
9226 * gcov.c (struct source_info): Remove typedef.
9227 (source_info::source_info): Add proper ctor.
9228 (accumulate_line_counts): Use struct, not it's typedef.
9229 (output_gcov_file): Likewise.
9230 (output_lines): Likewise.
9231 (main): Do not allocate an array.
9232 (output_intermediate_file): Use size of vector container.
9233 (process_file): Resize the vector.
9234 (generate_results): Do not preallocate, use newly added vector
9236 (release_structures): Do not release sources.
9237 (find_source): Use vector methods.
9238 (add_line_counts): Do not use typedef.
9240 2017-10-31 Martin Liska <mliska@suse.cz>
9242 * doc/gcov.texi: Document that.
9243 * gcov.c (add_line_counts): Mark lines with a non-executed
9245 (output_line_beginning): Handle such lines.
9246 (output_lines): Pass new argument.
9247 (output_intermediate_file): Print it in intermediate format.
9249 2017-10-31 Martin Liska <mliska@suse.cz>
9251 * color-macros.h: New file.
9252 * diagnostic-color.c: Factor out color related to macros to
9254 * doc/gcov.texi: Document -k option.
9255 * gcov.c (INCLUDE_STRING): Include string.h.
9256 (print_usage): Add -k option.
9257 (process_args): Parse it.
9258 (pad_count_string): New function.
9259 (output_line_beginning): Likewise.
9260 (DEFAULT_LINE_START): New macro.
9261 (output_lines): Support color output.
9263 2017-10-31 Martin Liska <mliska@suse.cz>
9265 PR gcov-profile/82633
9266 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
9267 their interaction with GCOV infrastructure.
9268 * configure.ac: Add -fkeep-{inline,static}-functions to
9270 * configure: Regenerate.
9272 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
9275 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
9277 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
9280 * config/rs6000/rs6000.md (allocate_stack): Force update interval
9281 into a register if it does not fit into an immediate offset field.
9283 2017-10-31 Olivier Hainque <hainque@adacore.com>
9285 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
9287 2017-10-31 Julia Koval <julia.koval@intel.com>
9289 * config.gcc: Add gfniintrin.h.
9290 * config/i386/gfniintrin.h: New.
9291 * config/i386/i386-builtin-types.def
9292 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
9293 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
9294 __builtin_ia32_vgf2p8affineinvqb_v32qi,
9295 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
9296 __builtin_ia32_vgf2p8affineinvqb_v16qi,
9297 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
9298 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
9299 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
9300 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
9301 V64QI_FTYPE_V64QI_V64QI_INT): New types.
9302 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
9303 * config/i386/immintrin.h: Include gfniintrin.h.
9304 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
9306 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
9308 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
9310 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9312 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
9313 (arm_ashldi3_1bit): Remove pattern.
9314 (ashrdi3): Remove shift by 1 expansion.
9315 (arm_ashrdi3_1bit): Remove pattern.
9316 (lshrdi3): Remove shift by 1 expansion.
9317 (arm_lshrdi3_1bit): Remove pattern.
9318 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
9319 cost of ashldi3 by 1.
9320 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
9321 (<shift>di3_neon): Likewise.
9323 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9325 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
9326 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
9327 and (*aarch64_simd_mov<VQ:mode>).
9328 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
9329 pattern alternative.
9330 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
9331 attributes to match pattern alternative.
9333 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
9335 * config.gcc (powerpc*-*-*): Add emmintrin.h.
9336 * config/rs6000/emmintrin.h: New file.
9337 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
9339 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9341 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
9342 * (movdi_vfp_cortexa8): Remove pattern.
9344 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9346 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
9348 (Specific, alpha*-dec-osf5.1): Remove.
9349 (Specific, mips-sgi-irix5): Remove.
9350 (Specific, mips-sgi-irix6): Remove.
9352 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9355 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
9356 arguments to clear_bit_region_be.
9358 2017-10-30 Jim Wilson <wilson@tuliptree.org>
9360 * gimplify.c: Include memmodel.h.
9362 2017-10-30 Martin Jambor <mjambor@suse.cz>
9364 * omp-grid.c (grid_attempt_target_gridification): Also insert a
9365 condition whether loop should be executed at all.
9367 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
9369 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9370 gimple folding of vec_madd() intrinsics.
9371 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
9372 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
9373 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
9375 2017-10-30 Richard Biener <rguenther@suse.de>
9377 PR tree-optimization/82762
9379 2017-10-23 Richard Biener <rguenther@suse.de>
9381 PR tree-optimization/82129
9383 2017-08-01 Richard Biener <rguenther@suse.de>
9385 PR tree-optimization/81181
9386 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9387 (compute_antic): ... end of iteration here.
9389 2017-10-30 Joseph Myers <joseph@codesourcery.com>
9391 * doc/invoke.texi (C Dialect Options): Document -std=c17,
9392 -std=iso9899:2017 and -std=gnu17.
9393 * doc/standards.texi (C Language): Document C17 support.
9394 * doc/cpp.texi (Overview): Mention -std=c17.
9395 (Standard Predefined Macros): Document C11 and C17 values of
9396 __STDC_VERSION__. Do not refer to C99 support as incomplete.
9397 * doc/extend.texi (Inline): Do not list individual options for
9398 standards newer than C99.
9399 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
9401 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
9404 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
9406 * asan.c (asan_finish_file): Align asan globals array by shadow
9409 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9412 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
9413 (struct store_immediate_info): Add bitregion_start and bitregion_end
9415 (store_immediate_info::store_immediate_info): Add brs and bre
9416 arguments and initialize bitregion_{start,end} from those.
9417 (struct merged_store_group): Add bitregion_start, bitregion_end,
9418 align_base and mask fields. Drop unnecessary struct keyword from
9419 struct store_immediate_info. Add do_merge method.
9420 (clear_bit_region_be): Use memset instead of loop storing zeros.
9421 (merged_store_group::do_merge): New method.
9422 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
9423 stores as long as the surrounding bitregions have no gaps.
9424 (merged_store_group::merge_overlapping): Use do_merge.
9425 (merged_store_group::apply_stores): Test that bitregion_{start,end}
9426 is byte aligned, rather than requiring that start and width are
9427 byte aligned. Drop unnecessary struct keyword from
9428 struct store_immediate_info. Allocate and populate also mask array.
9429 Make start of the arrays relative to bitregion_start rather than
9430 start and size them according to bitregion_{end,start} difference.
9431 (struct imm_store_chain_info): Drop unnecessary struct keyword from
9432 struct store_immediate_info.
9433 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
9434 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
9435 struct keyword from struct store_immediate_info.
9436 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
9437 between stores as long as the surrounding bitregions have no gaps.
9439 (struct split_store): Add orig non-static data member.
9440 (split_store::split_store): Initialize orig to false.
9441 (find_constituent_stmts): Return store_immediate_info *, non-NULL
9442 if there is exactly a single original stmt. Change stmts argument
9443 to pointer from reference, if NULL, don't push anything to it. Add
9444 first argument, use it to optimize skipping over orig stmts that
9445 are known to be before bitpos already. Simplify.
9446 (split_group): Return unsigned int count how many stores are or
9447 would be needed rather than a bool. Add allow_unaligned argument.
9448 Change split_stores argument from reference to pointer, if NULL,
9449 only do a dry run computing how many stores would be produced.
9450 Rewritten algorithm to use both alignment and misalign if
9451 !allow_unaligned and handle bitfield stores with gaps.
9452 (imm_store_chain_info::output_merged_store): Set start_byte_pos
9453 from bitregion_start instead of start. Compute allow_unaligned
9454 here, if true, do 2 split_group dry runs to compute which one
9455 produces fewer stores and prefer aligned if equal. Punt if
9456 new count is bigger or equal than original before emitting any
9457 statements, rather than during that. Remove no longer needed
9458 new_ssa_names tracking. Replace num_stmts with
9459 split_stores.length (). Use 32-bit stack allocated entries
9460 in split_stores auto_vec. Try to reuse original store lhs/rhs1
9461 if possible. Handle bitfields with gaps.
9462 (pass_store_merging::execute): Ignore bitsize == 0 stores.
9463 Compute bitregion_{start,end} for the stores and construct
9464 store_immediate_info with that. Formatting fixes.
9466 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
9469 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
9470 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
9472 2017-10-29 Jim Wilson <wilson@tuliptree.org>
9474 * gimplify.c: Include tm_p.h.
9476 * common.opt (gcoff): Re-add as ignored option.
9477 (gcoff1, gcoff2, gcoff3): Likewise.
9479 * Makefile.in (OBJS): Delete sdbout.o.
9480 (GTFILES): Delete $(srcdir)/sdbout.c.
9481 * debug.h: Delete sdb_debug_hooks.
9482 * final.c: Delete sdbout.h include.
9483 (final_scan_insn): Delete SDB_DEBUG check.
9484 (rest_of_clean_state): Likewise.
9485 * output.h: Delete sdb_begin_function_line.
9488 * toplev.c: Delete sdbout.h include.
9489 (process_options): Delete SDB_DEBUG check.
9490 * tree-core.h (tree_type_common): Delete pointer field of
9492 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
9493 TYPE_SYMTAB_POINTER.
9494 * tree.h (TYPE_SYMTAB_POINTER): Delete.
9495 (TYPE_SYMTAB_IS_POINTER): Delete.
9496 (TYPE_SYMTAB_IS_DIE): Renumber.
9497 * xcoffout.c: Refer to former sdbout.c file.
9498 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
9500 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
9501 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
9502 (Debugging Options): Delete -gcoff.
9503 (-gstabs): Delete SDB reference.
9505 (-gcoff@var{level}): Delete.
9506 * doc/passes.texi (Debugging information output): Delete SDB and
9507 sdbout.c references.
9508 * doc/tm.texi: Regenerate.
9509 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
9510 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
9512 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
9513 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
9514 (SDB_DEBUGGING_INFO): Delete.
9515 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
9516 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
9517 * target.def (output_source_filename): Delete COFF reference.
9519 * common.opt (gcoff): Delete.
9520 (gxcoff+): Update Negative chain.
9521 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
9523 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
9524 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
9525 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
9527 (expand_function_start): Change sdb reference to past tense.
9528 (expand_function_end): Change sdb reference to past tense.
9529 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
9530 * opts.c (debug_type_names): Delete coff entry.
9531 (common_handle_option): Delete OPT_gcoff case.
9532 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
9534 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
9535 * config/cris/cris.h: Delete SDB reference in comment.
9536 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
9537 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
9538 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
9539 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
9541 (ix86_expand_prologue): Likewise.
9542 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
9543 * config/ia64/ia64.h: Likewise.
9544 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
9545 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
9547 * config/mmix/mmix.h: Likewise.
9548 * config/nds32/nds32.c: Likewise.
9549 * config/stormy/storym16.h: Likewise.
9550 * config/visium/visium.h: Likewise.
9551 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
9553 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
9555 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
9556 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
9557 FRAME_POINTER_REGNUM point at high end of local var area.
9559 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
9561 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
9562 Move comment around. Do not reset best_edge for a copiable
9563 destination if the copy would cause a partition change.
9564 (better_edge_p): Remove redundant check.
9566 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9568 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
9570 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9572 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
9573 for math functions that have _Float<N> and _Float<N>X variants.
9574 (mathfn_built_in_2): Add support for math functions that have
9575 _Float<N> and _Float<N>X variants.
9576 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9577 (expand_builtin_mathfn_ternary): Add support for fma with
9578 _Float<N> and _Float<N>X variants.
9579 (expand_builtin): Likewise.
9580 (fold_builtin_3): Likewise.
9581 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9582 create math function _Float<N> and _Float<N>X variants as external
9584 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9585 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9586 the __builtin_ prefix and if not strict ansi, without the prefix.
9587 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9588 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9589 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9590 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9591 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9592 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9593 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9594 function signatures for fma _Float<N> and _Float<N>X variants.
9595 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9596 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9597 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9598 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9599 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9600 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9601 * gencfn-macros.c (print_case_cfn): Add support for math functions
9602 that have _Float<N> and _Float<N>X variants.
9603 (print_define_operator_list): Likewise.
9604 (fltfn_suffixes): Likewise.
9606 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9607 for math functions that have _Float<N> and _Float<N>X variants.
9608 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9609 and _Float<N>X variants.
9610 (COPYSIGN): Likewise.
9613 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9614 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9616 (integer_valued_read_call_p): Likewise.
9617 * fold-const-call.c (fold_const_call_ss): Likewise.
9618 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9619 _Float<N> and _Float<N>X variants.
9620 (fold_const_call_ssss): Add support for fma _Float<N> and
9621 _Float<N>X variants.
9622 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9623 support for copysign and fma _Float<N> and _Float<N>X variants.
9624 (backprop::process_builtin_call_use): Likewise.
9625 * tree-call-cdce.c (can_test_argument_range); Add support for
9626 sqrt _Float<N> and _Float<N>X variants.
9627 (edom_only_function): Likewise.
9628 (get_no_error_domain): Likewise.
9629 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9630 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9631 copysign _Float<N> and _Float<N>X variants.
9632 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9633 handled by machine independent code.
9634 (FMAF128): Likewise.
9635 * doc/cpp.texi (Common Predefined Macros): Document defining
9636 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9637 fma _Float<N> and _Float<N>X variants.
9639 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9642 * config/i386/i386-modes.def (CCFPU): Remove definition.
9643 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9644 (ix86_cc_modes_compatible): Ditto.
9645 (ix86_expand_carry_flag_compare): Ditto.
9646 (ix86_expand_int_movcc): Ditto.
9647 (ix86_expand_int_addcc): Ditto.
9648 (ix86_reverse_condition): Ditto.
9649 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9650 Return true/false for unordered/ordered fp comparisons.
9651 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9652 (ix86_prepare_fp_compare_args): Update for rename.
9653 (ix86_expand_fp_compare): Update for rename. Generate unordered
9654 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9655 (ix86_expand_sse_compare_and_jump): Ditto.
9656 * config/i386/predicates.md (fcmov_comparison_operator):
9657 Remove CCFPU mode handling.
9658 (ix86_comparison_operator): Ditto.
9659 (ix86_carry_flag_operator): Ditto.
9660 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9661 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9662 (*cmpu<mode>_cc_i387): Ditto.
9663 (FPCMP): Remove mode iterator.
9664 (unord): Remove mode attribute.
9665 (unord_subst): New define_subst transformation
9666 (unord): New define_subst attribute.
9668 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9669 (*cmpi<unord>xf_i387): Ditto.
9670 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9671 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9672 using unord_subst transformation.
9673 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9674 (round_saeonly): Also handle CCFP mode.
9675 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9676 Remove UNSPEC_SAHF unspec handling.
9678 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9680 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9682 2017-10-27 Jeff Law <law@redhat.com>
9684 * gimple-ssa-sprintf.c: Include domwalk.h.
9685 (class sprintf_dom_walker): New class, derived from dom_walker.
9686 (sprintf_dom_walker::before_dom_children): New function.
9687 (struct call_info): Moved into sprintf_dom_walker class
9688 (compute_formath_length, handle_gimple_call): Likewise.
9689 (sprintf_length::execute): Call the dominator walker rather
9690 than walking the statements.
9692 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9693 gimple statement locations.
9694 (check_array_bounds): Corresponding changes. Get the statement's
9695 location directly from wi->stmt.
9697 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9700 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9702 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9704 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9705 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9707 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9710 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9711 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9712 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9713 maybe_get_pool_constant.
9714 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9717 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9719 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9720 2.26 caveat. Update gas and gld versions.
9721 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9724 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9726 * cgraph.h (set_malloc_flag): Declare.
9727 * cgraph.c (set_malloc_flag_1): New function.
9728 (set_malloc_flag): Likewise.
9729 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9730 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9732 (read_ipa_call_summary): Add support for reading is_return_callee.
9733 (write_ipa_call_summary): Stream is_return_callee.
9734 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9735 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9736 ipa-prop.h, ipa-fnsummary.h.
9737 (pure_const_names): Change to static.
9738 (malloc_state_e): Define.
9739 (malloc_state_names): Define.
9740 (funct_state_d): Add field malloc_state.
9741 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9742 (check_retval_uses): New function.
9743 (malloc_candidate_p): Likewise.
9744 (analyze_function): Add support for malloc attribute.
9745 (pure_const_write_summary): Stream malloc_state.
9746 (pure_const_read_summary): Add support for reading malloc_state.
9747 (dump_malloc_lattice): New function.
9748 (propagate_malloc): New function.
9749 (warn_function_malloc): New function.
9750 (ipa_pure_const::execute): Call propagate_malloc and
9751 ipa_free_fn_summary.
9752 (pass_local_pure_const::execute): Add support for malloc attribute.
9753 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9754 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9756 2017-10-27 Martin Liska <mliska@suse.cz>
9758 PR gcov-profile/82457
9759 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9760 for fork-like functions to be properly instrumented.
9762 2017-10-27 Richard Biener <rguenther@suse.de>
9765 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9766 info when we redirected EH.
9768 2017-10-26 Michael Collison <michael.collison@arm.com>
9770 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9772 (<optab>_trunchf<GPI:mode>2: New pattern.
9773 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9774 * config/aarch64/iterators.md (wv): New mode attribute.
9775 (vf, VF): New mode attributes.
9776 (vgp, VGP): New mode attributes.
9777 (s): Update attribute with SImode and DImode prefixes.
9779 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9781 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9782 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9783 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9784 (nios2_option_overide): Initialize it. Don't allow R0-relative
9785 addressing with PIC.
9786 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9787 (nios2_symbolic_constant_p): Likewise.
9788 (nios2_legitimate_address_p): Likewise.
9789 (nios2_r0rel_section_name_p): New.
9790 (nios2_symbol_ref_in_r0rel_data_p): New.
9791 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9792 (r0rel_constant_p): New.
9793 (nios2_print_operand_address): Handle r0rel_constant_p.
9794 (nios2_cdx_narrow_form_p): Likewise.
9795 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9796 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9797 (Nios II Options): Document -mr0rel-sec.
9799 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9801 * config/nios2/nios2.c: Include xregex.h.
9802 (nios2_gprel_sec_regex): New.
9803 (nios2_option_overide): Initialize it. Don't allow GP-relative
9804 addressing with PIC.
9805 (nios2_small_section_name_p): Check for regex match.
9806 * config/nios2/nios2.opt (mgprel-sec=): New option.
9807 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9808 (Nios II Options): Document -mgprel-sec.
9810 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9812 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9814 2017-10-26 Tom de Vries <tom@codesourcery.com>
9816 PR tree-optimization/82707
9817 * gimple.c (gimple_copy): Fix unsharing of
9818 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9820 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
9822 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
9823 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
9824 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
9825 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
9826 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
9827 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
9828 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
9829 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
9830 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
9831 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
9832 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
9833 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
9834 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
9835 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
9836 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
9837 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
9838 _mm512_mask_cmpunord_ps_mask): New intrinsics.
9840 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9842 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
9844 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
9845 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
9846 warning to rs6000.c. Remove the Undocumented flag, since it has
9848 (-mabi=ibmlongdouble): Likewise.
9849 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
9850 already set, set the default format for long double.
9851 (rs6000_debug_reg_global): Print whether long double is IBM or
9853 (rs6000_option_override_internal): Rework setting long double
9854 format. Only warn if the user is changing the long double default
9855 and they did not use -Wno-psabi.
9856 * doc/invoke.texi (PowerPC options): Update the documentation for
9857 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
9859 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9860 Alan Hayward <alan.hayward@arm.com>
9861 David Sherwood <david.sherwood@arm.com>
9863 * rtl.h (wider_subreg_mode): New function.
9864 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
9865 rather than an unsigned int *.
9866 * ira-color.c (regno_max_ref_width): Replace with...
9867 (regno_max_ref_mode): ...this new variable.
9868 (coalesced_pseudo_reg_slot_compare): Update accordingly.
9869 Use wider_subreg_mode.
9870 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
9871 rather than an unsigned int *.
9872 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
9873 (process_alt_operands): Likewise.
9874 (invariant_p): Likewise.
9875 * lra-spills.c (assign_mem_slot): Likewise.
9876 (add_pseudo_to_slot): Likewise.
9877 * lra.c (collect_non_operand_hard_regs): Likewise.
9878 (add_regs_to_insn_regno_info): Likewise.
9879 * reload1.c (regno_max_ref_width): Replace with...
9880 (regno_max_ref_mode): ...this new variable.
9881 (reload): Update accordingly. Update call to
9882 ira_sort_regnos_for_alter_reg.
9883 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
9884 (init_eliminable_invariants): Update to use regno_max_ref_mode.
9885 (scan_paradoxical_subregs): Likewise.
9887 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9889 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
9890 (aarch64_frame): Add emit_frame_chain boolean.
9891 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9892 Move eh_return case to aarch64_layout_frame.
9893 (aarch64_layout_frame): Initialize emit_frame_chain.
9894 (aarch64_expand_prologue): Use emit_frame_chain.
9896 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9898 * config/aarch64/aarch64.c (aarch64_layout_frame):
9899 Ensure LR is always stored at the bottom of the callee-saves.
9900 Remove rarely used frame layout which saves callee-saves at top of
9901 frame, so the store of LR can be used as a valid probe in all cases.
9903 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9905 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
9906 Improve unaligned TImode/TFmode base/offset split.
9908 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9909 Alan Hayward <alan.hayward@arm.com>
9910 David Sherwood <david.sherwood@arm.com>
9912 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
9913 * combine.c (find_single_use_1): Likewise.
9914 (expand_field_assignment): Likewise.
9915 (move_deaths): Likewise.
9916 * lra-constraints.c (simplify_operand_subreg): Likewise.
9917 (curr_insn_transform): Likewise.
9918 * lra.c (collect_non_operand_hard_regs): Likewise.
9919 (add_regs_to_insn_regno_info): Likewise.
9920 * rtlanal.c (reg_referenced_p): Likewise.
9921 (covers_regno_no_parallel_p): Likewise.
9923 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9925 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
9926 Don't print any bits outside the precision of the value.
9927 * wide-int.cc (test_printing): Add some new tests.
9929 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9931 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
9932 supports -xbrace_comment option.
9933 * configure: Regenerate.
9934 * config.in: Regenerate.
9935 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
9936 (ASM_CPU_SPEC): Use it.
9938 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9940 * target.def (static_rtx_alignment): New hook.
9941 * targhooks.h (default_static_rtx_alignment): Declare.
9942 * targhooks.c (default_static_rtx_alignment): New function.
9943 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
9944 * doc/tm.texi: Regenerate.
9945 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
9946 instead of targetm.constant_alignment. Remove call to
9948 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9949 (cris_preferred_mininum_alignment): New function, split out from...
9950 (cris_constant_alignment): ...here.
9951 (cris_static_rtx_alignment): New function.
9952 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
9954 (ix86_constant_alignment): ...here.
9955 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9956 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9957 (mmix_static_rtx_alignment): New function.
9958 * config/spu/spu.c (spu_static_rtx_alignment): New function.
9959 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9961 2017-10-26 Tamar Christina <tamar.christina@arm.com>
9964 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
9965 Add flag_trapping_math and flag_fp_int_builtin_inexact.
9967 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
9969 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
9970 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
9971 mark as a sign-extending load.
9972 (local_pic_load_u): Define.
9974 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
9977 * fold-const.c (operand_equal_for_comparison_p): Also return true
9978 if ARG0 is a simple variant of ARG1 with narrower precision.
9979 (fold_ternary_loc): Always pass unstripped operands to the predicate.
9981 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
9983 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
9985 * i386.h (processor_costs): Add gather_static, gather_per_elt,
9986 scatter_static, scatter_per_elt.
9987 * x86-tune-costs.h: Add new cost entries.
9989 2017-10-25 Richard Biener <rguenther@suse.de>
9991 * tree-ssa-sccvn.h (vn_eliminate): Declare.
9992 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
9993 class pass_fre): Move to ...
9994 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
9995 class pass_fre): ... here and adjust for statistics.
9997 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10000 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10001 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10002 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10003 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10006 2017-10-25 Richard Biener <rguenther@suse.de>
10008 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10009 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10010 eliminate_push_avail, eliminate_insert): Move inside...
10011 (class eliminate_dom_walker): ... this class in preparation
10013 (fini_eliminate): Remove by merging with ...
10014 (eliminate): ... this function. Adjust for class changes.
10015 (pass_pre::execute): Remove fini_eliminate call.
10016 (pass_fre::execute): Likewise.
10018 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10021 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10022 (VPERMI2, VPERMI2I): New mode iterators.
10023 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10024 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10026 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10027 mode iterator. Remove 3 old define_insn patterns.
10028 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10029 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10030 VPERMI2 mode iterator, remove the other two expanders.
10031 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10032 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10033 instructions, remove the other two patterns.
10034 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10035 mode iterator, remove the other two patterns.
10036 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10037 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10038 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10039 and adjust argument order accordingly.
10040 (ix86_expand_vec_perm): Adjust caller.
10041 (expand_vec_perm_1): Likewise.
10042 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10043 (expand_vec_perm_vpermt2_vpshub2): ... this.
10044 (ix86_expand_vec_perm_const_1): Adjust caller.
10045 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10048 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10049 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10050 (vec_shl_<mode>): Remove unused expander.
10051 (avx512bw_<shift_insn><mode>3): New define_insn.
10052 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10053 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10055 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10058 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10061 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10063 PR rtl-optimization/82396
10064 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
10065 (autopref_multipass_init): Simplify initialization.
10066 (autopref_rank_data): Simplify sort order.
10067 * gcc/sched-int.h (autopref_multipass_data_): Remove
10068 multi_mem_insn_p, min_offset and max_offset.
10070 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10072 PR middle-end/60580
10073 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10074 Check special value of flag_omit_frame_pointer.
10075 (aarch64_can_eliminate): Likewise.
10076 (aarch64_override_options_after_change_1): Simplify handling of
10077 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
10079 2017-10-24 Richard Biener <rguenther@suse.de>
10081 PR tree-optimization/82697
10082 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
10083 zero for conditional load and unconditional store.
10085 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10087 * doc/install.texi: Document bootstrap-cet.
10089 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10092 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
10093 ENDBR instruction at function entrance if function is only
10096 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10099 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
10100 patterns to better describe from which operation the CF is computed.
10101 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
10102 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
10103 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
10104 is 0, use _0 suffixed expanders instead of emitting a comparison
10107 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
10109 * config/i386/i386.md(*movsf_internal, *movdf_internal):
10110 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
10112 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
10114 PR middle-end/82569
10115 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
10116 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
10117 * loop-iv.c (iv_get_reaching_def): Likewise.
10118 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
10119 variable is promoted and the partition contains undefined values.
10121 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10123 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
10125 (nios2_address_cost): Define.
10126 (nios2_legitimize_address): Recognize (exp + constant) directly.
10127 (TARGET_ADDRESS_COST): Define.
10129 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10131 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
10132 (nios2_symbolic_memory_operand_p): Declare.
10133 (nios2_split_large_constant): Declare.
10134 (nios2_split_symbolic_memory_operand): Declare.
10135 * config/nios2/nios2.c: Adjust includes.
10136 (nios2_symbolic_constant_allowed): New.
10137 (nios2_symbolic_constant_p): New.
10138 (nios2_plus_symbolic_constant_p): New.
10139 (nios2_valid_addr_expr_p): Recognize addresses involving
10140 symbolic constants.
10141 (nios2_legitimate_address_p): Likewise, also LO_SUM.
10142 (nios2_symbolic_memory_operand_p): New.
10143 (nios2_large_constant_p): New.
10144 (nios2_split_large_constant): New.
10145 (nios2_split_plus_large_constant): New.
10146 (nios2_split_symbolic_memory_operand): New.
10147 (nios2_legitimize_address): Code refactoring. Handle addresses
10148 involving symbolic constants.
10149 (nios2_emit_move_sequence): Likewise.
10150 (nios2_print_operand): Improve error output.
10151 (nios2_print_operand_address): Handle LO_SUM.
10152 (nios2_cdx_narrow_form_p): Likewise.
10153 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
10154 operands involving symbolic constants.
10155 (movhi_internal, movsi_internal): Likewise.
10156 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
10157 (extendhisi2, extendqi<mode>2): Likewise.
10159 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10161 * tree-pass.h (PROP_rtl_split_insns): Define.
10162 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
10164 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10166 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
10168 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10171 * target.def (const_not_ok_for_debug_p): Default to
10172 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
10173 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
10174 * targhooks.c (default_const_not_ok_for_debug_p): New function.
10175 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
10176 which targetm.const_not_ok_for_debug_p returned true.
10177 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
10179 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
10181 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
10182 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
10183 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
10185 (ix86_const_not_ok_for_debug_p): New function.
10186 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
10187 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
10189 2017-10-23 David Malcolm <dmalcolm@redhat.com>
10192 * system.h: Conditionally include "unique-ptr.h" if
10193 INCLUDE_UNIQUE_PTR is defined.
10194 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
10195 of defining INCLUDE_UNIQUE_PTR before including "system.h".
10197 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
10199 * config/rl78/rl78.md: New define_expand "subdi3".
10201 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
10204 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
10205 DF_REF_INSN if DF_REF_INSN_INFO is false.
10207 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
10209 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
10210 xmm_move instead of sse_move.
10211 (sse_store_index): New function.
10212 (ix86_register_move_cost): Be more sensible about mismatch stall;
10213 model AVX moves correctly; make difference between sse->integer and
10215 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
10216 moves; make difference between SSE and AVX.
10217 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
10218 and zmm_move. Increase size of sse load and store tables;
10219 add unaligned load and store tables; add ssemmx_to_integer.
10220 * x86-tune-costs.h: Update all entries according to real
10221 move latencies from Agner Fog's manual and chip documentation.
10223 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10226 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
10227 * config/i386/constraints.md (Wf): New constraint.
10228 * config/i386/i386.md (UNSPEC_SBB): New unspec.
10229 (cmp<dwi>_doubleword): Removed.
10230 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
10231 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
10232 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
10233 expand with cmp<dwi>_doubleword. For LTU and GEU use
10234 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
10236 * common.opt (gcolumn-info): Enable by default.
10237 * doc/invoke.texi (gcolumn-info): Document new default.
10239 2017-10-23 Richard Biener <rguenther@suse.de>
10241 PR tree-optimization/82672
10242 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
10243 Fold the stmt if we propagated into it.
10245 2017-10-23 Richard Biener <rguenther@suse.de>
10247 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
10248 (bitmap_remove_expr_from_set): ... this. All callers call this
10249 for non-constant values.
10250 (bitmap_set_subtract): Rename to...
10251 (bitmap_set_subtract_expressions): ... this. Adjust and
10253 (bitmap_set_contains_value): Remove superfluous check.
10254 (bitmap_set_replace_value): Inline into single caller ...
10255 (bitmap_value_replace_in_set): ... here and simplify.
10256 (dependent_clean): Merge into ...
10257 (clean): ... this using an overload. Adjust.
10258 (prune_clobbered_mems): Adjust.
10259 (compute_antic_aux): Likewise.
10260 (compute_partial_antic_aux): Likewise.
10262 2017-10-23 Richard Biener <rguenther@suse.de>
10264 PR tree-optimization/82129
10266 2017-08-01 Richard Biener <rguenther@suse.de>
10268 PR tree-optimization/81181
10269 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10270 (compute_antic): ... end of iteration here.
10272 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10274 * target.def (starting_frame_offset): New hook.
10275 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
10276 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
10277 * doc/tm.texi.in: Regenerate.
10278 * hooks.h (hook_hwi_void_0): Declare.
10279 * hooks.c (hook_hwi_void_0): New function.
10280 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
10281 STARTING_FRAME_OFFSET.
10282 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
10283 * reload1.c (reload): Likewise.
10284 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
10285 instead of STARTING_FRAME_OFFSET.
10286 * function.c (try_fit_stack_local): Likewise.
10287 (assign_stack_local_1): Likewise
10288 (instantiate_virtual_regs): Likewise.
10289 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
10290 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
10291 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
10292 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
10293 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
10294 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
10295 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
10296 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
10297 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
10298 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
10299 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
10300 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
10301 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
10302 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
10303 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
10304 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
10305 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
10306 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
10307 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
10308 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
10309 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
10310 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
10311 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
10312 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
10313 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
10314 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
10315 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
10316 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
10317 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
10318 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
10319 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
10320 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
10321 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
10322 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
10323 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
10324 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
10325 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
10326 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
10327 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
10328 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
10329 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
10330 * config/avr/avr.c (avr_starting_frame_offset): Make static and
10331 return a HOST_WIDE_INT.
10332 (avr_builtin_setjmp_frame_value): Use it instead of
10333 STARTING_FRAME_OFFSET.
10334 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10335 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
10336 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
10338 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10339 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
10340 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
10341 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10342 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
10343 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
10344 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10345 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
10346 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
10347 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10348 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
10349 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
10351 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10352 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
10353 * config/mips/mips.c (mips_compute_frame_info): Refer to
10354 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10355 (mips_starting_frame_offset): New function.
10356 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10357 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
10358 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
10359 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
10360 and return a HOST_WIDE_INT.
10361 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10362 (mmix_initial_elimination_offset): Refer to
10363 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10364 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
10365 * config/pa/pa.c (pa_starting_frame_offset): New function.
10366 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
10367 (pa_expand_prologue): Likewise.
10368 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10369 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
10370 !FRAME_GROWS_DOWNWARD handling to...
10371 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10372 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
10373 !FRAME_GROWS_DOWNWARD handling to...
10374 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10375 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
10376 !FRAME_GROWS_DOWNWARD handling to...
10377 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10378 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
10380 (rs6000_starting_frame_offset): New function.
10381 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
10382 !FRAME_GROWS_DOWNWARD handling to...
10383 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10384 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
10385 !FRAME_GROWS_DOWNWARD handling to...
10386 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10387 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
10388 !FRAME_GROWS_DOWNWARD handling to...
10389 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10390 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
10391 (rs6000_starting_frame_offset): New function.
10392 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
10393 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
10394 * config/vax/vax.c (vax_starting_frame_offset): New function.
10395 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
10396 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10397 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
10398 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
10399 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10400 * system.h (STARTING_FRAME_OFFSET): Poison.
10402 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10404 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10405 SCALAR_TYPE_MODE instead of TYPE_MODE.
10407 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10408 Alan Hayward <alan.hayward@arm.com>
10409 David Sherwood <david.sherwood@arm.com>
10411 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
10413 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10414 Alan Hayward <alan.hayward@arm.com>
10415 David Sherwood <david.sherwood@arm.com>
10417 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
10419 2017-10-23 Richard Biener <rguenther@suse.de>
10421 PR tree-optimization/82129
10422 * tree-ssa-pre.c (bitmap_set_and): Remove.
10423 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
10424 canonicalizing expressions in the set to those with lowest
10425 ID rather than taking that from the first edge.
10427 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10429 * combine.c (rtx_equal_for_field_assignment_p): Use
10430 byte_lowpart_offset.
10432 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10433 Alan Hayward <alan.hayward@arm.com>
10434 David Sherwood <david.sherwood@arm.com>
10436 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
10437 to a SUBREG_PROMOTED_VAR.
10439 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10440 Alan Hayward <alan.hayward@arm.com>
10441 David Sherwood <david.sherwood@arm.com>
10443 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
10444 (expand_debug_source_expr): Likewise.
10445 * combine.c (combine_simplify_rtx): Likewise.
10446 * cse.c (fold_rtx): Likewise.
10447 * optabs.c (expand_float): Likewise.
10448 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10449 (simplify_binary_operation_1): Likewise.
10451 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10452 Alan Hayward <alan.hayward@arm.com>
10453 David Sherwood <david.sherwood@arm.com>
10455 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
10456 (record_promoted_value): Likewise.
10457 * expr.c (expand_expr_real_2): Likewise.
10458 * ree.c (update_reg_equal_equiv_notes): Likewise.
10459 (combine_set_extension): Likewise.
10460 * rtlanal.c (low_bitmask_len): Likewise.
10461 * simplify-rtx.c (neg_const_int): Likewise.
10462 (simplify_binary_operation_1): Likewise.
10464 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10465 Alan Hayward <alan.hayward@arm.com>
10466 David Sherwood <david.sherwood@arm.com>
10468 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
10469 * regcprop.c (maybe_mode_change): Likewise.
10470 * reload1.c (alter_reg): Likewise.
10472 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10474 * inchash.h (inchash::hash::add_wide_int): New function.
10475 * lto-streamer-out.c (hash_tree): Use it.
10477 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10479 * inchash.h (inchash::hash::add_wide_int): Rename to...
10480 (inchash::hash::add_hwi): ...this.
10481 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
10482 (polymorphic_call_target_hasher::hash): Likewise.
10483 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
10484 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
10485 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
10486 * lto-streamer-out.c (hash_tree): Likewise.
10487 * optc-save-gen.awk: Likewise.
10488 * tree.c (add_expr): Likewise.
10490 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10493 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
10494 for ordered inequality comparisons even with TARGET_IEEE_FP.
10496 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10499 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
10500 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10501 Expand with cmp<dwi>_doubleword.
10503 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10505 * extend.texi: Add x86 specific to 'nocf_check' attribute.
10506 List CET intrinsics.
10507 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
10508 specific to -fcf-protection option.
10510 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10512 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
10513 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
10514 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
10515 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
10516 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
10517 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
10518 (extra_objs): Add cet.o for Linux/x86 targets.
10519 (tmake_file): Add i386/t-cet for Linux/x86 targets.
10520 * config/i386/cet.c: New file.
10521 * config/i386/cetintrin.h: Likewise.
10522 * config/i386/t-cet: Likewise.
10523 * config/i386/cpuid.h (bit_SHSTK): New.
10524 (bit_IBT): Likewise.
10525 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
10526 pass IBT and SHSTK bits.
10527 * config/i386/i386-builtin-types.def
10528 (VOID_FTYPE_UNSIGNED_PVOID): New.
10529 (VOID_FTYPE_UINT64_PVOID): Likewise.
10530 * config/i386/i386-builtin.def: Add CET intrinsics.
10531 * config/i386/i386-c.c (ix86_target_macros_internal): Add
10532 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
10533 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
10534 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
10536 * config/i386/i386.c (rest_of_insert_endbranch): New.
10537 (pass_data_insert_endbranch): Likewise.
10538 (pass_insert_endbranch): Likewise.
10539 (make_pass_insert_endbranch): Likewise.
10540 (ix86_notrack_prefixed_insn_p): Likewise.
10541 (ix86_target_string): Add -mibt, -mshstk flags.
10542 (ix86_option_override_internal): Add flag_cf_protection
10544 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
10545 (ix86_print_operand): Add 'notrack' prefix output.
10546 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
10547 (ix86_expand_builtin): Expand CET intrinsics.
10548 (x86_output_mi_thunk): Add 'endbranch' instruction.
10549 * config/i386/i386.h (TARGET_IBT): New.
10550 (TARGET_IBT_P): Likewise.
10551 (TARGET_SHSTK): Likewise.
10552 (TARGET_SHSTK_P): Likewise.
10553 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
10554 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
10555 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
10556 (builtin_setjmp_setup): New pattern.
10557 (builtin_longjmp): Likewise.
10558 (rdssp<mode>): Likewise.
10559 (incssp<mode>): Likewise.
10560 (saveprevssp): Likewise.
10561 (rstorssp): Likewise.
10562 (wrss<mode>): Likewise.
10563 (wruss<mode>): Likewise.
10564 (setssbsy): Likewise.
10565 (clrssbsy): Likewise.
10566 (nop_endbr): Likewise.
10567 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
10569 * config/i386/immintrin.h: Include <cetintrin.h>.
10570 * config/i386/linux-common.h
10571 (file_end_indicate_exec_stack_and_cet): New prototype.
10572 (TARGET_ASM_FILE_END): New.
10574 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10576 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10577 latencies instead of having separate table; make difference between
10578 integer and float costs.
10579 * i386.h (processor_costs): Remove scalar_stmt_cost,
10580 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10581 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10583 * x86-tune-costs.h: Remove entries which has been removed in
10584 procesor_costs from all tables; make cond_taken_branch_cost
10585 and cond_not_taken_branch_cost COST_N_INSNS based.
10587 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10589 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10591 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10593 * config/i386/i386.md (isa): Remove fma_avx512f.
10594 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10595 <avx512>_fmadd_<mode>_mask3<round_name>,
10596 <avx512>_fmsub_<mode>_mask<round_name>,
10597 <avx512>_fmsub_<mode>_mask3<round_name>,
10598 <avx512>_fnmadd_<mode>_mask<round_name>,
10599 <avx512>_fnmadd_<mode>_mask3<round_name>,
10600 <avx512>_fnmsub_<mode>_mask<round_name>,
10601 <avx512>_fnmsub_<mode>_mask3<round_name>,
10602 <avx512>_fmaddsub_<mode>_mask<round_name>,
10603 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10604 <avx512>_fmsubadd_<mode>_mask<round_name>,
10605 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10606 (*vec_widen_umult_even_v16si<mask_name>,
10607 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10608 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10610 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10612 * extend.texi: Add 'nocf_check' documentation.
10613 * gimple.texi: Add second parameter to
10614 gimple_build_call_from_tree.
10615 * invoke.texi: Add -fcf-protection documentation.
10616 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10618 2017-10-20 Richard Biener <rguenther@suse.de>
10620 PR tree-optimization/82473
10621 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10622 the largest input type.
10624 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10626 * c-attribs.c (handle_nocf_check_attribute): New function.
10627 (c_common_attribute_table): Add 'nocf_check' handling.
10628 * gimple-parser.c: Add second argument NULL to
10629 gimple_build_call_from_tree.
10630 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10631 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10633 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10634 * common.opt: Add fcf-protection flag.
10635 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10636 * flag-types.h: Add enum cf_protection_level.
10637 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10638 Add 'nocf_check' attribute propagation to gimple call.
10639 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10640 (gimple_build_call_from_tree): Update prototype.
10641 (gimple_call_nocf_check_p): New function.
10642 (gimple_call_set_nocf_check): Likewise.
10643 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10644 * ipa-icf.c: Add nocf_check attribute in statement hash.
10645 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10646 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10647 * toplev.c (process_options): Add flag_cf_protection handling.
10649 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10651 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10653 2017-10-20 Richard Biener <rguenther@suse.de>
10655 PR tree-optimization/82603
10656 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10657 remove false predicated stores.
10659 2017-10-20 Richard Biener <rguenther@suse.de>
10661 * graphite-isl-ast-to-gimple.c
10662 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10663 Remove return value and simplify, dump copied stmt after lhs
10665 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10666 Reduce dump verbosity.
10667 (gsi_insert_earliest): Likewise.
10668 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10669 * graphite.c (print_global_statistics): Adjust dumping.
10670 (print_graphite_scop_statistics): Likewise.
10671 (print_graphite_statistics): Do not dump loops here.
10672 (graphite_transform_loops): But here.
10674 2017-10-20 Nicolas Roche <roche@adacore.com>
10676 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10677 * configure: Regenerate.
10679 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10682 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10683 functions when optimizing replace GIMPLE_RETURN stmts with
10684 calls to __builtin_unreachable ().
10687 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10688 for -fsanitize=thread link of executables.
10689 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10690 link of executables.
10693 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10694 New mode iterators.
10695 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10696 define_insns for logical vector shifts to use VI248_AVX512BW
10697 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10698 condition, useless isa and prefix attributes. Change the first
10699 2 of these define_insns to ...
10700 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10701 define_insn for avx512vl.
10702 (<shift_insn><mode>3): ... and this, new define_insn without
10703 masking for non-avx512vl.
10706 * config/i386/sse.md (*andnot<mode>3,
10707 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10708 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10709 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10710 not applied use empty suffix even for TARGET_AVX512VL.
10711 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10712 is applied, supply evex,evex or evex,evex,evex instead of just
10715 2017-10-20 Julia Koval <julia.koval@intel.com>
10717 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10718 (OPTION_MASK_ISA_GFNI_UNSET): New.
10719 (ix86_handle_option): Handle OPT_mgfni.
10720 * config/i386/cpuid.h (bit_GFNI): New.
10721 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10722 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10723 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10724 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10725 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10726 * config/i386/i386.opt: Add mgfni.
10728 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10730 * config/msp430/msp430.c (msp430_option_override): Disable
10731 -fdelete-null-pointer-checks.
10732 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10734 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10736 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10737 of x87 and SSE instructions.
10739 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10741 * asan.c (create_cond_insert_point): Do not update edge count.
10742 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10743 (afdo_propagate_circuit): Likewise.
10744 (afdo_calculate_branch_prob): Likewise.
10745 (afdo_annotate_cfg): Likewise.
10746 * basic-block.h (struct edge_def): Remove count.
10747 (edge_def::count): New accessor.
10748 * bb-reorder.c (rotate_loop): Update.
10749 (find_traces_1_round): Update.
10750 (connect_traces): Update.
10751 (sanitize_hot_paths): Update.
10752 * cfg.c (unchecked_make_edge): Update.
10753 (make_single_succ_edge): Update.
10754 (check_bb_profile): Update.
10755 (dump_edge_info): Update.
10756 (update_bb_profile_for_threading): Update.
10757 (scale_bbs_frequencies_int): Update.
10758 (scale_bbs_frequencies_gcov_type): Update.
10759 (scale_bbs_frequencies_profile_count): Update.
10760 (scale_bbs_frequencies): Update.
10761 * cfganal.c (connect_infinite_loops_to_exit): Update.
10762 * cfgbuild.c (compute_outgoing_frequencies): Update.
10763 (find_many_sub_basic_blocks): Update.
10764 * cfgcleanup.c (try_forward_edges): Update.
10765 (try_crossjump_to_edge): Update
10766 * cfgexpand.c (expand_gimple_cond): Update
10767 (expand_gimple_tailcall): Update
10768 (construct_exit_block): Update
10769 * cfghooks.c (verify_flow_info): Update
10770 (redirect_edge_succ_nodup): Update
10771 (split_edge): Update
10772 (make_forwarder_block): Update
10773 (duplicate_block): Update
10774 (account_profile_record): Update
10775 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10776 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10777 * cfgloopmanip.c (scale_loop_profile): Update.
10779 (lv_adjust_loop_entry_edge): Update.
10780 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10781 (force_nonfallthru_and_redirect): Update.
10782 (purge_dead_edges): Update.
10783 (rtl_flow_call_edges_add): Update.
10784 * cgraphunit.c (init_lowered_empty_function): Update.
10785 (cgraph_node::expand_thunk): Update.
10786 * gimple-pretty-print.c (dump_probability): Update.
10787 (dump_edge_probability): Update.
10788 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10789 * haifa-sched.c (sched_create_recovery_edges): Update.
10790 * hsa-gen.c (convert_switch_statements): Update.
10791 * ifcvt.c (dead_or_predicable): Update.
10792 * ipa-inline-transform.c (inline_transform): Update.
10793 * ipa-split.c (split_function): Update.
10794 * ipa-utils.c (ipa_merge_profiles): Update.
10795 * loop-doloop.c (add_test): Update.
10796 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10797 * lto-streamer-in.c (input_cfg): Update.
10798 (input_function): Update.
10799 * lto-streamer-out.c (output_cfg): Update.
10800 * modulo-sched.c (sms_schedule): Update.
10801 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10802 * predict.c (maybe_hot_edge_p): Update.
10803 (unlikely_executed_edge_p): Update.
10804 (probably_never_executed_edge_p): Update.
10805 (dump_prediction): Update.
10806 (drop_profile): Update.
10807 (propagate_unlikely_bbs_forward): Update.
10808 (determine_unlikely_bbs): Update.
10809 (force_edge_cold): Update.
10810 * profile.c (compute_branch_probabilities): Update.
10811 * reg-stack.c (better_edge): Update.
10812 * shrink-wrap.c (handle_simple_exit): Update.
10813 * tracer.c (better_p): Update.
10814 * trans-mem.c (expand_transaction): Update.
10815 (split_bb_make_tm_edge): Update.
10816 * tree-call-cdce.c: Update.
10817 * tree-cfg.c (gimple_find_sub_bbs): Update.
10818 (gimple_split_edge): Update.
10819 (gimple_duplicate_sese_region): Update.
10820 (gimple_duplicate_sese_tail): Update.
10821 (gimple_flow_call_edges_add): Update.
10822 (insert_cond_bb): Update.
10823 (execute_fixup_cfg): Update.
10824 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
10825 * tree-complex.c (expand_complex_div_wide): Update.
10826 * tree-eh.c (lower_resx): Update.
10827 (unsplit_eh): Update.
10828 (cleanup_empty_eh_move_lp): Update.
10829 * tree-inline.c (copy_edges_for_bb): Update.
10830 (freqs_to_counts): Update.
10831 (copy_cfg_body): Update.
10832 * tree-ssa-dce.c (remove_dead_stmt): Update.
10833 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10834 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10835 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10836 (unloop_loops): Update.
10837 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10838 * tree-ssa-loop-split.c (connect_loops): Update.
10839 (split_loop): Update.
10840 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10841 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10842 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10843 * tree-ssa-reassoc.c (branch_fixup): Update.
10844 * tree-ssa-tail-merge.c (replace_block_by): Update.
10845 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10846 (compute_path_counts): Update.
10847 (update_profile): Update.
10848 (recompute_probabilities): Update.
10849 (update_joiner_offpath_counts): Update.
10850 (estimated_freqs_path): Update.
10851 (freqs_to_counts_path): Update.
10852 (clear_counts_path): Update.
10853 (ssa_fix_duplicate_block_edges): Update.
10854 (duplicate_thread_path): Update.
10855 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10856 (case_bit_test_cmp): Update.
10857 (collect_switch_conv_info): Update.
10858 (gen_inbound_check): Update.
10859 (do_jump_if_equal): Update.
10860 (emit_cmp_and_jump_insns): Update.
10861 * tree-tailcall.c (decrease_profile): Update.
10862 (eliminate_tail_call): Update.
10863 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10864 (vect_do_peeling): Update.
10865 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10866 * ubsan.c (ubsan_expand_null_ifn): Update.
10867 (ubsan_expand_ptr_ifn): Update.
10868 * value-prof.c (gimple_divmod_fixed_value): Update.
10869 (gimple_mod_pow2): Update.
10870 (gimple_mod_subtract): Update.
10871 (gimple_ic): Update.
10872 (gimple_stringop_fixed_value): Update.
10874 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10877 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
10879 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
10881 PR rtl-optimization/82395
10882 * ira-color.c (allocno_priority_compare_func): Fix comparison step
10883 based on non_spilled_static_chain_regno_p.
10885 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10887 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
10888 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
10889 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
10891 2017-10-19 Martin Sebor <msebor@redhat.com>
10893 PR tree-optimization/82596
10894 * tree.c (array_at_struct_end_p): Handle STRING_CST.
10896 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10898 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
10899 (get_mem_refs_of_builtin_call): Likewise.
10900 * builtins.c (expand_builtin_apply): Adjust call to
10901 allocate_dynamic_stack_space.
10902 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
10903 the third argument to allocate_dynamic_stack_space, otherwise -1.
10904 (expand_builtin): Deal with all alloca variants.
10905 (is_inexpensive_builtin): Likewise.
10906 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
10907 * calls.c (special_function_p): Deal with all alloca variants.
10908 (initialize_argument_information): Adjust call to
10909 allocate_dynamic_stack_space.
10910 (expand_call): Likewise.
10911 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
10912 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
10913 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
10914 use it for the stack usage computation.
10915 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
10916 * function.c (gimplify_parameters): Call build_alloca_call_expr.
10917 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
10918 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
10919 (in_loop_p): Remove first argument and useless check.
10920 (pass_walloca::execute): Remove useless test and adjust call to above.
10921 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
10922 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
10923 (gimplify_call_expr): Deal with all alloca variants.
10924 * hsa-gen.c (gen_hsa_alloca): Likewise.
10925 (gen_hsa_insns_for_call): Likewise.
10926 * ipa-pure-const.c (special_builtin_state): Likewise.
10927 * tree-chkp.c (chkp_build_returned_bound): Likewise.
10928 * tree-object-size.c (alloc_object_size): Likewise.
10929 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
10930 (call_may_clobber_ref_p_1): Likewise.
10931 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
10932 (ccp_fold_stmt): Likewise.
10933 (optimize_stack_restore): Likewise.
10934 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10935 (mark_all_reaching_defs_necessary_1): Likewise.
10936 (propagate_necessity): Likewise.
10937 (eliminate_unnecessary_stmts): Likewise.
10938 * tree.c (build_common_builtin_nodes): Build
10939 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
10940 (build_alloca_call_expr): New function.
10941 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
10942 (CASE_BUILT_IN_ALLOCA): Likewise.
10943 (build_alloca_call_expr): Declare.
10944 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
10946 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10949 * dwarf2out.c (new_die_raw): New static inline function.
10950 (new_die): Use it to create the DIE.
10951 (add_AT_external_die_ref): Likewise.
10952 (clone_die): Likewise.
10953 (clone_as_declaration): Likewise.
10954 (dwarf2out_vms_debug_main_pointer): Likewise.
10955 (base_type_die): Likewise. Remove early return for corner cases.
10956 Do not call add_pubtype on the DIE here.
10957 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
10958 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
10959 typedefs for base types with DW_AT_endianity. Make sure a DIE with
10960 native order exists for base types, attach the DIE manually and call
10961 add_pubtype on it. Do not equate a reverse order DIE to the type.
10963 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
10965 * config/arm/arm.c (align_ok_ldrd_strd): New function.
10966 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
10968 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
10970 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10972 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
10973 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
10974 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
10975 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
10976 * opts.c (sanitizer_opts): Add builtin.
10977 * ubsan.c (instrument_builtin): New function.
10978 (pass_ubsan::execute): Call it.
10979 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
10980 * doc/invoke.texi: Document -fsanitize=builtin.
10982 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
10983 builtins, store max (log2 (align), 0) into uchar field instead of
10984 align into uptr field.
10985 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
10986 store uchar 0 field instead of uptr 0 field.
10987 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
10988 instead of passing one address of struct with 2 locations pass
10989 two addresses of structs with 1 location each.
10990 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
10991 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
10992 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
10993 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
10994 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
10995 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
10996 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
10997 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
10999 2017-10-19 Martin Liska <mliska@suse.cz>
11002 * file-find.c (remove_prefix): Remove.
11003 * file-find.h (remove_prefix): Likewise.
11004 * gcc-ar.c: Remove smartness of lookup.
11006 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11008 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11009 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11010 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11012 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11015 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11016 (setcc + and to xor + setcc): New peephole2.
11018 2017-10-19 Tom de Vries <tom@codesourcery.com>
11020 * doc/sourcebuild.texi (Test Directives, Variants of
11021 dg-require-support): Add dg-require-stack-size.
11023 2017-10-19 Martin Liska <mliska@suse.cz>
11026 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11027 that have a large alignment.
11028 (gimplify_target_expr): Likewise.
11030 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11032 PR rtl-optimization/82602
11033 * ira.c (rtx_moveable_p): Return false for volatile asm.
11035 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11038 * config/i386/i386-modes.def (CCGZ): New CC mode.
11039 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11040 * config/i386/predicates.md (ix86_comparison_operator):
11042 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11043 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11044 with double-word subtraction.
11045 (put_condition_code): Handle CCGZmode.
11047 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11049 * wide-int.cc (debug (const wide_int &)): New.
11050 (debug (const wide_int *)): New.
11051 (debug (const widest_int &)): New.
11052 (debug (const widest_int *)): New.
11054 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11056 PR middle-end/82556
11057 * lra-constraints.c (curr_insn_transform): Use non-input operand
11058 instead of output one for matched reload.
11060 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11062 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11063 (tree-ssa-loop-ivopts.h): New header file.
11064 (struct builtin_info): New fields.
11065 (classify_builtin_1): Compute and record base and offset parts for
11066 memset builtin partition by calling strip_offset.
11067 (offset_cmp, fuse_memset_builtins): New functions.
11068 (finalize_partitions): Fuse adjacent memset partitions by calling
11070 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
11071 Expose the interface.
11072 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
11074 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11076 PR tree-optimization/82574
11077 * tree-loop-distribution.c (find_single_drs): New parameter. Check
11078 that data reference must be executed exactly once per iteration
11079 against the outermost loop in nest.
11080 (classify_partition): Update call to above function.
11082 2017-10-18 Richard Biener <rguenther@suse.de>
11084 PR tree-optimization/82591
11085 * graphite.c (graphite_transform_loops): Move code gen message
11087 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11088 Here. Handle scop_to_isl_ast failing.
11089 (scop_to_isl_ast): Limit the number of ISL operations.
11091 2017-10-18 Richard Biener <rguenther@suse.de>
11093 * graphite-isl-ast-to-gimple.c
11094 (translate_isl_ast_to_gimple::set_rename): Simplify.
11095 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
11096 (graphite_copy_stmts_from_block): ... here.
11097 (copy_bb_and_scalar_dependences): Simplify.
11098 (add_parameters_to_ivs_params): Canonicalize.
11099 (generate_entry_out_of_ssa_copies): Simplify.
11100 * graphite-sese-to-poly.c (extract_affine_name): Simplify
11101 by passing in ISL dimension.
11102 (parameter_index_in_region_1): Rename to ...
11103 (parameter_index_in_region): ... this.
11104 (extract_affine): Adjust assert, pass down parameter index.
11105 (add_param_constraints): Use range-info when available.
11106 (build_scop_context): Adjust.
11107 * sese.c (new_sese_info): Adjust.
11108 (free_sese_info): Likewise.
11109 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
11110 Remove unused typedefs.
11111 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
11113 2017-10-18 Martin Liska <mliska@suse.cz>
11115 * combine.c (simplify_compare_const): Add gcc_fallthrough.
11117 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11119 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
11120 (s390_sched_init): Do not reset s390_sched_state if we entered the
11121 current basic block via a fallthru edge and all others are unlikely.
11123 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11125 * config/s390/s390.c (NUM_SIDES): New variable.
11126 (LONGRUNNING_THRESHOLD): New variable.
11127 (LATENCY_FACTOR): New variable.
11128 (s390_sched_score): Decrease score for long-running instructions on
11130 (s390_sched_variable_issue): Perform bookkeeping for long-running
11133 2017-10-18 Richard Biener <rguenther@suse.de>
11135 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
11136 Simplify with removal of the parameter rename map.
11137 (set_rename): Likewise.
11138 (should_copy_to_new_region): Likewise.
11139 (graphite_copy_stmts_from_block): Likewise.
11140 (copy_bb_and_scalar_dependences): Remove initialization of
11141 unused copied_bb_map.
11142 (copy_def): Remove.
11143 (copy_internal_parameters): Likewise.
11144 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
11145 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
11146 Use INTEGRAL_TYPE_P.
11147 (parameter_index_in_region_1): Rename to ...
11148 (assign_parameter_index_in_region): ... this. Assert we have
11149 a parameter we handle.
11150 (scan_tree_for_params): Adjust.
11151 * sese.h (parameter_rename_map_t): Remove.
11152 (struct sese_info_t): Remove unused parameter_rename_map and
11153 copied_bb_map members.
11154 * sese.c (new_sese_info): Adjust.
11155 (free_sese_info): Likewise.
11157 2017-10-18 Martin Liska <mliska@suse.cz>
11160 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
11161 on an abnormal edge.
11163 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
11165 * doc/invoke.texi (ffunction-sections and fdata-sections):
11168 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11170 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
11171 the use statement can throw internally.
11173 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11175 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
11176 any RTX present on the RHS of a SET.
11177 * compare-elim.c (try_eliminate_compare): Restore comment.
11179 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11181 * langhooks.h (struct lang_hooks): Document that tree_size langhook
11182 may be also called on tcc_type nodes.
11183 * langhooks.c (lhd_tree_size): Likewise.
11185 2017-10-17 David Malcolm <dmalcolm@redhat.com>
11187 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
11188 format_warning_at_substring.
11189 (maybe_warn): Convert source_range * param to a location_t. Pass
11190 UNKNOWN_LOCATION rather than NULL to fmtwarn.
11191 (format_directive): Remove code to extract source_ranges and
11192 source_range * in favor of just a location_t.
11193 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
11195 * substring-locations.c (format_warning_va): Convert
11196 source_range * param to a location_t.
11197 (format_warning_at_substring): Likewise.
11198 * substring-locations.h (format_warning_va): Likewise.
11199 (format_warning_at_substring): Likewise.
11201 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
11203 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
11205 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
11206 and scatter/gather ops.
11208 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
11209 vec_gather_load and vec_scatter_store.
11210 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
11211 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
11212 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
11213 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
11214 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
11215 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
11217 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
11219 * reg-stack.c (compare_for_stack_reg): Add bool argument.
11220 Detect FTST instruction and handle its register pops. Only pop
11221 second operand if can_pop_second_op is true.
11222 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
11223 set can_pop_second_op to false in the compare_for_stack_reg call.
11225 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
11226 output_fp_compare for stack register operands.
11227 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
11228 instructions here. Do not emit stack register pops here. Assert
11229 that FCOMPP pops next to top stack register. Rewrite function.
11231 2017-10-17 Nathan Sidwell <nathan@acm.org>
11233 PR middle-end/82577
11234 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
11235 use DECL_ASSEMBLER_NAME_RAW.
11237 PR middle-end/82546
11238 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
11241 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
11242 Wilco Dijkstra <wilco.dijkstra@arm.com>
11244 * builtins.c (expand_builtin_update_setjmp_buf): Add a
11245 converstion to Pmode from the buf_addr.
11247 2017-10-17 Richard Biener <rguenther@suse.de>
11249 * graphite-dependences.c (scop_get_reads_and_writes): Change
11250 output parameters to references.
11252 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
11254 PR 71026/tree-optimization
11255 * fold-const.c (distribute_real_division): Removed.
11256 (fold_binary_loc): Remove calls to distribute_real_divison.
11258 2017-10-17 Richard Biener <rguenther@suse.de>
11260 * graphite-scop-detection.c
11261 (scop_detection::stmt_has_simple_data_refs_p): Always use
11262 the full nest as region.
11263 (try_generate_gimple_bb): Likewise.
11264 * sese.c (scalar_evolution_in_region): Simplify now that
11265 SCEV can handle instantiation in regions.
11266 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
11267 in the non-loop part of a function if requested.
11269 2017-10-17 Richard Biener <rguenther@suse.de>
11271 PR tree-optimization/82563
11272 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
11274 (graphite_regenerate_ast_isl): Call it.
11275 * graphite-scop-detection.c (build_scops): Remove entry edge split.
11277 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11279 PR tree-optimization/82549
11280 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
11281 Formatting fixes. Instead of calling make_bit_field_ref with negative
11284 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
11286 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
11287 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
11288 _mm_maskz_reduce_ss): New.
11289 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
11290 __builtin_ia32_reducess_mask): Ditto..
11291 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
11292 * config/i386/sse.md (reduces<mode>): Renamed to ...
11293 (reduces<mode><mask_scalar_name>): ... this.
11294 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
11296 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
11297 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
11299 2017-10-16 David Malcolm <dmalcolm@redhat.com>
11301 * Makefile.in (OBJS): Add unique-ptr-tests.o.
11302 * selftest-run-tests.c (selftest::run_tests): Call
11303 selftest::unique_ptr_tests_cc_tests.
11304 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
11305 * unique-ptr-tests.cc: New file.
11307 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
11310 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11312 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11313 (make_hard_regno_born, make_hard_regno_dead): Update
11314 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
11316 2017-10-16 Jeff Law <law@redhat.com>
11318 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11320 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
11322 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
11324 2017-10-16 Olivier Hainque <hainque@adacore.com>
11326 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
11327 with_cpu if we were configured for an e500v2 target cpu name.
11329 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
11331 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
11332 * doc/invoke.texi: Document +nodsp as a valid extension for
11335 2017-10-16 Martin Liska <mliska@suse.cz>
11337 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
11338 (test_set_range): Likewise.
11339 (test_range_functions): Rename to ...
11340 (test_bit_in_range): ... this.
11341 (sbitmap_c_tests): Add new test.
11343 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11345 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
11347 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
11348 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
11350 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11352 * config/aarch64/aarch64-builtins.c
11353 (aarch64_types_quadopu_lane_qualifiers): New.
11354 (TYPES_QUADOPU_LANE): New.
11355 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
11356 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
11357 (aarch64_<sur>dot_laneq<vsi2qi>): New.
11358 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
11359 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
11360 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
11361 (Vdottype, DOTPROD): New.
11362 (sur): Add SDOT and UDOT.
11364 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11366 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
11367 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
11368 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
11369 Add TARGET_DOTPROD.
11370 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
11371 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
11372 Enable TARGET_DOTPROD.
11373 (cortex-a75.cortex-a55): Likewise.
11374 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
11376 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11378 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
11379 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
11381 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
11383 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
11384 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
11385 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
11386 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
11387 * config/arm/types.md (neon_dot, neon_dot_q): New.
11388 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
11390 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11392 * config/arm/arm.h (TARGET_DOTPROD): New.
11393 * config/arm/arm.c (arm_arch_dotprod): New.
11394 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
11395 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
11396 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
11397 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
11398 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
11399 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
11400 * doc/invoke.texi (armv8.2-a): Document dotprod
11402 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11404 * i386.c (ix86_vec_cost): New function.
11405 (ix86_rtx_costs): Handle vector operations better.
11406 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
11407 * x86-tune-costs.h: Add new costs to all tables.
11409 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11411 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
11413 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
11414 divsd, sqrtss and sqrtsd
11415 * x86-tune-costs.h: Add new entries to all costs.
11416 (znver1_cost): Fix to match real instruction latencies.
11418 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11419 Michael Collison <michael.collison@arm.com>
11421 * compare-elim.c: Include emit-rtl.h.
11422 (can_merge_compare_into_arith): New function.
11423 (try_validate_parallel): Likewise.
11424 (try_merge_compare): Likewise.
11425 (try_eliminate_compare): Call the above when no previous clobber
11427 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
11430 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11432 PR middle-end/62263
11433 PR middle-end/82498
11434 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
11435 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
11437 PR middle-end/62263
11438 PR middle-end/82498
11439 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
11440 to be any operand_equal_p operands. For & (B - 1) require
11441 B to be power of 2. Recognize
11442 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
11444 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11447 * optabs.c (expand_memory_blockage): Fix call of
11448 targetm.have_memory_blockage.
11450 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11453 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
11454 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
11455 objects to extra_objs instead of overwriting it.
11457 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11459 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
11460 Use any_fp_register_operand as operand[3] predicate. Simplify
11461 equality test for operands[2] and operands[4] memory location.
11462 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
11463 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
11464 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
11465 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
11466 any_fp_register_operand as operand[1] predicate. Simplify
11467 equality test for operands[0] and operands[3] memory location.
11468 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
11469 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
11470 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
11472 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11474 * target-insns.def: Add memory_blockage.
11475 * optabs.c (expand_memory_blockage): New function.
11476 (expand_asm_memory_barrier): Rename ...
11477 (expand_asm_memory_blockage): ... to this.
11478 (expand_mem_thread_fence): Call expand_memory_blockage
11479 instead of expand_asm_memory_barrier.
11480 (expand_mem_singnal_fence): Ditto.
11481 (expand_atomic_load): Ditto.
11482 (expand_atomic_store): Ditto.
11483 * doc/md.texi (Standard Pattern Names For Generation):
11484 Document memory_blockage instruction pattern.
11486 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
11488 * config/rl78/rl78.c (rl78_emit_libcall): New function.
11489 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
11490 * config/rl78/rl78.md: New define_expand "adddi3".
11492 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11494 * cfghooks.c (verify_flow_info): Disable check that all probabilities
11497 2017-10-13 Jeff Law <law@redhat.com>
11499 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
11501 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11504 * internal-fn.c (expand_mul_overflow): If both operands have
11505 the same highpart of -1 or 0 and the topmost bit of lowpart
11506 is different, overflow is if res <= 0 rather than res < 0.
11508 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
11510 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
11511 TARGET_P9_VECTOR code for unaligned_load case.
11513 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11515 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
11517 2017-10-13 Nathan Sidwell <nathan@acm.org>
11519 * tree-core.h (tree_contains_struct): Make bool.
11520 * tree.c (tree_contains_struct): Likewise.
11521 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11522 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
11523 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
11524 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
11525 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
11527 2017-10-13 Richard Biener <rguenther@suse.de>
11529 * graphite-isl-ast-to-gimple.c
11530 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
11531 parameters and dominance check.
11532 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
11533 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
11534 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
11535 Do not update SSA form here or do intermediate IL verification.
11536 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
11537 (graphite_initialize): Remove check on the number of loops in
11538 the function and inline into graphite_transform_loops.
11539 (graphite_finalize): Inline into graphite_transform_loops.
11540 (graphite_transform_loops): Perform SSA update and IL verification
11542 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
11544 2017-10-13 Richard Biener <rguenther@suse.de>
11546 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
11547 graphite_expression_type_precision): Avoid global constructor
11549 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
11550 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
11551 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
11552 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
11553 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
11554 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
11556 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
11559 * config/i386/i386.h (ix86_red_zone_size): New.
11560 * config/i386/i386.md (push peephole2s): Replace
11561 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
11563 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11564 Alan Hayward <alan.hayward@arm.com>
11565 David Sherwood <david.sherwood@arm.com>
11567 * combine.c (can_change_dest_mode): Reject changes in
11568 REGMODE_NATURAL_SIZE.
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 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
11575 (expand_debug_source_expr): Likewise.
11576 * combine.c (combine_simplify_rtx): Likewise.
11577 * cse.c (fold_rtx): Likewise.
11578 * fwprop.c (canonicalize_address): Likewise.
11579 * targhooks.c (default_shift_truncation_mask): Likewise.
11581 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11582 Alan Hayward <alan.hayward@arm.com>
11583 David Sherwood <david.sherwood@arm.com>
11585 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11586 (widened_mode): Likewise.
11587 (expand_unop): Likewise.
11588 * ree.c (transform_ifelse): Likewise.
11589 (merge_def_and_ext): Likewise.
11590 (combine_reaching_defs): Likewise.
11591 * simplify-rtx.c (simplify_unary_operation_1): 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 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11598 * combine.c (gen_lowpart_for_combine): Likewise.
11599 * dwarf2out.c (rtl_for_decl_location): Likewise.
11600 * final.c (alter_subreg): Likewise.
11601 * rtlhooks.c (gen_lowpart_general): Likewise.
11602 (gen_lowpart_if_possible): Likewise.
11604 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11605 Alan Hayward <alan.hayward@arm.com>
11606 David Sherwood <david.sherwood@arm.com>
11608 * calls.c (expand_call): Use subreg_lowpart_offset.
11609 * cse.c (cse_insn): Likewise.
11610 * regcprop.c (copy_value): Likewise.
11611 (copyprop_hardreg_forward_1): Likewise.
11613 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11616 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11617 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11618 =Q constraints to +Q and into insn condition add check
11619 that operands[0] and operands[1] are equal.
11620 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11621 =Q constraints to +Q and into insn condition add check
11622 that operands[0] is equal to either operands[1] or operands[2].
11625 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11626 instead of handling MINUS_EXPR twice (once for each argument),
11627 canonicalize operand order and handle just once, use rtype where
11628 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11631 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11632 any values of __C while still being pattern recognizable as a simple
11633 rotate instruction.
11635 2017-10-13 Richard Biener <rguenther@suse.de>
11637 PR tree-optimization/82451
11639 2017-10-02 Richard Biener <rguenther@suse.de>
11641 PR tree-optimization/82355
11642 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11643 a mapping for the enclosing loop but avoid generating one for
11644 the loop tree root.
11645 (copy_bb_and_scalar_dependences): Remove premature codegen
11646 error on PHIs in blocks duplicated into multiple places.
11647 * graphite-scop-detection.c
11648 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11649 in the region use it as loop and nest to analyze the DR in.
11650 (try_generate_gimple_bb): Likewise.
11651 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11652 (add_loop_constraints): For blocks in a loop not in the region
11653 create a dimension with a single iteration.
11654 * sese.h (gbb_loop_at_index): Remove assert.
11656 * cfgloop.c (loop_preheader_edge): For the loop tree root
11657 return the single successor of the entry block.
11658 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11659 Reset the SCEV hashtable and niters.
11660 * graphite-scop-detection.c
11661 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11662 assert that we only have POLYNOMIAL_CHREC that vary in loops
11663 contained in the region.
11664 (scop_detection::graphite_can_represent_expr): Adjust.
11665 (scop_detection::stmt_has_simple_data_refs_p): For loops
11666 not in the region set loop to NULL. The nest is now the
11667 entry edge to the region.
11668 (try_generate_gimple_bb): Likewise.
11669 * sese.c (scalar_evolution_in_region): Adjust for
11670 instantiate_scev change.
11671 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11672 Make nest parameter the edge into the region.
11673 (create_data_ref): Likewise.
11674 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11675 entry edge into a region and adjust instantiate_scev calls.
11676 (create_data_ref): Likewise.
11677 (graphite_find_data_references_in_stmt): Likewise.
11678 (find_data_references_in_stmt): Pass the loop preheader edge
11679 from the nest argument.
11680 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11681 parameter the edge into the region.
11682 (instantiate_parameters): Use the loop preheader edge as entry.
11683 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11685 (get_instantiated_value_entry): Make instantiate_below parameter
11686 the edge into the region.
11687 (instantiate_scev_name): Likewise. Adjust dominance checks,
11688 when we cannot use loop-based instantiation instantiate by
11689 walking use-def chains.
11690 (instantiate_scev_poly): Adjust.
11691 (instantiate_scev_binary): Likewise.
11692 (instantiate_scev_convert): Likewise.
11693 (instantiate_scev_not): Likewise.
11694 (instantiate_array_ref): Remove.
11695 (instantiate_scev_3): Likewise.
11696 (instantiate_scev_2): Likewise.
11697 (instantiate_scev_1): Likewise.
11698 (instantiate_scev_r): Do not blindly handle N-operand trees.
11699 Do not instantiate array-refs. Handle all constants and invariants.
11700 (instantiate_scev): Make instantiate_below parameter
11701 the edge into the region.
11702 (resolve_mixers): Use the loop preheader edge for the region
11703 parameter to instantiate_scev_r.
11704 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11706 2017-10-13 Richard Biener <rguenther@suse.de>
11708 PR tree-optimization/82525
11709 * graphite-isl-ast-to-gimple.c
11710 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11712 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11713 Fail code generation when we cannot represent the isl integer.
11714 (binary_op_to_tree): Elide modulo operations that are no-ops
11715 in the type we code generate. Remove now superfluous code
11718 2017-10-13 Richard Biener <rguenther@suse.de>
11720 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11721 (scop_detection::harmful_loop_in_region): Remove premature
11722 IV type restriction.
11723 (scop_detection::graphite_can_represent_scev): We can handle
11724 pointer IVs just fine.
11726 2017-10-13 Alan Modra <amodra@gmail.com>
11728 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11729 "Clobbers and Scratch Registers". Add paragraph on
11730 alternative to clobbers for scratch registers and OpenBLAS
11733 2017-10-13 Alan Modra <amodra@gmail.com>
11735 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11736 example of a memory input for a string of known length. Move
11737 commentary out of table. Add a number of new examples
11738 covering array memory inputs.
11740 2017-10-12 Martin Liska <mliska@suse.cz>
11742 PR tree-optimization/82493
11743 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11744 (test_range_functions): New function.
11745 (sbitmap_c_tests): Likewise.
11746 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11747 * selftest.h (sbitmap_c_tests): New function.
11749 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11751 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11753 * config/rs6000/amo.h: Fix spacing issue.
11755 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11758 * config/i386/i386.md (*ashl<mode>3_mask_1,
11759 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11760 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11763 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11765 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11766 (profile_probability): Set max_probability
11767 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11769 * profile-count.c (profile_probability::differs_from_p): Do not
11770 rely on max_probaiblity == 10000
11772 2017-10-12 Jeff Law <law@redhat.com>
11774 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11777 2017-10-12 Martin Sebor <msebor@redhat.com>
11781 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11782 (handle_alias_pairs): Call it.
11783 * common.opt (-Wattribute-alias): New option.
11784 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11785 * doc/invoke.texi (-Wattribute-alias): Document.
11787 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11790 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11792 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11794 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11795 (make_hard_regno_born, make_hard_regno_dead): Update
11796 bb_killed_pseudos and bb_gen_pseudos.
11798 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11800 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11802 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11804 * config/alpha/alpha.c (alpha_split_conditional_move):
11805 Use std::swap instead of manually swapping.
11806 (alpha_stdarg_optimize_hook): Ditto.
11807 (alpha_canonicalize_comparison): Ditto.
11809 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11811 * tree-loop-distribution.c (struct builtin_info): New struct.
11812 (struct partition): Refactor fields into struct builtin_info.
11813 (partition_free): Free struct builtin_info.
11814 (build_size_arg_loc, build_addr_arg_loc): Delete.
11815 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11816 information from struct builtin_info.
11817 (find_single_drs): New function refactored from classify_partition.
11818 Also moved builtin validity checks to this function.
11819 (compute_access_range, alloc_builtin): New functions.
11820 (classify_builtin_st, classify_builtin_ldst): New functions.
11821 (classify_partition): Refactor code into functions find_single_drs,
11822 classify_builtin_st and classify_builtin_ldst.
11823 (distribute_loop): Don't do runtime alias check when distributing
11825 (find_seed_stmts_for_distribution): New function.
11826 (pass_loop_distribution::execute): Refactor code finding seed
11827 stmts into above function. Support distribution for the innermost
11828 two-level loop nest. Adjust dump information.
11830 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11832 * tree-loop-distribution.c: Adjust the general comment.
11833 (NUM_PARTITION_THRESHOLD): New macro.
11834 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
11835 (classify_partition): Skip builtin pattern of loop nest's inner loop.
11836 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
11837 in call to build_partition_graph.
11838 (finalize_partitions): New parameter. Make loop distribution more
11839 conservative by fusing more partitions.
11840 (distribute_loop): Don't do runtime alias check in case of loop nest
11842 (find_seed_stmts_for_distribution): New function.
11843 (prepare_perfect_loop_nest): New function.
11844 (pass_loop_distribution::execute): Refactor code finding seed stmts
11845 and loop nest into above functions. Support loop nest distribution.
11846 Adjust dump information accordingly.
11848 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11850 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
11851 and set PTYPE_SEQUENTIAL for merged partition.
11853 2017-10-12 Richard Biener <rguenther@suse.de>
11855 PR tree-optimization/69728
11857 2017-09-19 Richard Biener <rguenther@suse.de>
11859 PR tree-optimization/69728
11860 * graphite-sese-to-poly.c (schedule_error): New global.
11861 (add_loop_schedule): Handle empty domain by failing the
11863 (build_original_schedule): Handle schedule_error.
11865 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
11866 domain by returning an unchanged schedule.
11868 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11870 * genrecog.c (validate_pattern): For VEC_SELECT verify that
11871 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
11873 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
11875 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
11878 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11881 * expr.c (store_field): Don't optimize away bitsize == 0 store
11882 from CALL_EXPR with addressable return type.
11884 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11886 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
11887 * config/rs6000/rs6000.md (sel): Delete mode attribute.
11888 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
11889 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
11890 TARGET_ISEL instead of TARGET_ISEL<sel>.
11892 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
11894 * config/rs6000/rs6000.c
11895 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
11897 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11899 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
11901 (reg_or_cint_operand): Fix comment.
11902 (reg_or_zero_operand): New predicate.
11903 * config/rs6000/rs6000-protos.h (output_isel): Delete.
11904 * config/rs6000/rs6000.c (output_isel): Delete.
11905 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
11906 instead of reg_or_cint_operand. Output instruction directly (not via
11908 (isel_unsigned_<mode>): Ditto.
11909 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
11910 gpc_reg_operand. Add an instruction alternative for this. Output
11911 instruction directly.
11912 (*isel_reversed_unsigned_<mode>): Ditto.
11914 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11916 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
11917 (TARGET_CANONICALIZE_COMPARISON): Define.
11919 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
11922 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11923 Check whether the dest is REG before adding REG_EQUIV note.
11925 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11928 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11930 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11931 (make_hard_regno_born, make_hard_regno_dead): Update
11932 bb_killed_pseudos and bb_gen_pseudos.
11934 2017-10-11 Nathan Sidwell <nathan@acm.org>
11936 * incpath.h (enum incpath_kind): Name enum, prefix values.
11937 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
11938 * incpath.c (heads, tails): Use INC_MAX.
11939 (add_env_var_paths, add_standard_paths): Use incpath_kind.
11940 (merge_include_chains, split_quote_chain,
11941 register_include_chains): Update incpath_kind names.
11942 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
11943 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
11945 (add_framework_path, darwin_register_objc_includes): Likewise.
11946 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
11948 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11950 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
11951 Do not use float_operator operator predicate.
11952 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
11953 * config/i386/predicates.md (float_operator): Remove predicate.
11955 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11957 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
11958 (*jccxf_i387): Ditto.
11959 (*jcc<mode>_i387): Ditto.
11960 (*jccu<mode>_i387): Ditto.
11961 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
11962 (*jcc_*_i387 splitters): Remove.
11963 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
11964 * config/i386/i386.c (ix86_split_fp_branch): Remove.
11965 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
11968 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11970 * profile-count.h (slow_safe_scale_64bit): New function.
11971 (safe_scale_64bit): New inline.
11972 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
11973 * profile-count.c: Include wide-int.h
11974 (slow_safe_scale_64bit): New.
11976 2017-10-11 Nathan Sidwell <nathan@acm.org>
11978 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
11979 HAS_DECL_ASSEMBLER_NAME_P.
11980 * gimple-expr.c (gimple_decl_printable_name: Check
11981 HAS_DECL_ASSEMBLER_NAME_P too.
11982 * ipa-utils.h (type_in_anonymous_namespace_p): Check
11983 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
11984 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
11985 * passes.c (rest_of_decl_compilation): Check
11986 HAS_DECL_ASSEMBLER_NAME_P too.
11987 * recog.c (verify_changes): Likewise.
11988 * tree-pretty-print.c (dump_decl_name): Likewise.
11989 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
11991 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
11992 (SET_DECL_ASSEMBLER_NAME): Use it.
11993 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
11994 (COPY_DECL_ASSEMBLER_NAME): Likewise.
11995 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
11997 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11999 * config.gcc (i386, x86_64): Add extra objects.
12000 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12001 (ix86_min_insn_size): Declare.
12002 (ix86_issue_rate): Declare.
12003 (ix86_adjust_cost): Declare.
12004 (ia32_multipass_dfa_lookahead): Declare.
12005 (ix86_macro_fusion_p): Declare.
12006 (ix86_macro_fusion_pair_p): Declare.
12007 (ix86_bd_has_dispatch): Declare.
12008 (ix86_bd_do_dispatch): Declare.
12009 (ix86_core2i7_init_hooks): Declare.
12010 (ix86_atom_sched_reorder): Declare.
12011 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12012 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12013 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12014 (rip_relative_addr_p): Rename to ...
12015 (ix86_rip_relative_addr_p): ... this one; export.
12016 (memory_address_length): Update.
12017 (ix86_issue_rate): Move to x86-tune-sched.c.
12018 (ix86_flags_dependent): Move to x86-tune-sched.c.
12019 (ix86_agi_dependent): Move to x86-tune-sched.c.
12020 (exact_dependency_1): Move to x86-tune-sched.c.
12021 (exact_store_load_dependency): Move to x86-tune-sched.c.
12022 (ix86_adjust_cost): Move to x86-tune-sched.c.
12023 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12024 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12025 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12026 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12027 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12028 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12029 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12030 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12031 (min_insn_size): Rename to ...
12032 (ix86_min_insn_size): ... this one; export.
12033 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12034 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12035 (core2i7_first_cycle_multipass_backtrack): Move to
12036 x86-tune-sched-core.c.
12037 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12038 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12039 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12040 (ix86_avoid_jump_mispredicts): Update.
12041 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12042 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12043 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12044 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12045 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12046 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12047 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12048 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12049 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12050 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12051 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12052 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12053 (BIG): Move to ix86-tune-sched-bd.c.
12054 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12055 (enum insn_path): Move to ix86-tune-sched-bd.c.
12056 (get_mem_group): Move to ix86-tune-sched-bd.c.
12057 (is_cmp): Move to ix86-tune-sched-bd.c.
12058 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12059 (is_branch): Move to ix86-tune-sched-bd.c.
12060 (is_prefetch): Move to ix86-tune-sched-bd.c.
12061 (init_window): Move to ix86-tune-sched-bd.c.
12062 (allocate_window): Move to ix86-tune-sched-bd.c.
12063 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12064 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
12065 (process_end_window): Move to ix86-tune-sched-bd.c.
12066 (allocate_next_window): Move to ix86-tune-sched-bd.c.
12067 (find_constant): Move to ix86-tune-sched-bd.c.
12068 (get_num_immediates): Move to ix86-tune-sched-bd.c.
12069 (has_immediate): Move to ix86-tune-sched-bd.c.
12070 (get_insn_path): Move to ix86-tune-sched-bd.c.
12071 (get_insn_group): Move to ix86-tune-sched-bd.c.
12072 (count_num_restricted): Move to ix86-tune-sched-bd.c.
12073 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
12074 (add_insn_window): Move to ix86-tune-sched-bd.c.
12075 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
12076 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
12077 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
12078 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
12079 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
12080 (do_dispatch): Move to ix86-tune-sched-bd.c.
12081 (has_dispatch): Move to ix86-tune-sched-bd.c.
12082 * config/i386/t-i386: Add new object files.
12083 * config/i386/x86-tune-costs.h: New file.
12084 * config/i386/x86-tune-sched-atom.c: New file.
12085 * config/i386/x86-tune-sched-bd.c: New file.
12086 * config/i386/x86-tune-sched-core.c: New file.
12087 * config/i386/x86-tune-sched.c: New file.
12089 2017-10-11 Liu Hao <lh_mouse@126.com>
12091 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
12092 the generic version below instead.
12093 (should_colorize): Recognize Windows consoles as terminals
12095 * pretty-print.c [__MINGW32__] (write_all): New function.
12096 [__MINGW32__] (find_esc_head): Likewise.
12097 [__MINGW32__] (find_esc_terminator): Likewise.
12098 [__MINGW32__] (eat_esc_sequence): Likewise.
12099 [__MINGW32__] (mingw_ansi_fputs): New function that handles
12101 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
12104 2017-10-11 Richard Biener <rguenther@suse.de>
12106 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
12107 Properly call analyze_scalar_evolution with the loop of the stmt.
12109 2017-10-11 Richard Biener <rguenther@suse.de>
12111 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
12112 * tree-core.h (tree_base): Add chrec_var union member.
12113 * tree.h (CHREC_VAR): Remove.
12114 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
12115 * tree-chrec.h (build_polynomial_chrec): Adjust.
12116 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
12117 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
12119 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
12121 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
12122 * match.pd: ... here.
12123 ((T) X == (T) Y): Relax condition.
12125 2017-10-11 Bin Cheng <bin.cheng@arm.com>
12127 PR tree-optimization/82472
12128 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
12130 (break_alias_scc_partitions): Update postorder number.
12132 2017-10-11 Martin Liska <mliska@suse.cz>
12135 * opts.c (parse_no_sanitize_attribute): Do not use error_value
12137 * opts.h (parse_no_sanitize_attribute): Remove last argument.
12139 2017-10-11 Martin Liska <mliska@suse.cz>
12141 * print-rtl.c (print_insn): Move declaration of idbuf
12142 to same scope as name.
12144 2017-10-11 Martin Liska <mliska@suse.cz>
12149 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12150 volatile arguments.
12152 2017-10-11 Martin Liska <mliska@suse.cz>
12155 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12156 volatile arguments.
12158 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
12160 * config.gcc (default_gnu_indirect_function): Default to yes for
12161 arm*-*-linux* with glibc.
12163 2017-10-11 Richard Biener <rguenther@suse.de>
12165 * tree-scalar-evolution.c (get_scalar_evolution): Handle
12166 default-defs and types we do not want to analyze.
12167 (interpret_loop_phi): Replace unreachable code with an assert.
12168 (compute_scalar_evolution_in_loop): Remove and inline ...
12169 (analyze_scalar_evolution_1): ... here, replacing condition with
12170 what makes the intent clearer. Remove handling of cases
12171 get_scalar_evolution now handles.
12173 2017-10-10 Jim Wilson <wilson@tuliptree.org>
12175 PR rtl-optimization/81434
12176 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
12177 comment for main loop. In sched_group_found if, also add checks for
12178 pass and min_cost_group.
12180 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
12182 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
12183 (rs6000_insn_cost): New function.
12184 * config/rs6000/rs6000.md (cost): New attribute.
12186 2017-10-10 Jakub Jelinek <jakub@redhat.com>
12187 H.J. Lu <hongjiu.lu@intel.com>
12191 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
12192 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
12193 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
12194 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
12195 to OPTION_MASK_ISA_AVX512VL - builtins that have both
12196 OPTION_MASK_ISA_MMX and some other bit set require both
12197 mmx and the ISAs without the mmx bit.
12198 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
12199 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
12200 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
12201 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
12202 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
12203 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
12204 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
12205 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
12206 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
12207 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
12208 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
12209 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
12210 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
12211 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
12212 Add OPTION_MASK_ISA_MMX.
12214 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
12216 * config.gcc (armv7*-*-freebsd*): New target.
12217 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
12219 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
12221 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
12222 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
12225 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
12227 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
12228 a template parameter.
12229 (WIDE_INT_REF_FOR): Update accordingly.
12230 * tree.h (wi::int_traits <const_tree>): Delete.
12231 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
12232 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
12233 (wi::tree_to_wide_ref): New typedef.
12234 (wi::to_wide): New function.
12235 * calls.c (get_size_range): Use wi::to_wide when operating on
12236 trees as wide_ints.
12237 * cgraph.c (cgraph_node::create_thunk): Likewise.
12238 * config/i386/i386.c (ix86_data_alignment): Likewise.
12239 (ix86_local_alignment): Likewise.
12240 * dbxout.c (stabstr_O): Likewise.
12241 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
12242 * expr.c (const_vector_from_tree): Likewise.
12243 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
12244 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
12245 (fold_negate_expr_1, int_const_binop_1, const_binop)
12246 (fold_convert_const_int_from_real, optimize_bit_field_compare)
12247 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
12248 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
12249 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
12250 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
12251 (fold_not_const, round_up_loc): Likewise.
12252 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
12253 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
12254 (alloca_call_type): Likewise.
12255 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
12256 * godump.c (go_output_typedef): Likewise.
12257 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
12258 * internal-fn.c (get_min_precision): Likewise.
12259 * ipa-cp.c (ipcp_store_vr_results): Likewise.
12260 * ipa-polymorphic-call.c
12261 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
12262 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
12263 (ipa_modify_call_arguments): Likewise.
12264 * match.pd: Likewise.
12265 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
12266 * print-tree.c (print_node_brief, print_node): Likewise.
12267 * stmt.c (expand_case): Likewise.
12268 * stor-layout.c (layout_type): Likewise.
12269 * tree-affine.c (tree_to_aff_combination): Likewise.
12270 * tree-cfg.c (group_case_labels_stmt): Likewise.
12271 * tree-data-ref.c (dr_analyze_indices): Likewise.
12272 (prune_runtime_alias_test_list): Likewise.
12273 * tree-dump.c (dequeue_and_dump): Likewise.
12274 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
12275 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
12276 * tree-pretty-print.c (dump_generic_node): Likewise.
12277 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
12278 (simple_iv_with_niters): Likewise.
12279 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
12280 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
12281 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
12282 * tree-ssa-loop-niter.c (split_to_var_and_offset)
12283 (refine_value_range_using_guard, number_of_iterations_ne_max)
12284 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
12285 (get_cst_init_from_scev, record_nonwrapping_iv)
12286 (scev_var_range_cant_overflow): Likewise.
12287 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
12288 * tree-ssa-pre.c (compute_avail): Likewise.
12289 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
12290 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
12291 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
12292 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
12293 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
12294 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
12295 (dump_case_nodes, try_switch_expansion): Likewise.
12296 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
12297 (vect_do_peeling): Likewise.
12298 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
12299 * tree-vect-stmts.c (vectorizable_load): Likewise.
12300 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
12301 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
12302 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
12303 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
12304 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
12305 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
12306 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
12307 (evrp_dom_walker::before_dom_children): Likewise.
12308 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
12309 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
12310 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
12311 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
12312 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
12313 (get_range_pos_neg): Likewise.
12314 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
12315 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
12316 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
12317 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
12318 * config/avr/avr.c (avr_fold_builtin): Likewise.
12319 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
12320 * config/msp430/msp430.c (msp430_attr): Likewise.
12321 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
12322 * config/powerpcspe/powerpcspe-c.c
12323 (altivec_resolve_overloaded_builtin): Likewise.
12324 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
12325 (rs6000_expand_ternop_builtin): Likewise.
12326 * config/rs6000/rs6000-c.c
12327 (altivec_resolve_overloaded_builtin): Likewise.
12328 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
12329 (rs6000_expand_ternop_builtin): Likewise.
12330 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
12332 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12334 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
12335 when copying loop nest with only one inner loop.
12337 2017-10-10 Richard Biener <rguenther@suse.de>
12339 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
12340 blocks if SCEV is active.
12341 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
12343 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
12344 (scev_initialize): Assert we are not yet initialized.
12346 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12348 * tree-loop-distribution.c (generate_loops_for_partition): Remove
12349 inner loop's exit stmt by making it always exit the loop, otherwise
12350 we would generate an infinite empty loop.
12352 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12354 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
12355 renaming variables in new preheader if it's deleted.
12357 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12359 * tree-loop-distribution.c (struct partition): Remove unused field
12360 loops of the structure.
12361 (partition_alloc, partition_free): Ditto.
12362 (build_rdg_partition_for_vertex): Ditto.
12364 2017-10-09 Jeff Law <law@redhat.com>
12366 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
12367 return type to match prototype and documentation.
12369 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12371 * config/rs6000/rs6000.c (processor_costs): Move to ...
12372 * config/rs6000/rs6000.h: ... here.
12373 (rs6000_cost): Declare.
12375 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
12377 * except.c (setjmp_fn): New global variable.
12378 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
12379 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
12380 if DONT_USE_BUILTIN_SETJMP is defined.
12382 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12384 * target.def (insn_cost): New hook.
12385 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
12386 * doc/tm.texi: Regenerate.
12387 * rtlanal.c (insn_cost): Use the new hook.
12389 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12391 * combine.c (combine_validate_cost): Compute the new insn_cost,
12392 not just pattern_cost.
12393 (try_combine): Adjust comment.
12395 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12397 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
12399 * combine.c (uid_insn_cost): Adjust comment.
12400 (combine_validate_cost): Adjust comment. Use pattern_cost instead
12402 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
12403 * dse.c (find_shift_sequence): Ditto.
12404 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
12405 (bb_valid_for_noce_process_p): Use pattern_cost.
12406 * rtl.h (insn_rtx_cost): Delete.
12407 (pattern_cost): New prototype.
12408 (insn_cost): New prototype.
12409 * rtlanal.c (insn_rtx_cost): Rename to...
12410 (pattern_cost): ... this.
12413 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
12415 * config/i386/i386.md (*jcc_2): Remove insn pattern.
12416 (*jcc<mode>_0_r_i387): Ditto.
12417 (*jccxf_r_i387): Ditto.
12418 (*jcc<mode>_r_i387): Ditto.
12419 (*jccu<mode>_r_i387): Ditto.
12420 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
12421 (*jcc): Rename from *jcc_1.
12423 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12425 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
12426 deferred rescans after the lvx/stvx recombination pre-pass.
12428 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
12430 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
12431 memory operation instruction support.
12432 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
12433 (rs6000-ibm-aix[789]*): Likewise.
12434 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
12435 Document new functions.
12437 2017-10-09 Richard Biener <rguenther@suse.de>
12439 PR tree-optimization/82397
12440 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
12441 equality only for semantically equal trees.
12443 2017-10-09 Richard Biener <rguenther@suse.de>
12445 PR tree-optimization/82449
12446 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
12447 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
12448 allow constant addresses.
12449 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
12452 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12454 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
12457 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12460 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
12463 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12466 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
12468 2017-10-09 Jakub Jelinek <jakub@redhat.com>
12471 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
12472 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
12474 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
12476 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
12477 (WI_BINARY_PREDICATE_RESULT): Likewise.
12478 (wi::binary_traits::operator_result): New type.
12479 (wi::binary_traits::predicate_result): Likewise.
12480 (generic_wide_int::operator~, unary generic_wide_int::operator-)
12481 (generic_wide_int::operator==, generic_wide_int::operator!=)
12482 (generic_wide_int::operator&, generic_wide_int::and_not)
12483 (generic_wide_int::operator|, generic_wide_int::or_not)
12484 (generic_wide_int::operator^, generic_wide_int::operator+
12485 (binary generic_wide_int::operator-, generic_wide_int::operator*):
12487 (operator~, unary operator-, operator==, operator!=, operator&)
12488 (operator|, operator^, operator+, binary operator-, operator*): New
12490 * expr.c (get_inner_reference): Use wi::bit_and_not.
12491 * fold-const.c (fold_binary_loc): Likewise.
12492 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
12493 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
12494 (bit_value_binop): Likewise.
12495 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
12496 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
12497 (extract_range_from_binary_expr_1): Likewise.
12498 (masked_increment): Likewise.
12499 (simplify_bit_ops_using_ranges): Likewise.
12501 2017-10-09 Martin Jambor <mjambor@suse.cz>
12504 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
12505 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
12506 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
12508 (hsa_fixup_mov_insn_type): New function.
12509 (hsa_op_with_type::get_in_type): Use it.
12510 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
12511 immediates in an assert.
12512 (hsa_op_with_type::extend_int_to_32bit): New method.
12513 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
12514 types. Convert to dest type if necessary.
12515 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
12516 (reg_for_gimple_ssa): Pass false as min32int to
12517 hsa_type_for_scalar_tree_type.
12518 (gen_hsa_addr): Fixup type when creating addresable temporary.
12519 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
12520 (gen_hsa_unary_operation): Extend operands and convert to dest type if
12521 necessary. Call hsa_fixup_mov_insn_type.
12522 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
12523 extend operands and convert to dest type if necessary.
12524 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
12525 to dest type if necessary.
12526 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
12527 if conversion nt necessary and size matches.
12528 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
12529 to dest type if necessary.
12530 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
12531 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
12533 (gen_hsa_clrsb): Likewise.
12534 (gen_hsa_ffs): Likewise.
12535 (gen_hsa_divmod): Extend operands and convert to dest type if
12537 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
12539 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12541 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
12542 Remove empty default arguments. Use a brace block as output
12544 (conditional return): Ditto.
12546 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
12547 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
12549 (group_ending_nop): Ditto.
12550 (doloop_end): Ditto.
12551 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
12552 (splitters for those): Ditto.
12554 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12556 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
12557 a conditional jump (and the compare for it) so that pc_rtx is the
12559 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
12560 for the deleted and renamed ctr<mode>_internal[234] patterns.
12561 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
12562 Delete second conditional return pattern.
12563 (ctr<mode>_internal2): Delete this second bdnz pattern.
12564 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
12565 (ctr<mode>_internal4): Delete this second bdz pattern.
12567 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12569 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
12570 (always_initialized_rtx_for_ssa_name_p): New predicate.
12571 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
12572 (finish_out_of_ssa): Free new field of SA.
12573 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
12574 * tree-ssa-coalesce.c: Include tree-ssa.h.
12575 (get_parm_default_def_partitions): Remove extern keyword.
12576 (get_undefined_value_partitions): New function.
12577 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12578 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12579 uninitialized bits.
12580 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12582 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12584 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12586 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12588 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12589 for targets that preffer 128bit.
12591 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12593 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12595 2017-10-08 Olivier Hainque <hainque@adacore.com>
12597 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12598 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12600 (thumb_set_return_address): Likewise.
12602 2017-10-08 Olivier Hainque <hainque@adacore.com>
12604 * common/config/arm/arm-common.c (arm_except_unwind_info):
12605 Handle DWARF2_UNWIND_INFO.
12607 2017-10-07 Michael Collison <michael.collison@arm.com>
12609 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12612 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12614 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12615 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12616 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12617 defined, force the creation of a new block for a dispatch label.
12619 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12621 * invoke.texi (Wsuggest-attribute=cold): Document.
12622 * common.opt (Wsuggest-attribute=cold): New
12623 * ipa-pure-const.c (warn_function_cold): New function.
12624 * predict.c (compute_function_frequency): Use it.
12625 * predict.h (warn_function_cold): Declare.
12627 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12629 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12632 2017-10-06 Martin Liska <mliska@suse.cz>
12634 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12635 keyword for member functions.
12636 (struct sanopt_tree_couple): New struct.
12637 (struct sanopt_tree_couple_hash): New function.
12638 (struct sanopt_ctx): Add new hash_map.
12639 (has_dominating_ubsan_ptr_check): New function.
12640 (record_ubsan_ptr_check_stmt): Likewise.
12641 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12642 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12643 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12645 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12648 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12649 aarch64_simd_valid_immediate on CONST_VECTORs.
12650 (aarch64_reg_or_bic_imm): Likewise.
12652 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12654 PR rtl-optimization/82396
12655 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12657 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12659 * graphite-dependences.c (scop_get_reads): Move code to...
12660 (scop_get_must_writes): Move code to...
12661 (scop_get_may_writes): Move code to...
12662 (scop_get_reads_and_writes): ... here.
12663 (scop_get_dependences): Call scop_get_reads_and_writes.
12665 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12667 PR tree-optimization/82434
12668 * fold-const.h (can_native_encode_type_p,
12669 can_native_encode_string_p): Remove.
12670 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12671 don't encode anything, just return what would be otherwise returned.
12672 (native_encode_fixed, native_encode_complex, native_encode_vector):
12674 (native_encode_string): Likewise. Inline by hand
12675 can_native_encode_string_p.
12676 (can_native_encode_type_p): Remove.
12677 (can_native_encode_string_p): Remove.
12678 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12679 STRING_CSTs using can_native_encode_string_p, test all
12680 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12681 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12682 argument from native_encode_expr.
12683 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12684 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12687 2017-10-06 Richard Biener <rguenther@suse.de>
12689 PR tree-optimization/82397
12690 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12691 operand_equal_p but rely on data_ref_compare_tree for detecting
12693 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12694 to match up with dr_group_sort_cmp.
12696 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12699 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12701 * config/s390/s390-builtin-types.def: Regenerate.
12703 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12706 * config/s390/s390-builtin-types.def: Regenerate.
12707 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12708 Change flag from B_VXE to B_VX.
12709 (s390_vec_min_dbl): Remove B_VXE flag.
12711 2017-10-06 Richard Biener <rguenther@suse.de>
12713 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12714 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12715 translate_isl_ast_to_gimple::is_valid_rename,
12716 translate_isl_ast_to_gimple::get_rename,
12717 translate_isl_ast_to_gimple::get_def_bb_for_const,
12718 translate_isl_ast_to_gimple::get_new_name,
12719 translate_isl_ast_to_gimple::collect_all_ssa_names,
12720 translate_isl_ast_to_gimple::copy_loop_phi_args,
12721 translate_isl_ast_to_gimple::collect_all_ssa_names,
12722 translate_isl_ast_to_gimple::copy_loop_phi_args,
12723 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12724 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12725 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12726 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12727 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12728 translate_isl_ast_to_gimple::copy_cond_phi_args,
12729 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12730 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12731 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12732 translate_isl_ast_to_gimple::rename_uses,
12733 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12734 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12735 (set_rename_for_each_def): Likewise.
12736 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12737 here. Handle rewriting SCEV analyzable uses here.
12738 (copy_bb_and_scalar_dependences): Generate code for PHI
12740 (graphite_regenerate_ast_isl): Adjust.
12741 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12742 (add_write, add_read): New functions.
12743 (build_cross_bb_scalars_def): Use it and simplify.
12744 (build_cross_bb_scalars_use): Likewise.
12745 (graphite_find_cross_bb_scalar_vars): Inline into...
12746 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12747 simulating out-of-SSA. Compute liveout and add dependencies.
12748 (build_scops): Force an empty entry block.
12749 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12751 (sese_build_liveouts): Declare.
12752 (sese_trivially_empty_bb_p): Likewise.
12753 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12754 compute liveout and debug_liveout.
12755 (sese_bad_liveouts_use): Remove.
12756 (sese_reset_debug_liveouts_bb): Likewise.
12757 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12758 (sese_build_liveouts): Build liveout and debug_liveout and store
12760 (new_sese_info): Adjust.
12761 (free_sese_info): Likewise.
12762 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12763 do not build liveout here.
12764 (move_sese_in_condition): Adjust region entry.
12765 (scev_analyzable_p): Match up with chrec_apply requirements.
12766 (sese_trivially_empty_bb_p): New.
12767 * tree-into-ssa.c (get_reaching_def): Properly support generating
12768 default-defs for incremental rewrite of anonymous names.
12770 2017-10-06 Richard Biener <rguenther@suse.de>
12772 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12773 precision do not perform modulo reduction.
12775 2017-10-06 Richard Biener <rguenther@suse.de>
12777 PR tree-optimization/82436
12778 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12779 conservatively choose the vectorization factor when checking
12780 whether we can perform the required load permutation.
12781 (vect_transform_slp_perm_load): Assert when we may not fail.
12783 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12785 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12786 message for incompatible -msdata=* and -mcall-* options.
12788 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12790 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12791 rate for post-reload scheduling.
12793 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12795 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12797 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12799 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12800 to improve monte carlo in scimark.
12802 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12804 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12805 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12806 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12807 pentium4_cost, nocona_cost): Set reassociation width to 1.
12808 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12809 width to 2 for fp operations and 1 otherwise.
12810 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12812 (atom_cost): Set reassociation width to 2.
12813 (slm_cost, generic_cost): Set fp reassociation width
12814 to 2 and 1 otherwise.
12815 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12816 (core_cost): Set fp reassociation width to 4 and vector to 2.
12817 (ix86_reassociation_width): Rewrite using cost table; special case
12818 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12819 and TARGET_AVX128_OPTIMAL.
12820 * config/i386/i386.h (processor_costs): Add
12821 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
12822 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
12823 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
12824 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
12825 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
12826 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
12828 2017-10-05 Nathan Sidwell <nathan@acm.org>
12830 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
12832 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12834 * config/arm/arm.c (arm_test_fpu_data): New.
12835 (arm_run_selftests): Call arm_test_fpu_data.
12837 2017-10-04 Nathan Sidwell <nathan@acm.org>
12839 * toplev.c (toplev::main): Remove excess parens on pretty_printer
12841 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
12843 2017-10-04 Sudakshina Das <sudi.das@arm.com>
12845 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
12846 check type for aarch64_simd_valid_immediate.
12847 (aarch64_output_simd_mov_immediate): Update prototype.
12848 (aarch64_simd_valid_immediate): Update prototype.
12849 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
12850 support for ORR-immediate.
12851 (and<mode>3): modified pattern to add support for BIC-immediate.
12852 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
12853 now checks for valid immediate for BIC and ORR based on new enum
12855 (aarch64_output_simd_mov_immediate): Function now used to output
12856 BIC/ORR imm as well based on new enum argument.
12857 * config/aarch64/constraints.md (Do): New vector immediate constraint.
12859 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
12860 (aarch64_reg_or_bic_imm): Likewise.
12862 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12864 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
12865 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
12867 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12871 PR rtl-optimization/82396
12872 * haifa-sched.c (autopref_multipass_init): Simplify
12874 (autopref_rank_data): Simplify sort order.
12875 * sched-int.h (autopref_multipass_data_): Remove
12876 multi_mem_insn_p, min_offset and max_offset.
12878 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12880 * doc/sourcebuild.texi: Document vect_peeling_profitable.
12882 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12884 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
12885 vect_doubleint_cvt.
12887 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12889 * doc/sourcebuild.texi: Document vect_long_mult.
12891 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
12893 PR tree-optimization/82413
12894 * fold-const.c (build_range_check): Use widest_int when comparing
12895 the maximum ETYPE value with HIGH.
12897 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12899 PR rtl-optimization/82396
12900 * haifa-sched.c (autopref_multipass_init): Simplify
12902 (autopref_rank_data): Simplify sort order.
12903 * sched-int.h (autopref_multipass_data_): Remove
12904 multi_mem_insn_p, min_offset and max_offset.
12906 2017-10-04 Jakub Jelinek <jakub@redhat.com>
12908 PR tree-optimization/82381
12909 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
12910 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
12913 PR tree-optimization/82374
12914 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
12915 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
12916 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
12917 current_function_decl to the new decl.
12919 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
12921 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
12922 helper macro for IEEE float128 hardware built-in functions.
12923 (SQRTF128_ODD): Add built-in functions with the round-to-odd
12925 (TRUNCF128_ODD): Likewise.
12926 (ADDF128_ODD): Likewise.
12927 (SUBF128_ODD): Likewise.
12928 (MULF128_ODD): Likewise.
12929 (DIVF128_ODD): Likewise.
12930 (FMAF128_ODD): Likewise.
12931 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
12932 UNSPEC_TRUNC_ROUND_TO_ODD.
12933 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
12934 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
12935 floating point round to odd instructions.
12936 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
12937 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
12938 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
12939 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
12940 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
12941 (trunc<mode>sf2_hw): Change the truncate with round to odd
12942 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
12943 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
12944 to odd hardware instructions.
12945 (sub<mode>3_odd): Likewise.
12946 (mul<mode>3_odd): Likewise.
12947 (div<mode>3_odd): Likewise.
12948 (sqrt<mode>2_odd): Likewise.
12949 (fma<mode>4_odd): Likewise.
12950 (fms<mode>4_odd): Likewise.
12951 (nfma<mode>4_odd): Likewise.
12952 (nfms<mode>4_odd): Likewise.
12953 (trunc<mode>df2_odd): Change the truncate with round to odd
12954 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
12956 * doc/extend.texi (PowerPC built-in functions): Update documentation
12957 for existing IEEE float128-bit built-in functions. Add built-in
12958 functions that generate the IEEE 128-bit floating point round to
12961 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
12963 PR rtl-optimization/77729
12964 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
12965 to (X&(C1&~C2))|C2 transformations.
12967 2017-10-03 Martin Jambor <mjambor@suse.cz>
12969 PR tree-optimization/82363
12970 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
12971 mismatch, mark lacc written regardless of racc.
12973 2017-10-03 Jakub Jelinek <jakub@redhat.com>
12975 PR tree-optimization/82381
12976 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
12977 stmt_to_insert nor wheather SSA_NAMEs are default defs.
12978 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
12979 fallthrough into reassoc_stmt_dominates_stmt_p.
12982 * combine.c (combine_instructions): Don't combine in unreachable
12985 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
12988 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
12989 function to not use the have_cpu variable. Do not set cpu_index,
12990 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
12991 or the default cpu.
12992 (rs6000_valid_attribute_p): Remove duplicate initializations of
12993 old_optimize and func_optimize.
12994 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
12995 (rs6000_activate_target_options): Make global.
12996 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
12999 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13001 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13002 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13003 Set *poffset to 0 and *psize and *pmax_size to -1 if
13004 *poffset + *psize overflows in HOST_WIDE_INT.
13006 PR tree-optimization/82387
13007 PR tree-optimization/82388
13008 PR tree-optimization/82389
13009 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13010 instead of live_bytes non-NULL.
13012 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13015 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13018 2017-10-02 Richard Biener <rguenther@suse.de>
13020 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13021 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13022 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13024 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13026 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13027 requested precision matches the type's.
13028 * calls.c (alloc_max_size): Calculate the new candidate size as
13029 a widest_int and use wi::to_widest when comparing it with the
13030 current candidate size.
13031 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13032 zero rather than integer_zero_node.
13033 * match.pd: Check for a no-op conversion before using wi::add
13034 rather than after. Use tree_to_uhwi when summing small shift
13035 counts into an unsigned int.
13037 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13038 Alan Hayward <alan.hayward@arm.com>
13039 David Sherwood <david.sherwood@arm.com>
13042 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13043 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13044 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13045 POINTER_AND_FP_REGS.
13047 2017-10-02 Richard Biener <rguenther@suse.de>
13049 PR tree-optimization/82355
13050 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13051 a mapping for the enclosing loop but avoid generating one for
13052 the loop tree root.
13053 (copy_bb_and_scalar_dependences): Remove premature codegen
13054 error on PHIs in blocks duplicated into multiple places.
13055 * graphite-scop-detection.c
13056 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13057 in the region use it as loop and nest to analyze the DR in.
13058 (try_generate_gimple_bb): Likewise.
13059 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13060 (add_loop_constraints): For blocks in a loop not in the region
13061 create a dimension with a single iteration.
13062 * sese.h (gbb_loop_at_index): Remove assert.
13064 2017-10-01 Kevin Buettner <kevinb@redhat.com>
13066 * omp-expand.c (adjust_context_scope): New function.
13067 (expand_parallel_call): Call adjust_context_scope.
13069 2017-10-01 Jeff Law <law@redhat.com>
13071 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
13072 dom_opt_dom_walker class with direct access to private members.
13073 Add comments. Call test_for_singularity.
13074 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
13075 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
13076 m_dummy_cond anymore.
13077 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
13079 (pass_dominator:execute): Build the dummy_cond here and pass it
13080 to the dom_opt_dom_walker ctor.
13081 (test_for_singularity): New function.
13083 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
13084 Maya Rashish <coypu@sdf.org>
13086 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
13087 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
13088 (alpha*-*-netbsd*) Use nbsd_tm_file.
13089 (arm*-*-netbsdelf*) Likewise.
13090 (i[34567]86-*-netbsdelf*) Likewise.
13091 (x86_64-*-netbsd*) Likewise.
13092 (mips*-*-netbsd*) Likewise.
13093 (powerpc-*-netbsd*) Likewise.
13094 (sh*-*-netbsd*) Likewise.
13095 (sparc-*-netbsdelf*) Likewise.
13096 (sparc64-*-netbsd*) Likewise.
13097 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
13099 (vax-*-netbsdelf*) Likewise.
13100 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
13101 (UINT_FAST8_TYPE) Likewise.
13102 (INT_FAST16_TYPE) Check CHAR_FAST16.
13103 (UINT_FAST16_TYPE) Likewise.
13105 2017-09-30 Jakub Jelinek <jakub@redhat.com>
13108 * config/i386/i386.md
13109 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
13110 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
13111 *divmodsi4_zext_2): New define_insn_and_split.
13112 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
13113 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
13114 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
13115 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
13116 New define_insn_and_split.
13117 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
13118 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
13119 operands[1] having DImode when mode is SImode.
13121 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
13122 always SImode for DIV and MOD in REG_EQUAL notes.
13124 2017-09-29 Yury Gribov <tetra2005@gmail.com>
13126 PR middle-end/82319
13127 * match.pd: Fix handling of NaNs in pattern.
13129 2017-09-29 Jeff Law <law@redhat.com>
13131 * sbitmap.c (bitmap_bit_in_range_p): New function.
13132 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
13133 * tree-ssa-dse.c (live_bytes_read): New function.
13134 (dse_classify_store): Ignore reads of dead bytes.
13136 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
13137 typos and whitespace errors.
13138 * config/i386/predicates.md (address_no_seg_operand): Likewise.
13139 * config/s390/s390.c (s390_emit_prologue): Likewise.
13141 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13144 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
13145 with a symbol for LRA.
13147 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13149 PR rtl-optimization/82338
13150 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
13152 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
13154 * genmodes.c (calc_wider_mode): Suppress qsort macro.
13155 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
13156 (qsort_chk): Declare.
13157 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
13158 (qsort_chk): New function.
13160 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13162 PR tree-optimization/82337
13163 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
13164 phi definition if the PHI result appears in an abnormal PHI.
13165 (find_basis_for_base_expr): Don't record a basis if the LHS of the
13166 basis appears in an abnormal PHI.
13168 2017-09-29 Richard Biener <rguenther@suse.de>
13170 * graphite-isl-ast-to-gimple.c
13171 (translate_isl_ast_to_gimple::set_codegen_error): New function.
13172 (binary_op_to_tree): Use it.
13173 (get_rename_from_scev): Likewise.
13174 (copy_loop_phi_nodes): Likewise.
13175 (copy_bb_and_scalar_dependences): Likewise.
13176 (translate_pending_phi_nodes): Likewise.
13178 2017-09-29 Jakub Jelinek <jakub@redhat.com>
13181 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
13182 for movabsq $(i32 << shift), r64.
13184 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
13186 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
13187 index when encoding %esp as %rsp to avoid 0x67 prefix.
13189 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
13191 * config/i386/i386.md (*movsf_internal, *movdf_internal):
13192 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
13194 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
13196 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
13197 Extensions with more than 16 double VFP registers.
13198 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
13199 to_clear_mask and all code related to it. Replace the remaining
13200 entry by a sbitmap and adapt code accordingly.
13202 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
13204 * brig-builtins.def: Change pure attributes to const.
13206 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13208 * config.gcc (default_gnu_indirect_function): Default to yes for
13209 sparc*-*-linux* with glibc.
13211 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13213 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
13214 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
13215 when creating .init_array and .fini_array sections with priority
13218 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
13221 * config/aarch64/aarch64.c
13222 (aarch64_builtin_support_vector_misalignment): Always return false
13223 when misalignment is unknown.
13225 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
13227 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
13228 this function to return false if the definition used by the swap
13229 instruction is artificial, or if the memory address from which the
13230 constant value is loaded is not represented by a base address held
13231 in a register or if the base address register is a frame or stack
13232 pointer. Additionally, return false if the base address of the
13233 loaded constant is a SYMBOL_REF but is not considered to be a
13235 (replace_swapped_load_constant): New function.
13236 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
13237 loaded constant vector with a load of a swapped constant vector.
13239 2017-09-27 Carl Love <cel@us.ibm.com>
13241 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
13242 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
13243 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
13246 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
13248 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
13249 first, always call autopref_rank_data otherwise.
13251 2017-09-27 Richard Biener <rguenther@suse.de>
13253 * graphite-scop-detection.c (find_scop_parameters): Move
13254 loop bound handling ...
13255 (gather_bbs::before_dom_children): ... here, avoiding the need
13256 to build scop_info->loop_nest.
13257 (record_loop_in_sese): Remove.
13258 * sese.h (sese_info_t::loop_nest): Remove.
13259 * sese.c (new_sese_info): Do not allocate loop_nest.
13260 (free_sese_info): Do not free loop_nest.
13262 2017-09-27 Jakub Jelinek <jakub@redhat.com>
13265 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
13266 lhs from calls if the lhs has addressable type.
13268 2017-09-27 Richard Biener <rguenther@suse.de>
13270 * graphite.h (scop::max_alias_set): New member.
13271 * graphite-scop-detection.c: Remove references to non-existing
13272 --param in comments.
13273 (build_alias_sets): Record the maximum alias set used for drs.
13274 (build_scops): Support zero as unlimited for
13275 --param graphite-max-arrays-per-scop.
13276 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
13277 and inline into ...
13278 (build_poly_sr_1): ... here. Compute alias set based on the
13279 maximum alias set used for drs rather than
13280 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
13282 2017-09-27 Richard Biener <rguenther@suse.de>
13284 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
13285 --param loop-block-tile-size=0 to disable tiling.
13287 2017-09-27 Richard Biener <rguenther@suse.de>
13289 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
13290 (graphite-max-nb-scop-params): Document special value zero.
13291 * domwalk.h (dom_walker::STOP): New symbolical constant.
13292 (dom_walker::dom_walker): Add optional parameter for bb to
13294 (dom_walker::~dom_walker): Declare.
13295 (dom_walker::before_dom_children): Document STOP return value.
13296 (dom_walker::m_user_bb_to_rpo): New member.
13297 (dom_walker::m_bb_to_rpo): Likewise.
13298 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
13299 mapping here if not provided by the user.
13300 (dom_walker::~dom_walker): Free bb to RPO mapping if not
13301 provided by the user.
13302 (dom_walker::STOP): Define.
13303 (dom_walker::walk): Do not compute bb to RPO mapping here.
13304 Support STOP return value from before_dom_children to stop
13306 * graphite-optimize-isl.c (optimize_isl): If the schedule
13307 is the same still generate code if -fgraphite-identity
13308 or -floop-parallelize-all are given.
13309 * graphite-scop-detection.c: Include cfganal.h.
13310 (gather_bbs::gather_bbs): Get and pass through bb to RPO
13312 (gather_bbs::before_dom_children): Return STOP for BBs
13314 (build_scops): Compute bb to RPO mapping and pass it to
13315 the domwalk. Treat --param graphite-max-nb-scop-params=0
13316 as not limiting the number of params.
13317 * graphite.c (graphite_initialize): Remove limit on the
13318 number of basic-blocks in a function.
13319 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
13320 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
13321 default value of 10.
13323 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13325 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
13326 Adjust code to eliminate needing to do the shift right 32-bits
13327 operation after XSCVDPSPN.
13329 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13331 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
13332 ((X / Y) != 0 -> X >= Y): Likewise.
13334 2017-09-26 Carl Love <cel@us.ibm.com>
13336 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
13337 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
13338 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
13339 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
13340 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
13341 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
13342 definitions and overloading.
13343 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
13344 statement for P9V_BUILTIN_XST_LEN_R.
13345 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
13346 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
13347 define_expand and define_insn for the instructions and builtins.
13348 * doc/extend.texi: Update the built-in documentation file for the new
13349 built-in functions.
13350 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
13351 define_insn for the instructions
13353 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
13356 * gcc/config/netbsd-protos.h: New file.
13357 * gcc/config/netbsd.c: New file.
13358 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
13359 * gcc/config/t-netbsd: New file.
13360 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
13361 (tmake_file) Add t-netbsd.
13362 (extra_objs) Add netbsd.o.
13364 2017-09-26 Janus Weil <janus@gcc.gnu.org>
13368 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
13370 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13372 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
13373 sign extension from a vector register to a GPR by doing a 32-bit
13374 direct move and then an EXTSW.
13375 (extendsi<mode>2 splitter): Likewise.
13376 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
13377 right or vector extract after doing XSCVDPSPN. Use
13378 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
13380 (movdi_from_sf_zero_ext): Likewise.
13381 (reload_gpr_from_vsxsf): Likewise.
13382 (p8_mfvsrd_4_disf): Delete, no longer used.
13383 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
13384 then needing to move the SFmode to a GPR to use the XSCVDPSP
13385 instruction instead of FRSP and XSCVDPSPN.
13386 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
13387 it is adjacent to the other XSCVSPDP insns.
13388 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
13389 SFmode to be in traditional Altivec registers.
13390 (vsx_xscvdpspn): Eliminate useless alternative constraint.
13391 (vsx_xscvspdpn): Likewise.
13392 (vsx_xscvspdpn_scalar): Likewise.
13394 2017-09-26 Martin Jambor <mjambor@suse.cz>
13396 * tree-sra.c (compare_access_positions): Put integral types first,
13397 stabilize sorting of integral types, remove conditions putting
13398 non-full-precision integers last.
13399 (sort_and_splice_var_accesses): Disable scalarization if a
13400 non-integert would be represented by a non-full-precision integer.
13402 2017-09-26 Joseph Myers <joseph@codesourcery.com>
13404 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
13405 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
13406 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
13407 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
13408 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
13409 conditionals inside the function instead of around it. Call
13410 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
13411 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
13413 2017-09-26 Richard Biener <rguenther@suse.de>
13415 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
13417 (scop_detection::build_scop_breadth): ... this. Removed.
13418 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
13419 (scop_detection::harmful_stmt_in_bb): Likewise.
13420 (scop_detection::graphite_can_represent_stmt): Likewise.
13421 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
13422 (scop_detection::can_represent_loop): Remove recursion, fold in ...
13423 (scop_detection::can_represent_loop_1): ... this. Removed.
13424 (scop_detection::harmful_loop_in_region): Simplify after inlining
13425 the above and remove more quadraticness.
13426 (build_scops): Adjust.
13427 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
13430 2017-09-26 Jakub Jelinek <jakub@redhat.com>
13433 * config/i386/i386.c (ix86_print_operand_address_as): Only test
13434 REGNO (base) == SP_REG if base is a REG.
13436 PR middle-end/35691
13437 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
13438 if it is different SSA_NAME.
13439 (optimize_range_tests_cmp_bitwise): New function.
13440 (optimize_range_tests): Call it.
13442 2017-09-26 Richard Biener <rguenther@suse.de>
13444 PR tree-optimization/82321
13445 * graphite.c (canonicalize_loop_closed_ssa): Properly check
13446 for the def being inside the loop.
13448 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13450 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
13452 * config/s390/s390-builtins.def: Fix constraint on op4.
13454 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13456 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
13457 independent expanders.
13458 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
13459 ("vec_ordered", "vec_unordered"): New expanders.
13461 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13463 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
13466 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13468 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
13469 vec_unpacks_lo_v16qi.
13470 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
13472 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13474 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
13475 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
13476 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
13478 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13480 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
13482 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
13484 ("*vec_slb<mode>"): New insn pattern.
13485 ("vec_shr_<mode>"): New expander.
13486 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
13487 and force the shift count operand to V16QImode.
13488 ("vec_srb<mode>"): Set shift count mode to V16QI.
13490 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13492 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
13493 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
13494 ("vec_widen_smult_hi_<mode>"): New expander definitions.
13496 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
13499 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
13501 2017-09-26 Richard Biener <rguenther@suse.de>
13503 PR tree-optimization/82320
13504 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
13507 2017-09-25 Jeff Law <law@redhat.com>
13509 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
13510 prototype for new argument.
13511 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
13512 mostly extracted from rs6000_emit_allocate_stack.
13513 (rs6000_emit_probe_stack_range_stack_clash): New function.
13514 (rs6000_emit_allocate_stack): Call
13515 rs6000_emit_probe_stack_range_stack_clash as needed.
13516 (rs6000_emit_probe_stack_range): Add additional argument
13517 to call to gen_probe_stack_range{si,di}.
13518 (output_probe_stack_range): New.
13519 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
13520 (output_probe_stack_range_stack_clash): New.
13521 (rs6000_emit_prologue): Emit notes into dump file as requested.
13522 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
13523 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
13524 Add additional operand and pass it to output_probe_stack_range.
13526 2017-09-25 Bin Cheng <bin.cheng@arm.com>
13528 PR tree-optimization/82163
13529 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
13530 (checking_verify_loop_closed_ssa): New parameter.
13531 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
13532 (check_loop_closed_ssa_stmt): Delete.
13533 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
13534 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
13535 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
13538 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
13540 * brig-builtins.def: Treat HSAIL barrier builtins as
13541 setjmp/longjump style functions.
13543 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13545 * target.def (constant_alignment): New hook.
13546 * defaults.h (CONSTANT_ALIGNMENT): Delete.
13547 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
13548 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
13549 * doc/tm.texi: Regenerate.
13550 * targhooks.h (default_constant_alignment): Declare.
13551 (constant_alignment_word_strings): Likewise.
13552 * targhooks.c (default_constant_alignment): New function.
13553 (constant_alignment_word_strings): Likewise.
13554 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
13555 instead of CONSTANT_ALIGNMENT.
13556 * varasm.c (align_variable, get_variable_align, build_constant_desc)
13557 (force_const_mem): Likewise.
13558 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
13559 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
13560 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
13561 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13562 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
13564 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
13565 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13566 constant_alignment_word_strings.
13567 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
13568 (CONSTANT_ALIGNMENT): Likewise.
13569 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13570 (arm_constant_alignment): New function.
13571 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
13572 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13573 constant_alignment_word_strings.
13574 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
13575 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13576 constant_alignment_word_strings.
13577 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13578 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13579 (cris_constant_alignment): New function.
13580 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13581 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13582 (epiphany_constant_alignment): New function.
13583 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13584 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13585 constant_alignment_word_strings.
13586 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13587 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13588 constant_alignment_word_strings.
13589 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13590 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13591 constant_alignment_word_strings.
13592 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13593 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13594 * config/i386/i386.c (ix86_constant_alignment): Make static.
13595 Use the same interface as the target hook.
13596 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13597 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13598 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13599 constant_alignment_word_strings.
13600 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13601 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13602 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13603 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13604 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13605 constant_alignment_word_strings.
13606 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13607 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13608 constant_alignment_word_strings.
13609 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13610 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13611 constant_alignment_word_strings.
13612 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13613 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13615 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13616 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13617 * config/mips/mips.c (mips_constant_alignment): New function.
13618 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13619 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13620 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13621 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13622 (mmix_constant_alignment): Make static. Use the same interface
13623 as the target hook.
13624 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13625 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13626 constant_alignment_word_strings.
13627 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13628 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13629 constant_alignment_word_strings.
13630 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13631 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13632 constant_alignment_word_strings.
13633 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13634 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13635 (rs6000_constant_alignment): New function.
13636 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13637 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13638 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13639 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13640 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13641 (rs6000_constant_alignment): New function.
13642 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13643 * config/s390/s390.c (s390_constant_alignment): New function.
13644 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13645 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13646 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13647 constant_alignment_word_strings.
13648 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13649 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13650 (sparc_constant_alignment): New function.
13651 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13652 * config/spu/spu.c (spu_constant_alignment): New function.
13653 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13654 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13655 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13656 constant_alignment_word_strings.
13657 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13658 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13659 constant_alignment_word_strings.
13660 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13661 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13662 constant_alignment_word_strings.
13663 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13664 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13665 (visium_constant_alignment): New function.
13666 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13667 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13668 (xtensa_constant_alignment): New function.
13669 * system.h (CONSTANT_ALIGNMENT): Poison.
13671 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13673 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13674 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13675 (rs6000_builtin_valid_without_lhs): New helper function.
13676 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13677 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13679 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13681 * target.h (vec_perm_indices): Use unsigned short rather than
13683 (auto_vec_perm_indices): Likewise.
13684 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13685 Use unsigned int rather than unsigned char.
13686 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13688 2017-09-25 Richard Biener <rguenther@suse.de>
13690 * cfgloop.h (sort_sibling_loops): Declare.
13691 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13692 (sort_sibling_loops): New function sorting the sibling loop list
13694 * graphite.c (graphite_transform_loops): Sort sibling loops.
13696 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13698 * target.def (vec_perm_const_ok): Change sel parameter to
13700 * optabs-query.c (can_vec_perm_p): Update accordingly.
13701 * doc/tm.texi: Regenerate.
13702 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13703 auto_vec_perm_indices and remove separate nelt field.
13704 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13705 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13706 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13707 (aarch64_expand_vec_perm_const): Update accordingly.
13708 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13709 to vec_perm_indices.
13710 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13711 auto_vec_perm_indices and remove separate nelt field.
13712 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13713 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13714 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13716 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13717 to vec_perm_indices.
13718 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13719 sel to vec_perm_indices.
13720 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13721 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13722 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13724 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13727 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13730 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13731 on the FUNCTION_DECL function context if it has a DIE that is a
13734 2017-09-25 Richard Biener <rguenther@suse.de>
13736 PR tree-optimization/82285
13737 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13740 2017-09-25 Tom de Vries <tom@codesourcery.com>
13744 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13747 2017-09-25 Richard Biener <rguenther@suse.de>
13749 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13750 ISL errors other than isl_error_quota happen. Dump if the
13751 schedule is the same.
13752 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13753 errors instead of aborting inside ISL.
13755 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13758 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13759 of libgcc_eh for m64.
13760 * config/i386/darwin64.h: Likewise.
13762 2017-09-25 Richard Biener <rguenther@suse.de>
13764 PR middle-end/82144
13765 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13766 attribute for incomplete types nor twice for complete ones.
13768 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13771 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13772 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13775 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13778 * config/i386/i386.opt (mprefer-avx256): Use
13779 ix86_target_flags variable.
13780 * config/i386/i386.c (ix86_target_string): Move
13781 -mprefer-avx256 to flag2_opts.
13783 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13785 PR middle-end/35691
13786 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13787 and x != -1 | y != -1 into (x & y) != -1.
13789 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13791 * config.gcc: Add new case statement to set
13792 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13793 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13794 s390x-*-linux* case statements. Added aarch64 to the list of
13795 supported architectures.
13797 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13799 PR tree-optimization/82289
13800 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13801 STMT_VINFO_RELEVANT_P.
13803 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13804 Alan Hayward <alan.hayward@arm.com>
13805 David Sherwood <david.sherwood@arm.com>
13807 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
13808 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13809 (extract_range_from_binary_expr_1): Don't call
13810 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
13812 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13813 Alan Hayward <alan.hayward@arm.com>
13814 David Sherwood <david.sherwood@arm.com>
13816 * target.def (preferred_vector_alignment): New hook.
13817 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13819 * doc/tm.texi: Regenerate.
13820 * targhooks.h (default_preferred_vector_alignment): Declare.
13821 * targhooks.c (default_preferred_vector_alignment): New function.
13822 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
13824 (DR_TARGET_ALIGNMENT): New macro.
13825 (aligned_access_p): Update commentary.
13826 (vect_known_alignment_in_bytes): New function.
13827 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
13829 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
13830 Calculate the misalignment based on the target alignment rather than
13832 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
13833 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
13834 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
13835 the target alignment, rather than masking the element misalignment
13836 with the number of elements in a vector. Also use the target
13837 alignment when calculating the maximum number of peels.
13838 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
13839 instead of TYPE_ALIGN_UNIT.
13840 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
13841 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
13842 (vect_create_addr_base_for_vector_ref): Update call accordingly.
13843 (vect_create_data_ref_ptr): Likewise.
13844 (vect_setup_realignment): Realign by ANDing with
13845 -DR_TARGET_MISALIGNMENT.
13846 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
13847 the number of peels based on DR_TARGET_ALIGNMENT.
13848 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
13849 with the guaranteed alignment boundary when deciding whether
13851 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
13852 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
13853 (ensure_base_align): Remove stmt_info parameter. Get the
13854 target base alignment from DR_TARGET_ALIGNMENT.
13855 (vectorizable_store): Update call accordingly. Interpret
13856 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
13858 (vectorizable_load): Likewise.
13860 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13861 Alan Hayward <alan.hayward@arm.com>
13862 David Sherwood <david.sherwood@arm.com>
13864 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
13865 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
13866 (vect_enhance_data_refs_alignment): Likewise.
13868 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13870 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13871 error. Only quit immediately if parsing is complete.
13872 (BEGIN): Initialize fatal_err and parse_done.
13873 (begin fpu, end fpu): Check number of arguments.
13874 (begin arch, end arch): Likewise.
13875 (begin cpu, end cpu): Likewise.
13876 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13877 (optalias): Likewise.
13879 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13881 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13882 * config/arm/arm-isa.h: Delete. Move definitions to ...
13883 * arm-cpus.in: ... here. Use new feature and fgroup values.
13884 * config/arm/arm.c (arm_option_override): Use lower case for feature
13886 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13887 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13888 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13890 (print_isa_bits_for): New function.
13891 (gen_isa): New function.
13892 (gen_comm_data): Use print_isa_bits_for.
13893 (define feature): New keyword.
13894 (define fgroup): New keyword.
13895 * config/arm/t-arm (TM_H): Remove.
13896 (GTM_H): Add arm-isa.h.
13897 (arm-isa.h): Add rule to generate file.
13898 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13899 case for feature bit names.
13901 2017-09-22 Richard Biener <rguenther@suse.de>
13903 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
13905 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
13906 print of no dependency loops ...
13907 * graphite.c (graphite_transform_loops): ... here.
13908 (canonicalize_loop_closed_ssa_form): Work from inner to outer
13910 (same_close_phi_node, remove_duplicate_close_phi,
13911 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
13912 (canonicalize_loop_closed_ssa): ... here and simplify.
13913 * graphite-optimize-isl.c: Include tree-vectorizer.h.
13914 (optimize_isl): Use dump_printf_loc to tell when we stopped
13915 optimizing because of an ISL timeout.
13917 2017-09-22 Richard Biener <rguenther@suse.de>
13919 PR tree-optimization/82291
13920 * tree-if-conv.c (predicate_mem_writes): Make sure to
13921 remove writes in blocks predicated with false.
13923 2017-09-22 Richard Biener <rguenther@suse.de>
13925 * sese.c: Include cfganal.h.
13926 (if_region_set_false_region): Remove.
13927 (create_if_region_on_edge): Likewise.
13928 (move_sese_in_condition): Re-implement without destroying
13931 2017-09-22 Richard Biener <rguenther@suse.de>
13933 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
13934 Verify both BBs contain loop PHI nodes before dispatching to
13935 copy_loop_phi_args.
13936 (graphite_regenerate_ast_isl): Do not recompute dominators,
13937 do not verify three times. Restructure for clarity.
13938 * graphite-scop-detection.c (same_close_phi_node,
13939 remove_duplicate_close_phi, make_close_phi_nodes_unique,
13940 defined_in_loop_p, canonicalize_loop_closed_ssa,
13941 canonicalize_loop_closed_ssa_form): Simplify, remove excess
13942 checking and SSA rewrite, move to ...
13943 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
13944 (graphite_initialize): Do not pass in ctx, do not reset the
13945 SCEV cache, compute only dominators.
13946 (graphite_transform_loops): Allocate ISL ctx after
13947 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
13948 Maintain post-dominators only around build_scops.
13949 * sese.c (if_region_set_false_region): Make static. Free
13950 and recompute dominators.
13951 (move_sese_in_condition): Assert we don't get called with
13952 post-dominators computed.
13953 * sese.h (if_region_set_false_region): Remove.
13955 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
13957 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
13958 mode attribute for TARGET_AVX512VL.
13960 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
13962 * config/i386/i386.opt (mprefer-avx256): New option.
13963 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
13965 (ix86_preferred_simd_mode): Return 256-bit AVX modes
13966 for TARGET_PREFER_AVX256.
13967 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
13969 2017-09-21 Jeff Law <law@redhat.com>
13971 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
13972 Fix dump output if the only stack space is for pushed registers.
13974 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13976 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
13979 2017-09-21 Martin Sebor <msebor@redhat.com>
13982 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
13983 code (in C++) or code that triggers warnings.
13985 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
13987 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
13989 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
13991 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
13992 * sched-rgn.c: Ditto.
13993 * sel-sched-ir.c: Ditto.
13995 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
13997 * toplev.h (set_random_seed): Adjust return type.
13998 * toplev.c (init_local_tick): Move eager initialization of random_seed
13999 to get_random_seed. Adjust comment.
14000 (init_random_seed): Inline to get_random_seed, delete.
14001 (get_random_seed): Initialize random_seed lazily.
14002 (set_random_seed): Do not return previous value.
14003 (print_switch_value): Do not call get_random_seed.
14005 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14007 * cgraph.c (delete_function_version): New, broken out from...
14008 (cgraph_node::delete_function_version): ...here. Rename to
14009 cgraph_node::delete_function_version_by_decl. Update all uses.
14010 (cgraph_node::remove): Call delete_function_version.
14012 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14015 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14016 VAR_DECLs to which addressable non-volatile parameters are mapped
14017 and for id->retvar after the return value assignment. Clear
14018 id->retval and id->retbnd after inlining.
14020 2017-09-21 Richard Biener <rguenther@suse.de>
14022 PR tree-optimization/82276
14023 PR tree-optimization/82244
14024 * tree-vrp.c (build_assert_expr_for): Set
14025 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14027 (remove_range_assertions): Revert earlier change.
14029 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14032 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14034 2017-09-21 Richard Biener <rguenther@suse.de>
14036 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14037 Restore valid IL after code generation errors.
14038 * graphite.c (graphite_transform_loops): Diagnose code
14039 generation issues as MSG_MISSED_OPTIMIZATION and continue
14040 with processing SCOPs.
14042 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14043 Alan Hayward <alan.hayward@arm.com>
14044 David Sherwood <david.sherwood@arm.com>
14046 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14047 rather than choosing between plus_constant and gen_rtx_<CODE>.
14048 * expr.c (emit_push_insn): Likewise.
14049 (expand_expr_real_2): Likewise.
14051 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14052 Alan Hayward <alan.hayward@arm.com>
14053 David Sherwood <david.sherwood@arm.com>
14055 * loop-unroll.c (split_iv): Call copy_rtx on the step.
14057 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14058 Alan Hayward <alan.hayward@arm.com>
14059 David Sherwood <david.sherwood@arm.com>
14061 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14062 calling tree_to_uhwi.
14064 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14065 Alan Hayward <alan.hayward@arm.com>
14066 David Sherwood <david.sherwood@arm.com>
14068 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
14069 INTEGER_CST rather than a negative test for ADDR_EXPR.
14071 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14072 Alan Hayward <alan.hayward@arm.com>
14073 David Sherwood <david.sherwood@arm.com>
14075 * tree-vrp.c (extract_range_from_binary_expr_1): Check
14076 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
14078 2017-09-21 Richard Biener <rguenther@suse.de>
14080 PR tree-optimization/71351
14081 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
14082 graphite_create_new_loop_guard): Remove, fold remaining parts
14084 (translate_isl_ast_node_for): ... here and simplify.
14086 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14089 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
14090 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
14091 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
14092 alternative always use QI mode, for -Os imov (=R,R) alternative
14093 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
14096 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14097 Jeff Law <law@redhat.com>
14099 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
14100 (allocate_stack_space): New function, partially extracted from
14101 s390_emit_prologue.
14102 (s390_emit_prologue): Track offset to most recent stack probe.
14103 Code to allocate space moved into allocate_stack_space.
14104 Dump actions when no stack is allocated.
14105 (s390_prologue_plus_offset): New function.
14106 (s390_emit_stack_probe): Likewise.
14108 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
14110 * common.opt (Wa, Wl, Wp, g, gz=): Add
14112 (gno-column-info): Remove.
14113 (gcolumn-info): Drop RejectNegative.
14114 (gno-): New prefix.
14115 (gno-record-gcc-switches): Remove.
14116 (grecord-gcc-switches): Drop RejectNegative.
14117 (gno-split-dwarf): Remove.
14118 (gsplit-dwarf): Drop RejectNegative.
14119 (gno-strict-dwarf): Remove.
14120 (gstrict-dwarf): Drop RejectNegative.
14121 * config/darwin.opt (gfull, gused): Add RejectNegative.
14122 * dwarf2out.c (gen_producer_string): Drop
14123 gno-record-gcc-switches handler.
14124 * optc-gen.awk: Add g to prefixes with negative forms.
14125 * opts-common.c (remapping_prefix_p): New.
14126 (find_opt): Check it.
14127 (generate_canonical_option): Test g prefix.
14128 (option_map): Add -gno- mapping.
14129 (add_misspelling_candidates): Check remapping_prefix_p.
14131 2017-09-20 Jeff Law <law@redhat.com>
14133 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
14134 thinko in stack clash protection support.
14136 * explow.c (compute_stack_clash_protection_loop_data): Use
14137 CONST_INT_P instead of explicit test. Verify object is a
14138 CONST_INT_P before looking at INTVAL.
14139 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
14140 instead of explicit test.
14142 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
14145 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
14146 address instead of to r1 and r11.
14148 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
14150 * config.gcc: Support "knm".
14151 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
14152 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14154 * config/i386/i386.c (m_KNM): Define.
14155 (processor_target_table): Add "knm".
14157 (ix86_option_override_internal): Add "knm".
14158 (ix86_issue_rate): Add PROCESSOR_KNM.
14159 (ix86_adjust_cost): Ditto.
14160 (ia32_multipass_dfa_lookahead): Ditto.
14161 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
14162 (fold_builtin_cpu): Add M_INTEL_KNM.
14163 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
14164 (processor_type): Add PROCESSOR_KNM.
14165 * config/i386/x86-tune.def: Add m_KNM.
14166 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
14168 2017-09-20 Richard Biener <rguenther@suse.de>
14170 PR tree-optimization/80213
14171 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
14172 are allowed in empty BBs as well.
14173 (canonicalize_loop_closed_ssa): Also look for other complex
14175 (scop_detection::get_sese): Include the loop-closed PHI block
14177 (scop_detection::merge_sese): Remove code adding extra blocks.
14178 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
14179 (build_scops): Assert the final returned scop is invalid.
14181 2017-09-20 Richard Biener <rguenther@suse.de>
14183 PR tree-optimization/82264
14184 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
14186 (vn_phi_lookup): Likewise.
14187 (vn_phi_insert): Likewise.
14189 2017-09-20 Jakub Jelinek <jakub@redhat.com>
14191 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
14192 that fits into uhwi or shwi, add DW_AT_const_value regardless
14193 of early_dwarf without going through RTL, using add_AT_unsigned
14196 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
14197 (DEBUG_LTO_ABBREV_SECTION): Likewise.
14198 (DEBUG_LTO_MACINFO_SECTION): Likewise.
14199 (DEBUG_MACRO_SECTION): Likewise.
14200 (DEBUG_LTO_MACRO_SECTION): Likewise.
14201 (DEBUG_STR_DWO_SECTION): Likewise.
14202 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
14203 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
14204 (DEBUG_LTO_DWO_LINE_SECTION): Define.
14205 (DEBUG_LTO_LINE_STR_SECTION): Define.
14206 (init_sections_and_labels): Initialize debug_line_str_section
14207 variable. Initialize debug_loc_section for -gdwarf-5 to
14208 DEBUG_LOCLISTS_SECTION. Formatting fixes.
14210 2017-09-20 Richard Biener <rguenther@suse.de>
14212 * graphite-sese-to-poly.c (extract_affine): Properly handle
14213 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
14215 2017-09-20 Richard Biener <rguenther@suse.de>
14217 PR tree-optimization/81373
14218 * graphite-scop-detection.c (build_cross_bb_scalars_def):
14219 Force SESE live-out defs to be handled even if they are
14222 2017-09-19 Jeff Law <law@redhat.com>
14224 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
14225 nothing for stack adjustments with REG_STACK_CHECK.
14226 * sched-deps.c (parse_add_or_inc): Reject insns with
14227 REG_STACK_CHECK from dependency breaking.
14228 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
14229 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
14230 * reg-notes.def (STACK_CHECK): New note.
14232 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
14233 (ix86_expand_prologue): Dump stack clash info as needed.
14234 Call ix86_adjust_stack_and_probe_stack_clash as needed.
14236 * function.c (dump_stack_clash_frame_info): New function.
14237 * function.h (dump_stack_clash_frame_info): Prototype.
14238 (enum stack_clash_probes): New enum.
14240 * config/alpha/alpha.c (alpha_expand_prologue): Also check
14241 flag_stack_clash_protection.
14242 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
14243 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
14244 (arm_frame_pointer_required): Likewise.
14245 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
14246 (ia64_expand_prologue): Likewise.
14247 * config/mips/mips.c (mips_expand_prologue): Likewise.
14248 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
14249 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14250 (sparc_flat_expand_prologue): Likewise.
14251 * config/spu/spu.c (spu_expand_prologue): Likewise.
14253 * explow.c: Include "params.h".
14254 (anti_adjust_stack_and_probe_stack_clash): New function.
14255 (get_stack_check_protect): Likewise.
14256 (compute_stack_clash_protection_loop_data): Likewise.
14257 (emit_stack_clash_protection_loop_start): Likewise.
14258 (emit_stack_clash_protection_loop_end): Likewise.
14259 (allocate_dynamic_stack_space): Use get_stack_check_protect.
14260 Use anti_adjust_stack_and_probe_stack_clash.
14261 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
14262 (emit_stack_clash_protection_loop_start): Likewise.
14263 (emit_stack_clash_protection_loop_end): Likewise.
14264 * rtl.h (get_stack_check_protect): Prototype.
14265 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
14266 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
14267 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
14269 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
14271 * doc/tm.texi: Rebuilt.
14272 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
14273 get_stack_check_protect.
14274 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
14275 * config/arm/arm.c (arm_expand_prologue): Likewise.
14276 (arm_frame_pointer_required): Likewise.
14277 * config/i386/i386.c (ix86_expand_prologue): Likewise.
14278 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
14279 * config/mips/mips.c (mips_expand_prologue): Likewise.
14280 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
14281 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
14282 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14283 (sparc_flat_expand_prologue): Likewise.
14285 * common.opt (-fstack-clash-protection): New option.
14286 * flag-types.h (enum stack_check_type): Note difference between
14287 -fstack-check= and -fstack-clash-protection.
14288 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
14289 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
14290 * toplev.c (process_options): Issue warnings/errors for cases
14291 not handled with -fstack-clash-protection.
14292 * doc/invoke.texi (-fstack-clash-protection): Document new option.
14293 (-fstack-check): Note additional problem with -fstack-check=generic.
14294 Note that -fstack-check is primarily for Ada and refer users
14295 to -fstack-clash-protection for stack-clash-protection.
14296 Document new params for stack clash protection.
14298 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
14300 * config/i386/i386.c (ix86_split_long_move): Do not handle
14301 address used for LEA in a special way.
14303 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
14305 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
14307 2017-09-19 Martin Sebor <msebor@redhat.com>
14310 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
14311 of incompatible types.
14313 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
14315 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14316 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
14317 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14318 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
14320 2017-09-19 Richard Biener <rguenther@suse.de>
14322 PR tree-optimization/82244
14323 * tree-vrp.c (remove_range_assertions): Do not propagate
14324 a constant to abnormals but replace the assert with a copy.
14326 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
14328 PR rtl-optimization/57878
14329 PR rtl-optimization/68988
14330 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
14331 avoidance test involving non_reload_pseudos. Move frequency test
14332 below the general fragmentation avoidance test.
14334 2017-09-19 Richard Biener <rguenther@suse.de>
14336 PR tree-optimization/69728
14337 * graphite-sese-to-poly.c (schedule_error): New global.
14338 (add_loop_schedule): Handle empty domain by failing the
14340 (build_original_schedule): Handle schedule_error.
14342 2017-09-19 Richard Biener <rguenther@suse.de>
14344 * graphite-scop-detection.c (scop_detection::can_represent_loop):
14345 Do not iterate to sibling loops but only to siblings of inner
14348 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
14351 * config/m68k/m68k.md (moveq feeding equality comparison): Check
14352 that the registers are different.
14354 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
14356 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
14357 to processor_model and "amdfam17h" to arch_names_table.
14358 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
14360 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14363 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
14365 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14366 Alan Hayward <alan.hayward@arm.com>
14367 David Sherwood <david.sherwood@arm.com>
14369 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
14371 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
14373 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
14374 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
14375 vect_schedule_slp_instance.
14376 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
14377 Update call to vect_slp_analyze_node_operations. Simplify return
14379 (vect_slp_analyze_bb_1): Update call accordingly.
14380 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
14381 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
14382 (vect_schedule_slp): Update call accordingly.
14384 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14385 Alan Hayward <alan.hayward@arm.com>
14386 David Sherwood <david.sherwood@arm.com>
14388 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
14389 with types that aren't in fact scalar.
14391 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14393 * tree-vect-slp.c (vect_record_max_nunits): New function,
14395 (vect_build_slp_tree_1): ...here.
14396 (vect_build_slp_tree_2): Call it for phis too.
14398 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14400 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
14401 to vect_get_vec_def_for_operand when getting the mask operand.
14403 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14404 Alan Hayward <alan.hayward@arm.com>
14405 David Sherwood <david.sherwood@arm.com>
14407 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
14410 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14411 Alan Hayward <alan.hayward@arm.com>
14412 David Sherwood <david.sherwood@arm.com>
14414 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
14415 calculation for vector booleans.
14417 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14418 Alan Hayward <alan.hayward@arm.com>
14419 David Sherwood <david.sherwood@arm.com>
14421 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
14423 (vect_transform_stmt): ...here.
14424 (vect_analyze_stmt): Use it instead of calling
14425 vectorizable_live_operation directly.
14427 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
14429 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
14430 non-SIMT targets in acc vector loops.
14432 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
14434 * configure.ac: Add arc and check if assembler supports gdwarf2.
14435 * configure: Regenerate.
14437 2017-09-18 Richard Biener <rguenther@suse.de>
14439 PR tree-optimization/82220
14440 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
14441 epilogue niters from the min_profitable_iters compute.
14443 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14446 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
14447 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
14448 (ix86_init_pic_reg): Revert 2017-09-01 changes.
14450 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
14453 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
14454 switching to a new text section.
14456 2017-09-18 Richard Biener <rguenther@suse.de>
14458 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
14460 (build_alias_set): Reject aliases with no access function.
14462 2017-09-18 Richard Biener <rguenther@suse.de>
14464 PR tree-optimization/79622
14465 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
14467 (build_cross_bb_scalars_use): Likewise.
14469 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
14471 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
14473 2017-09-18 Alan Modra <amodra@gmail.com>
14476 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
14477 stack_pointer_rtx for count 0. Update comments. Break up
14478 large rtl expression.
14480 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
14482 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
14483 Increase to 20 bytes.
14484 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
14485 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
14486 or avx version of the stub.
14488 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
14491 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
14492 compute the minimum stack alignment. Also update preferred stack
14493 boundary for leaf functions.
14495 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
14497 PR tree-optimization/82228
14498 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
14501 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14503 * common/config/nds32/nds32-common.c
14504 (nds32_option_optimization_table): Refine formatting.
14505 (nds32_option_optimization_table): Use -fsched-pressure and
14506 -fomit-frame-pointer for specific optimization level.
14508 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14510 * config/nds32/nds32.c: Refine formatting and comments.
14511 * config/nds32/nds32.h: Likewise.
14512 * config/nds32/nds32.md: Likewise.
14513 * config/nds32/nds32-cost.c: Likewise.
14514 * config/nds32/nds32-isr.c: Likewise.
14515 * config/nds32/nds32-md-auxiliary.c: Likewise.
14516 * config/nds32/nds32-multiple.md: Likewise.
14517 * config/nds32/nds32-predicates.c: Likewise.
14519 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
14520 Jakub Jelinek <jakub@redhat.com>
14522 Add support for -std=c++2a.
14523 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
14525 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
14527 2017-09-15 Steve Ellcey <sellcey@cavium.com>
14530 * doc/extend.texi (Common Function Attributes): Add
14531 references to ARM, AArch64, and S/390 specific attributes.
14532 (Function Specific Option Pragmas): Add AArch64 and S/390
14533 to list of back ends that support the target pragma.
14535 2017-09-15 Nathan Sidwell <nathan@acm.org>
14537 * doc/standards.texi: Fix C++17 description. Update URLs for
14540 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14542 * common.opt (Wcast-align=strict): New warning option.
14543 * doc/invoke.texi: Document -Wcast-align=strict.
14545 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
14547 * cgraph.h (cgraph_thunk_info): Add comments.
14548 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
14549 assert for VIRTUAL_* arguments stricter.
14551 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
14553 PR tree-optimization/71026
14554 * match.pd: Move RDIV patterns from fold-const.c
14555 * fold-const.c (distribute_real_division): Removed.
14556 (fold_binary_loc): Remove calls to distribute_real_divison.
14558 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14560 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
14561 c++1z and gnu++1z as deprecated. Change other references to
14562 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
14563 Change -Wc++1z-compat to -Wc++17-compat.
14564 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
14565 * dwarf2out.c (highest_c_language): Handle C++17.
14566 (gen_compile_unit_die): Likewise.
14568 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14570 PR rtl-optimization/82192
14571 * combine.c (make_extraction): Don't look through non-paradoxical
14572 SUBREGs or TRUNCATE if pos + len is or might be bigger than
14575 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14576 Alan Hayward <alan.hayward@arm.com>
14577 David Sherwood <david.sherwood@arm.com>
14579 * target.def (function_arg_offset): New hook.
14580 * targhooks.h (default_function_arg_offset): Declare.
14581 * targhooks.c (default_function_arg_offset): New function.
14582 * function.c (locate_and_pad_parm): Use
14583 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14584 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14585 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14586 * doc/tm.texi: Regenerate.
14587 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14588 * config/spu/spu.c (spu_function_arg_offset): New function.
14589 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14590 * system.h (FUNCTION_ARG_OFFSET): Poison.
14592 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14593 Alan Hayard <alan.hayward@arm.com>
14594 David Sherwood <david.sherwood@arm.com>
14596 * target.def (truly_noop_truncation): New hook.
14597 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14598 than TRULY_NOOP_TRUNCATION.
14599 * hooks.h (hook_bool_uint_uint_true): Declare.
14600 * hooks.c (hook_bool_uint_uint_true): New function.
14601 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14602 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14603 * doc/tm.texi: Regenerate.
14604 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14605 rather than TRULY_NOOP_TRUNCATION in comments.
14606 (simplify_comparison): Likewise.
14607 (record_truncated_value): Likewise.
14608 * expmed.c (extract_bit_field_1): Likewise.
14609 (extract_split_bit_field): Likewise.
14610 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14611 instead of TRULY_NOOP_TRUNCATION.
14612 * function.c (assign_parm_setup_block): Likewise.
14613 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14614 * rtlhooks.c: Include target.h.
14615 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14616 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14617 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14618 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14619 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14620 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14621 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14622 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14623 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14624 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14625 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14626 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14627 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14628 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14629 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14630 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14631 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14632 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14633 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14634 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14635 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14636 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14637 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14638 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14639 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14640 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14641 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14642 rather than TRULY_NOOP_TRUNCATION in comments.
14643 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14644 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14645 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14646 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14647 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14648 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14649 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14650 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14651 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14652 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14653 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14654 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14655 rather than TRULY_NOOP_TRUNCATION in comments.
14656 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14657 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14658 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14659 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14660 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14661 TRULY_NOOP_TRUNCATION condition.
14662 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14663 (TRULY_NOOP_TRUNCATION): Delete.
14664 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14665 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14666 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14667 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14668 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14669 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14670 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14671 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14672 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14673 rather than TRULY_NOOP_TRUNCATION in comments.
14674 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14675 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14676 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14677 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14678 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14679 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14681 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14684 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14685 (*cmp_ior): Likewise.
14686 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14687 (*ior_scc_scc_cmp): Likewise.
14688 (*and_scc_scc): Likewise.
14689 (*and_scc_scc_cmp): Likewise.
14691 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14692 Alan Hayard <alan.hayward@arm.com>
14693 David Sherwood <david.sherwood@arm.com>
14695 * target.def (can_change_mode_class): New hook.
14696 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14697 (hard_regno_nregs): Likewise.
14698 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14699 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14700 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14701 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14702 (LOAD_EXTEND_OP): Update accordingly.
14703 * doc/tm.texi: Regenerate.
14704 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14705 CANNOT_CHANGE_MODE_CLASS.
14706 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14707 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14708 * combine.c (simplify_set): Update accordingly.
14709 * emit-rtl.c (validate_subreg): Likewise.
14710 * recog.c (general_operand): Likewise.
14711 * regcprop.c (mode_change_ok): Likewise.
14712 * reload1.c (choose_reload_regs): Likewise.
14713 (inherit_piecemeal_p): Likewise.
14714 * rtlanal.c (simplify_subreg_regno): Likewise.
14715 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14716 instead of CANNOT_CHANGE_MODE_CLASS.
14717 (reload_cse_simplify_operands): Likewise.
14718 * reload.c (push_reload): Use targetm.can_change_mode_class
14719 instead of CANNOT_CHANGE_MODE_CLASS.
14720 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14721 REG_CANNOT_CHANGE_MODE_P.
14722 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14723 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14724 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14725 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14726 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14727 (arm_can_change_mode_class): New function.
14728 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14729 than CANNOT_CHANGE_MODE_CLASS in comments.
14730 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14731 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14732 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14733 (ix86_can_change_mode_class): ...this new function, inverting the
14734 sense of the return value.
14735 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14736 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14737 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14738 (ia64_can_change_mode_class): New function.
14739 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14740 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14741 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14742 (m32c_can_change_mode_class): ...this new function, inverting the
14743 sense of the return value.
14744 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14745 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14746 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14747 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14748 (mips_can_change_mode_class): ...this new function, inverting the
14749 sense of the return value.
14750 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14751 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14752 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14753 (msp430_can_change_mode_class): New function.
14754 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14755 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14756 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14757 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14758 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14759 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14760 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14761 (pa_cannot_change_mode_class): Replace with...
14762 (pa_can_change_mode_class): ...this new function, inverting the
14763 sense of the return value.
14764 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14765 than CANNOT_CHANGE_MODE_CLASS in comments.
14766 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14767 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14768 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14769 (pdp11_cannot_change_mode_class): Replace with...
14770 (pdp11_can_change_mode_class): ...this new function, inverting the
14771 sense of the return value.
14772 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14773 * config/powerpcspe/powerpcspe-protos.h
14774 (rs6000_cannot_change_mode_class_ptr): Delete.
14775 * config/powerpcspe/powerpcspe.c
14776 (rs6000_cannot_change_mode_class_ptr): Delete.
14777 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14778 (rs6000_option_override_internal): Assign to
14779 targetm.can_change_mode_class instead of
14780 rs6000_cannot_change_mode_class_ptr.
14781 (rs6000_cannot_change_mode_class): Replace with...
14782 (rs6000_can_change_mode_class): ...this new function, inverting the
14783 sense of the return value.
14784 (rs6000_debug_cannot_change_mode_class): Replace with...
14785 (rs6000_debug_can_change_mode_class): ...this new function.
14786 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14787 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14788 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14789 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14790 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14792 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14793 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14794 (rs6000_option_override_internal): Assign to
14795 targetm.can_change_mode_class instead of
14796 rs6000_cannot_change_mode_class_ptr.
14797 (rs6000_cannot_change_mode_class): Replace with...
14798 (rs6000_can_change_mode_class): ...this new function, inverting the
14799 sense of the return value.
14800 (rs6000_debug_cannot_change_mode_class): Replace with...
14801 (rs6000_debug_can_change_mode_class): ...this new function.
14802 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14803 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14804 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14805 (s390_can_change_mode_class): ...this new function, inverting the
14806 sense of the return value.
14807 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14808 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14809 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14810 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14811 (sh_cannot_change_mode_class): Replace with...
14812 (sh_can_change_mode_class): ...this new function, inverting the
14813 sense of the return value.
14814 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14815 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14816 (sparc_can_change_mode_class): New function.
14817 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14818 * config/spu/spu.c (spu_can_change_mode_class): New function.
14819 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14820 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14821 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14822 (visium_can_change_mode_class): New function.
14823 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
14825 2017-09-15 Richard Biener <rguenther@suse.de>
14827 PR tree-optimization/82217
14828 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
14829 but not undefined case.
14831 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14834 * postreload.c (reload_cse_simplify_operands): Skip
14835 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
14837 2017-09-15 Richard Biener <rguenther@suse.de>
14839 PR tree-optimization/68823
14840 * graphite-scop-detection.c (build_alias_set): If we have a
14841 possible dependence check whether we can handle them by just
14842 looking at the DRs DR_ACCESS_FNs.
14843 (build_scops): If build_alias_set fails, fail the SCOP.
14845 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
14847 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
14848 to support float128 built-in functions that require the ISA 3.0
14850 (BU_FLOAT128_3_HW): Likewise.
14851 (SQRTF128): Add support for the IEEE 128-bit square root and fma
14852 built-in functions.
14853 (FMAF128): Likewise.
14855 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14856 support for built-in functions that need the ISA 3.0 IEEE 128-bit
14857 floating point instructions.
14858 (rs6000_invalid_builtin): Likewise.
14859 (rs6000_builtin_mask_names): Likewise.
14860 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
14861 (RS6000_BTM_FLOAT128_HW): Likewise.
14862 (RS6000_BTM_COMMON): Likewise.
14863 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
14865 * doc/extend.texi (RS/6000 built-in functions): Document the
14866 IEEE 128-bit floating point square root and fused multiply-add
14867 built-in functions.
14869 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
14871 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
14872 reg (r2) isn't in the set of registers defined in the prologue.
14874 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14875 Alan Hayward <alan.hayward@arm.com>
14876 David Sherwood <david.sherwood@arm.com>
14878 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
14879 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
14880 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
14881 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
14882 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
14884 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
14885 max_vectorization_factor.
14886 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
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_min_worthwhile_factor): Delete.
14893 (vect_worthwhile_without_simd_p): Declare.
14894 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
14895 (vectorizable_reduction): Use it.
14896 * tree-vect-stmts.c (vectorizable_shift): Likewise.
14897 (vectorizable_operation): Likewise.
14899 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14900 Alan Hayward <alan.hayward@arm.com>
14901 David Sherwood <david.sherwood@arm.com>
14903 * tree-vectorizer.h (vect_get_num_copies): New function.
14904 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
14905 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14906 (vectorizable_induction): Likewise.
14907 (vectorizable_live_operation): Likewise.
14908 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14909 (vectorizable_bswap): Likewise.
14910 (vectorizable_call): Likewise.
14911 (vectorizable_conversion): Likewise.
14912 (vectorizable_assignment): Likewise.
14913 (vectorizable_shift): Likewise.
14914 (vectorizable_operation): Likewise.
14915 (vectorizable_store): Likewise.
14916 (vectorizable_load): Likewise.
14917 (vectorizable_condition): Likewise.
14918 (vectorizable_comparison): Likewise.
14919 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
14921 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14922 Alan Hayward <alan.hayward@arm.com>
14923 David Sherwood <david.sherwood@arm.com>
14925 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
14926 of vect_init_vector.
14928 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14929 Alan Hayward <alan.hayward@arm.com>
14930 David Sherwood <david.sherwood@arm.com>
14932 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
14933 an inline wrapper that provides a location.
14934 (gimple_build_vector): Likewise.
14935 * gimple-fold.c (gimple_build_vector_from_val): New function.
14936 (gimple_build_vector): Likewise.
14937 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
14938 functions to build the initial value. Always return a gimple value.
14939 (get_initial_defs_for_reduction): Likewise. Only compute
14941 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
14942 vect_init_vector on the results from get_initial_def(s)_for_reduction.
14943 (vectorizable_induction): Use gimple_build_vector rather than
14946 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14947 Alan Hayward <alan.hayward@arm.com>
14948 David Sherwood <david.sherwood@arm.com>
14950 * target.h (vec_perm_indices): New typedef.
14951 (auto_vec_perm_indices): Likewise.
14952 * optabs-query.h: Include target.h
14953 (can_vec_perm_p): Take a vec_perm_indices *.
14954 * optabs-query.c (can_vec_perm_p): Likewise.
14955 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
14956 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14957 * tree-vect-generic.c (lower_vec_perm): Likewise.
14958 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
14959 (vect_grouped_load_supported): Likewise.
14960 (vect_shift_permute_load_chain): Likewise.
14961 (vect_permute_store_chain): Use auto_vec_perm_indices.
14962 (vect_permute_load_chain): Likewise.
14963 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
14964 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
14965 Update uses of can_vec_perm_p.
14966 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
14967 mode with a number of elements. Take a vec_perm_indices *.
14968 (vect_create_epilog_for_reduction): Update accordingly.
14969 Use auto_vec_perm_indices.
14970 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
14971 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
14972 (vect_transform_slp_perm_load): Likewise.
14973 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
14974 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
14975 (vect_gen_perm_mask_checked): Likewise.
14976 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
14977 (vect_gen_perm_mask_checked): Likewise.
14978 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
14979 (vectorizable_store): Likewise.
14980 (vectorizable_load): Likewise.
14981 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
14982 (vectorizable_bswap): Likewise.
14984 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14985 Alan Hayward <alan.hayward@arm.com>
14986 David Sherwood <david.sherwood@arm.com>
14988 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
14989 * tree.c (build_vector): Likewise.
14990 (build_vector_from_ctor): Update accordingly.
14991 (build_vector_from_val): Likewise.
14992 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14993 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14994 * tree-vect-generic.c (add_rshift): Likewise.
14995 (expand_vector_divmod): Likewise.
14996 (optimize_vector_constructor): Likewise.
14997 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
14998 (vect_transform_slp_perm_load): Likewise.
14999 (vect_schedule_slp_instance): Likewise.
15000 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15001 (vectorizable_call): Likewise.
15002 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15003 * expmed.c (make_tree): Likewise.
15004 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15005 a vector passed to build_vector.
15006 (fold_convert_const): Likewise.
15007 (exact_inverse): Likewise.
15008 (fold_ternary_loc): Likewise.
15009 (fold_relational_const): Likewise.
15010 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15011 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15012 (const_unop): Likewise. Store the reduction accumulator in a
15013 variable rather than an array.
15014 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15015 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15016 the new vector, rather than constructing it after the input arrays.
15017 (native_interpret_vector): Use auto_vec<tree> when building
15018 a vector passed to build_vector. Add elements in order.
15019 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15020 auto_vec<tree> when building a vector passed to build_vector.
15021 (vect_create_epilog_for_reduction): Likewise.
15022 (vectorizable_induction): Likewise.
15023 (get_initial_def_for_reduction): Likewise. Fix indentation of
15025 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15027 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15028 passed to build_vector.
15030 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15031 Alan Hayward <alan.hayward@arm.com>
15032 David Sherwood <david.sherwood@arm.com>
15034 * tree-core.h (tree_base::u): Add an "nelts" field.
15035 (tree_vector): Use VECTOR_CST_NELTS as the length.
15036 * tree.c (tree_size): Likewise.
15037 (make_vector): Initialize VECTOR_CST_NELTS.
15038 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15039 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15040 TYPE_VECTOR_SUBPARTS.
15041 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15042 as the number of units, setting it from VECTOR_CST_NELTS.
15043 (const_vector_from_tree): Likewise.
15044 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15045 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15046 (fold_negate_expr_1): Likewise.
15047 (fold_convert_const): Likewise.
15048 (const_binop): Likewise. Differentiate the number of output and
15050 (const_unop): Likewise.
15051 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15052 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15053 in cases that did the opposite.
15055 2017-09-14 Richard Biener <rguenther@suse.de>
15057 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15060 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15062 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15064 2017-09-14 Jakub Jelinek <jakub@redhat.com>
15067 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
15068 if and where to split a bb, except for splitting before debug insn
15069 sequences followed by non-label real insn. Delete debug insns
15070 in between basic blocks.
15072 * combine.c (make_compound_operation_int): Formatting fixes.
15074 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
15075 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15076 * config/netbsd.h (LINK_EH_SPEC): Likewise.
15077 * config/sol2.h (LINK_EH_SPEC): Likewise.
15078 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15079 * config/s390/linux.h (LINK_SPEC): Likewise.
15080 * config/freebsd.h (LINK_EH_SPEC): Likewise.
15081 * config/openbsd.h (LINK_EH_SPEC): Likewise.
15082 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15083 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
15084 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
15085 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15086 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
15087 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
15089 2017-09-13 Jakub Jelinek <jakub@redhat.com>
15091 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
15093 (ENDFILE_LINUX_SPEC): Likewise.
15094 (LINK_EH_SPEC): Likewise.
15095 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
15096 (LINK_OS_LINUX_SPEC32): Likewise.
15097 (LINK_OS_LINUX_SPEC64): Likewise.
15098 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
15099 (LINK_OS_LINUX_SPEC): Likewise.
15101 2017-09-13 Martin Liska <mliska@suse.cz>
15103 PR middle-end/82154
15104 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
15105 CASE_HIGH is NULL_TREE.
15107 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15108 Alan Hayward <alan.hayward@arm.com>
15109 David Sherwood <david.sherwood@arm.com>
15111 * target.def (secondary_memory_needed): New hook.
15112 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
15113 instead of SECONDARY_MEMORY_NEEDED.
15114 (secondary_memory_needed_mode): Likewise.
15115 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
15116 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
15117 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
15118 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
15119 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
15120 * doc/tm.texi: Regenerate.
15121 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
15122 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
15123 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15124 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
15125 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
15126 * config/i386/i386.c (inline_secondary_memory_needed): Put the
15127 mode argument first and change the reg_class arguments to reg_class_t.
15128 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
15129 Make static. Update the call to inline_secondary_memory_needed.
15130 (ix86_register_move_cost): Update the call to
15131 inline_secondary_memory_needed.
15132 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15133 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
15135 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
15136 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15138 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
15139 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
15140 * config/mips/mips.c (mips_secondary_memory_needed): Make static
15141 and match hook interface. Add comment from mips.h.
15142 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15143 * config/mmix/mmix.md (truncdfsf2): Refer to
15144 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15146 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
15147 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
15148 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15149 (pa_secondary_memory_needed): New function.
15150 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
15151 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
15152 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15153 (pdp11_secondary_memory_needed): Make static and match hook interface.
15154 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
15155 * config/powerpcspe/powerpcspe-protos.h
15156 (rs6000_secondary_memory_needed_ptr): Delete.
15157 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
15159 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15160 (rs6000_option_override_internal): Assign to
15161 targetm.secondary_memory_needed rather than
15162 rs6000_secondary_memory_needed_ptr.
15163 (rs6000_secondary_memory_needed): Match hook interface.
15164 (rs6000_debug_secondary_memory_needed): Likewise.
15165 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
15166 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
15167 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
15168 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15169 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
15170 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
15172 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
15173 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15174 (rs6000_option_override_internal): Assign to
15175 targetm.secondary_memory_needed rather than
15176 rs6000_secondary_memory_needed_ptr.
15177 (rs6000_secondary_memory_needed): Match hook interface.
15178 (rs6000_debug_secondary_memory_needed): Likewise.
15179 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
15180 * config/s390/s390.c (s390_secondary_memory_needed): New function.
15181 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15182 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
15183 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15184 (sparc_secondary_memory_needed): New function.
15185 * lra-constraints.c (check_and_process_move): Refer to
15186 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15188 (curr_insn_transform): Likewise.
15189 (process_alt_operands): Use targetm.secondary_memory_needed
15190 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15191 (check_secondary_memory_needed_p): Likewise.
15192 (choose_split_class): Likewise.
15193 * reload.c: Unconditionally include code that was previously
15194 conditional on SECONDARY_MEMORY_NEEDED.
15195 (push_secondary_reload): Use targetm.secondary_memory_needed
15196 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15197 (push_reload): Likewise.
15198 * reload1.c: Unconditionally include code that was previously
15199 conditional on SECONDARY_MEMORY_NEEDED.
15200 (choose_reload_regs): Use targetm.secondary_memory_needed
15201 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15202 (gen_reload): Likewise.
15203 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
15205 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15206 Alan Hayward <alan.hayward@arm.com>
15207 David Sherwood <david.sherwood@arm.com>
15209 * target.def (secondary_memory_needed_mode): New hook:
15210 * targhooks.c (default_secondary_memory_needed_mode): Declare.
15211 * targhooks.h (default_secondary_memory_needed_mode): New function.
15212 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
15213 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
15214 * doc/tm.texi: Regenerate.
15215 * lra-constraints.c (check_and_process_move): Use
15216 targetm.secondary_memory_needed_mode instead of
15217 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
15218 (curr_insn_transform): Likewise.
15219 * reload.c (get_secondary_mem): Likewise.
15220 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15221 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
15223 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15224 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15225 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
15226 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15227 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
15229 * config/powerpcspe/powerpcspe-protos.h
15230 (rs6000_secondary_memory_needed_mode): Delete.
15231 * config/powerpcspe/powerpcspe.c
15232 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15233 (rs6000_secondary_memory_needed_mode): Make static.
15234 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15235 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
15237 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15239 (rs6000_secondary_memory_needed_mode): Make static.
15240 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15241 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
15242 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15243 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15244 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15246 (sparc_secondary_memory_needed_mode): New function.
15247 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
15249 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
15251 * config/aarch64/constraints.md (Umq): New constraint.
15252 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15254 (mov<mode>): Update condition.
15256 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15258 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
15259 when bitposition is the same.
15261 2017-09-13 Richard Biener <rguenther@suse.de>
15263 * dwarf2out.c (output_die_symbol): Remove.
15264 (output_die): Do not output a DIEs symbol.
15266 2017-09-13 Richard Biener <rguenther@suse.de>
15268 PR middle-end/82128
15269 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
15270 default-def to avoid breaking iterator update with the weird
15271 interaction with cgraph_update_edges_for_call_stmt_node.
15273 2017-09-13 Richard Biener <rguenther@suse.de>
15275 * tree-cfg.c (verify_gimple_assign_binary): Add verification
15276 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
15277 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
15278 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
15280 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
15282 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
15283 Disable pc relative literal load irrespective of
15284 TARGET_FIX_ERR_A53_84341 for default.
15286 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
15288 * config/sparc/sparc.c (output_return): Output the source location of
15289 the insn in the delay slot, if any.
15290 (output_sibcall): Likewise.
15292 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
15295 * common.opt (-static-pie): New alias.
15296 (shared): Negate static-pie.
15297 (-no-pie): Update help text.
15299 (static-pie): New option.
15300 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
15301 -static-pie support.
15302 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
15303 (LINK_EH_SPEC): Likewise.
15304 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15305 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15306 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15307 * gcc.c (LINK_COMMAND_SPEC): Likewise.
15308 (init_gcc_specs): Likewise.
15309 (init_spec): Likewise.
15310 (display_help): Update help message for -pie.
15311 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
15314 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
15316 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
15317 (movdi_aarch64): Likewise.
15318 (movti_aarch64): Likewise.
15320 2017-09-12 Simon Wright <simon@pushface.org>
15323 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
15324 calculation of the minor version, always output as 0.
15326 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15329 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15330 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
15331 on it early, rather than manual conversion late. For
15332 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
15333 instead of performing manual conversion.
15335 2017-09-12 Carl Love <cel@us.ibm.com>
15337 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
15338 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
15340 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15341 VMULOSW): Add definitions.
15342 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15343 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15344 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15345 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15346 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15348 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15350 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
15352 (movti_aarch64): Likewise.
15353 (movdf_aarch64): Likewise.
15354 (movtf_aarch64): Likewise.
15355 (load_pairdi): Likewise.
15356 (store_pairdi): Likewise.
15357 (load_pairdf): Likewise.
15358 (store_pairdf): Likewise.
15359 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
15360 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
15361 (ldr_got_small_<mode>): Likewise.
15362 (ldr_got_small_28k_<mode>): Likewise.
15363 (ldr_got_tiny): Likewise.
15364 * config/aarch64/iterators.md (ldst_sz): New.
15365 (ldpstp_sz): Likewise.
15366 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
15368 (thunderx_load): Split load_8 to load_16.
15369 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
15371 (thunderx2t99_storepair_basic): Split store_8 to store_16.
15372 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
15373 (xgene1_store_pair): Split store_8 to store_16.
15374 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
15375 (falkor_st_0_st_sd): Split store_8 to store_16.
15377 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15379 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
15380 and store1/2/3/4 to store_4/8/12/16.
15381 * config/aarch64/aarch64.md: Update for rename.
15382 * config/arm/arm.md: Likewise.: Likewise.
15383 * config/arm/arm.c: Likewise.
15384 * config/arm/thumb1.md: Likewise.
15385 * config/arm/thumb2.md: Likewise.
15386 * config/arm/vfp.md: Likewise.
15387 * config/arm/arm-generic.md: Likewise.
15388 * config/arm/arm1020e.md: Likewise.
15389 * config/arm/arm1026ejs.md: Likewise.
15390 * config/arm/arm1136jfs.md: Likewise.
15391 * config/arm/arm926ejs.md: Likewise.
15392 * config/arm/cortex-a15.md: Likewise.
15393 * config/arm/cortex-a17.md: Likewise.
15394 * config/arm/cortex-a5.md: Likewise.
15395 * config/arm/cortex-a53.md: Likewise.
15396 * config/arm/cortex-a57.md: Likewise.
15397 * config/arm/cortex-a7.md: Likewise.
15398 * config/arm/cortex-a8.md: Likewise.
15399 * config/arm/cortex-a9.md: Likewise.
15400 * config/arm/cortex-m4.md: Likewise.
15401 * config/arm/cortex-m7.md: Likewise.
15402 * config/arm/cortex-r4.md: Likewise.
15403 * config/arm/exynos-m1.md: Likewise.
15404 * config/arm/fa526.md: Likewise.
15405 * config/arm/fa606te.md: Likewise.
15406 * config/arm/fa626te.md: Likewise.
15407 * config/arm/fa726te.md: Likewise.
15408 * config/arm/fmp626.md: Likewise.
15409 * config/arm/iwmmxt.md: Likewise.
15410 * config/arm/ldmstm.md: Likewise.
15411 * config/arm/marvell-pj4.md: Likewise.
15412 * config/arm/xgene1.md: Likewise.
15413 * config/aarch64/thunderx.md: Likewise.
15414 * config/aarch64/thunderx2t99.md: Likewise.
15415 * config/aarch64/falkor.md: Likewise.
15417 2017-09-12 Martin Liska <mliska@suse.cz>
15419 * attribs.c (private_lookup_attribute): New function.
15420 * attribs.h (private_lookup_attribute): Declared here.
15421 (lookup_attribute): Called from this place.
15423 2017-09-12 Richard Biener <rguenther@suse.de>
15425 PR tree-optimization/82157
15426 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
15427 stmts with side-effects.
15429 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15430 Alan Hayward <alan.hayward@arm.com>
15431 David Sherwood <david.sherwood@arm.com>
15433 * target.def (hard_regno_nregs): New hook.
15434 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
15435 * targhooks.h (default_hard_regno_nregs): Declare.
15436 * targhooks.c (default_hard_regno_nregs): New function.
15437 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
15438 (TARGET_HARD_REGNO_NREGS): ...this hook.
15439 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
15440 (CLASS_MAX_NREGS): Likewise.
15441 * doc/tm.texi: Regenerate.
15442 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
15443 instead of HARD_REGNO_NREGS.
15444 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
15445 HARD_REGNO_NREGS in the comment.
15446 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
15447 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
15448 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
15449 Return an unsigned int.
15450 (TARGET_HARD_REGNO_NREGS): Redefine.
15451 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
15452 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
15453 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15454 (arc_hard_regno_nregs): New function.
15455 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
15456 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
15457 (arm_hard_regno_nregs): New function.
15458 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
15459 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
15460 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
15461 (TARGET_HARD_REGNO_NREGS): Redefine.
15462 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
15463 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
15464 (HARD_REGNO_NREGS): Delete.
15465 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
15466 (cr16_hard_regno_nregs): New function.
15467 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
15468 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
15469 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
15470 (cris_hard_regno_nregs): New function.
15471 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
15472 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
15473 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15474 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
15475 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
15476 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
15477 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
15478 (frv_hard_regno_nregs): Make static. Take and return an
15480 (frv_class_max_nregs): Remove outdated copy of documentation.
15481 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
15482 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
15483 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
15484 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
15485 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
15486 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
15487 (TARGET_HARD_REGNO_NREGS): Redefine.
15488 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
15489 (CLASS_MAX_NREGS): Update comment.
15490 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
15491 (ia64_hard_regno_nregs): New function.
15492 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
15493 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
15494 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
15495 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
15496 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
15498 (m32c_hard_regno_nregs): Likewise. Make static.
15499 (TARGET_HARD_REGNO_NREGS): Redefine.
15500 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
15501 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
15502 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
15503 (m68k_hard_regno_nregs): New function.
15504 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
15505 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
15506 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
15507 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
15508 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
15509 Take and return an unsigned int.
15510 (TARGET_HARD_REGNO_NREGS): Redefine.
15511 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
15512 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15513 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
15514 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
15515 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
15516 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
15517 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
15518 (msp430_hard_regno_nregs): Make static. Take and return an
15520 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
15521 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
15522 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
15523 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
15524 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
15525 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
15526 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
15527 (TARGET_HARD_REGNO_NREGS): Redefine.
15528 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
15529 (PA_HARD_REGNO_NREGS): ...this.
15530 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
15531 (PA_HARD_REGNO_NREGS): ...this.
15532 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15533 (pa_hard_regno_nregs): New function.
15534 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
15535 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
15536 (pdp11_hard_regno_nregs): New function.
15537 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
15538 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
15539 (rs6000_hard_regno_nregs_hook): New function.
15540 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
15541 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
15542 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
15543 Take and return an unsigned int. Move earlier in file.
15544 (TARGET_HARD_REGNO_NREGS): Redefine.
15545 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
15546 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
15547 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
15548 (rl78_hard_regno_nregs): Make static. Take and return an
15550 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
15551 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
15552 (rs6000_hard_regno_nregs_hook): New function.
15553 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
15554 * config/rx/rx.c (rx_hard_regno_nregs): New function.
15555 (TARGET_HARD_REGNO_NREGS): Redefine.
15556 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
15557 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
15558 instead of HARD_REGNO_NREGS.
15559 (s390_hard_regno_nregs): New function.
15560 (s390_hard_regno_mode_ok): Add comment from s390.h.
15561 (TARGET_HARD_REGNO_NREGS): Redefine.
15562 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
15563 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
15564 (sh_hard_regno_nregs): New function.
15565 (sh_pass_in_reg_p): Use it.
15566 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
15567 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15568 (sparc_hard_regno_nregs): New function.
15569 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
15570 * config/spu/spu.c (spu_hard_regno_nregs): New function.
15571 (spu_function_arg_advance): Use it, supplying a valid register number.
15572 (TARGET_HARD_REGNO_NREGS): Redefine.
15573 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
15574 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
15575 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
15576 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15577 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15578 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15579 (CLASS_MAX_NREGS): Remove copy of old documentation.
15580 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15581 (visium_hard_regno_nregs): New function.
15582 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15583 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15584 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15585 xtensa_hard_regno_nregs): New function.
15586 * system.h (HARD_REGNO_NREGS): Poison.
15588 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15590 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15591 hard_regno_nregs instead of HARD_REGNO_NREGS.
15592 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15593 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15594 (c6x_expand_epilogue): Likewise.
15595 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15596 (frv_read_iacc_argument): Likewise.
15597 * config/sh/sh.c: Include regs.h.
15598 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15599 (regs_used): Likewise.
15600 (output_stack_adjust): Likewise.
15601 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15602 * expmed.c: Include regs.h.
15603 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15604 * ree.c: Include regs.h.
15605 (combine_reaching_defs): Use hard_regno_nregs instead of
15607 (add_removable_extension): Likewise.
15609 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15611 * regs.h (hard_regno_nregs): Turn into a function.
15612 (end_hard_regno): Update accordingly.
15613 * caller-save.c (setup_save_areas): Likewise.
15614 (save_call_clobbered_regs): Likewise.
15615 (replace_reg_with_saved_mem): Likewise.
15616 (insert_restore): Likewise.
15617 (insert_save): Likewise.
15618 * combine.c (can_change_dest_mode): Likewise.
15619 (move_deaths): Likewise.
15620 (distribute_notes): Likewise.
15621 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15622 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15623 (rs6000_split_multireg_move): Likewise.
15624 (rs6000_register_move_cost): Likewise.
15625 (rs6000_memory_move_cost): Likewise.
15626 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15627 (rs6000_split_multireg_move): Likewise.
15628 (rs6000_register_move_cost): Likewise.
15629 (rs6000_memory_move_cost): Likewise.
15630 * cselib.c (cselib_reset_table): Likewise.
15631 (cselib_lookup_1): Likewise.
15632 * emit-rtl.c (set_mode_and_regno): Likewise.
15633 * function.c (aggregate_value_p): Likewise.
15634 * ira-color.c (setup_profitable_hard_regs): Likewise.
15635 (check_hard_reg_p): Likewise.
15636 (calculate_saved_nregs): Likewise.
15637 (assign_hard_reg): Likewise.
15638 (improve_allocation): Likewise.
15639 (calculate_spill_cost): Likewise.
15640 * ira-emit.c (modify_move_list): Likewise.
15641 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15642 (ira_hard_reg_in_set_p): Likewise.
15643 * ira.c (setup_reg_mode_hard_regset): Likewise.
15644 (clarify_prohibited_class_mode_regs): Likewise.
15645 (check_allocation): Likewise.
15646 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15647 (lra_setup_reg_renumber): Likewise.
15648 (setup_try_hard_regno_pseudos): Likewise.
15649 (spill_for): Likewise.
15650 (assign_hard_regno): Likewise.
15651 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15652 * lra-constraints.c (in_class_p): Likewise.
15653 (lra_constraint_offset): Likewise.
15654 (simplify_operand_subreg): Likewise.
15655 (lra_constraints): Likewise.
15656 (split_reg): Likewise.
15657 (split_if_necessary): Likewise.
15658 (invariant_p): Likewise.
15659 (inherit_in_ebb): Likewise.
15660 * lra-lives.c (process_bb_lives): Likewise.
15661 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15662 (get_hard_regs): Likewise.
15663 (do_remat): Likewise.
15664 * lra-spills.c (assign_spill_hard_regs): Likewise.
15665 * mode-switching.c (create_pre_exit): Likewise.
15666 * postreload.c (reload_combine_recognize_pattern): Likewise.
15667 * recog.c (peep2_find_free_register): Likewise.
15668 * regcprop.c (kill_value_regno): Likewise.
15669 (set_value_regno): Likewise.
15670 (copy_value): Likewise.
15671 (maybe_mode_change): Likewise.
15672 (find_oldest_value_reg): Likewise.
15673 (copyprop_hardreg_forward_1): Likewise.
15674 * regrename.c (check_new_reg_p): Likewise.
15675 (regrename_do_replace): Likewise.
15676 * reload.c (push_reload): Likewise.
15677 (combine_reloads): Likewise.
15678 (find_dummy_reload): Likewise.
15679 (operands_match_p): Likewise.
15680 (find_reloads): Likewise.
15681 (find_equiv_reg): Likewise.
15682 (reload_adjust_reg_for_mode): Likewise.
15683 * reload1.c (count_pseudo): Likewise.
15684 (count_spilled_pseudo): Likewise.
15685 (find_reg): Likewise.
15686 (clear_reload_reg_in_use): Likewise.
15687 (free_for_value_p): Likewise.
15688 (allocate_reload_reg): Likewise.
15689 (choose_reload_regs): Likewise.
15690 (reload_adjust_reg_for_temp): Likewise.
15691 (emit_reload_insns): Likewise.
15692 (delete_output_reload): Likewise.
15693 * rtlanal.c (subreg_get_info): Likewise.
15694 * sched-deps.c (sched_analyze_reg): Likewise.
15695 * sel-sched.c (init_regs_for_mode): Likewise.
15696 (mark_unavailable_hard_regs): Likewise.
15697 (choose_best_reg_1): Likewise.
15698 (verify_target_availability): Likewise.
15699 * valtrack.c (dead_debug_insert_temp): Likewise.
15700 * var-tracking.c (track_loc_p): Likewise.
15701 (emit_note_insn_var_location): Likewise.
15702 * varasm.c (make_decl_rtl): Likewise.
15703 * reginfo.c (choose_hard_reg_mode): Likewise.
15704 (init_reg_modes_target): Refer directly to
15705 this_target_regs->x_hard_regno_nregs.
15707 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15709 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15710 instead of hard_regno_nregs.
15712 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15714 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15715 end_hard_regno instead of hard_regno_nregs.
15716 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15717 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15718 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15719 * ira-color.c (improve_allocation): Likewise.
15720 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15721 * lra-lives.c (mark_regno_live): Likewise.
15722 (mark_regno_dead): Likewise.
15723 * lra-remat.c (operand_to_remat): Likewise.
15724 * lra.c (collect_non_operand_hard_regs): Likewise.
15725 * postreload.c (reload_combine_note_store): Likewise.
15726 (move2add_valid_value_p): Likewise.
15727 * reload.c (regno_clobbered_p): Likewise.
15729 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15731 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15733 * config/v850/v850.c (v850_reorg): Likewise.
15734 * reload.c (refers_to_regno_for_reload_p): Likewise.
15735 (find_equiv_reg): Likewise.
15736 * reload1.c (reload_reg_reaches_end_p): Likewise.
15738 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15740 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15742 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15743 * config/arm/arm.c (output_move_neon): Likewise.
15744 (arm_attr_length_move_neon): Likewise.
15745 (neon_split_vcombine): Likewise.
15746 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15747 (c6x_mark_reg_written): Likewise.
15748 (c6x_dwarf_register_span): Likewise.
15749 * config/i386/i386.c (ix86_save_reg): Likewise.
15750 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15751 (rws_access_reg): Likewise.
15752 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15753 * mode-switching.c (create_pre_exit): Likewise.
15754 * ree.c (combine_reaching_defs): Likewise.
15755 (add_removable_extension): Likewise.
15756 * regcprop.c (find_oldest_value_reg): Likewise.
15757 (copyprop_hardreg_forward_1): Likewise.
15758 * reload.c (reload_inner_reg_of_subreg): Likewise.
15759 (push_reload): Likewise.
15760 (combine_reloads): Likewise.
15761 (find_dummy_reload): Likewise.
15762 (reload_adjust_reg_for_mode): Likewise.
15763 * reload1.c (find_reload_regs): Likewise.
15764 (forget_old_reloads_1): Likewise.
15765 (reload_reg_free_for_value_p): Likewise.
15766 (reload_adjust_reg_for_temp): Likewise.
15767 (emit_reload_insns): Likewise.
15768 (delete_output_reload): Likewise.
15769 * sel-sched.c (choose_best_reg_1): Likewise.
15770 (choose_best_pseudo_reg): Likewise.
15772 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15773 Alan Hayward <alan.hayward@arm.com>
15774 David Sherwood <david.sherwood@arm.com>
15776 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15777 * target.def (slow_unaligned_access): New hook.
15778 * targhooks.h (default_slow_unaligned_access): Declare.
15779 * targhooks.c (default_slow_unaligned_access): New function.
15780 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15781 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15782 * doc/tm.texi: Regenerate.
15783 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15784 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15785 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15787 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15788 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15790 (rs6000_slow_unaligned_access): New function.
15791 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15792 (expand_block_compare): Likewise.
15793 (expand_strn_compare): Likewise.
15794 (rs6000_rtx_costs): Likewise.
15795 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15796 (riscv_slow_unaligned_access): Likewise.
15797 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15798 (riscv_slow_unaligned_access_p): ...this and make static.
15799 (riscv_option_override): Update accordingly.
15800 (riscv_slow_unaligned_access): New function.
15801 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15802 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15803 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15804 (rs6000_slow_unaligned_access): New function.
15805 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15806 (rs6000_rtx_costs): Likewise.
15807 * config/rs6000/rs6000-string.c (expand_block_compare)
15808 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15809 of SLOW_UNALIGNED_ACCESS.
15810 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15811 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15812 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15813 of SLOW_UNALIGNED_ACCESS.
15814 * expmed.c (simple_mem_bitfield_p): Likewise.
15815 * expr.c (alignment_for_piecewise_move): Likewise.
15816 (emit_group_load_1): Likewise.
15817 (emit_group_store): Likewise.
15818 (copy_blkmode_from_reg): Likewise.
15819 (emit_push_insn): Likewise.
15820 (expand_assignment): Likewise.
15821 (store_field): Likewise.
15822 (expand_expr_real_1): Likewise.
15823 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15824 * lra-constraints.c (simplify_operand_subreg): Likewise.
15825 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15826 * gimple-ssa-store-merging.c: Likewise in block comment at start
15828 * tree-ssa-strlen.c: Include target.h.
15829 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
15830 of SLOW_UNALIGNED_ACCESS.
15831 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
15833 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15835 PR rtl-optimization/82185
15836 * expmed.c (emit_store_flag_int): Only test tem if it has been
15839 2017-09-12 Richard Biener <rguenther@suse.de>
15841 PR middle-end/82149
15842 * match.pd ((FTYPE) N CMP CST): Fix typo.
15844 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
15846 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
15848 (mips_near_type_p): Add 'short_call' attribute as a synonym
15850 * doc/extend.texi (short_call): Document new function attribute.
15852 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15855 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
15856 assertion check that in the condition.
15857 (get_atomic_generic_size): Likewise. Before testing if parameter
15858 has pointer type, if it has array type, call for C++
15859 default_conversion to perform array-to-pointer conversion.
15861 2017-09-12 Richard Biener <rguenther@suse.de>
15863 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
15864 for operations we cannot scalarize.
15866 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
15868 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
15869 vectors heap vectors. Clean up comments.
15870 Make visited_bbs a reference.
15871 (profitable_jump_thread_path): Make GC
15872 vectors heap vectors. Clean up comments.
15874 (convert_and_register_jump_thread_path): Make GC vectors heap
15876 (check_subpath_and_update_thread_path): Same. Clean up comments.
15877 Make visited_bbs a reference.
15878 (handle_phi): Abstract common code to to
15879 register_jump_thread_path_if_profitable.
15880 Rename VAR_BB to DEF_BB.
15882 Make GC vectors heap vectors.
15883 Make visited_bbs a reference.
15884 (handle_assignment): Same.
15885 (register_jump_thread_path_if_profitable): New.
15886 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
15888 Make GC vectors heap vectors. Clean up comments.
15889 Make visited_bbs a reference.
15890 (find_jump_threads_backwards): Make visited_bbs live in the stack.
15891 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
15894 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
15897 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
15898 words of E_DImode object are reachable by xtensa_uimm8x4 access.
15900 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
15902 Revert r251800 and r251799.
15904 2017-09-11 Martin Jambor <mjambor@suse.cz>
15907 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
15908 arguments in advance.
15909 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
15910 use it to find predecessor edges.
15911 (naive_outof_ssa): Collect vector of predecessors.
15913 2017-09-08 Jason Merrill <jason@redhat.com>
15915 PR c++/70029 - ICE with ref-qualifier and -flto
15916 * langhooks.h (struct lang_hooks_for_types): Add
15917 copy_lang_qualifiers.
15918 * attribs.c (build_type_attribute_qual_variant): Use it.
15919 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
15921 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
15922 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
15924 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
15927 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
15928 (*mulsi3_sp64): New instruction.
15929 (mulsi3): New expander.
15931 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
15933 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
15935 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15937 * sancov.c: Include memmodel.h.
15939 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
15942 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
15945 2017-09-07 Carl Love <cel@us.ibm.com>
15947 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
15948 the sldi instruction.
15950 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
15952 * sancov.c: Include tm_p.h.
15954 2017-09-07 Jakub Jelinek <jakub@redhat.com>
15957 * output.h (switch_to_other_text_partition): New declaration.
15958 * varasm.c (switch_to_other_text_partition): New function.
15959 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
15960 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
15961 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
15962 to the other text partition before emitting LCL label and switch back
15963 after emitting the word after it.
15965 2017-09-07 Richard Biener <rguenther@suse.de>
15967 * passes.def (pass_split_crit_edges): Remove instance before PRE.
15968 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
15969 critical edges here, after loop init.
15970 (pass_data_pre): Remove PROP_no_crit_edges flags.
15971 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
15972 for valueization of call args to avoid leaking VN_TOP.
15973 (visit_use): Assert we do not visit default defs.
15974 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
15975 Use error_mark_node to more easily detect leaking VN_TOP.
15976 All default-defs are varying, not VN_TOP. Mark them visited.
15977 (run_scc_vn): Make code match comment.
15979 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15981 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
15982 OPTION_MASK_FLOAT128_KEYWORD.
15983 (POWERPC_MASKS): Likewise.
15984 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
15985 support for the -mfloat128-type option, and make -mfloat128
15986 default on PowerPC Linux systems. Define or undefine
15987 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
15988 Define __float128 to be __ieee128 if IEEE 128-bit support is
15989 enabled, or undefine it.
15990 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
15991 Delete defining __FLOAT128_TYPE__.
15992 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
15993 -mfloat128-type option and make -mfloat128 default on PowerPC
15995 (TARGET_FLOAT128_TYPE): Likewise.
15996 (-mfloat128-type): Likewise.
15997 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15998 Delete the -mfloat128-type option and make -mfloat128 default on
15999 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16000 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16001 128-bit floating point is enabled. Change tests from using
16002 -mfloat128-type to -mfloat128.
16003 (rs6000_mangle_type): Use the correct mangling for the __float128
16004 type even if normal long double is restricted to 64-bits.
16005 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16007 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16008 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16009 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16011 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16012 documentation for -mfloat128.
16014 2017-09-06 Olivier Hainque <hainque@adacore.com>
16016 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16018 2017-09-06 Wish Wu <wishwu007@gmail.com>
16019 Jakub Jelinek <jakub@redhat.com>
16021 * asan.c (initialize_sanitizer_builtins): Add
16022 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16023 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16024 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16025 BT_FN_VOID_UINT64_PTR variables.
16026 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16027 (BT_FN_VOID_UINT16_UINT16): Likewise.
16028 (BT_FN_VOID_UINT32_UINT32): Likewise.
16029 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16030 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16031 (BT_FN_VOID_UINT64_PTR): Likewise.
16032 * common.opt (flag_sanitize_coverage): New variable.
16033 (fsanitize-coverage=trace-pc): Remove.
16034 (fsanitize-coverage=): Add.
16035 * flag-types.h (enum sanitize_coverage_code): New enum.
16036 * fold-const.c (fold_range_test): Disable non-short-circuit
16037 optimization if flag_sanitize_coverage.
16038 (fold_truth_andor): Likewise.
16039 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16040 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16041 (coverage_sanitizer_opts): New array.
16042 (get_closest_sanitizer_option): Add OPTS argument, handle also
16043 OPT_fsanitize_coverage_.
16044 (parse_sanitizer_options): Adjusted to also handle
16045 OPT_fsanitize_coverage_.
16046 (common_handle_option): Add OPT_fsanitize_coverage_.
16047 * sancov.c (instrument_comparison, instrument_switch): New function.
16048 (sancov_pass): Add trace-cmp support.
16049 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16050 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16051 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16052 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16053 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16054 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16055 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16056 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16057 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16058 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16060 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16062 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16063 error. Only quit immediately if parsing is complete.
16064 (BEGIN): Initialize fatal_err and parse_done.
16065 (begin fpu, end fpu): Check number of arguments.
16066 (begin arch, end arch): Likewise.
16067 (begin cpu, end cpu): Likewise.
16068 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
16069 (optalias): Likewise.
16071 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16073 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
16074 * config/arm/arm-isa.h: Delete. Move definitions to ...
16075 * arm-cpus.in: ... here. Use new feature and fgroup values.
16076 * config/arm/arm.c (arm_option_override): Use lower case for feature
16078 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
16079 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
16080 * config/arm/parsecpu.awk (END): Add new command 'isa'.
16082 (print_isa_bits_for): New function.
16083 (gen_isa): New function.
16084 (gen_comm_data): Use print_isa_bits_for.
16085 (define feature): New keyword.
16086 (define fgroup): New keyword.
16087 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
16088 (arm-isa.h): Add rule to generate file.
16089 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
16090 case for feature bit names.
16092 2017-09-06 Richard Biener <rguenther@suse.de>
16094 * tree-ssa-pre.c (NECESSARY): Remove.
16095 (create_expression_by_pieces): Do not touch pass-local flags.
16096 (insert_into_preds_of_block): Likewise.
16097 (do_pre_regular_insertion): Likewise.
16098 (eliminate_insert): Likewise.
16099 (eliminate_dom_walker::before_dom_children): Likewise.
16100 (fini_eliminate): Do not look at inserted_exprs.
16101 (mark_operand_necessary): Remove.
16102 (remove_dead_inserted_code): Replace with simple work-list
16103 algorithm based on inserted_exprs and SSA uses.
16104 (pass_pre::execute): Re-order fini_eliminate and
16105 remove_dead_inserted_code.
16107 2017-09-06 Olivier Hainque <hainque@adacore.com>
16109 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
16110 for VxWorks 7. Adjust surrounding comments.
16112 2017-09-06 Richard Biener <rguenther@suse.de>
16114 * gimple-ssa-strength-reduction.c
16115 (find_candidates_dom_walker::before_dom_children): Also allow
16118 2017-09-06 Richard Biener <rguenther@suse.de>
16120 PR tree-optimization/82108
16121 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
16122 for gap in the non-permutation SLP case.
16124 2017-09-06 Martin Jambor <mjambor@suse.cz>
16126 PR tree-optimization/82078
16127 * tree-sra.c (sort_and_splice_var_accesses): Move call to
16128 add_access_to_work_queue...
16129 (build_accesses_from_assign): ...here.
16130 (propagate_all_subaccesses): Make sure racc is the group
16131 representative, if there is one.
16133 2017-09-06 Jakub Jelinek <jakub@redhat.com>
16135 PR middle-end/82095
16136 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
16139 2017-09-06 Richard Biener <rguenther@suse.de>
16141 * gimple-ssa-strength-reduction.c
16142 (find_candidates_dom_walker::before_doom_children): Use a
16143 type and not a mode check.
16145 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
16148 * config/arm/predicates.md (arm_general_adddi_operand): Create new
16150 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
16152 2017-09-05 Jeff Law <law@redhat.com>
16154 PR tree-optimization/64910
16155 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
16156 cases where we have 3 or more operands.
16158 2017-09-05 Jakub Jelinek <jakub@redhat.com>
16160 PR middle-end/81768
16161 * omp-low.c (lower_omp_for): Recompute tree invariant if
16162 gimple_omp_for_initial/final is ADDR_EXPR.
16164 PR middle-end/81768
16165 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
16166 into gimple val before gimplification fo the COND_EXPR.
16168 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
16170 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
16171 REGION_COPY argument.
16172 (thread_through_all_blocks): Remove unused argument to
16173 duplicate_thread_path.
16175 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16176 Alan Hayward <alan.hayward@arm.com>
16177 David Sherwood <david.sherwood@arm.com>
16179 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
16180 Take a scalar_mode rather than a machine_mode.
16181 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16182 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
16183 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16184 (aarch64_gen_adjusted_ldpstp): Likewise.
16185 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
16187 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16188 Alan Hayward <alan.hayward@arm.com>
16189 David Sherwood <david.sherwood@arm.com>
16191 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
16192 Take a scalar_int_mode instead of a machine_mode.
16193 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16194 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16195 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16196 (aarch64_simd_attr_length_rglist): Delete.
16197 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
16198 a scalar_int_mode instead of a machine_mode.
16199 (aarch64_add_offset): Likewise.
16200 (aarch64_internal_mov_immediate): Likewise
16201 (aarch64_add_constant_internal): Likewise.
16202 (aarch64_add_constant): Likewise.
16203 (aarch64_movw_imm): Likewise.
16204 (aarch64_rtx_arith_op_extract_p): Likewise.
16205 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16206 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16207 Remove assert that the mode isn't a vector.
16208 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16209 (aarch64_expand_mov_immediate): Update calls after above changes.
16210 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
16211 (aarch64_and_bitmask_imm): Check for scalar integer modes.
16212 (aarch64_move_imm): Likewise.
16213 (aarch64_can_const_movi_rtx_p): Likewise.
16214 (aarch64_strip_extend): Likewise.
16215 (aarch64_extr_rtx_p): Likewise.
16216 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
16217 a CONST_INT when the mode parameter is VOIDmode.
16218 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
16220 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16222 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
16223 * stor-layout.c (bitwise_mode_for_mode): Likewise.
16224 (bitwise_type_for_mode): Update accordingly.
16226 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16228 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
16229 * stor-layout.c (mode_for_size_tree): Likewise.
16230 (mode_for_array): Update accordingly.
16231 (layout_decl): Likewise.
16232 (compute_record_mode): Likewise. Only set the mode once.
16234 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16236 * target.def (get_mask_mode): Change return type to opt_mode.
16238 * doc/tm.texi: Regenerate.
16239 * targhooks.h (default_get_mask_mode): Return an opt_mode.
16240 * targhooks.c (default_get_mask_mode): Likewise.
16241 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16242 * optabs-query.c (can_vec_mask_load_store_p): Update use of
16243 targetm.get_mask_mode.
16244 * tree.c (build_truth_vector_type): Likewise.
16246 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16248 * machmode.h (mode_for_vector): Return an opt_mode.
16249 * stor-layout.c (mode_for_vector): Likewise.
16250 (mode_for_int_vector): Update accordingly.
16251 (layout_type): Likewise.
16252 * config/i386/i386.c (emit_memmov): Likewise.
16253 (ix86_expand_set_or_movmem): Likewise.
16254 (ix86_expand_vector_init): Likewise.
16255 (ix86_get_mask_mode): Likewise.
16256 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
16258 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
16259 * expmed.c (extract_bit_field_1): Likewise.
16260 * expr.c (expand_expr_real_2): Likewise.
16261 * optabs-query.c (can_vec_perm_p): Likewise.
16262 (can_vec_mask_load_store_p): Likewise.
16263 * optabs.c (expand_vec_perm): Likewise.
16264 * targhooks.c (default_get_mask_mode): Likewise.
16265 * tree-vect-stmts.c (vectorizable_store): Likewise.
16266 (vectorizable_load): Likewise.
16267 (get_vectype_for_scalar_type_and_size): Likewise.
16269 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16271 * machmode.h (mode_for_int_vector): New function.
16272 * stor-layout.c (mode_for_int_vector): Likewise.
16273 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
16274 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
16275 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16276 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16277 (s390_expand_vcond): Likewise.
16279 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16281 * machmode.h (opt_machine_mode): New type.
16282 (opt_mode<T>): Allow construction from anything that can be
16284 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
16285 (mode_for_size): Return an opt_machine_mode.
16286 * stor-layout.c (mode_for_size): Likewise.
16287 (mode_for_size_tree): Update call accordingly.
16288 (bitwise_mode_for_mode): Likewise.
16289 (make_fract_type): Likewise.
16290 (make_accum_type): Likewise.
16291 * caller-save.c (replace_reg_with_saved_mem): Update call
16293 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16294 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16295 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16296 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16297 * expmed.c (extract_bit_field_1): Likewise.
16298 * reload.c (get_secondary_mem): Likewise.
16299 * varasm.c (assemble_integer): Likewise.
16300 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
16303 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16305 * machmode.h (decimal_float_mode_for_size): New function.
16306 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
16307 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
16308 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
16309 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
16310 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
16311 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
16313 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16315 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
16316 (get_builtin_sync_mode): Likewise.
16317 (expand_ifn_atomic_compare_exchange): Likewise.
16318 (expand_builtin_atomic_clear): Likewise.
16319 (expand_builtin_atomic_test_and_set): Likewise.
16320 (fold_builtin_atomic_always_lock_free): Likewise.
16321 * calls.c (compute_argument_addresses): Likewise.
16322 (emit_library_call_value_1): Likewise.
16323 (store_one_arg): Likewise.
16324 * combine.c (combine_instructions): Likewise.
16325 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
16326 * config/arm/arm.c (arm_function_value): Likewise.
16327 (aapcs_allocate_return_reg): Likewise.
16328 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
16329 * config/i386/i386.c (construct_container): Likewise.
16330 (ix86_gimplify_va_arg): Likewise.
16331 (ix86_expand_sse_cmp): Likewise.
16332 (emit_memmov): Likewise.
16333 (emit_memset): Likewise.
16334 (expand_small_movmem_or_setmem): Likewise.
16335 (ix86_expand_pextr): Likewise.
16336 (ix86_expand_pinsr): Likewise.
16337 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
16338 * config/microblaze/microblaze.c (microblaze_block_move_straight):
16340 * config/mips/mips.c (mips_function_value_1) Likewise.
16341 (mips_block_move_straight): Likewise.
16342 (mips_expand_ins_as_unaligned_store): Likewise.
16343 * config/powerpcspe/powerpcspe.c
16344 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16345 (rs6000_darwin64_record_arg_flush): Likewise.
16346 * config/rs6000/rs6000.c
16347 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16348 (rs6000_darwin64_record_arg_flush): Likewise.
16349 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
16350 (sparc_function_value_1): Likewise.
16351 * config/spu/spu.c (adjust_operand): Likewise.
16352 (spu_emit_branch_or_set): Likewise.
16353 (arith_immediate_p): Likewise.
16354 * emit-rtl.c (gen_lowpart_common): Likewise.
16355 * expr.c (expand_expr_real_1): Likewise.
16356 * function.c (assign_parm_setup_block): Likewise.
16357 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
16358 * reload1.c (alter_reg): Likewise.
16359 * stor-layout.c (mode_for_vector): Likewise.
16360 (layout_type): Likewise.
16362 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16364 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
16365 (spu_convert_move): Likewise.
16366 * lower-subreg.c (resolve_simple_move): Likewise.
16368 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16371 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
16372 define_insn to a define_expand.
16373 (altivec_vsum2sws_direct): New define_insn.
16374 (altivec_vsumsws): Convert from a define_insn to a define_expand.
16376 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16378 * config/arm/arm.c (arm_option_params_internal): Improve setting of
16381 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
16386 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
16387 realign stack if stack alignment needed is less than incoming
16390 2017-09-05 Marek Polacek <polacek@redhat.com>
16393 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
16396 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16398 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
16400 2017-09-05 Richard Biener <rguenther@suse.de>
16402 PR tree-optimization/82084
16403 * fold-const.c (can_native_encode_string_p): Handle wide characters.
16405 2017-09-05 Richard Biener <rguenther@suse.de>
16407 PR tree-optimization/82102
16408 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
16410 2017-09-05 Martin Liska <mliska@suse.cz>
16412 PR tree-optimization/82032
16413 * tree-cfg.c (generate_range_test): New function.
16414 * tree-cfg.h (generate_range_test): Declared here.
16415 * tree-cfgcleanup.c (convert_single_case_switch): New function.
16416 (cleanup_control_expr_graph): Use it.
16417 * tree-switch-conversion.c (try_switch_expansion): Remove
16419 (emit_case_nodes): Use generate_range_test.
16421 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16424 * config/i386/i386.md (*<btsc><mode>_mask): Add
16425 TARGET_USE_BT to insn constraint.
16426 (*btr<mode>_mask): Ditto.
16428 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
16430 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
16431 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
16433 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
16436 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
16437 TARGET_NEON and TARGET_IWMMXT.
16438 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
16439 TARGET_NEON and TARGET_IWMMXT.
16440 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
16442 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16444 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
16445 (ix86_rewrite_tls_address): Ditto.
16446 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
16447 (ix86_rewrite_tls_address_1): Ditto.
16448 (ix86_rewrite_tls_address): Ditto.
16449 * config/i386/predicates.md (tls_address_pattern): New predicate.
16450 * config/i386/i386.md (TLS address splitter): New splitter.
16452 2017-09-04 Richard Biener <rguenther@suse.de>
16454 PR tree-optimization/82084
16455 * fold-const.h (can_native_encode_string_p): Declare.
16456 * fold-const.c (can_native_encode_string_p): Factor out from ...
16457 (native_encode_string): ... here.
16458 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
16459 vectorizing stores from constants we later cannot handle.
16461 2017-09-04 Marek Polacek <polacek@redhat.com>
16464 * doc/invoke.texi: Update -Wtautological-compare documentation.
16466 2017-09-04 Jeff Law <law@redhat.com>
16468 PR tree-optimization/64910
16469 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
16470 swap the first and last operand if the last is a constant.
16472 2017-09-04 Marek Polacek <polacek@redhat.com>
16475 * convert.c (do_narrow): When sanitizing signed integer overflows,
16476 bail out for signed types.
16477 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
16479 2017-09-04 Richard Biener <rguenther@suse.de>
16481 PR tree-optimization/82060
16482 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16483 Move devirtualization after stmt folding and before EH/AB/noreturn
16484 cleanup to get the stmt refs canonicalized. Use a bool instead
16485 of gimple_modified_p since that doesn't work for NOPs. Schedule
16486 NOPs generated by folding for removal.
16488 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16489 Alan Hayward <alan.hayward@arm.com>
16490 David Sherwood <david.sherwood@arm.com>
16492 * coretypes.h (pad_direction): New enum.
16493 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
16494 (FUNCTION_ARG_PADDING): Likewise.
16495 * target.def (function_arg_padding): New hook.
16496 * targhooks.h (default_function_arg_padding): Declare.
16497 * targhooks.c (default_function_arg_padding): New function.
16498 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
16499 (TARGET_FUNCTION_ARG_PADDING): ...this.
16500 * doc/tm.texi: Regenerate.
16501 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16502 instead of direction.
16503 (compute_argument_addresses): Likewise.
16504 (load_register_parameters): Likewise.
16505 (emit_library_call_value_1): Likewise.
16506 (store_one_arg): Use targetm.calls.function_arg_padding instead
16507 of FUNCTION_ARG_PADDING.
16508 (must_pass_in_stack_var_size_or_pad): Likewise.
16509 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
16510 (emit_group_store): Likewise.
16511 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
16512 instead of FUNCTION_ARG_PADDING.
16513 (emit_push_insn): Likewise, and propagate enum change throughout
16515 * function.h (direction): Delete.
16516 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
16518 * function.c (assign_parm_find_stack_rtl): Likewise.
16519 (assign_parm_setup_block_p): Likewise.
16520 (assign_parm_setup_block): Likewise.
16521 (gimplify_parameters): Likewise.
16522 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
16523 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
16525 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
16526 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16527 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
16528 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
16529 (aarch64_function_arg_padding): ...this new function.
16530 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
16531 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16532 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
16533 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16534 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
16535 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16536 (arm_pad_arg_upward): Replace with...
16537 (arm_function_arg_padding): ...this new function.
16538 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
16540 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
16541 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
16542 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16543 (ia64_hpux_function_arg_padding): Replace with...
16544 (ia64_function_arg_padding): ...this new function. Use pad_direction
16545 instead of direction. Check for TARGET_HPUX.
16546 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
16547 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16548 (iq2000_function_arg_padding): New function.
16549 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
16550 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
16551 (mips_function_arg_padding): ...this new function.
16552 (mips_pad_reg_upward): Update accordingly.
16553 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16554 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
16555 targetm.calls.function_arg_padding.
16556 (FUNCTION_ARG_PADDING): Delete.
16557 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16558 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
16559 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16560 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
16561 (nios2_block_reg_padding): Return pad_direction instead of direction.
16562 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
16563 instead of direction.
16564 (nios2_function_arg_padding): Likewise. Make static.
16565 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16566 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
16567 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
16568 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
16569 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16570 (pa_function_arg_padding): Make static. Return pad_direction instead
16572 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
16573 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16574 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
16575 instead of direction. Use targetm.calls.function_arg_padding.
16576 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16577 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16578 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16579 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16580 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16582 (function_arg_padding): Rename to...
16583 (rs6000_function_arg_padding): ...this. Make static. Return
16584 pad_direction instead of direction.
16585 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16586 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16587 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16588 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16589 instead of direction. Use targetm.calls.function_arg_padding.
16590 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16591 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16592 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16593 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16594 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16595 (function_arg_padding): Rename to...
16596 (rs6000_function_arg_padding): ...this. Make static. Return
16597 pad_direction instead of direction.
16598 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16599 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16600 * config/s390/s390.c (s390_function_arg_padding): New function.
16601 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16602 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16603 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16604 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16605 (function_arg_padding): Rename to...
16606 (sparc_function_arg_padding): ...this. Make static. Return
16607 pad_direction instead of direction.
16608 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16609 * config/spu/spu.c (spu_function_arg_padding): New function.
16610 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16611 * system.h (FUNCTION_ARG_PADDING): Poison.
16613 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16614 Alan Hayward <alan.hayward@arm.com>
16615 David Sherwood <david.sherwood@arm.com>
16617 * target.def (modes_tieable_p): New hook.
16618 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16619 (TARGET_MODES_TIEABLE_P): ...this.
16620 * doc/tm.texi.in: Regenerate.
16621 * hooks.h (hook_bool_mode_mode_true): Declare.
16622 * hooks.c (hook_bool_mode_mode_true): New function.
16623 * combine.c (subst): Use targetm.modes_tieable_p instead of
16625 * dse.c (find_shift_sequence): Likewise.
16626 * expmed.c (extract_low_bits): Likewise.
16627 * lower-subreg.c: Include target.h.
16628 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16630 * rtlanal.c (rtx_cost): Likewise.
16631 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16632 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16633 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16634 (TARGET_MODES_TIEABLE_P): Redefine.
16635 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16636 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16637 (TARGET_MODES_TIEABLE_P): Redefine.
16638 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16639 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16640 (arc_modes_tieable_p): New function.
16641 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16642 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16643 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16644 (arm_modes_tieable_p): Make static.
16645 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16646 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16647 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16648 (TARGET_MODES_TIEABLE_P): Redefine.
16649 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16650 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16651 (TARGET_MODES_TIEABLE_P): Redefine.
16652 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16653 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16654 (cr16_modes_tieable_p): New function.
16655 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16656 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16657 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16658 (TRULY_NOOP_TRUNCATION): Update comment.
16659 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16660 (TRULY_NOOP_TRUNCATION): Update comment.
16661 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16662 (frv_modes_tieable_p): New function.
16663 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16664 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16665 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16666 (TARGET_MODES_TIEABLE_P): Redefine.
16667 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16668 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16669 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16670 (TARGET_MODES_TIEABLE_P): Redefine.
16671 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16672 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16673 (ia64_modes_tieable_p): New function.
16674 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16675 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16676 (iq2000_modes_tieable_p): New function.
16677 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16678 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16679 (lm32_modes_tieable_p): New function.
16680 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16681 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16682 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16683 (TARGET_MODES_TIEABLE_P): Redefine.
16684 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16685 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16686 (m32r_modes_tieable_p): New function.
16687 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16688 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16689 (m68k_modes_tieable_p): New function.
16690 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16691 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16692 (mcore_modes_tieable_p): New function.
16693 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16694 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16696 (TARGET_MODES_TIEABLE_P): Redefine.
16697 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16698 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16699 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16700 (TARGET_MODES_TIEABLE_P): Redefine.
16701 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16702 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16703 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16704 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16705 (mn10300_modes_tieable_p): ...this and make static.
16706 (TARGET_MODES_TIEABLE_P): Redefine.
16707 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16708 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16709 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16710 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16711 (msp430_modes_tieable_p): Make static.
16712 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16713 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16714 (TARGET_MODES_TIEABLE_P): Redefine.
16715 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16716 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16717 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16718 (TARGET_MODES_TIEABLE_P): Redefine.
16719 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16720 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16721 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16722 (TARGET_MODES_TIEABLE_P): Redefine.
16723 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16724 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16725 (pdp11_modes_tieable_p): New function.
16726 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16727 * config/powerpcspe/powerpcspe.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/powerpcspe/powerpcspe.md: Update comment.
16731 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16732 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16733 (TARGET_MODES_TIEABLE_P): Redefine.
16734 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16735 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16736 (rl78_modes_tieable_p): New function.
16737 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16738 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16739 (rs6000_modes_tieable_p): New function.
16740 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16741 * config/rs6000/rs6000.md: Update comment.
16742 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16743 * config/rx/rx.c (rx_modes_tieable_p): New function.
16744 (TARGET_MODES_TIEABLE_P): Redefine.
16745 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16746 * config/s390/s390.c (s390_modes_tieable_p): New function.
16747 (TARGET_MODES_TIEABLE_P): Redefine.
16748 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16749 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16750 (sh_modes_tieable_p): New function.
16751 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16752 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16753 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16754 (sparc_modes_tieable_p): Make static.
16755 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16756 * config/spu/spu.c (spu_modes_tieable_p): New function.
16757 (TARGET_MODES_TIEABLE_P): Redefine.
16758 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16759 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16760 (TARGET_MODES_TIEABLE_P): Redefine.
16761 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16762 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16763 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16764 * config/v850/v850.c (v850_modes_tieable_p): New function.
16765 (TARGET_MODES_TIEABLE_P): Redefine.
16766 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16767 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16768 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16769 (visium_modes_tieable_p): New function.
16770 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16771 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16772 (xtensa_modes_tieable_p): New function.
16773 * system.h (MODES_TIEABLE_P): Poison.
16775 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16776 Alan Hayward <alan.hayward@arm.com>
16777 David Sherwood <david.sherwood@arm.com>
16779 * target.def (hard_regno_mode_ok): New hook.
16780 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16781 (TARGET_HARD_REGNO_MODE_OK): ...this.
16782 * doc/tm.texi.in: Regenerate.
16783 * hooks.h (hook_bool_uint_mode_true): Declare.
16784 * hooks.c (hook_bool_uint_mode_true): New function.
16785 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16786 HARD_REGNO_MODE_OK.
16787 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16788 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16789 instead of HARD_REGNO_MODE_OK.
16790 * caller-save.c: Include target.h.
16791 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16792 HARD_REGNO_MODE_OK.
16793 * combine.c (can_combine_p): Likewise.
16794 (combinable_i3pat): Likewise.
16795 (can_change_dest_mode): Likewise.
16796 * expr.c (init_expr_target): Likewise.
16797 (convert_move): Likewise.
16798 (convert_modes): Likewise.
16799 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16800 (setup_prohibited_mode_move_regs): Likewise.
16801 * ira.h (target_ira): Likewise.
16802 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16803 * lra-constraints.c (process_alt_operands): Likewise.
16804 (split_reg): Likewise.
16805 * recog.c (peep2_find_free_register): Likewise.
16806 * ree.c (combine_reaching_defs): Likewise.
16807 * regcprop.c (maybe_mode_change): Likewise.
16808 * reginfo.c (init_reg_sets_1): Likewise.
16809 (choose_hard_reg_mode): Likewise.
16810 (simplifiable_subregs): Likewise.
16811 * regrename.c (check_new_reg_p): Likewise.
16812 * reload.c (find_valid_class): Likewise.
16813 (find_valid_class_1): Likewise.
16814 (reload_inner_reg_of_subreg): Likewise.
16815 (push_reload): Likewise.
16816 (combine_reloads): Likewise.
16817 (find_dummy_reload): Likewise.
16818 (find_reloads): Likewise.
16819 * reload1.c (find_reg): Likewise.
16820 (set_reload_reg): Likewise.
16821 (allocate_reload_reg): Likewise.
16822 (choose_reload_regs): Likewise.
16823 (reload_adjust_reg_for_temp): Likewise.
16824 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
16825 (simplify_subreg_regno): Likewise.
16826 * sel-sched.c (init_regs_for_mode): Likewise.
16827 * varasm.c (make_decl_rtl): Likewise.
16828 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
16829 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
16830 HARD_REGNO_MODE_OK.
16831 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
16832 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
16833 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16834 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
16835 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
16836 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16837 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
16838 (arc_mode_class): Delete.
16839 (HARD_REGNO_MODE_OK): Delete.
16840 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16841 (arc_hard_regno_mode_ok): Rename old array to...
16842 (arc_hard_regno_mode_ok_modes): ...this.
16843 (arc_conditional_register_usage): Update accordingly.
16844 (arc_mode_class): Make static.
16845 (arc_hard_regno_mode_ok): New function.
16846 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
16847 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
16848 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16849 (arm_hard_regno_mode_ok): Make static.
16850 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
16851 HARD_REGNO_MODE_OK.
16852 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
16853 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
16854 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
16856 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16857 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
16858 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
16859 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
16860 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
16861 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16862 * config/bfin/predicates.md (valid_reg_operand): Use
16863 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
16864 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
16865 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
16866 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16867 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
16868 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
16869 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16870 (cr16_hard_regno_mode_ok): Make static and return a bool.
16871 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
16872 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16873 (cris_hard_regno_mode_ok): New function.
16874 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
16875 (epiphany_mode_class): Delete.
16876 (HARD_REGNO_MODE_OK): Delete.
16877 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
16878 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16879 (hard_regno_mode_ok): Rename to...
16880 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
16881 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
16882 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
16883 HARD_REGNO_MODE_OK.
16884 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
16885 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
16886 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16887 (frv_hard_regno_mode_ok): Make static and return a bool.
16888 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
16889 HARD_REGNO_MODE_OK.
16890 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
16891 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
16892 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
16893 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
16895 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16896 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
16897 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
16898 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
16900 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16901 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
16902 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16903 (ia64_hard_regno_mode_ok): New function.
16904 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
16905 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16906 (iq2000_hard_regno_mode_ok): New function.
16907 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
16908 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16909 (lm32_hard_regno_mode_ok): New function.
16910 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
16911 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
16912 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
16913 instead of HARD_REGNO_MODE_OK.
16914 (m32c_hard_regno_ok): Rename to...
16915 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
16916 (m32c_cannot_change_mode_class): Update accordingly.
16917 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16918 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
16919 (m32r_mode_class): Delete.
16920 (HARD_REGNO_MODE_OK): Delete.
16921 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16922 (m32r_hard_regno_mode_ok): Rename to...
16923 (m32r_hard_regno_modes): ...this.
16924 (m32r_mode_class): Make static.
16925 (m32r_hard_regno_mode_ok): New function.
16926 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
16927 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
16928 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16929 (m68k_hard_regno_mode_ok): Make static.
16930 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
16931 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16932 (mcore_hard_regno_mode_ok): New function.
16933 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
16934 (HARD_REGNO_MODE_OK): Delete.
16935 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
16937 (microblaze_hard_regno_mode_ok_p): ...this and make static.
16938 (microblaze_hard_regno_mode_ok): New function.
16939 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16940 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
16941 (mips_hard_regno_mode_ok): Delete.
16942 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
16943 (mips_hard_regno_mode_ok_p): ...this and make static.
16944 (mips_hard_regno_mode_ok_p): Rename to...
16945 (mips_hard_regno_mode_ok_uncached): ...this.
16946 (mips_hard_regno_mode_ok): New function.
16947 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
16948 of HARD_REGNO_MODE_OK.
16949 (mips_option_override): Update after above name changes.
16950 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16951 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
16952 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
16953 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
16954 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
16955 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16956 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
16957 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
16958 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
16959 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16960 (msp430_hard_regno_mode_ok): Make static and return a bool.
16961 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
16962 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
16963 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
16965 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16966 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
16967 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
16968 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
16969 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
16970 (PA_HARD_REGNO_MODE_OK): ...this
16971 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
16972 (PA_HARD_REGNO_MODE_OK): ...this.
16973 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16974 (pa_hard_regno_mode_ok): New function.
16975 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
16976 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16977 (pdp11_hard_regno_mode_ok): New function.
16978 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
16979 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
16981 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
16983 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16984 (rs6000_hard_regno_mode_ok): Rename to...
16985 (rs6000_hard_regno_mode_ok_uncached): ...this.
16986 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16987 (rs6000_hard_regno_mode_ok): New function.
16988 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
16989 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
16990 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
16991 (riscv_hard_regno_mode_ok): ...this and make static.
16992 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16993 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
16994 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
16995 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16996 (rl78_hard_regno_mode_ok): Make static and return bool.
16997 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
16998 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17000 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17001 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17002 (rs6000_hard_regno_mode_ok): Rename to...
17003 (rs6000_hard_regno_mode_ok_uncached): ...this.
17004 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17005 (rs6000_hard_regno_mode_ok): New function.
17006 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17007 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17008 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17009 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17010 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17011 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17012 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17013 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17014 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17015 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17016 (sh_hard_regno_mode_ok): Make static.
17017 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17018 instead of HARD_REGNO_MODE_OK.
17019 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17020 (sparc_mode_class): Delete.
17021 (HARD_REGNO_MODE_OK): Delete.
17022 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17023 (hard_regno_mode_classes): Make static.
17024 (sparc_mode_class): Likewise.
17025 (sparc_hard_regno_mode_ok): New function.
17026 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17027 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17028 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17030 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17031 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17032 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17033 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17034 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17035 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17036 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17037 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17038 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17039 (visium_hard_regno_mode_ok): New function.
17040 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17041 instead of HARD_REGNO_MODE_OK.
17042 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17043 (HARD_REGNO_MODE_OK): Delete.
17044 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17045 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17046 (xtensa_option_override): Update accordingly.
17047 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17048 (xtensa_hard_regno_mode_ok): New function.
17049 * system.h (HARD_REGNO_MODE_OK): Poison.
17051 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17052 Alan Hayward <alan.hayward@arm.com>
17053 David Sherwood <david.sherwood@arm.com>
17055 * target.def (hard_regno_call_part_clobbered): New hook.
17056 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17057 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17058 * doc/tm.texi: Regenerate.
17059 * hooks.h (hook_bool_uint_mode_false): Declare.
17060 * hooks.c (hook_bool_uint_mode_false): New function.
17061 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17062 * cselib.c (cselib_process_insn): Use
17063 targetm.hard_regno_call_part_clobbered instead of
17064 HARD_REGNO_CALL_PART_CLOBBERED.
17065 * ira-conflicts.c (ira_build_conflicts): Likewise.
17066 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17067 * lra-constraints.c (need_for_call_save_p): Likewise.
17068 * lra-lives.c: Include target.h.
17069 (check_pseudos_live_through_calls): Use
17070 targetm.hard_regno_call_part_clobbered instead of
17071 HARD_REGNO_CALL_PART_CLOBBERED.
17072 * regcprop.c: Include target.h.
17073 (copyprop_hardreg_forward_1): Use
17074 targetm.hard_regno_call_part_clobbered instead of
17075 HARD_REGNO_CALL_PART_CLOBBERED.
17076 * reginfo.c (choose_hard_reg_mode): Likewise.
17077 * regrename.c (check_new_reg_p): Likewise.
17078 * reload.c (find_equiv_reg): Likewise.
17079 * reload1.c (emit_reload_insns): Likewise.
17080 * sched-deps.c (deps_analyze_insn): Likewise.
17081 * sel-sched.c (init_regs_for_mode): Likewise.
17082 (mark_unavailable_hard_regs): Likewise.
17083 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
17084 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17085 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
17087 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17088 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17089 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
17091 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
17093 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17094 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17095 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
17097 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17098 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17099 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
17101 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17102 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
17104 * config/powerpcspe/powerpcspe.c
17105 (rs6000_hard_regno_call_part_clobbered): New function.
17106 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17107 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17108 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
17110 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17111 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17112 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
17114 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17115 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17116 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
17118 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17119 Alan Hayward <alan.hayward@arm.com>
17120 David Sherwood <david.sherwood@arm.com>
17122 * rtl.h (subreg_memory_offset): Declare.
17123 * emit-rtl.c (subreg_memory_offset): New function.
17124 * expmed.c (store_bit_field_1): Use it.
17125 * expr.c (undefined_operand_subword_p): Likewise.
17126 * simplify-rtx.c (simplify_subreg): Likewise.
17128 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
17130 PR rtl-optimization/57448
17133 * optabs.c (expand_atomic_load): Place compiler memory barriers if
17134 using atomic_load pattern.
17135 (expand_atomic_store): Likewise.
17137 2017-09-04 Jakub Jelinek <jakub@redhat.com>
17140 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
17141 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
17142 handling. Use replace_call_with_value with NULL instead of
17143 gsi_replace, unlink_stmt_vdef and release_defs.
17145 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
17148 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
17150 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17153 * rtl.h (emit_library_call_value_1): Declare.
17154 (emit_library_call): Replace declaration with a series of overloads.
17155 Remove the parameter count argument.
17156 (emit_library_call_value): Likewise.
17157 * calls.c (emit_library_call_value_1): Make global. Replace varargs
17158 with an "rtx_mode_t *".
17159 (emit_library_call_value): Delete.
17160 (emit_library_call): Likewise.
17161 * asan.c (asan_emit_stack_protection): Update calls accordingly.
17162 (asan_emit_allocas_unpoison): Likewise.
17163 * builtins.c (expand_builtin_powi): Likewise.
17164 (expand_asan_emit_allocas_unpoison): Likewise.
17165 * cfgexpand.c (expand_main_function): Likewise.
17166 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
17167 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
17168 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
17169 * config/arm/arm.c (arm_trampoline_init): Likewise.
17170 (arm_call_tls_get_addr): Likewise.
17171 (arm_expand_divmod_libfunc): Likewise.
17172 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
17173 (smulsi3_highpart): Likewise.
17174 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
17175 (c6x_expand_compare): Likewise.
17176 (c6x_expand_movmem): Likewise.
17177 * config/frv/frv.c (frv_trampoline_init): Likewise.
17178 * config/i386/i386.c (ix86_trampoline_init): Likewise.
17179 (ix86_expand_divmod_libfunc): Likewise.
17180 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
17181 (ia64_expand_compare): Likewise.
17182 (ia64_profile_hook): Likewise.
17183 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
17184 (nonlocal_goto): Likewise.
17185 (restore_stack_nonlocal): Likewise.
17186 * config/m32r/m32r.c (block_move_call): Likewise.
17187 (m32r_trampoline_init): Likewise.
17188 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
17189 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
17190 (m68k_call_m68k_read_tp): Likewise.
17191 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
17192 (microblaze_expand_divide): Likewise.
17193 * config/mips/mips.h (mips_args): Likewise.
17194 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
17195 (MIPS_ICACHE_SYNC): Likewise.
17196 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
17197 (nios2_trampoline_init): Likewise.
17198 * config/pa/pa.c (hppa_tls_call): Likewise.
17199 (pa_trampoline_init): Likewise.
17200 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
17201 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
17202 (expand_strn_compare): Likewise.
17203 (rs6000_generate_compare): Likewise.
17204 (rs6000_expand_float128_convert): Likewise.
17205 (output_profile_hook): Likewise.
17206 (rs6000_trampoline_init): Likewise.
17207 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
17208 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
17209 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
17210 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
17211 (rs6000_generate_compare): Likewise.
17212 (rs6000_expand_float128_convert): Likewise.
17213 (output_profile_hook): Likewise.
17214 (rs6000_trampoline_init): Likewise.
17215 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
17216 * config/sh/sh.c (sh_trampoline_init): Likewise.
17217 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
17218 (sparc_emit_float_lib_cmp): Likewise.
17219 (sparc32_initialize_trampoline): Likewise.
17220 (sparc64_initialize_trampoline): Likewise.
17221 (sparc_profile_hook): Likewise.
17222 * config/spu/spu.c (ea_load_store): Likewise.
17223 * config/spu/spu.md (floatunssidf2): Likewise.
17224 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
17225 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
17226 * config/visium/visium.c (expand_block_move_4): Likewise.
17227 (expand_block_move_2): Likewise.
17228 (expand_block_move_1): Likewise.
17229 (expand_block_set_4): Likewise.
17230 (expand_block_set_2): Likewise.
17231 (expand_block_set_1): Likewise.
17232 (visium_trampoline_init): Likewise.
17233 (visium_profile_hook): Likewise.
17234 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
17235 (xtensa_setup_frame_addresses): Likewise.
17236 (xtensa_trampoline_init): Likewise.
17237 * except.c (sjlj_emit_function_enter): Likewise.
17238 (sjlj_emit_function_exit): Likewise.
17239 * explow.c (allocate_dynamic_stack_space): Likewise.
17240 (probe_stack_range): Likewise.
17241 * expr.c (convert_mode_scalar): Likewise.
17242 * optabs.c (expand_binop): Likewise.
17243 (expand_twoval_binop_libfunc): Likewise.
17244 (expand_unop): Likewise.
17245 (prepare_cmp_insn): Likewise.
17246 (prepare_float_lib_cmp): Likewise.
17247 (expand_float): Likewise.
17248 (expand_fix): Likewise.
17249 (expand_fixed_convert): Likewise.
17250 (maybe_emit_sync_lock_test_and_set): Likewise.
17251 (expand_atomic_compare_and_swap): Likewise.
17252 (expand_mem_thread_fence): Likewise.
17253 (expand_atomic_fetch_op): Likewise.
17255 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
17257 * doc/generic.texi (OpenACC): Adjust URL.
17258 * doc/invoke.texi (C Dialect Options): Ditto.
17260 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
17262 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
17263 predicate for operand 1. Add (m,<S>) constraint.
17264 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
17265 Prevent memory operand 1 with register operand 2.
17267 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
17269 PR rtl-optimization/82024
17270 * combine.c (try_combine): If the combination result is a PARALLEL,
17271 and we only need to retain the SET in there that would be placed
17272 at I2, check that we can place that at I3 instead, before doing so.
17274 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17277 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
17279 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
17280 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
17283 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
17284 Jeff Law <law@redhat.com>
17286 * varasm.c (bss_initializer_p): Do not put constants into .bss
17287 (categorize_decl_for_section): Handle bss_initializer_p returning
17288 false when DECL_INITIAL is NULL.
17290 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17293 * config/s390/s390.c (s390_can_inline_p): New function.
17295 2017-09-01 Jeff Law <law@redhat.com>
17297 PR tree-optimization/82052
17298 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17299 Always initialize the returned slot after a hash table miss
17300 when INSERT is true.
17302 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
17304 * config/s390/s390.md (mem_signal_fence): Remove.
17305 * doc/md.texi (mem_signal_fence): Remove.
17306 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
17308 * target-insns.def (mem_signal_fence): Remove.
17310 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17313 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
17316 * asan.c (create_odr_indicator): Strip name encoding from assembler
17317 name before appending it after __odr_asan_.
17319 2017-09-01 Martin Liska <mliska@suse.cz>
17321 PR tree-optimization/82059
17322 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
17323 frequency only when an edge is redirected.
17325 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17327 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
17328 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
17329 (arc_conditional_register_usage): Remove ARC600 lp_count
17331 (arc_file_start): Emit Tag_ARC_CPU_variation.
17332 (arc_can_use_doloop_p): New conditions to use ZOLs.
17333 (hwloop_fail): New function.
17334 (hwloop_optimize): Likewise.
17335 (hwloop_pattern_reg): Likewise.
17336 (arc_doloop_hooks): New struct, to be used with reorg_loops.
17337 (arc_reorg_loops): New function, calls reorg_loops.
17338 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
17339 (arc600_corereg_hazard): Remove ZOL checking, case handled by
17341 (arc_loop_hazard): Remove function, functionality moved into
17343 (arc_hazard): Remove arc_loop_hazard call.
17344 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
17345 into hwloop_optimize.
17346 (arc_label_align): Remove ZOL handling.
17347 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
17348 * config/arc/arc.md (doloop_begin): Remove pattern.
17349 (doloop_begin_i): Likewise.
17350 (doloop_end_i): Likewise.
17351 (doloop_fallback): Likewise.
17352 (doloop_fallback_m): Likewise.
17353 (doloop_end): Reimplement expand.
17354 (arc_lp): New pattern for LP instruction.
17355 (loop_end): New pattern.
17356 (loop_fail): Likewise.
17357 (decrement_and_branch_until_zero): Likewise.
17358 * config/arc/arc.opt (mlpc-width): New option.
17359 * doc/invoke.texi (mlpc-width): Document option.
17361 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17363 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
17364 (arc_ccfsm_advance): Fix checking for delay slots.
17365 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
17367 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17369 * config/arc/arc.md (movqi_insn): Add stores to save constant long
17371 (movhi_insn): Update store instruction constraint which are saving
17372 6-bit short immediates.
17373 (movsi_insn): Consider also short scaled load operations.
17374 (zero_extendhisi2_i): Use Usd constraint instead of T.
17375 (extendhisi2_i): Add q constraint.
17376 (arc_clzsi2): Add type and length attributes.
17377 (arc_ctzsi2): Likewise.
17378 * config/arc/constraints.md (Usc): Update constraint, the
17379 assembler can parse two relocations for a single instruction.
17381 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17383 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
17384 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
17386 2017-08-31 Olivier Hainque <hainque@adacore.com>
17388 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
17389 match as powerpc-wrs-vxworks*.
17391 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
17393 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
17394 register constraint for by-element operand.
17395 (aarch64_mls_elt_merge<mode>): Likewise.
17397 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17399 * config/arc/arc.c (arc_can_follow_jump): Check for short
17402 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17404 * config.gcc: Use g.opt for arc.
17405 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
17406 functionality moved to ...
17407 (legitimate_scaled_address_p): New function, ...here.
17408 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
17409 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
17410 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
17412 (arc_override_options): Handle G option.
17413 (arc_output_pic_addr_const): Correct function definition.
17414 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
17415 (arc_decl_anon_ns_mem_p): Delete.
17416 (arc_in_small_data_p): Overhaul this function to take into
17417 consideration the value given via G option.
17418 (arc_rewrite_small_data_1): Renamed and corrected old
17419 arc_rewrite_small_data function.
17420 (arc_rewrite_small_data): New function.
17421 (small_data_pattern): Don't use pic_offset_table_rtx.
17422 * config/arc/arc.h (CC1_SPEC): Recognize G option.
17423 * config/arc/simdext.md (movmisalignv2hi): Use
17424 prepare_move_operands function.
17426 (movmisalign*): Likewise.
17427 * doc/invoke.texi (ARC options): Document -G option.
17429 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17431 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
17433 * config/arc/arc.c (arc_print_operand): Output scalled address for
17434 sdata whenever is possible.
17435 (arc_in_small_data_p): Allow sdata for 64bit datum when double
17436 load/stores are available.
17437 (compact_sda_memory_operand): Check for the alignment required by
17438 code density instructions.
17439 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
17441 * config/arc/constraints.md (Usd): Update constraint.
17442 (Us0): New constraint.
17443 (Usc): Update constraint.
17445 2017-08-31 Richard Biener <rguenther@suse.de>
17447 PR middle-end/82054
17448 * dwarf2out.c (dwarf2out_early_global_decl): Process each
17449 function only once.
17451 2017-08-31 Tamar Christina <tamar.christina@arm.com>
17453 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
17454 Resize type_signature.
17456 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17457 Alan Hayward <alan.hayward@arm.com>
17458 David Sherwood <david.sherwood@arm.com>
17460 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
17461 subregs whose inner modes can be stored in GPRs.
17462 (aarch64_classify_index): Likewise.
17464 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17465 Alan Hayward <alan.hayward@arm.com>
17466 David Sherwood <david.sherwood@arm.com>
17468 * config/aarch64/iterators.md (V_cmp_result): Rename to...
17469 (V_INT_EQUIV): ...this.
17470 (v_cmp_result): Rename to...
17471 (v_int_equiv): ...this.
17472 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
17473 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
17474 (copysign<mode>3): Likewise.
17475 (aarch64_simd_bsl<mode>_internal): Likewise.
17476 (aarch64_simd_bsl<mode>): Likewise.
17477 (vec_cmp<mode><mode>): Likewise.
17478 (vcond<mode><mode>): Likewise.
17479 (vcond<v_cmp_mixed><mode>): Likewise.
17480 (vcondu<mode><v_cmp_mixed>): Likewise.
17481 (aarch64_cm<optab><mode>): Likewise.
17482 (aarch64_cmtst<mode>): Likewise.
17483 (aarch64_fac<optab><mode>): Likewise.
17484 (vec_perm_const<mode>): Likewise.
17485 (vcond_mask_<mode><v_cmp_result>): Rename to...
17486 (vcond_mask_<mode><v_int_equiv>): ...this.
17487 (vec_cmp<mode><v_cmp_result>): Rename to...
17488 (vec_cmp<mode><v_int_equiv>): ...this.
17490 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17491 Alan Hayward <alan.hayward@arm.com>
17492 David Sherwood <david.sherwood@arm.com>
17494 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
17496 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
17497 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
17498 (UNSPEC_LD4_DREG): New unspecs.
17499 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
17500 (aarch64_ld2<mode>_dreg_be): Replace with...
17501 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
17503 (aarch64_ld3<mode>_dreg_le)
17504 (aarch64_ld3<mode>_dreg_be): Replace with...
17505 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
17507 (aarch64_ld4<mode>_dreg_le)
17508 (aarch64_ld4<mode>_dreg_be): Replace with...
17509 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
17512 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17514 PR tree-optimization/81987
17515 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
17516 insert an initializer in a location not dominated by the stride
17519 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17521 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
17522 on the entire header of the finally block in the fallthru case.
17524 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17526 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
17528 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
17530 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
17531 rs6000_emit_move_from_cr and call renamed function.
17532 (rs6000_emit_prologue): Call renamed functions.
17533 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
17534 movesi_from_cr, remove volatile CRs.
17536 2017-08-30 Jon Beniston <jon@beniston.com>
17537 Richard Biener <rguenther@suse.de>
17539 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
17540 of VECTOR_MODE_P check.
17541 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17542 element vector types.
17544 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17546 * df.h (df_read_modify_subreg_p): Remove in favor of...
17547 * rtl.h (read_modify_subreg_p): ...this new function. Take a
17548 const_rtx instead of an rtx.
17549 * cprop.c (local_cprop_find_used_regs): Update accordingly.
17550 * df-problems.c (df_word_lr_mark_ref): Likewise.
17551 * ira-lives.c (mark_pseudo_reg_live): Likewise.
17552 (mark_pseudo_reg_dead): Likewise.
17553 (mark_ref_dead): Likewise.
17554 * reginfo.c (init_subregs_of_mode): Likewise.
17555 * sched-deps.c (sched_analyze_1): Likewise.
17556 * df-scan.c (df_def_record_1): Likewise.
17557 (df_uses_record): Likewise.
17558 (df_read_modify_subreg_p): Remove in favor of...
17559 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
17560 const_rtx instead of an rtx.
17562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17563 Alan Hayward <alan.hayward@arm.com>
17564 David Sherwood <david.sherwood@arm.com>
17566 * rtl.h (partial_subreg_p): New function.
17567 * caller-save.c (save_call_clobbered_regs): Use it.
17568 * calls.c (expand_call): Likewise.
17569 * combine.c (combinable_i3pat): Likewise.
17570 (simplify_set): Likewise.
17571 (make_extraction): Likewise.
17572 (make_compound_operation_int): Likewise.
17573 (gen_lowpart_or_truncate): Likewise.
17574 (force_to_mode): Likewise.
17575 (make_field_assignment): Likewise.
17576 (reg_truncated_to_mode): Likewise.
17577 (record_truncated_value): Likewise.
17578 (move_deaths): Likewise.
17579 * cse.c (record_jump_cond): Likewise.
17580 (cse_insn): Likewise.
17581 * cselib.c (cselib_lookup_1): Likewise.
17582 * expmed.c (extract_bit_field_using_extv): Likewise.
17583 * function.c (assign_parm_setup_reg): Likewise.
17584 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17585 * ira-build.c (create_insn_allocnos): Likewise.
17586 * lra-coalesce.c (merge_pseudos): Likewise.
17587 * lra-constraints.c (match_reload): Likewise.
17588 (simplify_operand_subreg): Likewise.
17589 (curr_insn_transform): Likewise.
17590 * lra-lives.c (process_bb_lives): Likewise.
17591 * lra.c (new_insn_reg): Likewise.
17592 (lra_substitute_pseudo): Likewise.
17593 * regcprop.c (mode_change_ok): Likewise.
17594 (maybe_mode_change): Likewise.
17595 (copyprop_hardreg_forward_1): Likewise.
17596 * reload.c (push_reload): Likewise.
17597 (find_reloads): Likewise.
17598 (find_reloads_subreg_address): Likewise.
17599 * reload1.c (alter_reg): Likewise.
17600 (eliminate_regs_1): Likewise.
17601 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17603 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17605 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17606 back to if statements, including unpack.
17608 2017-08-30 Martin Liska <mliska@suse.cz>
17610 PR inline-asm/82001
17611 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17613 (func_checker::compare_asm_inputs_outputs): ... this function.
17614 (func_checker::compare_gimple_asm): Use the function to compare
17615 also ASM constrains.
17616 * ipa-icf-gimple.h: Rename the function.
17618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17619 Alan Hayward <alan.hayward@arm.com>
17620 David Sherwood <david.sherwood@arm.com>
17622 * coretypes.h (complex_mode): New type.
17623 * gdbhooks.py (build_pretty_printer): Handle it.
17624 * machmode.h (complex_mode): New class.
17625 (complex_mode::includes_p): New function.
17626 (is_complex_int_mode): Likewise.
17627 (is_complex_float_mode): Likewise.
17628 * genmodes.c (get_mode_class): Handle complex mode classes.
17629 * function.c (expand_function_end): Use is_complex_int_mode.
17631 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17632 Alan Hayward <alan.hayward@arm.com>
17633 David Sherwood <david.sherwood@arm.com>
17635 * coretypes.h (scalar_mode_pod): New typedef.
17636 * gdbhooks.py (build_pretty_printer): Handle it.
17637 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17638 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17639 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17640 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17641 as_a <scalar_mode>.
17643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17644 Alan Hayward <alan.hayward@arm.com>
17645 David Sherwood <david.sherwood@arm.com>
17647 * machmode.h (mode_for_vector): Take a scalar_mode instead
17649 * stor-layout.c (mode_for_vector): Likewise.
17650 * explow.c (promote_mode): Use as_a <scalar_mode>.
17651 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
17653 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17654 Alan Hayward <alan.hayward@arm.com>
17655 David Sherwood <david.sherwood@arm.com>
17657 * target.def (preferred_simd_mode): Take a scalar_mode
17658 instead of a machine_mode.
17659 * targhooks.h (default_preferred_simd_mode): Likewise.
17660 * targhooks.c (default_preferred_simd_mode): Likewise.
17661 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17662 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17663 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17664 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17665 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17666 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17667 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17668 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17670 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17671 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17672 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17673 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17674 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17675 * doc/tm.texi: Regenerate.
17676 * optabs-query.c (can_vec_mask_load_store_p): Return false for
17679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17680 Alan Hayward <alan.hayward@arm.com>
17681 David Sherwood <david.sherwood@arm.com>
17683 * target.def (scalar_mode_supported_p): Take a scalar_mode
17684 instead of a machine_mode.
17685 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17686 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17687 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17688 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17689 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17690 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17691 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17692 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17693 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17694 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17695 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17696 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17697 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17698 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17700 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17701 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17702 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17703 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17704 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17706 * doc/tm.texi: Regenerate.
17708 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17709 Alan Hayward <alan.hayward@arm.com>
17710 David Sherwood <david.sherwood@arm.com>
17712 * coretypes.h (opt_scalar_mode): New typedef.
17713 * gdbhooks.py (build_pretty_printers): Handle it.
17714 * machmode.h (mode_iterator::get_2xwider): Add overload for
17716 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17718 * expr.c (convert_mode_scalar): Likewise.
17719 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17720 * optabs.c (expand_float): Likewise.
17721 (expand_fix): Likewise.
17722 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17725 Alan Hayward <alan.hayward@arm.com>
17726 David Sherwood <david.sherwood@arm.com>
17728 * optabs.c (expand_float): Explicitly check for scalars before
17729 using a branching expansion.
17730 (expand_fix): Likewise.
17732 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17733 Alan Hayward <alan.hayward@arm.com>
17734 David Sherwood <david.sherwood@arm.com>
17736 * expr.c (convert_mode): Split scalar handling out into...
17737 (convert_mode_scalar): ...this new function. Treat the modes
17740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17741 Alan Hayward <alan.hayward@arm.com>
17742 David Sherwood <david.sherwood@arm.com>
17744 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17746 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17748 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17749 Alan Hayward <alan.hayward@arm.com>
17750 David Sherwood <david.sherwood@arm.com>
17752 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17753 rather than a machine_mode.
17754 (fixed_from_string): Likewise.
17755 (fixed_convert): Likewise.
17756 (fixed_convert_from_int): Likewise.
17757 (fixed_convert_from_real): Likewise.
17758 (real_convert_from_fixed): Likewise.
17759 * fixed-value.c (fixed_from_double_int): Likewise.
17760 (fixed_from_string): Likewise.
17761 (fixed_convert): Likewise.
17762 (fixed_convert_from_int): Likewise.
17763 (fixed_convert_from_real): Likewise.
17764 (real_convert_from_fixed): Likewise.
17765 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17768 Alan Hayward <alan.hayward@arm.com>
17769 David Sherwood <david.sherwood@arm.com>
17771 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17772 of separate mode class checks. Do not allow vector modes here.
17773 (immed_wide_int_const): Use as_a <scalar_mode>.
17774 * explow.c (trunc_int_for_mode): Likewise.
17775 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17776 (wi::shwi): Likewise.
17777 (wi::min_value): Likewise.
17778 (wi::max_value): Likewise.
17779 * dwarf2out.c (loc_descriptor): Likewise.
17780 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17781 for CONST_WIDE_INT.
17783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17784 Alan Hayward <alan.hayward@arm.com>
17785 David Sherwood <david.sherwood@arm.com>
17787 * tree.h (SCALAR_TYPE_MODE): New macro.
17788 * expr.c (expand_expr_addr_expr_1): Use it.
17789 (expand_expr_real_2): Likewise.
17790 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17791 (fold_convert_const_fixed_from_int): Likewise.
17792 (fold_convert_const_fixed_from_real): Likewise.
17793 (native_encode_fixed): Likewise
17794 (native_encode_complex): Likewise
17795 (native_encode_vector): Likewise.
17796 (native_interpret_fixed): Likewise.
17797 (native_interpret_real): Likewise.
17798 (native_interpret_complex): Likewise.
17799 (native_interpret_vector): Likewise.
17800 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17801 (simd_clone_adjust_argument_types): Likewise.
17802 (simd_clone_init_simd_arrays): Likewise.
17803 (simd_clone_adjust): Likewise.
17804 * stor-layout.c (layout_type): Likewise.
17805 * tree.c (build_minus_one_cst): Likewise.
17806 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17807 * tree-inline.c (estimate_move_cost): Likewise.
17808 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17809 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17810 (vectorizable_reduction): Likewise.
17811 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17812 (vect_recog_mixed_size_cond_pattern): Likewise.
17813 (check_bool_pattern): Likewise.
17814 (adjust_bool_pattern): Likewise.
17815 (search_type_for_mask_1): Likewise.
17816 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17817 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17818 (vectorizable_load): Likewise.
17819 (vectorizable_store): Likewise.
17820 * ubsan.c (ubsan_encode_value): Likewise.
17821 * varasm.c (output_constant): Likewise.
17823 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17824 Alan Hayward <alan.hayward@arm.com>
17825 David Sherwood <david.sherwood@arm.com>
17827 * coretypes.h (scalar_mode): New class.
17828 * machmode.h (scalar_mode): Likewise.
17829 (scalar_mode::includes_p): New function.
17830 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
17831 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
17832 * genmodes.c (get_mode_class): Handle remaining scalar modes.
17833 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
17834 * expmed.c (store_bit_field_1): Likewise.
17835 (extract_bit_field_1): Likewise.
17836 * expr.c (write_complex_part): Likewise.
17837 (read_complex_part): Likewise.
17838 (emit_move_complex_push): Likewise.
17839 (expand_expr_real_2): Likewise.
17840 * function.c (assign_parm_setup_reg): Likewise.
17841 (assign_parms_unsplit_complex): Likewise.
17842 * optabs.c (expand_binop): Likewise.
17843 * rtlanal.c (subreg_get_info): Likewise.
17844 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17845 * varasm.c (output_constant_pool_2): Likewise.
17847 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17848 Alan Hayward <alan.hayward@arm.com>
17849 David Sherwood <david.sherwood@arm.com>
17851 * expmed.c (extract_high_half): Use scalar_int_mode and remove
17853 (expmed_mult_highpart_optab): Likewise.
17854 (expmed_mult_highpart): Likewise.
17856 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17857 Alan Hayward <alan.hayward@arm.com>
17858 David Sherwood <david.sherwood@arm.com>
17860 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
17861 instead of a machine_mode.
17862 (builtin_memset_read_str): Likewise.
17863 * builtins.c (c_readstr): Likewise.
17864 (builtin_memcpy_read_str): Likewise.
17865 (builtin_strncpy_read_str): Likewise.
17866 (builtin_memset_read_str): Likewise.
17867 (builtin_memset_gen_str): Likewise.
17868 (expand_builtin_signbit): Use scalar_int_mode for local variables.
17869 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
17870 instead of a machine_mode.
17871 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
17873 (make_extraction): Likewise.
17874 (try_widen_shift_mode): Take and return scalar_int_modes instead
17876 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
17877 a scalar_int_mode instead of a machine_mode.
17878 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
17879 (avr_addr_space_pointer_mode): Likewise.
17880 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
17881 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
17882 (msp430_unwind_word_mode): Likewise.
17883 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
17884 (spu_addr_space_pointer_mode): Likewise.
17885 (spu_addr_space_address_mode): Likewise.
17886 (spu_libgcc_cmp_return_mode): Likewise.
17887 (spu_libgcc_shift_count_mode): Likewise.
17888 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
17889 (rl78_addr_space_pointer_mode): Likewise.
17890 (fl78_unwind_word_mode): Likewise.
17891 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
17893 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
17894 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
17895 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
17896 (mips_valid_pointer_mode): Likewise.
17897 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
17898 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
17899 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
17901 (ft32_addr_space_address_mode): Likewise.
17902 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
17903 scalar_int_mode instead of a machine_mode.
17904 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
17906 (m32c_addr_space_address_mode): Likewise.
17907 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
17908 (rs6000_eh_return_filter_mode): Likewise.
17909 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
17910 (rs6000_eh_return_filter_mode): Likewise.
17911 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
17912 (s390_libgcc_shift_count_mode): Likewise.
17913 (s390_unwind_word_mode): Likewise.
17914 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
17916 * target.def (mode_rep_extended): Likewise.
17917 (valid_pointer_mode): Likewise.
17918 (addr_space.valid_pointer_mode): Likewise.
17919 (eh_return_filter_mode): Return a scalar_int_mode rather than
17921 (libgcc_cmp_return_mode): Likewise.
17922 (libgcc_shift_count_mode): Likewise.
17923 (unwind_word_mode): Likewise.
17924 (addr_space.pointer_mode): Likewise.
17925 (addr_space.address_mode): Likewise.
17926 * doc/tm.texi: Regenerate.
17927 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
17929 (do_jump): Use scalar_int_mode for local variables.
17930 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
17931 rather than a machine_mode.
17932 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
17933 (scompare_loc_descriptor_wide): Likewise.
17934 (scompare_loc_descriptor_narrow): Likewise.
17935 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
17937 * except.c (sjlj_emit_dispatch_table): Likewise.
17938 (expand_builtin_eh_copy_values): Likewise.
17939 * explow.c (convert_memory_address_addr_space_1): Likewise.
17940 Take a scalar_int_mode rather than a machine_mode.
17941 (convert_memory_address_addr_space): Take a scalar_int_mode rather
17942 than a machine_mode.
17943 (memory_address_addr_space): Use scalar_int_mode for local variables.
17944 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
17945 rather than a machine_mode.
17946 * expmed.c (mask_rtx): Likewise.
17947 (init_expmed_one_conv): Likewise.
17948 (expand_mult_highpart_adjust): Likewise.
17949 (extract_high_half): Likewise.
17950 (expmed_mult_highpart_optab): Likewise.
17951 (expmed_mult_highpart): Likewise.
17952 (expand_smod_pow2): Likewise.
17953 (expand_sdiv_pow2): Likewise.
17954 (emit_store_flag_int): Likewise.
17955 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
17957 (extract_low_bits): Likewise.
17958 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
17960 * expr.c (pieces_addr::adjust): Likewise.
17961 (can_store_by_pieces): Likewise.
17962 (store_by_pieces): Likewise.
17963 (clear_by_pieces_1): Likewise.
17964 (expand_expr_addr_expr_1): Likewise.
17965 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
17966 (expand_expr_real_1): Likewise.
17967 (try_casesi): Likewise.
17968 * final.c (shorten_branches): Likewise.
17969 * fold-const.c (fold_convert_const_int_from_fixed): Change the
17970 type of "mode" to machine_mode.
17971 * internal-fn.c (expand_arith_overflow_result_store): Take a
17972 scalar_int_mode rather than a machine_mode.
17973 (expand_mul_overflow): Use scalar_int_mode for local variables.
17974 * loop-doloop.c (doloop_modify): Likewise.
17975 (doloop_optimize): Likewise.
17976 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
17977 than a machine_mode.
17978 (expand_doubleword_shift_condmove): Likewise.
17979 (expand_doubleword_shift): Likewise.
17980 (expand_doubleword_clz): Likewise.
17981 (expand_doubleword_popcount): Likewise.
17982 (expand_doubleword_parity): Likewise.
17983 (expand_absneg_bit): Use scalar_int_mode for local variables.
17984 (prepare_float_lib_cmp): Likewise.
17985 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
17986 rather than a machine_mode.
17987 (convert_memory_address_addr_space): Likewise.
17988 (get_mode_bounds): Likewise.
17989 (get_address_mode): Return a scalar_int_mode rather than a
17991 * rtlanal.c (get_address_mode): Likewise.
17992 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
17993 than a machine_mode.
17994 * targhooks.c (default_mode_rep_extended): Likewise.
17995 (default_valid_pointer_mode): Likewise.
17996 (default_addr_space_valid_pointer_mode): Likewise.
17997 (default_eh_return_filter_mode): Return a scalar_int_mode rather
17998 than a machine_mode.
17999 (default_libgcc_cmp_return_mode): Likewise.
18000 (default_libgcc_shift_count_mode): Likewise.
18001 (default_unwind_word_mode): Likewise.
18002 (default_addr_space_pointer_mode): Likewise.
18003 (default_addr_space_address_mode): Likewise.
18004 * targhooks.h (default_eh_return_filter_mode): Likewise.
18005 (default_libgcc_cmp_return_mode): Likewise.
18006 (default_libgcc_shift_count_mode): Likewise.
18007 (default_unwind_word_mode): Likewise.
18008 (default_addr_space_pointer_mode): Likewise.
18009 (default_addr_space_address_mode): Likewise.
18010 (default_mode_rep_extended): Take a scalar_int_mode rather than
18012 (default_valid_pointer_mode): Likewise.
18013 (default_addr_space_valid_pointer_mode): Likewise.
18014 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18016 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18017 rather than a machine_mode.
18018 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18019 for local variables.
18020 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18021 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18022 than a machine_mode.
18024 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18025 Alan Hayward <alan.hayward@arm.com>
18026 David Sherwood <david.sherwood@arm.com>
18028 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18029 the mode argument to scalar_int_mode.
18030 (do_jump_by_parts_zero_rtx): Likewise.
18031 (do_jump_by_parts_equality_rtx): Likewise.
18032 (do_jump_by_parts_greater): Take a mode argument.
18033 (do_jump_by_parts_equality): Likewise.
18034 (do_jump_1): Update calls accordingly.
18036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18037 Alan Hayward <alan.hayward@arm.com>
18038 David Sherwood <david.sherwood@arm.com>
18040 * is-a.h (safe_dyn_cast): New function.
18041 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18042 (jump_table_for_label): Likewise.
18043 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18044 instead of an rtx_insn *.
18045 (shorten_branches): Use dyn_cast instead of LABEL_P and
18046 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18047 rtx_jump_table_data::get_data_mode.
18048 (final_scan_insn): Likewise.
18050 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18051 Alan Hayward <alan.hayward@arm.com>
18052 David Sherwood <david.sherwood@arm.com>
18054 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18055 trying to combine a full-register integer set with a subreg
18058 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18059 Alan Hayward <alan.hayward@arm.com>
18060 David Sherwood <david.sherwood@arm.com>
18062 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18063 that is always either address_mode or pointer_mode.
18065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18066 Alan Hayward <alan.hayward@arm.com>
18067 David Sherwood <david.sherwood@arm.com>
18069 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
18070 when the two are known to be equal.
18072 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18073 Alan Hayward <alan.hayward@arm.com>
18074 David Sherwood <david.sherwood@arm.com>
18076 * simplify-rtx.c (simplify_const_unary_operation): Use
18077 is_a <scalar_int_mode> instead of checking for a nonzero
18078 precision. Forcibly convert op_mode to a scalar_int_mode
18079 in that case. More clearly differentiate the operand and
18080 result modes and use the former when deciding what the value
18081 of a count-bits operation should be. Use is_int_mode instead
18082 of checking for a MODE_INT. Remove redundant check for whether
18083 this mode has a zero precision.
18085 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18086 Alan Hayward <alan.hayward@arm.com>
18087 David Sherwood <david.sherwood@arm.com>
18089 * optabs.c (widen_leading): Change the type of the mode argument
18090 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
18091 (widen_bswap): Likewise.
18092 (expand_parity): Likewise.
18093 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
18094 (expand_ffs): Likewise.
18095 (epand_unop): Check for scalar integer modes before calling the
18098 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18099 Alan Hayward <alan.hayward@arm.com>
18100 David Sherwood <david.sherwood@arm.com>
18102 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
18104 (expand_expr_real_1): Update call accordingly.
18106 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18107 Alan Hayward <alan.hayward@arm.com>
18108 David Sherwood <david.sherwood@arm.com>
18110 * expmed.c (store_bit_field_using_insv): Add op0_mode and
18111 value_mode arguments. Use scalar_int_mode internally.
18112 (store_bit_field_1): Rename the new integer mode from imode
18113 to op0_mode and use it instead of GET_MODE (op0). Update calls
18114 to store_split_bit_field, store_bit_field_using_insv and
18115 store_fixed_bit_field.
18116 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
18117 Use scalar_int_mode internally. Use a bit count rather than a mode
18118 when calculating the largest bit size for get_best_mode.
18119 Update calls to store_split_bit_field and store_fixed_bit_field_1.
18120 (store_fixed_bit_field_1): Add mode and value_mode arguments.
18121 Remove assertion that OP0 has a scalar integer mode.
18122 (store_split_bit_field): Add op0_mode and value_mode arguments.
18123 Update calls to extract_fixed_bit_field.
18124 (extract_bit_field_using_extv): Add an op0_mode argument.
18125 Use scalar_int_mode internally.
18126 (extract_bit_field_1): Rename the new integer mode from imode to
18127 op0_mode and use it instead of GET_MODE (op0). Update calls to
18128 extract_split_bit_field, extract_bit_field_using_extv and
18129 extract_fixed_bit_field.
18130 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
18131 to extract_split_bit_field and extract_fixed_bit_field_1.
18132 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
18133 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
18134 on the target mode.
18135 (extract_split_bit_field): Add an op0_mode argument. Update call
18136 to extract_fixed_bit_field.
18138 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18139 Alan Hayward <alan.hayward@arm.com>
18140 David Sherwood <david.sherwood@arm.com>
18142 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
18143 * explow.c (hard_function_value): Likewise.
18144 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
18145 convert_to_mode call outside the loop.
18146 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
18147 for the mode iterator. Require the mode specified by max_pieces
18149 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
18151 (copy_blkmode_to_reg): Likewise.
18152 (set_storage_via_setmem): Likewise.
18153 * optabs.c (prepare_cmp_insn): Likewise.
18154 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18155 * stor-layout.c (finish_bitfield_representative): Likewise.
18157 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18158 Alan Hayward <alan.hayward@arm.com>
18159 David Sherwood <david.sherwood@arm.com>
18161 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
18162 * expr.c (convert_move): Use them.
18163 (convert_modes): Likewise.
18164 (store_expr_with_bounds): Likewise.
18166 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18167 Alan Hayward <alan.hayward@arm.com>
18168 David Sherwood <david.sherwood@arm.com>
18170 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
18171 parameter for the mode of "x". Remove the "known_x", "known_mode"
18172 and "known_ret" arguments. Change the type of the mode argument
18173 to scalar_int_mode.
18174 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
18175 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
18176 (reg_num_sign_bit_copies_for_combine): Likewise.
18177 * rtlanal.c (nonzero_bits1): Likewise.
18178 (num_sign_bit_copies1): Likewise.
18179 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
18180 (reg_num_sign_bit_copies_general): Likewise.
18181 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
18182 (reg_nonzero_bits_general): Likewise.
18184 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18185 Alan Hayward <alan.hayward@arm.com>
18186 David Sherwood <david.sherwood@arm.com>
18188 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
18189 than in subroutines. Return 1 for non-integer modes.
18190 (cached_num_sign_bit_copies): Change the type of the mode parameter
18191 to scalar_int_mode.
18192 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
18193 classes. Handle CONST_INT_P first and then check whether X also
18194 has a scalar integer mode. Check the same thing for inner registers
18195 of a SUBREG and for values that are being extended or truncated.
18197 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18198 Alan Hayward <alan.hayward@arm.com>
18199 David Sherwood <david.sherwood@arm.com>
18201 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
18202 in subroutines. Return the mode mask for non-integer modes.
18203 (cached_nonzero_bits): Change the type of the mode parameter
18204 to scalar_int_mode.
18205 (nonzero_bits1): Likewise. Remove early exit for other mode
18206 classes. Handle CONST_INT_P first and then check whether X
18207 also has a scalar integer mode.
18209 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18210 Alan Hayward <alan.hayward@arm.com>
18211 David Sherwood <david.sherwood@arm.com>
18213 * expr.c (widest_int_mode_for_size): Make the comment match the code.
18214 Return a scalar_int_mode and assert that the size is greater than
18216 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
18217 (op_by_pieces_d::op_by_pieces_d): Likewise.
18218 (op_by_pieces_d::run): Likewise.
18219 (can_store_by_pieces): Likewise.
18221 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18222 Alan Hayward <alan.hayward@arm.com>
18223 David Sherwood <david.sherwood@arm.com>
18225 * combine.c (extract_left_shift): Add a mode argument and update
18227 (make_compound_operation_int): Change the type of the mode parameter
18228 to scalar_int_mode and update the call to extract_left_shift.
18230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18231 Alan Hayward <alan.hayward@arm.com>
18232 David Sherwood <david.sherwood@arm.com>
18234 * combine.c (simplify_and_const_int): Change the type of the mode
18235 parameter to scalar_int_mode.
18236 (simplify_and_const_int_1): Likewise. Update recursive call.
18238 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18239 Alan Hayward <alan.hayward@arm.com>
18240 David Sherwood <david.sherwood@arm.com>
18242 * combine.c (simplify_compare_const): Check that the mode is a
18243 scalar_int_mode (rather than VOIDmode) before testing its
18245 (simplify_comparison): Move COMPARISON_P handling out of the
18246 loop and restrict the latter part of the loop to scalar_int_modes.
18247 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
18248 and when considering SUBREG_REGs. Use is_int_mode instead of
18249 checking GET_MODE_CLASS against MODE_INT.
18251 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18252 Alan Hayward <alan.hayward@arm.com>
18253 David Sherwood <david.sherwood@arm.com>
18255 * combine.c (try_widen_shift_mode): Move check for equal modes to...
18256 (simplify_shift_const_1): ...here. Use scalar_int_mode for
18257 shift_unit_mode and for modes involved in scalar shifts.
18259 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18260 Alan Hayward <alan.hayward@arm.com>
18261 David Sherwood <david.sherwood@arm.com>
18263 * combine.c (force_int_to_mode): New function, split out from...
18264 (force_to_mode): ...here. Keep xmode up-to-date and use it
18265 instead of GET_MODE (x).
18267 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18268 Alan Hayward <alan.hayward@arm.com>
18269 David Sherwood <david.sherwood@arm.com>
18271 * optabs-query.h (extraction_insn::struct_mode): Change type to
18272 opt_scalar_int_mode and update comment.
18273 (extraction_insn::field_mode): Change type to scalar_int_mode.
18274 (extraction_insn::pos_mode): Likewise.
18275 * combine.c (make_extraction): Update accordingly.
18276 * optabs-query.c (get_traditional_extraction_insn): Likewise.
18277 (get_optab_extraction_insn): Likewise.
18278 * recog.c (simplify_while_replacing): Likewise.
18279 * expmed.c (narrow_bit_field_mem): Change the type of the mode
18280 parameter to opt_scalar_int_mode.
18282 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18283 Alan Hayward <alan.hayward@arm.com>
18284 David Sherwood <david.sherwood@arm.com>
18286 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
18287 to a scalar_int_mode instead of a machine_mode.
18288 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
18289 (get_best_mode): Return a boolean and use a pointer argument to store
18290 the selected mode. Replace the limit mode parameter with a bit limit.
18291 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
18292 for the values returned by bit_field_mode_iterator::next_mode.
18293 (store_bit_field): Update call to get_best_mode.
18294 (store_fixed_bit_field): Likewise.
18295 (extract_fixed_bit_field): Likewise.
18296 * expr.c (optimize_bitfield_assignment_op): Likewise.
18297 * fold-const.c (optimize_bit_field_compare): Likewise.
18298 (fold_truth_andor_1): Likewise.
18299 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
18300 Update for new type of m_mode.
18301 (get_best_mode): As above.
18303 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18304 Alan Hayward <alan.hayward@arm.com>
18305 David Sherwood <david.sherwood@arm.com>
18307 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
18308 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
18309 (store_bit_field): Check is_a <scalar_int_mode> before calling
18310 strict_volatile_bitfield_p.
18311 (extract_bit_field): Likewise.
18313 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18314 Alan Hayward <alan.hayward@arm.com>
18315 David Sherwood <david.sherwood@arm.com>
18317 * target.def (cstore_mode): Return a scalar_int_mode.
18318 * doc/tm.texi: Regenerate.
18319 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
18320 * targhooks.h (default_cstore_mode): Likewise.
18321 * targhooks.c (default_cstore_mode): Likewise, using a forced
18323 * expmed.c (emit_cstore): Expect the target of the cstore to be
18326 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18327 Alan Hayward <alan.hayward@arm.com>
18328 David Sherwood <david.sherwood@arm.com>
18330 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
18332 (niter_desc): Likewise mode.
18333 (iv_analyze): Add a mode parameter.
18335 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
18336 and change its type to scalar_int_mode.
18337 * loop-iv.c: Update commentary at head of file.
18338 (iv_constant): Pass the mode paraeter before the rtx it describes
18339 and change its type to scalar_int_mode. Remove VOIDmode handling.
18340 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
18341 (iv_extend): Likewise.
18342 (shorten_into_mode): Likewise.
18343 (iv_add): Use scalar_int_mode.
18344 (iv_mult): Likewise.
18345 (iv_shift): Likewise.
18346 (canonicalize_iv_subregs): Likewise.
18347 (get_biv_step_1): Pass the outer_mode parameter before the rtx
18348 it describes and change its mode to scalar_int_mode. Also change
18349 the type of the returned inner_mode to scalar_int_mode.
18350 (get_biv_step): Likewise, turning outer_mode from a pointer
18351 into a direct parameter. Update call to get_biv_step_1.
18352 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
18353 iv_constant and get_biv_step.
18354 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
18355 and change its type to scalar_int_mode. Don't initialise iv->mode
18356 to VOIDmode and remove later checks for its still being VOIDmode.
18357 Update calls to iv_analyze_op and iv_analyze_expr. Check
18358 is_a <scalar_int_mode> when changing the mode under consideration.
18359 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
18360 Update call to iv_analyze_expr.
18361 (iv_analyze_op): Add a mode parameter. Reject subregs whose
18362 inner register is not also a scalar_int_mode. Update call to
18364 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
18365 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
18366 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
18367 separate mode class checks. Update calls to iv_analyze. Remove
18368 fix-up of VOIDmodes after iv_analyze_biv.
18369 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
18370 don't have a scalar_int_mode. Update call to biv_p.
18372 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18373 Alan Hayward <alan.hayward@arm.com>
18374 David Sherwood <david.sherwood@arm.com>
18376 * cfgexpand.c (convert_debug_memory_address): Use
18377 as_a <scalar_int_mode>.
18378 * combine.c (expand_compound_operation): Likewise.
18379 (make_extraction): Likewise.
18380 (change_zero_ext): Likewise.
18381 (simplify_comparison): Likewise.
18382 * cse.c (cse_insn): Likewise.
18383 * dwarf2out.c (minmax_loc_descriptor): Likewise.
18384 (mem_loc_descriptor): Likewise.
18385 (loc_descriptor): Likewise.
18386 * expmed.c (init_expmed_one_mode): Likewise.
18387 (synth_mult): Likewise.
18388 (emit_store_flag_1): Likewise.
18389 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
18390 of a comparison with size.
18391 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
18392 (reduce_to_bit_field_precision): Likewise.
18393 * function.c (expand_function_end): Likewise.
18394 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
18395 * loop-doloop.c (doloop_modify): Likewise.
18396 * optabs.c (expand_binop): Likewise.
18397 (expand_unop): Likewise.
18398 (expand_copysign_absneg): Likewise.
18399 (prepare_cmp_insn): Likewise.
18400 (maybe_legitimize_operand): Likewise.
18401 * recog.c (const_scalar_int_operand): Likewise.
18402 * rtlanal.c (get_address_mode): Likewise.
18403 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18404 (simplify_cond_clz_ctz): Likewise.
18405 * tree-nested.c (get_nl_goto_field): Likewise.
18406 * tree.c (build_vector_type_for_mode): Likewise.
18407 * var-tracking.c (use_narrower_mode): Likewise.
18409 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18410 Alan Hayward <alan.hayward@arm.com>
18411 David Sherwood <david.sherwood@arm.com>
18413 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
18414 * builtins.c (expand_builtin_signbit): Use it.
18415 * cfgexpand.c (expand_debug_expr): Likewise.
18416 * dojump.c (do_jump): Likewise.
18417 (do_compare_and_jump): Likewise.
18418 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
18419 * expmed.c (make_tree): Likewise.
18420 * expr.c (expand_expr_real_2): Likewise.
18421 (expand_expr_real_1): Likewise.
18422 (try_casesi): Likewise.
18423 * fold-const-call.c (fold_const_call_ss): Likewise.
18424 * fold-const.c (unextend): Likewise.
18425 (extract_muldiv_1): Likewise.
18426 (fold_single_bit_test): Likewise.
18427 (native_encode_int): Likewise.
18428 (native_encode_string): Likewise.
18429 (native_interpret_int): Likewise.
18430 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
18431 * internal-fn.c (expand_addsub_overflow): Likewise.
18432 (expand_neg_overflow): Likewise.
18433 (expand_mul_overflow): Likewise.
18434 (expand_arith_overflow): Likewise.
18435 * match.pd: Likewise.
18436 * stor-layout.c (layout_type): Likewise.
18437 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18438 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
18439 * tree-ssanames.c (get_range_info): Likewise.
18440 * tree-switch-conversion.c (array_value_type) Likewise.
18441 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
18442 (vect_recog_divmod_pattern): Likewise.
18443 (vect_recog_mixed_size_cond_pattern): Likewise.
18444 * tree-vrp.c (extract_range_basic): Likewise.
18445 (simplify_float_conversion_using_ranges): Likewise.
18446 * tree.c (int_fits_type_p): Likewise.
18447 * ubsan.c (instrument_bool_enum_load): Likewise.
18448 * varasm.c (mergeable_string_section): Likewise.
18449 (narrowing_initializer_constant_valid_p): Likewise.
18450 (output_constant): Likewise.
18452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18453 Alan Hayward <alan.hayward@arm.com>
18454 David Sherwood <david.sherwood@arm.com>
18456 * machmode.h (NARROWEST_INT_MODE): New macro.
18457 * expr.c (alignment_for_piecewise_move): Use it instead of
18458 GET_CLASS_NARROWEST_MODE (MODE_INT).
18459 (push_block): Likewise.
18460 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
18462 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
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 * postreload.c (move2add_valid_value_p): Change the type of the
18469 mode parameter to scalar_int_mode.
18470 (move2add_use_add2_insn): Add a mode parameter and use it instead
18472 (move2add_use_add3_insn): Likewise.
18473 (reload_cse_move2add): Update accordingly.
18475 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18476 Alan Hayward <alan.hayward@arm.com>
18477 David Sherwood <david.sherwood@arm.com>
18479 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
18481 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
18482 * optabs.c (expand_unop): Likewise.
18484 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18485 Alan Hayward <alan.hayward@arm.com>
18486 David Sherwood <david.sherwood@arm.com>
18488 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
18489 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
18490 (popcount_loc_descriptor): Likewise.
18491 (bswap_loc_descriptor): Likewise.
18492 (rotate_loc_descriptor): Likewise.
18493 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
18494 calling the functions above.
18496 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18497 Alan Hayward <alan.hayward@arm.com>
18498 David Sherwood <david.sherwood@arm.com>
18500 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
18502 (try_combine): Likewise.
18503 (simplify_if_then_else): Likewise.
18504 * cse.c (cse_insn): Likewise.
18505 * dwarf2out.c (mem_loc_descriptor): Likewise.
18506 * emit-rtl.c (gen_lowpart_common): Likewise.
18507 * simplify-rtx.c (simplify_truncation): Likewise.
18508 (simplify_binary_operation_1): Likewise.
18509 (simplify_const_relational_operation): Likewise.
18510 (simplify_ternary_operation): Likewise.
18511 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
18513 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18514 Alan Hayward <alan.hayward@arm.com>
18515 David Sherwood <david.sherwood@arm.com>
18517 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
18518 * reload.c (push_reload): Likewise.
18519 (find_reloads): Likewise.
18521 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18522 Alan Hayward <alan.hayward@arm.com>
18523 David Sherwood <david.sherwood@arm.com>
18525 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
18526 (make_compound_operation_int): Likewise.
18527 (change_zero_ext): Likewise.
18528 * expr.c (convert_move): Likewise.
18529 (convert_modes): Likewise.
18530 * fwprop.c (forward_propagate_subreg): Likewise.
18531 * loop-iv.c (get_biv_step_1): Likewise.
18532 * optabs.c (widen_operand): Likewise.
18533 * postreload.c (move2add_valid_value_p): Likewise.
18534 * recog.c (simplify_while_replacing): Likewise.
18535 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18536 (simplify_binary_operation_1): Likewise. Remove redundant
18537 mode equality check.
18539 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18540 Alan Hayward <alan.hayward@arm.com>
18541 David Sherwood <david.sherwood@arm.com>
18543 * combine.c (combine_simplify_rtx): Add checks for
18544 is_a <scalar_int_mode>.
18545 (simplify_if_then_else): Likewise.
18546 (make_field_assignment): Likewise.
18547 (simplify_comparison): Likewise.
18548 * ifcvt.c (noce_try_bitop): Likewise.
18549 * loop-invariant.c (canonicalize_address_mult): Likewise.
18550 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18552 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18553 Alan Hayward <alan.hayward@arm.com>
18554 David Sherwood <david.sherwood@arm.com>
18556 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
18557 is_a <scalar_int_mode> instead of != BLKmode.
18559 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18560 Alan Hayward <alan.hayward@arm.com>
18561 David Sherwood <david.sherwood@arm.com>
18563 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
18564 instead of != VOIDmode.
18565 * combine.c (if_then_else_cond): Likewise.
18566 (change_zero_ext): Likewise.
18567 * dwarf2out.c (mem_loc_descriptor): Likewise.
18568 (loc_descriptor): Likewise.
18569 * rtlanal.c (canonicalize_condition): Likewise.
18570 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
18572 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18573 Alan Hayward <alan.hayward@arm.com>
18574 David Sherwood <david.sherwood@arm.com>
18576 * simplify-rtx.c (simplify_binary_operation_1): Use
18577 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18580 Alan Hayward <alan.hayward@arm.com>
18581 David Sherwood <david.sherwood@arm.com>
18583 * wide-int.h (int_traits<unsigned char>) New class.
18584 (int_traits<unsigned short>) Likewise.
18585 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18586 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18588 * combine.c (set_nonzero_bits_and_sign_copies): Use
18589 is_a <scalar_int_mode>.
18590 (find_split_point): Likewise.
18591 (combine_simplify_rtx): Likewise.
18592 (simplify_logical): Likewise.
18593 (expand_compound_operation): Likewise.
18594 (expand_field_assignment): Likewise.
18595 (make_compound_operation): Likewise.
18596 (extended_count): Likewise.
18597 (change_zero_ext): Likewise.
18598 (simplify_comparison): Likewise.
18599 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18600 (ucompare_loc_descriptor): Likewise.
18601 (minmax_loc_descriptor): Likewise.
18602 (mem_loc_descriptor): Likewise.
18603 (loc_descriptor): Likewise.
18604 * expmed.c (init_expmed_one_mode): Likewise.
18605 * lra-constraints.c (lra_constraint_offset): Likewise.
18606 * optabs.c (prepare_libcall_arg): Likewise.
18607 * postreload.c (move2add_note_store): Likewise.
18608 * reload.c (operands_match_p): Likewise.
18609 * rtl.h (load_extend_op): Likewise.
18610 * rtlhooks.c (gen_lowpart_general): Likewise.
18611 * simplify-rtx.c (simplify_truncation): Likewise.
18612 (simplify_unary_operation_1): Likewise.
18613 (simplify_binary_operation_1): Likewise.
18614 (simplify_const_binary_operation): Likewise.
18615 (simplify_const_relational_operation): Likewise.
18616 (simplify_subreg): Likewise.
18617 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18618 * var-tracking.c (adjust_mems): Likewise.
18619 (prepare_call_arguments): Likewise.
18621 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18622 Alan Hayward <alan.hayward@arm.com>
18623 David Sherwood <david.sherwood@arm.com>
18625 * machmode.h (is_int_mode): New fuction.
18626 * combine.c (find_split_point): Use it.
18627 (combine_simplify_rtx): Likewise.
18628 (simplify_if_then_else): Likewise.
18629 (simplify_set): Likewise.
18630 (simplify_shift_const_1): Likewise.
18631 (simplify_comparison): Likewise.
18632 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18633 * cse.c (notreg_cost): Likewise.
18634 (cse_insn): Likewise.
18635 * cselib.c (cselib_lookup_1): Likewise.
18636 * dojump.c (do_jump_1): Likewise.
18637 (do_compare_rtx_and_jump): Likewise.
18638 * dse.c (get_call_args): Likewise.
18639 * dwarf2out.c (rtl_for_decl_init): Likewise.
18640 (native_encode_initializer): Likewise.
18641 * expmed.c (emit_store_flag_1): Likewise.
18642 (emit_store_flag): Likewise.
18643 * expr.c (convert_modes): Likewise.
18644 (store_field): Likewise.
18645 (expand_expr_real_1): Likewise.
18646 * fold-const.c (fold_read_from_constant_string): Likewise.
18647 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18648 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18649 * optabs.c (expand_binop): Likewise.
18650 (expand_unop): Likewise.
18651 (expand_abs_nojump): Likewise.
18652 (expand_one_cmpl_abs_nojump): Likewise.
18653 * simplify-rtx.c (mode_signbit_p): Likewise.
18654 (val_signbit_p): Likewise.
18655 (val_signbit_known_set_p): Likewise.
18656 (val_signbit_known_clear_p): Likewise.
18657 (simplify_relational_operation_1): Likewise.
18658 * tree.c (vector_type_mode): Likewise.
18660 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18661 Alan Hayward <alan.hayward@arm.com>
18662 David Sherwood <david.sherwood@arm.com>
18664 * machmode.h (smallest_mode_for_size): Fix formatting.
18665 (smallest_int_mode_for_size): New function.
18666 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18667 instead of smallest_mode_for_size.
18668 * combine.c (make_extraction): Likewise.
18669 * config/arc/arc.c (arc_expand_movmem): Likewise.
18670 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18671 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18672 * config/s390/s390.c (s390_expand_insv): Likewise.
18673 * config/sparc/sparc.c (assign_int_registers): Likewise.
18674 * config/spu/spu.c (spu_function_value): Likewise.
18675 (spu_function_arg): Likewise.
18676 * coverage.c (get_gcov_type): Likewise.
18677 (get_gcov_unsigned_t): Likewise.
18678 * dse.c (find_shift_sequence): Likewise.
18679 * expmed.c (store_bit_field_1): Likewise.
18680 * expr.c (convert_move): Likewise.
18681 (store_field): Likewise.
18682 * internal-fn.c (expand_arith_overflow): Likewise.
18683 * optabs-query.c (get_best_extraction_insn): Likewise.
18684 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18685 * stor-layout.c (layout_type): Likewise.
18686 (initialize_sizetypes): Likewise.
18687 * targhooks.c (default_get_mask_mode): Likewise.
18688 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18691 Alan Hayward <alan.hayward@arm.com>
18692 David Sherwood <david.sherwood@arm.com>
18694 * machmode.h (opt_mode::else_blk): New function.
18695 (int_mode_for_mode): Declare.
18696 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18697 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18699 * cfgexpand.c (expand_debug_expr): Likewise.
18700 * combine.c (gen_lowpart_or_truncate): Likewise.
18701 (gen_lowpart_for_combine): Likewise.
18702 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18703 * config/avr/avr.c (avr_to_int_mode): Likewise.
18704 (avr_out_plus_1): Likewise.
18705 (avr_out_plus): Likewise.
18706 (avr_out_round): Likewise.
18707 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18708 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18709 (s390_expand_vcond): Likewise.
18710 * config/spu/spu.c (spu_split_immediate): Likewise.
18711 (spu_expand_mov): Likewise.
18712 * dse.c (get_stored_val): Likewise.
18713 * expmed.c (store_bit_field_1): Likewise.
18714 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18716 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18717 (extract_low_bits): Likewise.
18718 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18719 handling rather than repeating the check.
18720 (emit_group_store): Likewise.
18721 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18722 * optabs.c (expand_absneg_bit): Likewise.
18723 (expand_copysign_absneg): Likewise.
18724 (expand_copysign_bit): Likewise.
18725 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18726 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18727 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18728 * var-tracking.c (prepare_call_arguments): Likewise.
18729 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18730 int_mode_for_mode instead of mode_for_size.
18731 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18733 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18734 Alan Hayward <alan.hayward@arm.com>
18735 David Sherwood <david.sherwood@arm.com>
18737 * machmode.h (int_mode_for_size): New function.
18738 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18739 instead of mode_for_size.
18740 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18742 * combine.c (expand_field_assignment): Use int_mode_for_size
18743 instead of mode_for_size.
18744 (make_extraction): Likewise.
18745 (simplify_shift_const_1): Likewise.
18746 (simplify_comparison): Likewise.
18747 * dojump.c (do_jump): Likewise.
18748 * dwarf2out.c (mem_loc_descriptor): Likewise.
18749 * emit-rtl.c (init_derived_machine_modes): Likewise.
18750 * expmed.c (flip_storage_order): Likewise.
18751 (convert_extracted_bit_field): Likewise.
18752 * expr.c (copy_blkmode_from_reg): Likewise.
18753 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18754 * internal-fn.c (expand_mul_overflow): Likewise.
18755 * lower-subreg.c (simple_move): Likewise.
18756 * optabs-libfuncs.c (init_optabs): Likewise.
18757 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18758 * tree.c (vector_type_mode): Likewise.
18759 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18760 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18761 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18762 * tree-vect-stmts.c (vectorizable_load): Likewise.
18763 (vectorizable_store): Likewise.
18765 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18766 Alan Hayward <alan.hayward@arm.com>
18767 David Sherwood <david.sherwood@arm.com>
18769 * coretypes.h (pod_mode): New type.
18770 (scalar_int_mode_pod): New typedef.
18771 * machmode.h (pod_mode): New class.
18772 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18773 * genmodes.c (emit_mode_int_n): Update accordingly.
18774 * lower-subreg.h (target_lower_subreg): Change type to
18775 scalar_int_mode_pod.
18776 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18777 scalar_int_mode_pod.
18779 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18780 Alan Hayward <alan.hayward@arm.com>
18781 David Sherwood <david.sherwood@arm.com>
18783 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18784 machine_mode to scalar_int_mode.
18785 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18786 (rs6000_option_override_internal): Remove cast to int.
18787 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18788 machine_mode to scalar_int_mode.
18789 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18790 (rs6000_option_override_internal): Remove cast to int.
18791 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18792 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18794 * config/s390/s390.c (s390_expand_builtin): Likewise.
18795 * coretypes.h (scalar_int_mode): New type.
18796 (opt_scalar_int_mode): New typedef.
18797 * machmode.h (scalar_int_mode): New class.
18798 (scalar_int_mode::includes_p): New function.
18799 (byte_mode): Change type to scalar_int_mode.
18800 (word_mode): Likewise.
18801 (ptr_mode): Likewise.
18802 * emit-rtl.c (byte_mode): Likewise.
18803 (word_mode): Likewise.
18804 (ptr_mode): Likewise.
18805 (init_derived_machine_modes): Update accordingly.
18806 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18807 and MODE_PARTIAL_INT.
18808 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18809 opt_scalar_int_mode.
18811 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18812 Alan Hayward <alan.hayward@arm.com>
18813 David Sherwood <david.sherwood@arm.com>
18815 * target.def (libgcc_floating_mode_supported_p): Take a
18817 * doc/tm.texi: Regenerate.
18818 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18820 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18821 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
18824 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18825 Alan Hayward <alan.hayward@arm.com>
18826 David Sherwood <david.sherwood@arm.com>
18828 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
18829 * doc/tm.texi: Regenerate.
18830 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
18831 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
18832 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
18833 * targhooks.h (default_floatn_mode): Likewise.
18834 * targhooks.c (default_floatn_mode): Likewise.
18835 * tree.c (build_common_tree_nodes): Update accordingly.
18837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18838 Alan Hayward <alan.hayward@arm.com>
18839 David Sherwood <david.sherwood@arm.com>
18841 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
18842 (mode_iterator::iterate_p): Likewise.
18843 (mode_iterator::get_wider): Likewise.
18844 * expr.c (init_expr_target): Use 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 * coretypes.h (opt_scalar_float_mode): New typedef.
18851 * machmode.h (float_mode_for_size): New function.
18852 * emit-rtl.c (double_mode): Delete.
18853 (init_emit_once): Use float_mode_for_size.
18854 * stor-layout.c (layout_type): Likewise.
18855 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
18857 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18858 Alan Hayward <alan.hayward@arm.com>
18859 David Sherwood <david.sherwood@arm.com>
18861 * output.h (assemble_real): Take a scalar_float_mode.
18862 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
18863 * config/arm/arm.md (consttable_4): Likewise.
18864 (consttable_8): Likewise.
18865 (consttable_16): Likewise.
18866 * config/mips/mips.md (consttable_float): Likewise.
18867 * config/s390/s390.c (s390_output_pool_entry): Likewise.
18868 * varasm.c (assemble_real): Take a scalar_float_mode.
18869 (output_constant_pool_2): Update accordingly.
18870 (output_constant): Likewise.
18872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18873 Alan Hayward <alan.hayward@arm.com>
18874 David Sherwood <david.sherwood@arm.com>
18876 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
18877 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
18878 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
18879 (native_encode_real): Likewise.
18880 (native_interpret_real): Likewise.
18881 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
18882 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18884 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18885 Alan Hayward <alan.hayward@arm.com>
18886 David Sherwood <david.sherwood@arm.com>
18888 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
18889 <scalar_float_mode>. Simplify.
18890 (gen_extend_conv_libfunc): Likewise.
18892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18893 Alan Hayward <alan.hayward@arm.com>
18894 David Sherwood <david.sherwood@arm.com>
18896 * coretypes.h (scalar_float_mode): New type.
18897 * machmode.h (mode_traits::from_int): Use machine_mode if
18898 USE_ENUM_MODES is defined.
18899 (is_a): New function.
18901 (dyn_cast): Likewise.
18902 (scalar_float_mode): New class.
18903 (scalar_float_mode::includes_p): New function.
18904 (is_float_mode): Likewise.
18905 * gdbhooks.py (MachineModePrinter): New class.
18906 (build_pretty_printer): Use it for scalar_float_mode.
18907 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
18908 (format_helper::format_helper): Turn into a template.
18909 * genmodes.c (get_mode_class): New function.
18910 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
18911 or machine_mode if none.
18912 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
18913 as_a <scalar_float_mode>.
18914 * dwarf2out.c (mem_loc_descriptor): Likewise.
18915 (insert_float): Likewise.
18916 (add_const_value_attribute): Likewise.
18917 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18918 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
18919 (expand_unop): Update accordingly.
18920 (expand_abs_nojump): Likewise.
18921 (expand_copysign_absneg): Take a scalar_float_mode.
18922 (expand_copysign_bit): Likewise.
18923 (expand_copysign): Update accordingly.
18925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18926 Alan Hayward <alan.hayward@arm.com>
18927 David Sherwood <david.sherwood@arm.com>
18929 * coretypes.h (opt_mode): New class.
18930 * machmode.h (opt_mode): Likewise.
18931 (opt_mode::else_void): New function.
18932 (opt_mode::require): Likewise.
18933 (opt_mode::exists): Likewise.
18934 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
18935 (GET_MODE_2XWIDER_MODE): Likewise.
18936 (mode_iterator::get_wider): Update accordingly.
18937 (mode_iterator::get_2xwider): Likewise.
18938 (mode_iterator::get_known_wider): Likewise, turning into a template.
18939 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
18940 forcing a wider mode to exist.
18941 * config/cr16/cr16.h (LONG_REG_P): Likewise.
18942 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18943 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
18944 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
18945 * lower-subreg.c (init_lower_subreg): Likewise.
18946 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
18947 on the final iteration.
18948 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
18949 a wider mode exists before asking for a move pattern.
18950 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
18951 forcing a wider mode to exist.
18952 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
18953 returning false if no such mode exists.
18954 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
18955 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
18956 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
18957 Avoid checking for a MODE_INT if we already know the mode is not a
18959 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
18960 forcing a wider mode to exist.
18961 (expmed_mult_highpart_optab): Likewise.
18962 (expmed_mult_highpart): Likewise.
18963 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
18965 * lto-streamer-in.c (lto_input_mode_table): Likewise.
18966 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
18967 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
18968 * internal-fn.c (expand_mul_overflow): Update use of
18969 GET_MODE_2XWIDER_MODE.
18970 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18971 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
18972 GET_MODE_WIDER_MODE.
18973 (convert_plusminus_to_widen): Likewise.
18974 * tree-switch-conversion.c (array_value_type): Likewise.
18975 * var-tracking.c (emit_note_insn_var_location): Likewise.
18976 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18977 Return false inside rather than outside the loop if no wider mode
18979 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
18980 and GET_MODE_2XWIDER_MODE
18981 (can_compare_p): Use else_void.
18982 * gdbhooks.py (OptMachineModePrinter): New class.
18983 (build_pretty_printer): Use it for opt_mode.
18985 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18986 Alan Hayward <alan.hayward@arm.com>
18987 David Sherwood <david.sherwood@arm.com>
18989 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
18990 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
18992 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18993 Alan Hayward <alan.hayward@arm.com>
18994 David Sherwood <david.sherwood@arm.com>
18996 * machmode.h (mode_traits): New structure.
18997 (get_narrowest_mode): New function.
18998 (mode_iterator::start): Likewise.
18999 (mode_iterator::iterate_p): Likewise.
19000 (mode_iterator::get_wider): Likewise.
19001 (mode_iterator::get_known_wider): Likewise.
19002 (mode_iterator::get_2xwider): Likewise.
19003 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19004 (FOR_EACH_MODE): Likewise.
19005 (FOR_EACH_MODE_FROM): Likewise.
19006 (FOR_EACH_MODE_UNTIL): Likewise.
19007 (FOR_EACH_WIDER_MODE): Likewise.
19008 (FOR_EACH_2XWIDER_MODE): Likewise.
19009 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19010 * combine.c (simplify_comparison): Likewise
19011 * config/i386/i386.c (type_natural_mode): Likewise.
19012 * cse.c (cse_insn): Likewise.
19013 * dse.c (find_shift_sequence): Likewise.
19014 * emit-rtl.c (init_derived_machine_modes): Likewise.
19015 (init_emit_once): Likewise.
19016 * explow.c (hard_function_value): Likewise.
19017 * expmed.c (extract_fixed_bit_field_1): Likewise.
19018 (extract_bit_field_1): Likewise.
19019 (expand_divmod): Likewise.
19020 (emit_store_flag_1): Likewise.
19021 * expr.c (init_expr_target): Likewise.
19022 (convert_move): Likewise.
19023 (alignment_for_piecewise_move): Likewise.
19024 (widest_int_mode_for_size): Likewise.
19025 (emit_block_move_via_movmem): Likewise.
19026 (copy_blkmode_to_reg): Likewise.
19027 (set_storage_via_setmem): Likewise.
19028 (compress_float_constant): Likewise.
19029 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19030 * optabs-query.c (get_best_extraction_insn): Likewise.
19031 * optabs.c (expand_binop): Likewise.
19032 (expand_twoval_unop): Likewise.
19033 (expand_twoval_binop): Likewise.
19034 (widen_leading): Likewise.
19035 (widen_bswap): Likewise.
19036 (expand_parity): Likewise.
19037 (expand_unop): Likewise.
19038 (prepare_cmp_insn): Likewise.
19039 (prepare_float_lib_cmp): Likewise.
19040 (expand_float): Likewise.
19041 (expand_fix): Likewise.
19042 (expand_sfix_optab): Likewise.
19043 * postreload.c (move2add_use_add2_insn): Likewise.
19044 * reg-stack.c (reg_to_stack): Likewise.
19045 * reginfo.c (choose_hard_reg_mode): Likewise.
19046 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19047 * stor-layout.c (mode_for_size): Likewise.
19048 (smallest_mode_for_size): Likewise.
19049 (mode_for_vector): Likewise.
19050 (finish_bitfield_representative): Likewise.
19051 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19052 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19053 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19054 * var-tracking.c (prepare_call_arguments): Likewise.
19056 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19057 Alan Hayward <alan.hayward@arm.com>
19058 David Sherwood <david.sherwood@arm.com>
19060 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19061 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19062 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19063 * machmode.h (mode_size): Move earlier in file.
19064 (mode_precision): Likewise.
19065 (mode_inner): Likewise.
19066 (mode_nunits): Likewise.
19067 (mode_unit_size): Likewise.
19068 (unit_unit_precision): Likewise.
19069 (mode_wider): Likewise.
19070 (mode_2xwider): Likewise.
19071 (machine_mode): New class.
19072 (mode_to_bytes): New function.
19073 (mode_to_bits): Likewise.
19074 (mode_to_precision): Likewise.
19075 (mode_to_inner): Likewise.
19076 (mode_to_unit_size): Likewise.
19077 (mode_to_unit_precision): Likewise.
19078 (mode_to_nunits): Likewise.
19079 (GET_MODE_SIZE): Use mode_to_bytes.
19080 (GET_MODE_BITSIZE): Use mode_to_bits.
19081 (GET_MODE_PRECISION): Use mode_to_precision.
19082 (GET_MODE_INNER): Use mode_to_inner.
19083 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
19084 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
19085 (GET_MODE_NUNITS): Use mode_to_nunits.
19086 * system.h (ALWAYS_INLINE): New macro.
19087 * config/powerpcspe/powerpcspe-c.c
19088 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
19089 int for arg1_mode and arg2_mode.
19091 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19092 Alan Hayward <alan.hayward@arm.com>
19093 David Sherwood <david.sherwood@arm.com>
19095 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19096 Prefix mode names with E_ in case statements.
19097 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19098 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
19099 (aarch64_split_simd_move): Likewise.
19100 (aarch64_gen_storewb_pair): Likewise.
19101 (aarch64_gen_loadwb_pair): Likewise.
19102 (aarch64_gen_store_pair): Likewise.
19103 (aarch64_gen_load_pair): Likewise.
19104 (aarch64_get_condition_code_1): Likewise.
19105 (aarch64_constant_pool_reload_icode): Likewise.
19106 (get_rsqrte_type): Likewise.
19107 (get_rsqrts_type): Likewise.
19108 (get_recpe_type): Likewise.
19109 (get_recps_type): Likewise.
19110 (aarch64_gimplify_va_arg_expr): Likewise.
19111 (aarch64_simd_container_mode): Likewise.
19112 (aarch64_emit_load_exclusive): Likewise.
19113 (aarch64_emit_store_exclusive): Likewise.
19114 (aarch64_expand_compare_and_swap): Likewise.
19115 (aarch64_gen_atomic_cas): Likewise.
19116 (aarch64_emit_bic): Likewise.
19117 (aarch64_emit_atomic_swap): Likewise.
19118 (aarch64_emit_atomic_load_op): Likewise.
19119 (aarch64_evpc_trn): Likewise.
19120 (aarch64_evpc_uzp): Likewise.
19121 (aarch64_evpc_zip): Likewise.
19122 (aarch64_evpc_ext): Likewise.
19123 (aarch64_evpc_rev): Likewise.
19124 (aarch64_evpc_dup): Likewise.
19125 (aarch64_gen_ccmp_first): Likewise.
19126 (aarch64_gen_ccmp_next): Likewise.
19127 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
19128 (alpha_emit_xfloating_libcall): Likewise.
19129 (emit_insxl): Likewise.
19130 (alpha_arg_type): Likewise.
19131 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
19132 (arc_preferred_simd_mode): Likewise.
19133 (arc_secondary_reload): Likewise.
19134 (get_arc_condition_code): Likewise.
19135 (arc_print_operand): Likewise.
19136 (arc_legitimate_constant_p): Likewise.
19137 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19138 * config/arc/arc.md (casesi_load): Likewise.
19139 (casesi_compact_jump): Likewise.
19140 * config/arc/predicates.md (proper_comparison_operator): Likewise.
19141 (cc_use_register): Likewise.
19142 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19143 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19144 (arm_init_iwmmxt_builtins): Likewise.
19145 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
19146 (neon_expand_vector_init): Likewise.
19147 (arm_attr_length_move_neon): Likewise.
19148 (maybe_get_arm_condition_code): Likewise.
19149 (arm_emit_vector_const): Likewise.
19150 (arm_preferred_simd_mode): Likewise.
19151 (arm_output_iwmmxt_tinsr): Likewise.
19152 (thumb1_output_casesi): Likewise.
19153 (thumb2_output_casesi): Likewise.
19154 (arm_emit_load_exclusive): Likewise.
19155 (arm_emit_store_exclusive): Likewise.
19156 (arm_expand_compare_and_swap): Likewise.
19157 (arm_evpc_neon_vuzp): Likewise.
19158 (arm_evpc_neon_vzip): Likewise.
19159 (arm_evpc_neon_vrev): Likewise.
19160 (arm_evpc_neon_vtrn): Likewise.
19161 (arm_evpc_neon_vext): Likewise.
19162 (arm_validize_comparison): Likewise.
19163 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
19164 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
19165 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
19166 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
19167 (c6x_preferred_simd_mode): Likewise.
19168 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
19169 (epiphany_rtx_costs): Likewise.
19170 * config/epiphany/predicates.md (proper_comparison_operator):
19172 * config/frv/frv.c (condexec_memory_operand): Likewise.
19173 (frv_emit_move): Likewise.
19174 (output_move_single): Likewise.
19175 (output_condmove_single): Likewise.
19176 (frv_hard_regno_mode_ok): Likewise.
19177 (frv_matching_accg_mode): Likewise.
19178 * config/h8300/h8300.c (split_adds_subs): Likewise.
19179 (h8300_rtx_costs): Likewise.
19180 (h8300_print_operand): Likewise.
19181 (compute_mov_length): Likewise.
19182 (output_logical_op): Likewise.
19183 (compute_logical_op_length): Likewise.
19184 (compute_logical_op_cc): Likewise.
19185 (h8300_shift_needs_scratch_p): Likewise.
19186 (output_a_shift): Likewise.
19187 (compute_a_shift_length): Likewise.
19188 (compute_a_shift_cc): Likewise.
19189 (expand_a_rotate): Likewise.
19190 (output_a_rotate): Likewise.
19191 * config/i386/i386.c (classify_argument): Likewise.
19192 (function_arg_advance_32): Likewise.
19193 (function_arg_32): Likewise.
19194 (function_arg_64): Likewise.
19195 (function_value_64): Likewise.
19196 (ix86_gimplify_va_arg): Likewise.
19197 (ix86_legitimate_constant_p): Likewise.
19198 (put_condition_code): Likewise.
19199 (split_double_mode): Likewise.
19200 (ix86_avx256_split_vector_move_misalign): Likewise.
19201 (ix86_expand_vector_logical_operator): Likewise.
19202 (ix86_split_idivmod): Likewise.
19203 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
19204 (ix86_build_const_vector): Likewise.
19205 (ix86_build_signbit_mask): Likewise.
19206 (ix86_match_ccmode): Likewise.
19207 (ix86_cc_modes_compatible): Likewise.
19208 (ix86_expand_branch): Likewise.
19209 (ix86_expand_sse_cmp): Likewise.
19210 (ix86_expand_sse_movcc): Likewise.
19211 (ix86_expand_int_sse_cmp): Likewise.
19212 (ix86_expand_vec_perm_vpermi2): Likewise.
19213 (ix86_expand_vec_perm): Likewise.
19214 (ix86_expand_sse_unpack): Likewise.
19215 (ix86_expand_int_addcc): Likewise.
19216 (ix86_split_to_parts): Likewise.
19217 (ix86_vectorize_builtin_gather): Likewise.
19218 (ix86_vectorize_builtin_scatter): Likewise.
19219 (avx_vpermilp_parallel): Likewise.
19220 (inline_memory_move_cost): Likewise.
19221 (ix86_tieable_integer_mode_p): Likewise.
19222 (x86_maybe_negate_const_int): Likewise.
19223 (ix86_expand_vector_init_duplicate): Likewise.
19224 (ix86_expand_vector_init_one_nonzero): Likewise.
19225 (ix86_expand_vector_init_one_var): Likewise.
19226 (ix86_expand_vector_init_concat): Likewise.
19227 (ix86_expand_vector_init_interleave): Likewise.
19228 (ix86_expand_vector_init_general): Likewise.
19229 (ix86_expand_vector_set): Likewise.
19230 (ix86_expand_vector_extract): Likewise.
19231 (emit_reduc_half): Likewise.
19232 (ix86_emit_i387_round): Likewise.
19233 (ix86_mangle_type): Likewise.
19234 (ix86_expand_round_sse4): Likewise.
19235 (expand_vec_perm_blend): Likewise.
19236 (canonicalize_vector_int_perm): Likewise.
19237 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
19238 (expand_vec_perm_1): Likewise.
19239 (expand_vec_perm_interleave3): Likewise.
19240 (expand_vec_perm_even_odd_pack): Likewise.
19241 (expand_vec_perm_even_odd_1): Likewise.
19242 (expand_vec_perm_broadcast_1): Likewise.
19243 (ix86_vectorize_vec_perm_const_ok): Likewise.
19244 (ix86_expand_vecop_qihi): Likewise.
19245 (ix86_expand_mul_widen_hilo): Likewise.
19246 (ix86_expand_sse2_abs): Likewise.
19247 (ix86_expand_pextr): Likewise.
19248 (ix86_expand_pinsr): Likewise.
19249 (ix86_preferred_simd_mode): Likewise.
19250 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
19251 * config/i386/sse.md (*andnot<mode>3): Likewise.
19252 (<mask_codefor><code><mode>3<mask_name>): Likewise.
19253 (*<code><mode>3): Likewise.
19254 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
19255 (ia64_expand_atomic_op): Likewise.
19256 (ia64_arg_type): Likewise.
19257 (ia64_mode_to_int): Likewise.
19258 (ia64_scalar_mode_supported_p): Likewise.
19259 (ia64_vector_mode_supported_p): Likewise.
19260 (expand_vec_perm_broadcast): Likewise.
19261 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
19262 (iq2000_function_arg_advance): Likewise.
19263 (iq2000_function_arg): Likewise.
19264 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
19265 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
19266 (m68k_libcall_value): Likewise.
19267 (m68k_function_value): Likewise.
19268 (sched_attr_op_type): Likewise.
19269 * config/mcore/mcore.c (mcore_output_move): Likewise.
19270 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
19272 (microblaze_function_arg): Likewise.
19273 * config/mips/mips.c (mips16_build_call_stub): Likewise.
19274 (mips_print_operand): Likewise.
19275 (mips_mode_ok_for_mov_fmt_p): Likewise.
19276 (mips_vector_mode_supported_p): Likewise.
19277 (mips_preferred_simd_mode): Likewise.
19278 (mips_expand_vpc_loongson_even_odd): Likewise.
19279 (mips_expand_vec_unpack): Likewise.
19280 (mips_expand_vi_broadcast): Likewise.
19281 (mips_expand_vector_init): Likewise.
19282 (mips_expand_vec_reduc): Likewise.
19283 (mips_expand_msa_cmp): Likewise.
19284 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
19285 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
19286 (cc_flags_for_mode): Likewise.
19287 * config/msp430/msp430.c (msp430_print_operand): Likewise.
19288 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
19289 (nds32_output_casesi_pc_relative): Likewise.
19290 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19291 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
19292 (nvptx_gen_unpack): Likewise.
19293 (nvptx_gen_pack): Likewise.
19294 (nvptx_gen_shuffle): Likewise.
19295 (nvptx_gen_wcast): Likewise.
19296 (nvptx_preferred_simd_mode): Likewise.
19297 * config/pa/pa.c (pa_secondary_reload): Likewise.
19298 * config/pa/predicates.md (base14_operand): Likewise.
19299 * config/powerpcspe/powerpcspe-c.c
19300 (altivec_resolve_overloaded_builtin): Likewise.
19301 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
19303 (rs6000_preferred_simd_mode): Likewise.
19304 (output_vec_const_move): Likewise.
19305 (rs6000_expand_vector_extract): Likewise.
19306 (rs6000_split_vec_extract_var): Likewise.
19307 (reg_offset_addressing_ok_p): Likewise.
19308 (rs6000_legitimate_offset_address_p): Likewise.
19309 (rs6000_legitimize_address): Likewise.
19310 (rs6000_emit_set_const): Likewise.
19311 (rs6000_const_vec): Likewise.
19312 (rs6000_emit_move): Likewise.
19313 (spe_build_register_parallel): Likewise.
19314 (rs6000_darwin64_record_arg_recurse): Likewise.
19315 (swap_selector_for_mode): Likewise.
19316 (spe_init_builtins): Likewise.
19317 (paired_init_builtins): Likewise.
19318 (altivec_init_builtins): Likewise.
19319 (do_load_for_compare): Likewise.
19320 (rs6000_generate_compare): Likewise.
19321 (rs6000_expand_float128_convert): Likewise.
19322 (emit_load_locked): Likewise.
19323 (emit_store_conditional): Likewise.
19324 (rs6000_output_function_epilogue): Likewise.
19325 (rs6000_handle_altivec_attribute): Likewise.
19326 (rs6000_function_value): Likewise.
19327 (emit_fusion_gpr_load): Likewise.
19328 (emit_fusion_p9_load): Likewise.
19329 (emit_fusion_p9_store): Likewise.
19330 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
19331 (fusion_gpr_mem_load): Likewise.
19332 (fusion_addis_mem_combo_load): Likewise.
19333 (fusion_addis_mem_combo_store): Likewise.
19334 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
19335 (fusion_gpr_mem_load): Likewise.
19336 (fusion_addis_mem_combo_load): Likewise.
19337 (fusion_addis_mem_combo_store): Likewise.
19338 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19340 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
19341 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19342 (rs6000_preferred_simd_mode): Likewise.
19343 (output_vec_const_move): Likewise.
19344 (rs6000_expand_vector_extract): Likewise.
19345 (rs6000_split_vec_extract_var): Likewise.
19346 (reg_offset_addressing_ok_p): Likewise.
19347 (rs6000_legitimate_offset_address_p): Likewise.
19348 (rs6000_legitimize_address): Likewise.
19349 (rs6000_emit_set_const): Likewise.
19350 (rs6000_const_vec): Likewise.
19351 (rs6000_emit_move): Likewise.
19352 (rs6000_darwin64_record_arg_recurse): Likewise.
19353 (swap_selector_for_mode): Likewise.
19354 (paired_init_builtins): Likewise.
19355 (altivec_init_builtins): Likewise.
19356 (rs6000_expand_float128_convert): Likewise.
19357 (emit_load_locked): Likewise.
19358 (emit_store_conditional): Likewise.
19359 (rs6000_output_function_epilogue): Likewise.
19360 (rs6000_handle_altivec_attribute): Likewise.
19361 (rs6000_function_value): Likewise.
19362 (emit_fusion_gpr_load): Likewise.
19363 (emit_fusion_p9_load): Likewise.
19364 (emit_fusion_p9_store): Likewise.
19365 * config/rx/rx.c (rx_gen_move_template): Likewise.
19366 (flags_from_mode): Likewise.
19367 * config/s390/predicates.md (s390_alc_comparison): Likewise.
19368 (s390_slb_comparison): Likewise.
19369 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
19370 (s390_vector_mode_supported_p): Likewise.
19371 (s390_cc_modes_compatible): Likewise.
19372 (s390_match_ccmode_set): Likewise.
19373 (s390_canonicalize_comparison): Likewise.
19374 (s390_emit_compare_and_swap): Likewise.
19375 (s390_branch_condition_mask): Likewise.
19376 (s390_rtx_costs): Likewise.
19377 (s390_secondary_reload): Likewise.
19378 (__SECONDARY_RELOAD_CASE): Likewise.
19379 (s390_expand_cs): Likewise.
19380 (s390_preferred_simd_mode): Likewise.
19381 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
19382 * config/sh/sh.c (sh_print_operand): Likewise.
19383 (dump_table): Likewise.
19384 (sh_secondary_reload): Likewise.
19385 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19386 * config/sh/sh.md (casesi_worker_1): Likewise.
19387 (casesi_worker_2): Likewise.
19388 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
19389 (fcc_comparison_operator): Likewise.
19390 * config/sparc/sparc.c (sparc_expand_move): Likewise.
19391 (emit_soft_tfmode_cvt): Likewise.
19392 (sparc_preferred_simd_mode): Likewise.
19393 (output_cbranch): Likewise.
19394 (sparc_print_operand): Likewise.
19395 (sparc_expand_vec_perm_bmask): Likewise.
19396 (vector_init_bshuffle): Likewise.
19397 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
19398 (spu_vector_mode_supported_p): Likewise.
19399 (spu_expand_insv): Likewise.
19400 (spu_emit_branch_or_set): Likewise.
19401 (spu_handle_vector_attribute): Likewise.
19402 (spu_builtin_splats): Likewise.
19403 (spu_builtin_extract): Likewise.
19404 (spu_builtin_promote): Likewise.
19405 (spu_expand_sign_extend): Likewise.
19406 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
19407 (tilegx_simd_int): Likewise.
19408 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
19409 (tilepro_simd_int): Likewise.
19410 * config/v850/v850.c (const_double_split): Likewise.
19411 (v850_print_operand): Likewise.
19412 (ep_memory_offset): Likewise.
19413 * config/vax/vax.c (vax_rtx_costs): Likewise.
19414 (vax_output_int_move): Likewise.
19415 (vax_output_int_add): Likewise.
19416 (vax_output_int_subtract): Likewise.
19417 * config/visium/predicates.md (visium_branch_operator): Likewise.
19418 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
19419 (visium_print_operand_address): Likewise.
19420 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19421 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
19422 (xtensa_expand_conditional_branch): Likewise.
19423 (xtensa_copy_incoming_a7): Likewise.
19424 (xtensa_output_literal): Likewise.
19425 * dfp.c (decimal_real_maxval): Likewise.
19426 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19429 Alan Hayward <alan.hayward@arm.com>
19430 David Sherwood <david.sherwood@arm.com>
19432 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
19433 (mode_nunits_inline): Likewise.
19434 (mode_inner_inline): Likewise.
19435 (mode_unit_size_inline): Likewise.
19436 (mode_unit_precision_inline): Likewise.
19437 (emit_insn_modes_h): Likewise. Also emit a #define of the
19439 (emit_mode_wider): Add an E_ prefix to mode names.
19440 (emit_mode_complex): Likewise.
19441 (emit_mode_inner): Likewise.
19442 (emit_mode_adjustments): Likewise.
19443 (emit_mode_int_n): Likewise.
19444 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
19445 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
19446 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
19447 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
19448 (CRC32_BUILTIN, ENTRY): Likewise.
19449 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
19450 (aarch64_pop_regs): Likewise.
19451 (aarch64_process_components): Likewise.
19452 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
19453 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
19454 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
19455 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
19456 * config/arm/arm.c (arm_init_libfuncs): Likewise.
19457 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
19459 * config/i386/i386-builtin.def (pcmpestr): Likewise.
19460 (pcmpistr): Likewise.
19461 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
19462 * config/mmix/mmix.c (mmix_output_condition): Likewise.
19463 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19465 * config/rl78/rl78.c (mduc_regs): Likewise.
19466 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19467 (htm_expand_builtin): Likewise.
19468 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
19469 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
19470 * config/xtensa/xtensa.c (print_operand): Likewise.
19471 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
19472 (NUM_MODE_VECTOR_INT): Likewise.
19473 * genoutput.c (null_operand): Likewise.
19474 (output_operand_data): Likewise.
19475 * genrecog.c (print_parameter_value): Likewise.
19476 * lra.c (debug_operand_data): Likewise.
19478 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19480 * dwarf2out.c (scompare_loc_descriptor_wide)
19481 (scompare_loc_descriptor_narrow): New functions, split out from...
19482 (scompare_loc_descriptor): ...here.
19483 * expmed.c (emit_store_flag_int): New function, split out from...
19484 (emit_store_flag): ...here.
19486 2017-08-30 Richard Biener <rguenther@suse.de>
19488 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
19489 (dwarf2out_early_finish): Move setting of AT_pubnames from
19490 early debug output to early finish.
19492 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19494 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
19495 and -mdata-region to the assembler.
19497 2017-08-30 Richard Biener <rguenther@suse.de>
19499 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
19501 (gen_subprogram_die): Add DW_AT_object_pointer only early.
19502 (dwarf2out_early_global_decl): Only generate a DIE for the
19503 abstract origin if it doesn't already exist or is a declaration DIE.
19504 (resolve_addr): Do not add the linkage name twice when
19505 generating a stub DIE for the DW_TAG_GNU_call_site target.
19507 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19509 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19510 Use machine_mode rather than int for arg1_mode.
19512 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19515 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
19516 that the second argument of the built-in functions to unpack
19517 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
19518 switch statement instead a lot of if statements.
19519 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
19520 Allow 64-bit values to be in Altivec registers as well as
19521 traditional floating point registers.
19522 (pack<mode>, FMOVE128_VSX iterator): Likewise.
19524 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
19526 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
19527 MAX_REGS_PER_ADDRESS == 1.
19529 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
19531 * config/i386/i386.opt (flag_fentry): Do not init to -1.
19532 * config/i386/i386.c (ix86_option_override_internal): Simplify
19533 setting of opts->x_flag_entry.
19535 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19536 Jakub Jelinek <jakub@redhat.com>
19537 Richard Biener <rguenther@suse.de>
19539 PR tree-optimization/81503
19540 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
19541 folded constant fits in the target type; reorder tests for clarity.
19543 2017-08-29 Martin Liska <mliska@suse.cz>
19545 * passes.def: Include pass_lower_switch.
19546 * stmt.c (dump_case_nodes): Remove and move to
19547 tree-switch-conversion.
19548 (case_values_threshold): Likewise.
19549 (expand_switch_as_decision_tree_p): Likewise.
19550 (emit_case_decision_tree): Likewise.
19551 (expand_case): Likewise.
19552 (balance_case_nodes): Likewise.
19553 (node_has_low_bound): Likewise.
19554 (node_has_high_bound): Likewise.
19555 (node_is_bounded): Likewise.
19556 (emit_case_nodes): Likewise.
19557 (struct simple_case_node): New struct.
19558 (add_case_node): Remove.
19559 (emit_case_dispatch_table): Use vector instead of case_list.
19560 (reset_out_edges_aux): Remove.
19561 (compute_cases_per_edge): Likewise.
19562 (expand_case): Build list of simple_case_node.
19563 (expand_sjlj_dispatch_table): Use it.
19564 * tree-switch-conversion.c (struct case_node): Moved from
19565 stmt.c and adjusted.
19566 (emit_case_nodes): Likewise.
19567 (node_has_low_bound): Likewise.
19568 (node_has_high_bound): Likewise.
19569 (node_is_bounded): Likewise.
19570 (case_values_threshold): Likewise.
19571 (reset_out_edges_aux): Likewise.
19572 (compute_cases_per_edge): Likewise.
19573 (add_case_node): Likewise.
19574 (dump_case_nodes): Likewise.
19575 (balance_case_nodes): Likewise.
19576 (expand_switch_as_decision_tree_p): Likewise.
19577 (emit_jump): Likewise.
19578 (emit_case_decision_tree): Likewise.
19579 (try_switch_expansion): Likewise.
19580 (do_jump_if_equal): Likewise.
19581 (emit_cmp_and_jump_insns): Likewise.
19582 (fix_phi_operands_for_edge): New function.
19583 (record_phi_operand_mapping): Likewise.
19584 (class pass_lower_switch): New pass.
19585 (pass_lower_switch::execute): New function.
19586 (make_pass_lower_switch): Likewise.
19587 (conditional_probability):
19588 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19589 * tree-pass.h: Add make_pass_lower_switch.
19591 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19594 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19597 2017-08-29 Richard Biener <rguenther@suse.de>
19599 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19600 we do not add a DW_AT_inline attribute twice.
19601 (gen_subprogram_die): Remove code setting DW_AT_inline on
19602 DECL_ABSTRACT_P nodes.
19604 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19606 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19607 calls to internal functions.
19608 (gimplify_modify_expr): Likewise.
19609 * tree-call-cdce.c (use_internal_fn): Likewise.
19610 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19611 (convert_to_divmod): Set the nothrow flag.
19612 * tree-if-conv.c (predicate_mem_writes): Likewise.
19613 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19614 (vectorizable_call): Likewise.
19615 (vectorizable_store): Likewise.
19616 (vectorizable_load): Likewise.
19617 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19618 (vect_recog_mask_conversion_pattern): Likewise.
19620 2017-08-29 Martin Liska <mliska@suse.cz>
19623 * gcc.c (driver_handle_option): Add new argument.
19624 * opts-common.c (handle_option): Pass
19625 target_option_override_hook.
19626 * opts-global.c (lang_handle_option): Add new option.
19627 (set_default_handlers): Add new argument.
19628 (decode_options): Likewise.
19629 * opts.c (target_handle_option): Likewise.
19630 (common_handle_option): Call target_option_override_hook.
19631 * opts.h (struct cl_option_handler_func): Add hook for
19632 target option override.
19633 (struct cl_option_handlers): Likewise.
19634 (set_default_handlers): Add new argument.
19635 (decode_options): Likewise.
19636 (common_handle_option): Likewise.
19637 (target_handle_option): Likewise.
19638 * toplev.c (toplev::main): Pass targetm.target_option.override
19641 2017-08-29 Richard Biener <rguenther@suse.de>
19642 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19644 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19645 life to the active subtree.
19647 2017-08-28 Jeff Law <law@redhat.com>
19649 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19650 derive_equivalences.
19651 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19652 Code moved into....
19653 (edge_info::derive_equivalences): New private member function
19655 * tree-ssa-dom.c (class edge_info): Changed from a struct
19656 to a class. Add ctor/dtor, methods and data members.
19657 (edge_info::edge_info): Renamed from allocate_edge_info.
19658 Initialize additional members.
19659 (edge_info::~edge_info): New.
19660 (free_dom_edge_info): Delete the edge info.
19661 (record_edge_info): Use new class & associated member functions.
19662 Tighten forms for testing for edge equivalences.
19663 (record_temporary_equivalences): Iterate over the simple
19664 equivalences rather than assuming there's only one per edge.
19665 (cprop_into_successor_phis): Iterate over the simple
19666 equivalences rather than assuming there's only one per edge.
19667 (optimize_stmt): Use operand_equal_p rather than pointer
19668 equality for mini-DSE code.
19670 2017-08-28 Nathan Sidwell <nathan@acm.org>
19672 * gcc.c (execute): Fold SIGPIPE handling into switch
19673 statement. Adjust internal error message.
19675 2017-08-28 Richard Biener <rguenther@suse.de>
19678 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19679 Do nothing for removed DIEs.
19681 2017-08-28 Richard Biener <rguenther@suse.de>
19683 PR tree-optimization/81977
19684 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19687 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19690 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19691 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19692 using targetm.gen_mem_thread_fence.
19694 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19697 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19698 predicate to register_operand. Reorder operands.
19699 (*btr<mode>): Ditto.
19700 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19701 (*btr<mode>_mask): Ditto.
19703 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19705 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19706 * config/rs6000/xmmintrin.h: New file.
19707 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19709 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19712 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19713 parameter and_insn and return it.
19714 (recombine_lvx_pattern): Insert a copy to ensure availability of
19715 the base register of the copied masking operation at the point of
19716 the instruction replacement.
19717 (recombine_stvx_pattern): Likewise.
19719 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19721 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19722 undocumented switches.
19723 (-mpower9-dform-vector): Likewise.
19724 (-mpower9-dform): Likewise.
19725 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19726 comments to delete references to -mpower9-dform* switches.
19727 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19728 Delete reference to -mpower9-dform* switches, test for
19729 -mpower9-vector instead.
19730 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19731 (OTHER_P9_VECTOR_MASKS): Likewise.
19732 (POWERPC_MASKS): Likewise.
19733 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19734 tests against -mpower9-dform* to -mpower9-vector. Delete code
19735 that checked for -mpower9-dform* consistancy with other options.
19736 Add test for -mpower9-misc to enable other power9 switches.
19737 (rs6000_init_hard_regno_mode_ok): Likewise.
19738 (rs6000_option_override_internal): Likewise.
19739 (rs6000_emit_prologue): Likewise.
19740 (rs6000_emit_epilogue): Likewise.
19741 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19742 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19743 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19745 (emit_fusion_p9_store): Likewise.
19746 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19747 resetting these macros if the assembler does not support ISA 3.0
19749 (TARGET_P9_DFORM_VECTOR): Likewise.
19750 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19751 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19753 2017-08-25 Alan Modra <amodra@gmail.com>
19755 PR rtl-optimization/81747
19756 * cse.c (cse_extended_basic_block): Don't attempt to record
19757 equivalences for degenerate conditional branches that branch
19758 to their fall-through.
19760 2017-08-24 Martin Sebor <msebor@redhat.com>
19762 PR middle-end/81908
19763 * gimple-fold.c (size_must_be_zero_p): New function.
19764 (gimple_fold_builtin_memory_op): Call it.
19766 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19768 * config/rs6000/mm_malloc.h: New file.
19770 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19772 PR tree-optimization/81913
19773 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19774 analysis when either IVs in condition can wrap.
19776 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19778 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19779 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19781 2017-08-24 Richard Biener <rguenther@suse.de>
19784 * targhooks.c (default_target_can_inline_p): Properly
19785 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19786 is present and always compare.
19787 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19788 infer -mfpmath=sse from TARGET_SSE_P.
19789 (ix86_can_inline_p): Properly use target_option_default_node when
19790 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19792 2017-08-24 Richard Biener <rguenther@suse.de>
19795 * dwarf2out.c (output_die): Handle flag_generate_offload like
19797 (output_comp_unit): Likewise.
19798 (gen_array_type_die): Likewise.
19799 (dwarf2out_early_finish): Likewise.
19800 (note_variable_value_in_expr): Likewise.
19801 (dwarf2out_finish): Likewise. Adjust assert.
19802 * cgraphunit.c (symbol_table::compile): Move setting of
19803 flag_generate_offload earlier ...
19804 (symbol_table::finalize_compilation_unit): ... here, before
19805 early debug finalization.
19807 2017-08-24 Richard Biener <rguenther@suse.de>
19809 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19810 and ipa-fnsummary.h.
19811 (ix86_can_inline_p): When ix86_fpmath flags do not match
19812 check whether the callee uses FP math at all.
19814 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19816 PR middle-end/81931
19817 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19818 instead of TYPE_PRECISION.
19820 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19821 Alan Hayward <alan.hayward@arm.com>
19822 David Sherwood <david.sherwood@arm.com>
19824 * combine.c (make_extraction): Use subreg_offset_from_lsb.
19826 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
19828 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
19830 (ix86_frame::stack_realign_allocate): New field.
19831 (struct machine_frame_state): Modify comments.
19832 (machine_frame_state::sp_realigned_fp_end): New field.
19833 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
19834 layout calculation.
19835 (sp_valid_at): Add assertion to assure no attempt to access invalid
19836 offset of a realigned stack.
19837 (fp_valid_at): Likewise.
19838 (choose_baseaddr): Modify comments.
19839 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
19840 ix86_expand_prologue.
19841 (ix86_expand_prologue): Modify stack realignment and allocation.
19842 (ix86_expand_epilogue): Modify comments.
19843 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
19844 avx2_runtime, avx512f, and avx512f_runtime.
19846 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
19848 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
19849 (mstackrealign): Do not init to -1.
19850 * config/i386/i386.c (ix86_option_override_internal):
19851 Check opts_set, not opts when setting default value of
19852 opts->x_ix86_force_align_arg_pointer.
19854 2017-08-23 Richard Biener <rguenther@suse.de>
19856 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
19857 lang_hooks.decl_printable_name.
19858 * print-rtl-function.c (print_rtx_function): Likewise.
19859 * tree-pretty-print.c (dump_function_header): Likewise.
19861 2017-08-23 Richard Biener <rguenther@suse.de>
19864 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
19865 -g0 at compile-time.
19867 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19869 PR middle-end/19706
19870 * doc/sourcebuild.texi (Other hardware attributes):
19873 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19875 PR middle-end/19706
19876 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
19877 Fix single-use check.
19879 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
19881 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
19883 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19885 * doc/install.texi: Modify to add more details on running selected
19888 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19890 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
19891 is combined with -mabi=ms.
19892 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
19895 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19897 PR tree-optimization/81488
19898 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
19899 and cached_basis fields.
19900 (MAX_SPREAD): New constant.
19901 (alloc_cand_and_find_basis): Initialize new fields.
19902 (clear_visited): New function.
19903 (create_phi_basis_1): Rename from create_phi_basis, set visited
19904 and cached_basis fields.
19905 (create_phi_basis): New wrapper function.
19906 (phi_add_costs_1): Rename from phi_add_costs, add spread
19907 parameter, set visited field, short-circuit when limits reached.
19908 (phi_add_costs): New wrapper function.
19909 (record_phi_increments_1): Rename from record_phi_increments, set
19911 (record_phi_increments): New wrapper function.
19912 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
19913 (phi_incr_cost): New wrapper function.
19914 (all_phi_incrs_profitable_1): Rename from
19915 all_phi_incrs_profitable, set visited field.
19916 (all_phi_incrs_profitable): New wrapper function.
19918 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19919 Alan Hayward <alan.hayward@arm.com>
19920 David Sherwood <david.sherwood@arm.com>
19922 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
19923 that takes the outer and inner modes.
19924 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
19925 comparison as the canonical test for a paradoxical subreg.
19926 * combine.c (simplify_set): Use paradoxical_subreg_p.
19927 (make_extraction): Likewise.
19928 (force_to_mode): Likewise.
19929 (rtx_equal_for_field_assignment_p): Likewise.
19930 (gen_lowpart_for_combine): Likewise.
19931 (simplify_comparison): Likewise.
19932 * cse.c (equiv_constant): Likewise.
19933 * expmed.c (store_bit_field_1): Likewise.
19934 * final.c (alter_subreg): Likewise.
19935 * fwprop.c (propagate_rtx): Likewise.
19936 (forward_propagate_subreg): Likewise.
19937 * ira-conflicts.c (ira_build_conflicts): Likewise.
19938 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
19939 * lra-constraints.c (curr_insn_transform): Likewise.
19940 (split_reg): Likewise.
19941 * lra-eliminations.c (move_plus_up): Likewise.
19942 (lra_eliminate_regs_1): Likewise.
19943 * recog.c (general_operand): Likewise.
19944 * ree.c (combine_reaching_defs): Likewise.
19945 * reload.c (push_reload): Likewise.
19946 (find_reloads): Likewise.
19947 * reload1.c (elimination_effects): Likewise.
19948 (compute_reload_subreg_offset): Likewise.
19949 (choose_reload_regs): Likewise.
19950 * rtlanal.c (subreg_lsb_1): Likewise.
19951 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19952 (simplify_subreg): Likewise.
19953 * var-tracking.c (track_loc_p): Likewise.
19954 * emit-rtl.c (byte_lowpart_offset): Likewise.
19955 (paradoxical_subreg_p): Delete out-of-line definition.
19957 2017-08-22 Jeff Law <law@redhat.com>
19959 PR tree-optimization/81741
19960 PR tree-optimization/71947
19961 * tree-ssa-dom.c: Include tree-inline.h.
19962 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
19963 equivalences if one is more expensive to compute than the other.
19964 * tree-ssa-scopedtables.h (class const_or_copies): Make
19965 record_const_or_copy_raw method private.
19966 (class avail_exprs_stack): New method simplify_binary_operation.
19967 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
19968 avail_exprs_stack::simplify_binary_operation as needed.
19969 (avail_exprs_stack::simplify_binary_operation): New function.
19971 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
19973 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
19974 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
19975 (DOT_SYMBOLS): Likewise.
19976 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
19977 (RELOCATABLE_NEEDS_FIXUP): Likewise.
19978 (RS6000_ABI_NAME): Likewise.
19979 (TARGET_CMODEL): Likewise.
19980 (TOC_SECTION_ASM_OP): Likewise.
19981 (SET_CMODEL): New macro.
19982 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
19984 2017-08-22 Richard Biener <rguenther@suse.de>
19986 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
19987 to allow for free-lang-data replacements similar to verify_type_variant.
19989 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19990 Alan Hayward <alan.hayward@arm.com>
19991 David Sherwood <david.sherwood@arm.com>
19993 * config/aarch64/aarch64.md (casesi): Use DImode rather than
19994 VOIDmode for the LABEL_REF.
19996 2017-08-22 Richard Biener <rguenther@suse.de>
19998 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20000 2017-08-22 Richard Biener <rguenther@suse.de>
20002 * common.opt (feliminate-dwarf2-dups): Ignore.
20003 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20004 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20005 same_die_p_wrap, compute_section_prefix,
20006 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20007 (comdat_symbol_id, comdat_symbol_number): Likewise.
20008 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20010 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20011 (output_die): Mark unreachable path unreachable.
20012 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20013 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20014 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20015 (dwarf2out_early_finish): Likewise.
20017 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20019 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20021 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20024 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20025 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20026 (avr_attribute_table) <io, io_low, address>: Initialize
20027 .decl_required with true.
20029 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20031 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20032 undocumented debugging options.
20033 (-mvsx-scalar-double): Likewise.
20034 (-mallow-df-permute): Likewise.
20035 (-mvectorize-builtins): Likewise.
20036 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20037 (rs6000_builtin_vectorized_function): Likewise.
20038 (rs6000_builtin_md_vectorized_function): Likewise.
20039 (rs6000_opt_vars): Likewise.
20041 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20044 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20045 (*btrq_imm): Rename from *btrq.
20046 (*btcq_imm): Rename from *btcq.
20047 (btsc): New code attribute.
20048 (*<btsc><mode>): New insn pattern.
20049 (*btr<mode>): Ditto.
20050 (*<btsc><mode>_mask): New insn_and_split pattern.
20051 (*btr<mode>_mask): Ditto.
20053 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20054 Alan Hayward <alan.hayward@arm.com>
20055 David Sherwood <david.sherwood@arm.com>
20057 * function.c (pad_below): Simplify padding calculation.
20059 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20060 Alan Hayward <alan.hayward@arm.com>
20061 David Sherwood <david.sherwood@arm.com>
20063 * target.def (function_prologue): Remove frame size argument.
20064 (function_epilogue): Likewise.
20065 * doc/tm.texi: Regenerate.
20066 * final.c (final_start_function): Update call to function_prologue.
20067 (final_end_function): Update call to function_epilogue.
20068 (default_function_pro_epilogue): Remove frame size argument.
20069 * output.h (default_function_pro_epilogue): Likewise.
20070 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
20071 (arm_output_function_prologue): Likewise.
20072 * config/frv/frv.c (frv_function_prologue): Likewise.
20073 (frv_function_epilogue): Likewise.
20074 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
20075 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
20076 (ia64_output_function_epilogue): Likewise.
20077 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
20078 (m32r_output_function_epilogue): Likewise.
20079 * config/microblaze/microblaze.c (microblaze_function_prologue)
20080 (microblaze_function_epilogue): Likewise.
20081 * config/mips/mips.c (mips_output_function_prologue): Likewise.
20082 (mips_output_function_epilogue): Likewise.
20083 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
20084 (mmix_target_asm_function_epilogue): Likewise.
20085 * config/msp430/msp430.c (msp430_start_function): Likewise.
20086 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
20087 (nds32_asm_function_epilogue): Likewise.
20088 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
20089 * config/pa/pa.c (pa_output_function_prologue): Likewise.
20090 (pa_output_function_epilogue): Likewise.
20091 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
20092 (rs6000_output_function_epilogue): Likewise.
20093 * config/rl78/rl78.c (rl78_start_function): Likewise.
20094 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
20095 (rs6000_output_function_epilogue): Likewise.
20096 * config/rx/rx.c (rx_output_function_prologue): Likewise.
20097 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
20098 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
20099 (sparc_asm_function_epilogue): Likewise.
20101 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20103 * tree.h (type_has_mode_precision_p): New function.
20104 * convert.c (convert_to_integer_1): Use it.
20105 * expr.c (expand_expr_real_2): Likewise.
20106 (expand_expr_real_1): Likewise.
20107 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
20108 * match.pd: Likewise.
20109 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
20110 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
20111 * tree-tailcall.c (process_assignment): Likewise.
20112 * tree-vect-loop.c (vectorizable_reduction): Likewise.
20113 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
20114 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
20115 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
20116 (vectorizable_assignment): Likewise.
20117 (vectorizable_shift): Likewise.
20118 (vectorizable_operation): Likewise.
20119 * tree-vrp.c (register_edge_assert_for_2): Likewise.
20121 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
20123 * match.pd: Add pow (C, x) simplification.
20125 2017-08-21 Richard Biener <rguenther@suse.de>
20127 PR tree-optimization/81900
20128 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
20129 for blocks with abnormal predecessors.
20130 (compute_antic): Do not set visited flag prematurely.
20132 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
20135 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
20137 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20139 * stor-layout.h (vector_type_mode): Move to...
20140 * tree.h (vector_type_mode): ...here.
20141 * stor-layout.c (vector_type_mode): Move to...
20142 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
20144 2017-08-21 Richard Biener <rguenther@suse.de>
20146 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
20147 register_external_die hooks.
20148 (debug_false_tree_charstarstar_uhwistar): Declare.
20149 (debug_nothing_tree_charstar_uhwi): Likewise.
20150 * debug.c (do_nothing_debug_hooks): Adjust.
20151 (debug_false_tree_charstarstar_uhwistar): New do nothing.
20152 (debug_nothing_tree_charstar_uhwi): Likewise.
20153 * dbxout.c (dbx_debug_hooks): Adjust.
20154 (xcoff_debug_hooks): Likewise.
20155 * sdbout.c (sdb_debug_hooks): Likewise.
20156 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20157 * dwarf2out.c (macinfo_label_base): New global.
20158 (dwarf2out_register_external_die): New function for the
20159 register_external_die hook.
20160 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
20161 (dwarf2_debug_hooks): Use them.
20162 (dwarf2_lineno_debug_hooks): Adjust.
20163 (struct die_struct): Add with_offset flag.
20164 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
20165 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
20166 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
20167 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
20168 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
20169 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
20170 defining section names for the early LTO debug variants.
20171 (reset_indirect_string): New helper.
20172 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
20173 (print_dw_val): Add support for offsetted symbol references.
20174 (get_ultimate_context): Split out from is_cxx.
20175 (is_cxx): Use get_ultimate_context.
20176 (is_fortran): Add decl overload.
20177 (compute_comp_unit_symbol): Split out worker from
20178 compute_section_prefix.
20179 (compute_section_prefix): Call compute_comp_unit_symbol and
20180 set comdat_type_p here.
20181 (output_die): Skip DIE symbol output for the LTO added one.
20182 Handle DIE symbol references with offset.
20183 (output_comp_unit): Guard section name mangling properly.
20184 For LTO debug sections emit a symbol at the section beginning
20185 which we use to refer to its DIEs.
20186 (add_abstract_origin_attribute): For DIEs registered via
20187 dwarf2out_register_external_die directly refer to the early
20188 DIE rather than indirectly through the shadow one we created.
20189 Remove obsolete call to dwarf2out_abstract_function for
20190 non-function/block origins.
20191 (gen_array_type_die): When generating early LTO debug do
20192 not emit DW_AT_string_length.
20193 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
20194 late when in LTO. As suggested place a gcc_unreachable for
20195 the DECL_ABSTRACT_P case.
20196 (gen_subprogram_die): Avoid another specification DIE
20197 for early built declarations/definitions for the late LTO case.
20198 (gen_variable_die): Add type references for late duplicated VLA dies
20200 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
20201 we have the abstract instance already.
20202 (process_scope_var): Adjust decl DIE contexts in LTO which
20203 first puts them in limbo.
20204 (gen_decl_die): Do not generate type DIEs late apart from
20205 types for VLAs or for decls we do not yet have a DIE. Do not
20206 call dwarf2out_abstract_function late.
20207 (dwarf2out_early_global_decl): Make sure to create DIEs
20208 for abstract instances of a decl first.
20209 (dwarf2out_late_global_decl): Adjust comment.
20210 (output_macinfo_op): With multiple macro sections use
20211 macinfo_label_base to distinguish labels.
20212 (output_macinfo): Likewise. Update macinfo_label_base.
20213 Pass in the line info label.
20214 (note_variable_value_in_expr): When generating LTO resolve
20215 all variable values here by generating DIEs as needed.
20216 (init_sections_and_labels): Add early LTO debug flag parameter
20217 and generate different sections and names if set. Add generation
20218 counter for the labels so we can have multiple of them.
20219 (reset_dies): Helper to allow DIEs to be output multiple times.
20220 (dwarf2out_finish): When outputting DIEs to the fat part of an
20221 LTO object first reset DIEs.
20222 (dwarf2out_early_finish): Output early DIEs when generating LTO.
20223 (modified_type_die): Check for decl_ultimate_origin being self
20225 (gen_type_die_with_usage): Likewise.
20226 (gen_typedef_die): Allow decl_ultimate_origin being self.
20227 (set_decl_abstract_flags): Remove.
20228 (set_block_abstract_flags): Likewise.
20229 (dwarf2out_abstract_function): Treat the early generated DIEs
20230 as the abstract copy and only add DW_AT_inline and
20231 DW_AT_artificial here and call set_decl_origin_self.
20232 If the DIE has an abstract origin don't do anything.
20233 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
20234 if we have none yet (Go fails to build one, PR78628).
20235 (variably_modified_type_p): Prevent endless recursion for Ada
20236 cyclic pointer types.
20237 * lto-streamer-in.c: Include debug.h.
20238 (dref_queue): New global.
20239 (lto_read_tree_1): Stream in DIE references.
20240 (lto_input_tree): Register DIE references.
20241 (input_function): Stream DECL_DEBUG_ARGS.
20242 * lto-streamer-out.c: Include debug.h.
20243 (lto_write_tree_1): Output DIE references.
20244 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
20245 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
20246 (output_function): Stream DECL_DEBUG_ARGS.
20247 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
20248 Stream DECL_ABSTRACT_ORIGIN.
20249 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
20250 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
20251 DECL_CONTEXT for file-scope decls.
20252 * lto-streamer.h (struct dref_entry): Declare.
20253 (dref_queue): Likewise.
20254 * cfgexpand.c (pass_expand::execute): Do not call the
20255 outlining_inline_function hook here.
20256 * lto-wrapper.c (debug_obj): New global.
20257 (tool_cleanup): Unlink it if required.
20258 (debug_objcopy): New function.
20259 (run_gcc): Handle early debug sections in the IL files by
20260 extracting them to separate files, partially linkin them and
20261 feeding the result back as result to the linker.
20262 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
20263 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
20264 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
20265 sections into a separate segment.
20266 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
20268 (darwin_asm_dwarf_section): Likewise.
20269 (darwin_asm_output_dwarf_offset): Likewise.
20270 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
20272 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20273 Alan Hayward <alan.hayward@arm.com>
20274 David Sherwood <david.sherwood@arm.com>
20276 * read-md.h (md_reader::record_potential_iterator_use): Replace
20277 pointer argument with an rtx and an index.
20278 * read-rtl.c (iterator_group::apply_iterator): Likewise.
20279 (apply_mode_iterator): Likewise.
20280 (apply_code_iterator): Likewise.
20281 (apply_int_iterator): Likewise.
20282 (apply_subst_iterator): Likewise.
20283 (record_iterator_use): Likewise.
20284 (record_attribute_use): Likewise.
20285 (md_reader::record_potential_iterator_use): Likewise. Update calls
20286 to record_iterator_use and apply_iterator.
20287 (iterator_use): Replace ptr with x and index.
20288 (attribute_use): Likewise.
20289 (apply_attribute_uses): Update calls to apply_iterator.
20290 (apply_iterators): Likewise. Update initialization of iterator_use.
20291 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
20292 and record_potential_iterator_use.
20293 (rtx_reader::read_rtx_operand): Likewise.
20295 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20296 Alan Hayward <alan.hayward@arm.com>
20297 David Sherwood <david.sherwood@arm.com>
20299 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
20302 2017-08-21 Richard Biener <rguenther@suse.de>
20304 PR middle-end/81884
20305 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
20306 at struct end conservatively when comparing common bases.
20308 2017-08-21 Richard Biener <rguenther@suse.de>
20310 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
20311 (mem_ref_in_stmt): Remove.
20312 (determine_max_movement): Use ref index to get at the reference.
20313 (invariantness_dom_walker::before_dom_children): Deal with
20314 lim data already initialized.
20315 (gather_mem_refs_stmt): Initialize lim data and record ref index.
20317 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20319 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
20320 (TARGET_ISA_ROUND): Ditto.
20321 (TARGET_ROUND): Ditto.
20322 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
20323 * config/i386/i386.md: Ditto.
20324 * config/i386/sse.md: Ditto.
20325 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
20326 with OPTION_MASK_ISA_SSE4_1.
20328 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20331 * doc/extend.texi (x86 Built-in Functions): Correct the name of
20332 __builtin_ia32_lzcnt_u16.
20334 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
20337 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
20338 (rs6000_set_current_function): Rewrite function to use it.
20340 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
20343 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
20344 and TYPE_WARN_IF_NOT_ALIGN.
20345 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
20346 (handle_warn_if_not_align): New.
20347 (place_union_field): Call handle_warn_if_not_align.
20348 (place_field): Call handle_warn_if_not_align.
20349 Copy TYPE_WARN_IF_NOT_ALIGN.
20350 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
20351 (layout_type): Likewise.
20352 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
20354 (tree_decl_common): Add warn_if_not_align.
20355 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
20356 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
20357 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
20358 (DECL_WARN_IF_NOT_ALIGN): Likewise.
20359 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
20360 * doc/extend.texi: Document warn_if_not_aligned attribute.
20361 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
20363 2017-08-17 Martin Liska <mliska@suse.cz>
20366 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
20367 (get_data_dependence): Use it as pointer type.
20368 (distribute_loop): Likewise.
20370 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20372 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
20373 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
20374 (p8_vmrgow_<mode>_direct): New define_insn.
20375 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
20376 handle endianness for vmrgew and vmrgow permute patterns.
20378 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20380 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
20381 * config/rs6000/rs6000-cpus.def: Remove comment.
20382 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
20383 (POWERPC_MASKS): Likewise.
20384 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
20385 use of TARGET_VSX_TIMODE.
20386 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
20387 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
20388 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
20389 (rs6000_option_override_internal): Remove dead code.
20390 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
20391 (rs6000_legitimize_reload_address): Likewise.
20392 (rs6000_legitimate_address_p): Likewise.
20393 (rs6000_opt_masks): Delete "vsx-timode".
20394 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
20395 from function comment.
20396 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
20397 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
20398 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
20400 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
20401 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
20402 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
20403 (VSX_TI): Likewise.
20405 (define_peephole2): Likewise.
20407 2017-08-17 Martin Sebor <msebor@redhat.com>
20410 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
20411 past the end of an array.
20412 (test_pp_format): Add test cases.
20414 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
20416 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
20417 missing ECF_NOTHROW flags.
20419 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20422 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
20423 operands residing in integer registers.
20424 (*vsx_le_perm_load_<mode>): Likewise.
20425 (*vsx_le_perm_store_<mode>): Likewise.
20426 (define_peephole2): Add peepholes to optimize the above.
20428 2017-08-17 Marek Polacek <polacek@redhat.com>
20430 PR middle-end/81814
20431 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
20432 to mimic what shorten_compare did. Change the return type to bool.
20433 (fold_cond_expr_with_comparison): Update call to
20434 operand_equal_for_comparison_p.
20435 (fold_ternary_loc): Likewise.
20437 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
20439 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
20441 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
20443 2017-08-17 Richard Biener <rguenther@suse.de>
20445 * tree-ssa-structalias.c (solve_graph): When propagating
20446 to successors update the graphs succ edges and avoid duplicate work.
20448 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
20451 * config/i386/i386.c (ix86_option_override_internal): Save target
20452 specific options after ix86_stack_protector_guard_reg was changed.
20454 2017-08-17 Richard Biener <rguenther@suse.de>
20456 PR tree-optimization/81827
20457 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
20458 (new_var_info): Initialize it conservatively.
20459 (get_call_vi): Mark register vars.
20460 (new_scalar_tmp_constraint_exp): Likewise.
20461 (handle_rhs_call): Likewise.
20462 (handle_const_call): Likewise.
20463 (create_function_info_for): Likewise.
20464 (solve_constraints): Sort varinfos to separate register from
20465 non-register vars to pack points-to solution bitmaps during
20468 2017-08-17 Marek Polacek <polacek@redhat.com>
20470 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
20472 2017-08-17 Richard Biener <rguenther@suse.de>
20474 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
20475 to true when overflow is undefined and we saturated the result.
20477 2017-08-17 Alan Modra <amodra@gmail.com>
20480 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
20481 Don't use store multiple if only one reg needs saving.
20482 (interesting_frame_related_regno): New function.
20483 (rs6000_frame_related): Don't emit frame info for regs that
20485 (rs6000_emit_epilogue): Likewise.
20487 2017-08-16 Nathan Sidwell <nathan@acm.org>
20489 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
20490 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
20491 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
20492 * tree.c (free_lang_data_in_type): Use else-if chain. Always
20493 clear TYPE_LANG_1. Remove obsolete member-function stripping.
20494 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
20495 (verify_type): Adjust for TYPE_BINFO move.
20496 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
20497 process TYPE_BINFO directly.
20498 (hash_tree): Likewise.
20499 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20501 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20504 2017-08-16 David Malcolm <dmalcolm@redhat.com>
20506 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
20508 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
20511 * config/i386/i386.md (*anddi_1_btr): Change predicates of
20512 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
20513 Add ix86_binary_operator_ok to insn constraint.
20514 (*iordi_1_bts): Ditto.
20515 (*xordi_1_btc): Ditto.
20516 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
20517 Update corresponding peephole2 pattern.
20521 2017-08-16 Bin Cheng <bin.cheng@arm.com>
20523 PR tree-optimization/81832
20524 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
20525 copy loop header which has IFN_LOOP_DIST_ALIAS call.
20527 2017-08-16 Marek Polacek <polacek@redhat.com>
20530 * fold-const.c (fold_indirect_ref_1): Restore original behavior
20531 regarding size_zero_node.
20533 2017-08-16 Martin Liska <mliska@suse.cz>
20536 * config.gcc: Respect previously set extra_objs in case
20539 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
20541 PR tree-optimization/81835
20542 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
20543 the phi SSA_NAME. Check that the condition in a COND_EXPR does
20544 not depend on the phi.
20546 2017-08-16 Alan Modra <amodra@gmail.com>
20548 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
20551 2017-08-16 Alan Modra <amodra@gmail.com>
20553 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
20554 (save_reg_p): ..into this. Update all callers.
20555 (first_reg_to_save): Simplify.
20557 2017-08-16 Alan Modra <amodra@gmail.com>
20559 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
20562 2017-08-15 Joseph Myers <joseph@codesourcery.com>
20566 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
20567 constant count if that count is less than 32.
20569 2017-08-15 Nathan Sidwell <nathan@acm.org>
20571 * gcc.c (execute): Emit friendlier message if inferior is killed
20572 by an external cause.
20574 2017-08-15 Richard Biener <rguenther@suse.de>
20576 PR tree-optimization/81790
20577 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20578 CONSTRUCTORs from simplifying and VN.
20580 2017-08-14 Martin Sebor <msebor@redhat.com>
20582 * builtin-attrs.def: Add comments.
20584 2017-08-14 Martin Sebor <msebor@redhat.com>
20587 * doc/extend.texi (attribute nonstring): Document new attribute.
20589 2017-08-14 Martin Sebor <msebor@redhat.com>
20592 * tree-diagnostic.c (default_tree_printer): Handle %G.
20593 * gimple-pretty-print.h (percent_G_format): Declare new function.
20594 * gimple-pretty-print.c (percent_G_format): Define.
20595 * tree-pretty-print.c (percent_K_format): Add argument.
20597 2017-08-14 Martin Sebor <msebor@redhat.com>
20599 PR translation/79998
20600 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20601 Remove a stray space.
20603 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20606 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20607 (*iordi_1_bts): Ditto.
20608 (*xordi_1_btc): Ditto.
20610 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20613 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20614 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20616 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20617 quoted strings, and make more translator-friendly.
20618 (darwin_rs6000_override_options): Likewise.
20619 (rs6000_option_override_internal): Likewise.
20620 (rs6000_return_in_memory): Fix overlong line.
20621 (init_cmulative_args): Use quoted strings, and make more
20622 translator-friendly.
20623 (rs6000_pass_by_reference): Fix overlong line.
20624 (def_builtin): Use quoted strings.
20625 (altivec_expand_predicate_builtin): Use quoted strings, and make
20626 more translator-friendly.
20627 (htm_expand_builtin): Use quoted strings.
20628 (cpu_expand_builtin): Use quoted strings, and make more
20629 translator-friendly.
20630 (altivec_expand_builtin): Likewise.
20631 (paired_expand_predicate_builtin): Likewise.
20632 (rs6000_invalid_builtin): Likewise.
20633 (builtin_function_type): Use quoted strings.
20634 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20635 more translator-friendly.
20636 (rs6000_trampoline_init): Likewise.
20637 (rs6000_handle_altivec_attribute): Likewise.
20638 (rs6000_inner_target_options): Use quoted strings.
20639 (rs6000_disable_incompatible_switches): Likewise.
20640 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20641 strings, and make more translator-friendly.
20642 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20644 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20646 PR tree-optimization/81799
20647 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20648 cond_expr to simple gimple operand.
20650 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20652 PR middle-end/46932
20653 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20655 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20659 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20660 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20661 TARGET_GASISR_PROLOGUES.
20662 * config/avr/avr.c (avr_option_override): Same.
20663 (avr_pass_pre_proep::execute): Same.
20665 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20668 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20669 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20671 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20673 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20674 define_insn_and_split. Split to a memory load from 0 in
20675 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20676 using PTR mode iterator.
20677 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20678 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20679 (*add_tp_<mode>): Redefine as define_insn_and_split.
20680 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20681 address space. Merge with *add_tp_x32 using PTR mode iterator.
20682 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20683 Split to an add with a memory load from 0 in
20684 DEFAULT_TLS_SEG_REG address space.
20686 2017-08-12 Andrew Pinski <apinski@cavium.com>
20688 * config/aarch64/aarch64-option-extensions.def (rdma):
20689 Fix feature string to what Linux prints out in /proc/cpuinfo.
20691 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20694 * dwarf2out.c (modified_type_die): For C typedef types that have
20695 an ultimate origin, process the ultimate origin instead of the
20697 (gen_typedef_die): Assert that input DECLs have no ultimate
20699 (gen_type_die_with_usage): For typedef variants that have an
20700 ultimate origin, just call gen_decl_die on the original DECL.
20701 (process_scope_var): Avoid creating DIEs for local typedefs and
20702 concrete static variables.
20704 2017-08-12 Alan Modra <amodra@gmail.com>
20708 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20709 match gnu-user.h startfile.
20710 (ENDFILE_LINUX_SPEC): Similarly.
20712 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20715 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20717 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20719 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20720 * config/aarch64/aarch64.md (mov<mode>): Change.
20721 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20722 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20723 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20725 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20727 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20728 for storage order barriers.
20730 2017-08-11 Martin Liska <mliska@suse.cz>
20733 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20734 variables of void type.
20736 2017-08-11 Martin Liska <mliska@suse.cz>
20738 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20739 TARGET_SUPPORTS_ALIASES.
20740 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20741 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20742 (optimize_weakref): Likewise.
20743 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20744 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20745 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20747 2017-08-11 Martin Liska <mliska@suse.cz>
20750 * config/i386/i386.c (make_resolver_func): Do complete
20751 refactoring of the function.
20753 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20756 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20757 * config/i386/i386.c (ix86_stack_protect_guard): Use
20758 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20759 * doc/invoke.texi (x86 Options): Document
20760 -mstack-protector-guard-symbol= option.
20762 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20764 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20765 * config/i386/i386.c (ix86_split_stack_guard): New function.
20766 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20767 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20768 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20769 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20770 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20771 (split_stack_space_check): Call ix86_split_stack_guard.
20773 2017-08-10 Martin Sebor <msebor@redhat.com>
20775 * print-tree.c (print_node): Print location using the established
20777 Replace spaces with colons.
20778 (debug_raw, debug): Ditto.
20780 2017-08-10 Martin Sebor <msebor@redhat.com>
20783 * pretty-print.c (pp_format): Correct the handling of %s precision.
20785 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20788 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20790 (ix86_finalize_stack_frame_flags): This. Also clear
20791 frame_pointer_needed if -fno-omit-frame-pointer is used without
20793 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20794 with ix86_finalize_stack_frame_flags.
20795 (ix86_expand_epilogue): Likewise.
20796 (ix86_expand_split_stack_prologue): Likewise.
20797 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20799 2017-08-10 Martin Liska <mliska@suse.cz>
20802 * c-attribs.c (handle_target_attribute):
20803 Report warning for an empty string argument of target attribute.
20805 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20808 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20810 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20811 or DECL_NONLOCAL labels.
20812 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20813 or DECL_NONLOCAL labels here.
20815 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20817 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20818 to indicate when early gimple folding has been disabled.
20819 (rs6000_gimple_fold_builtin): Add debug content.
20820 (rs6000_invalid_builtin): Fix whitespace.
20821 (rs6000_expand_builtin): Fix whitespace.
20822 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
20824 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
20827 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
20828 SAVE_MULTIPLE if not all the registers that saves, should be saved.
20830 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
20832 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
20833 (qdf24xx): Likewise.
20834 * config/aarch64/aarch64.md: Include falkor.md.
20835 * config/aarch64/falkor.md: New.
20837 2017-08-09 Marek Polacek <polacek@redhat.com>
20840 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
20841 * diagnostic.c (emit_diagnostic): Add a comment.
20842 (emit_diagnostic_valist): New function.
20844 2017-08-09 Marek Polacek <polacek@redhat.com>
20847 * input.c (make_location): New overload.
20848 * input.h (make_location): Declare.
20850 2017-08-08 Alan Modra <amodra@gmail.com>
20851 H.J. Lu <hongjiu.lu@intel.com>
20854 * gcc.c (NO_PIE_SPEC): Delete.
20855 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
20857 (LINK_PIE_SPEC): ..to here.
20858 (LINK_COMMAND_SPEC): Support -no-pie.
20859 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
20860 chain of crtbegin*.o selection, update for PIE_SPEC changes and
20862 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
20863 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
20864 (ENDFILE_CRTEND_SPEC): Similarly.
20866 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
20869 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
20870 (mstack-protector-guard-offset=): Ditto.
20871 * config/i386/i386.c (ix86_option_override): Handle
20872 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
20874 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
20875 ix86_stack_protect_guard_offset variables.
20876 (TARGET_STACK_PROTECT_GUARD): Always define.
20877 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
20878 and -mstack-protector-guard-offset= options.
20880 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20882 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
20883 boundary case for the last candidate.
20885 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20887 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
20888 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
20890 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20892 PR middle-end/19706
20893 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
20894 * config/aarch64/aarch64-builtins.c
20895 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
20896 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
20897 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
20899 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20900 Andrew Pinski <pinskia@gmail.com>
20902 PR middle-end/19706
20903 * internal-fn.def (XORSIGN): New.
20904 * optabs.def (xorsign_optab): New.
20905 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
20906 (convert_expand_mult_copysign): New.
20907 (pass_optimize_widening_mul::execute): Call
20908 convert_expand_mult_copysign.
20910 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20912 PR tree-optimization/81354
20913 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
20914 Insert on edges rather than explicitly creating landing pads.
20915 (analyze_candidates_and_replace): Commit edge inserts.
20917 2017-08-08 Richard Biener <rguenther@suse.de>
20919 PR middle-end/81719
20920 * tree-ssa-loop-niter.c: Include tree-dfa.h.
20921 (expand_simple_operations): Also look through ADDR_EXPRs with
20922 MEM_REF bases treating them as POINTER_PLUS_EXPR.
20924 2017-08-08 Richard Biener <rguenther@suse.de>
20926 PR tree-optimization/81723
20927 * tree-vect-slp.c (struct bst_traits): New hash traits.
20928 (bst_fail): New global.
20929 (vect_build_slp_tree_2): New worker, split out from ...
20930 (vect_build_slp_tree): ... this now wrapping it with using
20931 bst_fail set to cache SLP tree build fails. Properly handle
20933 (vect_analyze_slp_instance): Allocate and free bst_fail.
20935 2017-08-08 Martin Liska <mliska@suse.cz>
20938 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
20939 LABEL_DECLs that can be from a different function.
20941 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20943 PR tree-optimization/81744
20944 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
20945 loop's number of iterations.
20947 2017-08-08 Martin Liska <mliska@suse.cz>
20949 * asan.c: Include header files.
20950 * attribs.c (build_decl_attribute_variant): New function moved
20952 (build_type_attribute_qual_variant): Likewise.
20953 (cmp_attrib_identifiers): Likewise.
20954 (simple_cst_list_equal): Likewise.
20955 (omp_declare_simd_clauses_equal): Likewise.
20956 (attribute_value_equal): Likewise.
20957 (comp_type_attributes): Likewise.
20958 (build_type_attribute_variant): Likewise.
20959 (lookup_ident_attribute): Likewise.
20960 (remove_attribute): Likewise.
20961 (merge_attributes): Likewise.
20962 (merge_type_attributes): Likewise.
20963 (merge_decl_attributes): Likewise.
20964 (merge_dllimport_decl_attributes): Likewise.
20965 (handle_dll_attribute): Likewise.
20966 (attribute_list_equal): Likewise.
20967 (attribute_list_contained): Likewise.
20968 * attribs.h (lookup_attribute): New function moved from tree.[ch].
20969 (lookup_attribute_by_prefix): Likewise.
20970 * bb-reorder.c: Include header files.
20971 * builtins.c: Likewise.
20972 * calls.c: Likewise.
20973 * cfgexpand.c: Likewise.
20974 * cgraph.c: Likewise.
20975 * cgraphunit.c: Likewise.
20976 * convert.c: Likewise.
20977 * dwarf2out.c: Likewise.
20978 * final.c: Likewise.
20979 * fold-const.c: Likewise.
20980 * function.c: Likewise.
20981 * gimple-expr.c: Likewise.
20982 * gimple-fold.c: Likewise.
20983 * gimple-pretty-print.c: Likewise.
20984 * gimple.c: Likewise.
20985 * gimplify.c: Likewise.
20986 * hsa-common.c: Likewise.
20987 * hsa-gen.c: Likewise.
20988 * internal-fn.c: Likewise.
20989 * ipa-chkp.c: Likewise.
20990 * ipa-cp.c: Likewise.
20991 * ipa-devirt.c: Likewise.
20992 * ipa-fnsummary.c: Likewise.
20993 * ipa-inline.c: Likewise.
20994 * ipa-visibility.c: Likewise.
20996 * lto-cgraph.c: Likewise.
20997 * omp-expand.c: Likewise.
20998 * omp-general.c: Likewise.
20999 * omp-low.c: Likewise.
21000 * omp-offload.c: Likewise.
21001 * omp-simd-clone.c: Likewise.
21002 * opts-global.c: Likewise.
21003 * passes.c: Likewise.
21004 * predict.c: Likewise.
21005 * sancov.c: Likewise.
21006 * sanopt.c: Likewise.
21007 * symtab.c: Likewise.
21008 * toplev.c: Likewise.
21009 * trans-mem.c: Likewise.
21010 * tree-chkp.c: Likewise.
21011 * tree-eh.c: Likewise.
21012 * tree-into-ssa.c: Likewise.
21013 * tree-object-size.c: Likewise.
21014 * tree-parloops.c: Likewise.
21015 * tree-profile.c: Likewise.
21016 * tree-ssa-ccp.c: Likewise.
21017 * tree-ssa-live.c: Likewise.
21018 * tree-ssa-loop.c: Likewise.
21019 * tree-ssa-sccvn.c: Likewise.
21020 * tree-ssa-structalias.c: Likewise.
21021 * tree-ssa.c: Likewise.
21022 * tree-streamer-in.c: Likewise.
21023 * tree-vectorizer.c: Likewise.
21024 * tree-vrp.c: Likewise.
21025 * tsan.c: Likewise.
21026 * ubsan.c: Likewise.
21027 * varasm.c: Likewise.
21028 * varpool.c: Likewise.
21029 * tree.c: Remove functions moved to attribs.[ch].
21030 * tree.h: Likewise.
21031 * config/aarch64/aarch64.c: Add attrs.h header file.
21032 * config/alpha/alpha.c: Likewise.
21033 * config/arc/arc.c: Likewise.
21034 * config/arm/arm.c: Likewise.
21035 * config/avr/avr.c: Likewise.
21036 * config/bfin/bfin.c: Likewise.
21037 * config/c6x/c6x.c: Likewise.
21038 * config/cr16/cr16.c: Likewise.
21039 * config/cris/cris.c: Likewise.
21040 * config/darwin.c: Likewise.
21041 * config/epiphany/epiphany.c: Likewise.
21042 * config/fr30/fr30.c: Likewise.
21043 * config/frv/frv.c: Likewise.
21044 * config/ft32/ft32.c: Likewise.
21045 * config/h8300/h8300.c: Likewise.
21046 * config/i386/winnt.c: Likewise.
21047 * config/ia64/ia64.c: Likewise.
21048 * config/iq2000/iq2000.c: Likewise.
21049 * config/lm32/lm32.c: Likewise.
21050 * config/m32c/m32c.c: Likewise.
21051 * config/m32r/m32r.c: Likewise.
21052 * config/m68k/m68k.c: Likewise.
21053 * config/mcore/mcore.c: Likewise.
21054 * config/microblaze/microblaze.c: Likewise.
21055 * config/mips/mips.c: Likewise.
21056 * config/mmix/mmix.c: Likewise.
21057 * config/mn10300/mn10300.c: Likewise.
21058 * config/moxie/moxie.c: Likewise.
21059 * config/msp430/msp430.c: Likewise.
21060 * config/nds32/nds32-isr.c: Likewise.
21061 * config/nds32/nds32.c: Likewise.
21062 * config/nios2/nios2.c: Likewise.
21063 * config/nvptx/nvptx.c: Likewise.
21064 * config/pa/pa.c: Likewise.
21065 * config/pdp11/pdp11.c: Likewise.
21066 * config/powerpcspe/powerpcspe.c: Likewise.
21067 * config/riscv/riscv.c: Likewise.
21068 * config/rl78/rl78.c: Likewise.
21069 * config/rx/rx.c: Likewise.
21070 * config/s390/s390.c: Likewise.
21071 * config/sh/sh.c: Likewise.
21072 * config/sol2.c: Likewise.
21073 * config/sparc/sparc.c: Likewise.
21074 * config/spu/spu.c: Likewise.
21075 * config/stormy16/stormy16.c: Likewise.
21076 * config/tilegx/tilegx.c: Likewise.
21077 * config/tilepro/tilepro.c: Likewise.
21078 * config/v850/v850.c: Likewise.
21079 * config/vax/vax.c: Likewise.
21080 * config/visium/visium.c: Likewise.
21081 * config/xtensa/xtensa.c: Likewise.
21083 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21086 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
21087 constraints since the -mupper-regs-* switches have been
21089 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
21090 into a vector from a double word element that was extracted from
21091 another vector, and eliminate extra XXPERMDI instructions.
21092 (vsx_concat_<mode>_2): Likewise.
21093 (vsx_concat_<mode>_3): Likewise.
21094 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
21095 concat to allow optimizing inserts from previous extracts.
21097 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
21099 * config/i386/i386.c (ix86_stack_protect_guard): Generate
21100 memory reference to a SSP offset in TLS address space.
21101 (ix86_print_operand) <case '@'>: Remove.
21102 (ix86_print_operand_punct_valid_p): Remove '@' code.
21103 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
21104 UNSPEC_SP_TLS_TEST.
21105 (stack_tls_protect_set_<mode>): Remove.
21106 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
21107 (stack_tls_protect_test_<mode>): Remove.
21108 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
21110 2017-08-07 Olivier Hainque <hainque@adacore.com>
21113 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
21115 2017-08-07 Douglas Rupp <rupp@adacore.com>
21117 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
21118 variable was referenced as multidir in command.
21120 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21123 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
21126 2017-08-07 Martin Liska <mliska@suse.cz>
21128 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
21129 * config/rl78/rl78.c: Add include of attribs.h.
21130 * config/sh/sh.c: Likewise.
21131 * config/v850/v850.c: Likewise.
21133 2017-08-07 Tom de Vries <tom@codesourcery.com>
21135 PR middle-end/78266
21136 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
21138 2017-08-07 Martin Liska <mliska@suse.cz>
21140 * config/mips/mips.c: Include attribs.h.
21142 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
21145 * doc/invoke.texi: Document change in behvaior for -Ofast for
21148 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
21150 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
21152 (aarch64_pop_regs): Likewise.
21153 (aarch64_gen_load_pair): Likewise.
21154 (aarch64_save_callee_saves): Likewise.
21155 (aarch64_restore_callee_saves): Likewise.
21157 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21159 * config/i386/i386.c: Revert the last change.
21161 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21164 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21166 (ix86_finalize_stack_frame_flags): This. Also clear
21167 frame_pointer_needed if -fno-omit-frame-pointer is used without
21169 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21170 with ix86_finalize_stack_frame_flags.
21171 (ix86_expand_epilogue): Likewise.
21172 (ix86_expand_split_stack_prologue): Likewise.
21174 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21177 * config/i386/i386.c (get_builtin_code_for_version): Set priority
21178 to P_AES for Westmere.
21180 2017-08-07 Jonathan Yong <10walls@gmail.com>
21182 * config/i386/mingw.opt (fset-stack-executable): Removed.
21183 * config/i386/cygming.opt (fset-stack-executable): Moved
21185 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
21187 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
21189 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
21191 2017-08-07 Marek Polacek <polacek@redhat.com>
21193 PR middle-end/81737
21194 * fold-const.c (fold_indirect_ref_1): Check type_domain.
21196 2017-08-07 Martin Liska <mliska@suse.cz>
21198 * attribs.h (canonicalize_attr_name): New function.
21199 (cmp_attribs): Move from c-format.c and adjusted.
21200 (is_attribute_p): Moved from tree.h.
21201 * tree-inline.c: Add new includes.
21202 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
21203 (private_is_attribute_p): Remove.
21204 (private_lookup_attribute): Likewise.
21205 (private_lookup_attribute_by_prefix): Simplify.
21206 (remove_attribute): Use is_attribute_p.
21207 * tree.h: Remove removed declarations.
21209 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21211 PR middle-end/81698
21212 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
21213 instead of computing it in the function. Formatting fix.
21214 (expand_case): Don't rely on default_edge being the first edge,
21215 clear it if removing it, pass default_edge to
21216 emit_case_dispatch_table.
21217 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
21220 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
21222 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
21223 insn in the function, emit NOP after the insn.
21225 2017-08-06 Tom de Vries <tom@codesourcery.com>
21227 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
21230 2017-08-06 Tom de Vries <tom@codesourcery.com>
21232 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
21235 2017-08-04 Yury Gribov <tetra2005@gmail.com>
21237 PR tree-optimization/57371
21238 * match.pd: New pattern.
21240 2017-08-04 Marek Polacek <polacek@redhat.com>
21242 PR middle-end/81695
21243 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
21244 perform the computation in offset_int.
21246 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21248 PR tree-optimization/81136
21249 * tree-vectorizer.h: Include tree-hash-traits.h.
21250 (vec_base_alignments): New typedef.
21251 (vec_info): Add a base_alignments field.
21252 (vect_record_base_alignments): Declare.
21253 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
21255 (DR_IS_CONDITIONAL_IN_STMT): New macro.
21256 (create_data_ref): Add an is_conditional_in_stmt argument.
21257 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
21258 the is_conditional_in_stmt field.
21259 (data_ref_loc): Add an is_conditional_in_stmt field.
21260 (get_references_in_stmt): Set the is_conditional_in_stmt field.
21261 (find_data_references_in_stmt): Update call to create_data_ref.
21262 (graphite_find_data_references_in_stmt): Likewise.
21263 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
21264 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
21265 (vect_record_base_alignment): New function.
21266 (vect_record_base_alignments): Likewise.
21267 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
21268 for nested statements even if we fail to compute a misalignment.
21269 Use pooled base alignments for unconditional references.
21270 (vect_find_same_alignment_drs): Compare base addresses instead
21272 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
21273 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
21275 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21277 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
21278 Add an explicit name for the enum. Use auto_vec for slp_instances
21279 and grouped_stores.
21280 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
21282 (_bb_vec_info): Add a constructor and destructor.
21283 (vinfo_for_stmt): Return NULL for uids of -1 as well.
21284 (destroy_loop_vec_info): Delete.
21285 (vect_destroy_datarefs): Likewise.
21286 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
21287 (vec_info::vec_info): New function.
21288 (vec_info::~vec_info): Likewise.
21289 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
21290 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
21291 destroy_loop_vec_info.
21292 * tree-vect-loop.c (new_loop_vec_info): Replace with...
21293 (_loop_vec_info::_loop_vec_info): ...this.
21294 (destroy_loop_vec_info): Replace with...
21295 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
21296 the stmt_vec_infos. Leave handling of vec_info information to its
21297 destructor. Remove explicit vector releases.
21298 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
21299 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
21300 * tree-vect-slp.c (new_bb_vec_info): Replace with...
21301 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
21302 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
21303 (destroy_bb_vec_info): Replace with...
21304 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
21305 information to its destructor.
21306 (vect_slp_analyze_bb_1): Use new and delete instead of
21307 new_bb_vec_info and destroy_bb_vec_info.
21308 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
21311 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21313 * tree-data-ref.h (subscript): Add access_fn field.
21314 (data_dependence_relation): Add could_be_independent_p.
21315 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
21316 (same_access_functions): Move to tree-data-ref.c.
21317 * tree-data-ref.c (ref_contains_union_access_p): New function.
21318 (access_fn_component_p): Likewise.
21319 (access_fn_components_comparable_p): Likewise.
21320 (dr_analyze_indices): Add a reference to access_fn_component_p.
21321 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
21323 (constant_access_functions): Likewise.
21324 (add_other_self_distances): Likewise.
21325 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
21326 (initialize_data_dependence_relation): Use XCNEW and remove
21327 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
21328 of access functions that have the same type. Allow the
21329 subsequence to end with different bases in some circumstances.
21330 Record the chosen access functions in SUB_ACCESS_FN.
21331 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
21332 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
21333 (subscript_dependence_tester_1): Likewise dra and drb.
21334 (build_classic_dist_vector): Update calls accordingly.
21335 (subscript_dependence_tester): Likewise.
21336 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
21337 DDR_COULD_BE_INDEPENDENT_P.
21338 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
21339 comp_alias_ddrs instead of may_alias_ddrs.
21340 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
21342 (vect_analyze_data_ref_dependence): Use it if
21343 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
21344 distance vectors if that fails.
21345 (dependence_distance_ge_vf): New function.
21346 (vect_prune_runtime_alias_test_list): Use it. Don't clear
21347 LOOP_VINFO_MAY_ALIAS_DDRS.
21349 2017-08-04 Richard Biener <rguenther@suse.de>
21351 PR middle-end/81705
21352 * fold-const.c (fold_binary_loc): Properly restrict
21353 minus_var0 && minus_var1 case when associating undefined overflow
21356 2017-08-04 Tom de Vries <tom@codesourcery.com>
21358 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
21360 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21362 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21363 Don't start diagnostic messages with a capital letter.
21364 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21366 (rs6000_invalid_builtin): Likewise.
21367 (rs6000_trampoline_init): Likewise.
21369 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21372 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
21373 after setting changeable df flags.
21375 2017-08-03 Richard Biener <rguenther@suse.de>
21377 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
21378 up if the use is in USE - X.
21380 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
21382 * toplev.c (dumpfile.h): New include.
21383 (internal_error_reentered): New static function. Use it...
21384 (internal_error_function): ...here to handle reentered internal_error.
21386 2017-08-03 Richard Biener <rguenther@suse.de>
21388 PR middle-end/81148
21389 * fold-const.c (split_tree): Add minus_var and minus_con
21390 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
21391 here but always use minus_*.
21392 (associate_trees): Assert we never associate with MINUS_EXPR
21393 and NULL first operand. Do not recurse for PLUS_EXPR operands
21394 when associating as MINUS_EXPR either.
21395 (fold_binary_loc): Track minus_var and minus_con.
21397 2017-08-03 Tom de Vries <tom@codesourcery.com>
21400 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
21401 ACCEL_COMPILER, apply finish_options on
21402 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
21404 2017-08-03 Tom de Vries <tom@codesourcery.com>
21407 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
21408 function_entry_patch_area_size > 0.
21410 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21413 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
21414 instead of 10??LU, perform unit multiplication in wide_int,
21415 don't change alloc_object_size_limit if the limit is larger
21418 PR tree-optimization/81655
21419 PR tree-optimization/81588
21420 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
21421 the case when ranges[i].low and high are 1 for unsigned type with
21424 PR middle-end/81052
21425 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
21426 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
21428 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21430 * tree-vrp.h: Add include guard.
21432 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
21435 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
21436 (ud2): New insn pattern.
21437 * config/i386/i386.c (ix86_expand_epilogue):
21438 For naked functions, generate ud2 instead of trap insn.
21440 2017-08-02 Marek Polacek <polacek@redhat.com>
21443 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
21445 2017-08-02 Tom de Vries <tom@codesourcery.com>
21446 Cesar Philippidis <cesar@codesourcery.com>
21448 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
21449 Add missing edge probabilities.
21451 2017-08-02 Tamar Christina <tamar.christina@arm.com>
21453 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
21454 Correct endianness.
21456 2017-08-02 Jakub Jelinek <jakub@redhat.com>
21458 PR middle-end/79499
21459 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
21460 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
21461 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
21462 prologue_seq sequences - if any.
21464 2017-08-02 Richard Biener <rguenther@suse.de>
21466 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
21467 via vectors if supported, integer extracts via punning if supported
21468 or otherwise vector extracts.
21470 2017-08-02 Richard Biener <rguenther@suse.de>
21472 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
21474 (bitmap_insert_into_set): ... this.
21476 2017-08-02 Richard Biener <rguenther@suse.de>
21478 PR tree-optimization/81633
21480 2015-08-17 Alan Hayward <alan.hayward@arm.com>
21482 PR tree-optimization/71752
21483 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
21485 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
21487 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
21488 (machine_function::call_ms2sysv_pad_out): Remove field.
21489 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
21490 (ix86_compute_frame_layout): Likewise.
21492 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
21495 * config/i386/i386.c (ix86_set_func_type): Disallow naked
21496 attribute with interrupt attribute.
21498 2017-08-01 Andrew Pinski <apinski@cavium.com>
21500 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
21501 BIT_INSERT_EXPR's operand 1
21502 to see if the types precision matches.
21504 2017-08-01 Martin Liska <mliska@suse.cz>
21506 PR middle-end/70140
21507 * builtins.c (expand_builtin_memcpy_args): Remove.
21508 (expand_builtin_memcpy): Call newly added function
21509 expand_builtin_memory_copy_args.
21510 (expand_builtin_memcpy_with_bounds): Likewise.
21511 (expand_builtin_mempcpy): Remove last argument.
21512 (expand_builtin_mempcpy_with_bounds): Likewise.
21513 (expand_builtin_memory_copy_args): New function created from
21514 expand_builtin_mempcpy_args with small modifications.
21515 (expand_builtin_mempcpy_args): Remove.
21516 (expand_builtin_stpcpy): Remove unused argument.
21517 (expand_builtin): Likewise.
21518 (expand_builtin_with_bounds): Likewise.
21520 2017-08-01 Martin Liska <mliska@suse.cz>
21523 Make mempcpy more optimal (PR middle-end/70140).
21525 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21528 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
21529 __builtin_vec_cmpne verify both arguments are compatible vectors
21530 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
21531 verify arg1_type is a pointer or array type. For __builtin_vec_st,
21532 move computation of aligned to after checking the argument types.
21536 * config/rs6000/vsx.md (vextract_fp_from_shorth,
21537 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
21540 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
21541 Doug Rupp <rupp@adacore.com>
21542 Olivier Hainque <hainque@adacore.com>
21544 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
21545 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
21547 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
21548 for TARGET_OS_CPP_BUILTIN.
21549 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
21550 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
21552 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
21553 passing required abi options to the assembler for EABI configurations.
21554 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
21555 of .text.hot and .text.unlikely sections for kernel modules when
21556 using ARM style exceptions.
21557 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
21558 options. Add EXTRA_CC1_SPEC.
21559 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
21561 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
21563 (ARM_TARGET2_DWARF_FORMAT): Define.
21564 * config/arm/t-vxworks: Adjust multilib control to removal of the
21565 Diab command line options.
21567 2017-08-01 Martin Liska <mliska@suse.cz>
21569 PR gcov-profile/81561
21570 * gcov.c (unblock): Make unblocking safe as we need to preserve
21571 index correspondence of blocks and block_lists.
21573 2017-08-01 Richard Biener <rguenther@suse.de>
21575 PR tree-optimization/81181
21576 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21577 (compute_antic): ... end of iteration here.
21579 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21581 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21582 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21583 (ftree-slp-vectorize): Likewise.
21584 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21585 can no longer be set independent of flag_tree_loop_vectorize.
21586 * omp-general.c (emp_max_vf): Likewise.
21587 * opts.c (enable_fdo_optimizations): Remove references to
21588 flag_tree_vectorize, these are now implicit.
21589 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21590 and leave it for the options machinery.
21592 2017-08-01 Martin Liska <mliska@suse.cz>
21594 PR middle-end/70140
21595 * builtins.c (expand_builtin_memcpy_args): Remove.
21596 (expand_builtin_memcpy): Call newly added function
21597 expand_builtin_memory_copy_args.
21598 (expand_builtin_memcpy_with_bounds): Likewise.
21599 (expand_builtin_mempcpy): Remove last argument.
21600 (expand_builtin_mempcpy_with_bounds): Likewise.
21601 (expand_builtin_memory_copy_args): New function created from
21602 expand_builtin_mempcpy_args with small modifications.
21603 (expand_builtin_mempcpy_args): Remove.
21604 (expand_builtin_stpcpy): Remove unused argument.
21605 (expand_builtin): Likewise.
21606 (expand_builtin_with_bounds): Likewise.
21608 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21611 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21612 print "ds:" only for immediates in generic address space.
21614 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21617 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21618 (ix86_function_ok_for_sibcall): Return false for naked functions.
21620 2017-08-01 Richard Biener <rguenther@suse.de>
21622 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21623 (compute_antic): Seed worklist with exit block predecessors.
21624 * cfganal.c (dfs_find_deadend): For a cycle return the source
21625 of the edge closing it.
21627 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21629 * config/aarch64/aarch64.c
21630 (aarch64_can_const_movi_rtx_p): Move 0 check.
21632 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21634 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21635 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21637 * match.pd: Ditto in address comparison pattern.
21639 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21641 PR tree-optimization/81627
21642 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21643 closed ssa form for store-store chain.
21645 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21647 PR tree-optimization/81620
21648 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21649 for store-store chain.
21651 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21653 PR tree-optimization/81588
21654 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21655 ranges[i].in_p, invert comparison code ccode. For >/>=,
21656 swap rhs1 and rhs2 and comparison code unconditionally,
21657 for </<= don't do that. Don't swap rhs1/rhs2 again if
21658 ranges[i].in_p, instead invert comparison code ccode if
21659 opcode or oe->rank is BIT_IOR_EXPR.
21662 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21663 a direct optab to conversion optab.
21664 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21665 with GET_MODE_INNER as last argument instead of optab_handler.
21666 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21667 vector extraction if possible and optab is available.
21668 * expr.c (store_constructor): Use convert_optab_handler instead
21669 of optab_handler. Use vector initialization from smaller
21670 vectors if possible and optab is available.
21671 * tree-vect-stmts.c (vectorizable_load): Likewise.
21672 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21673 now have two modes.
21674 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21675 of vec_init from half-sized vectors with the same element mode.
21676 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21677 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21678 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21679 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21680 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21681 after mode in gen_vec_extract* calls.
21682 (vec_extract<mode>): Renamed to ...
21683 (vec_extract<mode><ssescalarmodelower>): ... this.
21684 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21685 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21686 element mode after mode in gen_vec_init* calls.
21687 (VEC_INIT_HALF_MODE): New mode iterator.
21688 (vec_init<mode>): Renamed to ...
21689 (vec_init<mode><ssescalarmodelower>): ... this.
21690 (vec_init<mode><ssehalfvecmodelower>): New expander.
21691 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21692 (vec_extractv2sfsf): ... this.
21693 (vec_initv2sf): Renamed to ...
21694 (vec_initv2sfsf): ... this.
21695 (vec_extractv2si): Renamed to ...
21696 (vec_extractv2sisi): ... this.
21697 (vec_initv2si): Renamed to ...
21698 (vec_initv2sisi): ... this.
21699 (vec_extractv4hi): Renamed to ...
21700 (vec_extractv4hihi): ... this.
21701 (vec_initv4hi): Renamed to ...
21702 (vec_initv4hihi): ... this.
21703 (vec_extractv8qi): Renamed to ...
21704 (vec_extractv8qiqi): ... this.
21705 (vec_initv8qi): Renamed to ...
21706 (vec_initv8qiqi): ... this.
21707 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21708 (vec_init<mode>): Renamed to ...
21709 (vec_init<mode><VEC_base_l>): ... this.
21710 (vec_extract<mode>): Renamed to ...
21711 (vec_extract<mode><VEC_base_l>): ... this.
21712 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21713 (vec_initv2sfsf): ... this.
21714 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21715 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21716 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21717 element mode after mode in gen_vec_init* calls.
21718 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21719 (vec_init<mode><Vel>): ... this.
21720 (vec_extract<mode>): Renamed to ...
21721 (vec_extract<mode><Vel>): ... this.
21722 * config/aarch64/iterators.md (Vel): New mode attribute.
21723 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21724 Add element mode after mode in gen_vec_extract* calls.
21725 * config/s390/vector.md (non_vec_l): New mode attribute.
21726 (vec_extract<mode>): Renamed to ...
21727 (vec_extract<mode><non_vec_l>): ... this.
21728 (vec_init<mode>): Renamed to ...
21729 (vec_init<mode><non_vec_l>): ... this.
21730 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21731 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21733 * config/arm/iterators.md (V_elem_l): New mode attribute.
21734 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21735 (vec_extract<mode><V_elem_l>): ... this.
21736 (vec_extractv2di): Renamed to ...
21737 (vec_extractv2didi): ... this.
21738 (vec_init<mode>): Renamed to ...
21739 (vec_init<mode><V_elem_l>): ... this.
21740 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21741 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21742 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21743 Add element mode after gen_vec_extract* calls.
21744 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21745 (vec_init<mode><unitmode>): ... this.
21746 (vec_extract<mode>): Renamed to ...
21747 (vec_extract<mode><unitmode>): ... this.
21748 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21749 (vec_init<mode><unitmode>): ... this.
21750 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21751 (vec_initv2sfsf): ... this.
21752 (vec_extractv2sf): Renamed to ...
21753 (vec_extractv2sfsf): ... this.
21754 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21755 Add element mode after gen_vec_extract* calls.
21756 * config/mips/mips.md (unitmode): New mode iterator.
21757 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21758 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21759 * config/spu/spu.md (inner_l): New mode attribute.
21760 (vec_init<mode>): Renamed to ...
21761 (vec_init<mode><inner_l>): ... this.
21762 (vec_extract<mode>): Renamed to ...
21763 (vec_extract<mode><inner_l>): ... this.
21764 * config/sparc/sparc.md (veltmode): New mode iterator.
21765 (vec_init<VMALL:mode>): Renamed to ...
21766 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21767 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21768 (vec_initv2sisi): ... this.
21769 (vec_initv2sf): Renamed to ...
21770 (vec_initv2sfsf): ... this.
21771 (vec_extractv2sf): Renamed to ...
21772 (vec_extractv2sfsf): ... this.
21773 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21774 (vec_init<mode>): Renamed to ...
21775 (vec_init<mode><VEC_base_l>): ... this.
21776 (vec_extract<mode>): Renamed to ...
21777 (vec_extract<mode><VEC_base_l>): ... this.
21778 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21779 (vec_initv2sfsf): ... this.
21780 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21781 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21782 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21783 gen_vec_init* calls.
21785 2017-08-01 Richard Biener <rguenther@suse.de>
21787 PR tree-optimization/81297
21788 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21789 TREE_OVERFLOW from INTEGER_CSTs.
21791 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21793 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21795 2017-07-31 Carl Love <cel@us.ibm.com>
21797 * config/rs6000/rs6000-c: Add support for built-in functions
21798 vector signed char vec_xl_be (signed long long, signed char *);
21799 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21800 vector signed int vec_xl_be (signed long long, signed int *);
21801 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21802 vector signed long long vec_xl_be (signed long long, signed long long *);
21803 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21804 vector signed short vec_xl_be (signed long long, signed short *);
21805 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21806 vector double vec_xl_be (signed long long, double *);
21807 vector float vec_xl_be (signed long long, float *);
21808 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21809 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21810 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21812 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21813 (altivec_expand_builtin): Add switch statement to call
21814 altivec_expand_xl_be for each builtin.
21815 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21816 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21817 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21818 __builtin_vsx_le_be_v16qi.
21819 * doc/extend.texi: Update the built-in documentation file for the
21820 new built-in functions.
21822 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21825 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
21827 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21829 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21831 * config.gcc: Add z14.
21832 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
21833 CPU model numbers for z13s and z14.
21834 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
21836 * config/s390/s390-opts.h (enum processor_type): Rename
21837 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21838 * config/s390/s390.c (processor_table): Add field for CPU name to
21839 be passed to Binutils.
21840 (s390_asm_output_machine_for_arch): Use the new field in
21841 processor_table for Binutils.
21842 (s390_expand_builtin): Replace arch12 with z14.
21843 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21844 (s390_get_sched_attrmask): Likewise.
21845 (s390_get_unit_mask): Likewise.
21846 * config/s390/s390.opt: Add z14 to processor_type enum.
21848 2017-07-31 Martin Jambor <mjambor@suse.cz>
21851 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
21852 regardless of optimization level.
21854 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
21855 Martin Liska <mliska@suse.cz>
21857 * predict.def: Remove old comment and adjust probability.
21858 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
21859 PREDICT statements.
21861 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21864 * config/i386/i386.c (ix86_function_naked): New function.
21865 (ix86_can_use_return_insn_p): Return false for naked functions.
21866 (ix86_expand_prologue): Skip prologue for naked functions.
21867 (ix86_expand_epilogue): Skip epilogue for naked functions
21868 and emit trap instruction.
21869 (ix86_warn_func_return): New function.
21870 (ix86_attribute_table): Add "naked" attribute specification.
21871 (TARGET_WARN_FUNC_RETURN): Define.
21872 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
21874 2017-07-31 Martin Liska <mliska@suse.cz>
21876 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
21877 (dump_gimple_bb_header): Always dump BB info.
21878 (pp_cfg_jump): Do not append info about BB when dumping a jump.
21880 2017-07-31 Martin Liska <mliska@suse.cz>
21883 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
21884 also with current_function_decl non-null equality.
21886 2017-07-31 Jakub Jelinek <jakub@redhat.com>
21889 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
21890 change type to the element type, instead add eltype variable and
21891 use it where we are interested in the element type.
21893 PR tree-optimization/81603
21894 * ipa-polymorphic-call.c
21895 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
21896 offset arithmetic in offset_int, bail out if the resulting bit offset
21897 doesn't fit into shwi.
21899 2017-07-31 Martin Liska <mliska@suse.cz>
21901 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
21902 (gimplify_save_expr): Fix comment.
21904 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
21907 * config/i386/i386.c (ix86_function_arg): Update arguments for
21909 (ix86_compute_frame_layout): Set the initial stack offset to
21910 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
21911 INCOMING_FRAME_SP_OFFSET.
21912 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
21913 stack before exception handler returns.
21914 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
21915 the 'ERROR_CODE' for exception handler.
21917 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
21919 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
21920 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
21921 (ASM_OUTPUT_REG_POP): Ditto.
21922 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
21923 instead of asm_fprintf to output pure string.
21925 2017-07-29 Jakub Jelinek <jakub@redhat.com>
21927 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
21928 to imported_module_or_decl hook.
21929 (debug_nothing_tree_tree_tree_bool): Remove.
21930 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
21931 * debug.c (do_nothing_debug_hooks): Use
21932 debug_nothing_tree_tree_tree_bool_bool instead of
21933 debug_nothing_tree_tree_tree_bool.
21934 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21935 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
21936 * sdbout.c (sdb_debug_hooks): Likewise.
21937 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
21938 (gen_namespace_die): Add DW_AT_export_symbols attribute if
21940 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
21941 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
21942 attribute, don't add anything.
21944 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21946 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
21947 (fold_build2_stat_loc): Likewise.
21948 (fold_build3_stat_loc): Likewise.
21949 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
21950 (fold_build1_loc): Remove macro.
21951 (fold_build2_loc): Likewise.
21952 (fold_build3_loc): Likewise.
21954 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21956 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
21957 (gimple_build_debug_bind_source_stat): Likewise.
21958 * gimple.h (gimple_build_debug_bind): Remove macro.
21959 (gimple_build_debug_bind_source): Likewise.
21961 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21963 * bitmap.c (bitmap_alloc): Adjust.
21964 (bitmap_gc_alloc): Likewise.
21965 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
21967 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21969 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
21970 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
21971 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
21972 (bitmap_gc_alloc_stat): Likewise.
21973 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
21975 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21977 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
21978 * rtl.h (shallow_copy_rtx): Remove macro.
21980 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21982 * emit-rtl.c (gen_raw_REG): Adjust.
21983 * gengenrtl.c (gendef): Likewise.
21984 * rtl.c (rtx_alloc_stat): Remove _stat from name.
21985 * rtl.h (rtx_alloc): Remove macro.
21987 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21989 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
21990 (build_tree_list_stat): Likewise.
21991 * tree.h (build_tree_list): Remove macro.
21992 (build_tree_list_vec): Likewise.
21994 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21996 * tree.c (make_vector_stat): Remove _stat from name.
21997 (build_vector_stat): Likewise.
21998 * tree.h (make_vector_stat): Remove macro.
21999 (build_vector_stat): Likewise.
22001 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22003 * tree.h (build_var_debug_value): Remove prototype.
22005 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22007 * tree.c (tree_cons_stat): Remove _stat from name.
22008 * tree.h (tree_cons): Remove macro.
22010 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22012 * tree.c (build_vl_exp_stat): Remove _stat from name.
22013 * tree.h (build_vl_exp): Remove macro.
22015 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22017 * tree.c (build_decl_stat): Remove _stat from name.
22018 * tree.h (build_decl): Remove macro.
22020 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22022 * gimple.c (gimple_build_with_ops_stat): Adjust.
22023 (gimple_alloc_stat): Remove _stat from name.
22024 * gimple.h (gimple_alloc): Remove macro.
22026 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22028 * tree.c (make_tree_vec_stat): Remove _stat from name.
22029 (grow_tree_vec_stat): Likewise.
22030 * tree.h (make_tree_vec_stat): Adjust prototype.
22031 (grow_tree_vec_stat): Likewise.
22032 (make_tree_vec): Remove macro.
22033 (grow_tree_vec): Likewise.
22035 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22037 * fold-const.c (fold_build1_stat_loc): Adjust.
22038 (fold_build2_stat_loc): Likewise.
22039 (fold_build3_stat_loc): Likewise.
22040 * tree.c (build0_stat): Remove _stat from name.
22041 (build1_stat): Likewise.
22042 (build2_stat): Likewise.
22043 (build3_stat): Likewise.
22044 (build4_stat): Likewise.
22045 (build5_stat): Likewise.
22046 * tree.h (build1_loc): Remove macro, and rename _stat function
22048 (build2_loc): Likewise.
22049 (build3_loc): Likewise.
22050 (build4_loc): Likewise.
22051 (build5_loc): Likewise.
22053 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22055 * tree.c (make_int_cst_stat): Remove _stat from name.
22056 * tree.h (make_int_cst_stat): Adjust prototype.
22057 (make_int_cst): Remove macro.
22059 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22061 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22062 * tree.h (make_tree_binfo_stat): Adjust prototype.
22063 (make_tree_binfo): Remove.
22065 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22067 * tree.c (copy_node_stat): Rename to copy_node.
22068 (build_distinct_type_copy): Adjust.
22069 * tree.h (copy_node_stat): Adjust prototype.
22070 (copy_node): Remove macro.
22072 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22074 * tree.c (make_node_stat): rename to make_node.
22075 (build_tree_list_stat): Adjust.
22076 (build0_stat): Likewise.
22077 (build2_stat): Likewise.
22078 (build3_stat): Likewise.
22079 (build4_stat): Likewise.
22080 (build5_stat): Likewise.
22081 (build_decl_stat): Likewise.
22082 * tree.h (make_node_stat): Adjust prototype.
22083 (make_node): remove macro.
22085 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
22087 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
22088 (PPC_FEATURE2_SCV): Likewise.
22089 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
22091 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22093 * config/aarch64/aarch64.c
22094 (aarch64_internal_mov_immediate): Add new special pattern.
22095 * config/aarch64/aarch64.md (*movdi_aarch64):
22096 Add reg/32bit const mov case.
22098 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22099 Richard Sandiford <richard.sandiford@linaro.org>
22101 * config/aarch64/aarch64.md (mov<mode>): Generalize.
22102 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
22103 Add integer and movi cases.
22104 (movi-split-hf-df-sf split, fp16): New.
22105 (enabled): Added TARGET_FP_F16INST.
22106 * config/aarch64/iterators.md (GPF_HF): New.
22107 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
22109 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22111 * config/aarch64/aarch64.c
22112 (aarch64_simd_container_mode): Add prototype.
22113 (aarch64_expand_mov_immediate): Add HI support.
22114 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
22115 (aarch64_can_const_movi_rtx_p): New.
22116 (aarch64_preferred_reload_class):
22117 Remove restrictions of using FP registers for certain SIMD operations.
22118 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
22119 (aarch64_valid_floating_const): Add integer move validation.
22120 (aarch64_simd_imm_scalar_p): Remove.
22121 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
22122 (aarch64_legitimate_constant_p): Expand list of supported cases.
22123 * config/aarch64/aarch64-protos.h
22124 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
22125 (aarch64_reinterpret_float_as_int): New.
22126 (aarch64_simd_imm_scalar_p): Remove.
22127 * config/aarch64/constraints.md (Uvi): New.
22128 (Dd): Split into Ds and new Dd.
22129 * config/aarch64/aarch64.md (*movsi_aarch64):
22131 (*movdi_aarch64): Add SIMD mov case.
22133 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22135 * tree-predcom.c: (struct chain): Handle store-store chain in which
22136 stores for elimination only store loop invariant values.
22137 (execute_pred_commoning_chain): Ditto.
22138 (prepare_initializers_chain_store_elim): Ditto.
22139 (prepare_finalizers): Ditto.
22140 (is_inv_store_elimination_chain): New function.
22141 (initialize_root_vars_store_elim_1): New function.
22143 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22145 * tree-predcom.c: Revise general description of the pass.
22146 (enum chain_type): New enum type for store elimination.
22147 (struct chain): New field supporting store elimination.
22148 (struct component): Ditto.
22149 (dump_chain): Dump store-stores chain.
22150 (release_chain): Release resources.
22151 (split_data_refs_to_components): Compute and create component
22152 contains only stores for elimination.
22153 (get_chain_last_ref_at): New function.
22154 (make_invariant_chain): Initialization.
22155 (make_rooted_chain): Specify chain type in parameter and record it.
22156 (add_looparound_copies): Skip for store-stores chain.
22157 (determine_roots_comp): Compute type of chain and pass it to
22159 (initialize_root_vars_store_elim_2): New function.
22160 (finalize_eliminated_stores): New function.
22161 (remove_stmt): Handle store for elimination.
22162 (execute_pred_commoning_chain): Execute predictive commoning on
22163 store-store chains.
22164 (determine_unroll_factor): Skip unroll for store-stores chain.
22165 (prepare_initializers_chain_store_elim): New function.
22166 (prepare_initializers_chain): Hanlde store-store chain.
22167 (prepare_finalizers_chain, prepare_finalizers): New function.
22168 (tree_predictive_commoning_loop): Return integer value indicating
22169 if loop is unrolled or lcssa form is corrupted.
22170 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
22172 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22174 * tree-predcom.c (initialize_root): Delete.
22175 (execute_pred_commoning_chain): Initialize root vars and replace
22176 reference of non-combined chain directly, rather than call above
22179 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22181 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
22182 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
22184 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22186 * tree-predcom.c (struct chain): New field init_seq.
22187 (release_chain): Release init_seq.
22188 (prepare_initializers_chain): Record intialization stmts in above
22190 (insert_init_seqs): New function.
22191 (tree_predictive_commoning_loop): Call insert_init_seqs.
22193 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22195 * tree-predcom.c (determine_roots_comp): Skip trivial components.
22197 2017-07-28 Richard Biener <rguenther@suse.de>
22199 * match.pd: Remove superfluous :c.
22200 * genmatch.c (simplify::id): Add member.
22201 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
22203 (current_id): New global.
22204 (dt_node::parent): Move from ...
22205 (dt_operand::parent): ... here. Add for_id member.
22206 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
22207 (decision_tree::find_node): Relax order requirement when
22208 merging DT_TRUE nodes to ones inbetween the current simplify
22209 and the one we try to merge with. Add diagnostic whenever
22210 we need to enforce pattern order by not merging.
22211 (decision_tree::insert): Set current_id.
22212 (decision_tree::print_node): Dump parent node and for_id.
22213 (parser::last_id): Add member.
22214 (parser::push_simplify): Assign unique id.
22215 (parser::parser): Initialize last_id.
22217 2017-07-28 Martin Liska <mliska@suse.cz>
22220 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
22221 gimple_build_debug_bind.
22223 2017-07-28 Richard Biener <rguenther@suse.de>
22225 PR tree-optimization/81502
22226 * match.pd: Add pattern combining BIT_INSERT_EXPR with
22228 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
22230 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
22231 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
22232 for BIT_FIELD_REF args.
22233 * fold-const.c (make_bit_field_ref): Likewise.
22234 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
22236 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22239 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
22240 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
22241 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
22242 Or it into SANITIZER_UNDEFINED.
22243 * ubsan.c: Include gimple-fold.h and varasm.h.
22244 (ubsan_expand_ptr_ifn): New function.
22245 (instrument_pointer_overflow): New function.
22246 (maybe_instrument_pointer_overflow): New function.
22247 (instrument_object_size): Formatting fix.
22248 (pass_ubsan::execute): Call instrument_pointer_overflow
22249 and maybe_instrument_pointer_overflow.
22250 * internal-fn.c (expand_UBSAN_PTR): New function.
22251 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
22252 * sanitizer.def (__ubsan_handle_pointer_overflow,
22253 __ubsan_handle_pointer_overflow_abort): New builtins.
22254 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
22255 * internal-fn.def (UBSAN_PTR): New internal function.
22256 * opts.c (sanitizer_opts): Add pointer-overflow.
22257 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
22258 * fold-const.c (build_range_check): Compute pointer range check in
22259 integral type if pointer arithmetics would be needed. Formatting
22262 2017-07-28 Martin Liska <mliska@suse.cz>
22265 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
22266 parameters that are of a variable-length.
22268 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
22270 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
22272 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
22273 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
22274 (CRT_CALL_STATIC_FUNCTION): Likewise.
22275 (ASM_DEFAULT_SPEC): New define.
22276 (ASM_SPEC32): Likewise.
22277 (ASM_SPEC64): Likewise.
22278 (ASM_SPEC_COMMON): Likewise.
22279 (ASM_SPEC): Likewise.
22280 (INVALID_64BIT): Likewise.
22281 (LINK_OS_DEFAULT_SPEC): Likewise.
22282 (LINK_OS_SPEC32): Likewise.
22283 (LINK_OS_SPEC64): Likewise.
22284 (POWERPC_LINUX): Likewise.
22285 (PTRDIFF_TYPE): Likewise.
22286 (RESTORE_FP_PREFIX): Likewise.
22287 (RESTORE_FP_SUFFIX): Likewise.
22288 (SAVE_FP_PREFIX): Likewise.
22289 (SAVE_FP_SUFFIX): Likewise.
22290 (SIZE_TYPE): Likewise.
22291 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
22292 (TARGET_64BIT): Likewise.
22293 (TARGET_64BIT): Likewise.
22294 (TARGET_AIX): Likewise.
22295 (WCHAR_TYPE_SIZE): Likewise.
22296 (WCHAR_TYPE): Undefine.
22297 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
22298 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
22299 (CPP_OS_RTEMS_SPEC): Delete.
22300 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
22301 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
22303 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
22305 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22307 PR tree-optimization/81578
22308 * tree-parloops.c (build_new_reduction): Bail out if
22309 reduction_code isn't one of the standard OpenMP reductions.
22310 Move the details printing after that decision.
22312 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22314 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
22315 related to reload_in_progress.
22316 (splat_input_operand): Likewise.
22317 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
22319 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
22321 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
22322 (TARGET_INSTANTIATE_DECLS): Likewise.
22323 (legitimate_indexed_address_p): Delete reload_in_progress code.
22324 (rs6000_debug_legitimate_address_p): Likewise.
22325 (rs6000_eliminate_indexed_memrefs): Likewise.
22326 (rs6000_emit_le_vsx_store): Likewise.
22327 (rs6000_emit_move_si_sf_subreg): Likewise.
22328 (rs6000_emit_move): Likewise.
22329 (register_to_reg_type): Likewise.
22330 (rs6000_pre_atomic_barrier): Likewise.
22331 (rs6000_machopic_legitimize_pic_address): Likewise.
22332 (rs6000_allocate_stack_temp): Likewise.
22333 (rs6000_address_for_fpconvert): Likewise.
22334 (rs6000_address_for_altivec): Likewise.
22335 (rs6000_secondary_memory_needed_rtx): Delete function.
22336 (rs6000_check_sdmode): Likewise.
22337 (rs6000_alloc_sdmode_stack_slot): Likewise.
22338 (rs6000_instantiate_decls): Likewise.
22339 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
22340 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
22341 Delete reload_in_progress.
22342 (*vec_reload_and_plus_<mptrsize>): Likewise.
22343 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
22344 (vsx_div_v2di): Likewise.
22345 (vsx_udiv_v2di): Likewise.
22347 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22349 * config/rs6000/rs6000.opt (mlra): Replace with stub.
22350 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
22351 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
22352 (rs6000_debug_reg_global): Delete print of LRA status.
22353 (rs6000_option_override_internal): Delete dead LRA related code.
22354 (rs6000_lra_p): Delete function.
22355 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
22357 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
22359 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
22360 * config/riscv/rtems.h: New file.
22362 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22363 Sudakshina Das <sudi.das@arm.com>
22365 * config/aarch64/aarch64.md
22366 (define_split for and<mode>3nr_compare): Move
22367 non aarch64_logical_operand to a register.
22368 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
22369 register immediate operand to a register.
22370 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
22372 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22374 PR middle-end/81564
22375 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
22377 2017-07-27 Richard Biener <rguenther@suse.de>
22379 PR tree-optimization/81573
22380 PR tree-optimization/81494
22381 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
22382 multi defuse cycle case.
22384 2017-07-27 Richard Biener <rguenther@suse.de>
22386 PR tree-optimization/81571
22387 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
22390 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
22392 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
22393 earlier and only if MASK_FPU is set. Adjust formatting.
22395 2017-07-27 Martin Liska <mliska@suse.cz>
22397 * opt-functions.awk: Add validation of value of Init.
22398 * optc-gen.awk: Pass new argument.
22400 2017-07-27 Martin Liska <mliska@suse.cz>
22402 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
22403 Fix wrong condition.
22405 2017-07-27 Martin Liska <mliska@suse.cz>
22407 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
22408 BBs and edges seen by autoFDO.
22410 2017-07-27 Richard Biener <rguenther@suse.de>
22412 PR tree-optimization/81502
22413 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
22414 with incompatible but same sized type.
22415 (execute_update_addresses_taken): Likewise.
22417 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
22419 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
22420 flag_tree_loop_vectorize rather than flag_tree_vectorize.
22422 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22425 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
22426 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
22427 Change s_operand to memory_operand.
22429 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
22431 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
22432 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
22433 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
22434 Emit instructions rather than returning an expression. Handle TFmode
22435 and KFmode by casting to TImode.
22436 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
22437 (rs6000_emit_le_vsx_store): Likewise.
22438 * config/rs6000/vsx.md (VSX_TI): New iterator.
22439 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
22440 (*vsx_le_undo_permute_<mode>): Likewise.
22441 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
22442 emit the split sequence.
22443 (*vsx_le_perm_store_<mode>): Likewise.
22445 2017-07-27 Jakub Jelinek <jakub@redhat.com>
22447 PR tree-optimization/81555
22448 PR tree-optimization/81556
22449 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
22450 if true, force CHANGED for the recursive invocation.
22451 (reassociate_bb): Remember original length of ops array, pass
22452 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
22454 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
22455 attributes for noipa attribute. For naked attribute use
22456 lookup_attribute first before lookup_attribute_spec.
22457 * final.c (rest_of_handle_final): Disable IPA RA for functions with
22459 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
22460 for functions with noipa attribute.
22461 (cgraph_externally_visible_p): Return true for functions with noipa
22463 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
22464 for functions with noipa attribute.
22465 * doc/extend.texi: Document noipa function attribute.
22466 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
22467 also for functions with noipa attribute.
22468 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
22470 2017-07-26 Andrew Pinski <apinski@cavium.com>
22472 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
22473 vec_unalign_load_cost and vec_unalign_store_cost.
22475 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
22477 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
22478 -mvsx-small-integer option.
22479 (ISA_3_0_MASKS_IEEE): Likewise.
22480 (OTHER_VSX_VECTOR_MASKS): Likewise.
22481 (POWERPC_MASKS): Likewise.
22482 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
22483 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
22484 code, only testing for DImode being allowed in non-VSX floating
22486 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
22487 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
22489 (rs6000_option_override_internal): Delete -mvsx-small-integer.
22490 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
22491 TARGET_P8_VECTOR test.
22492 (rs6000_secondary_reload_simple_move): Likewise.
22493 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
22494 since TARGET_P9_VECTOR was already tested.
22495 (rs6000_opt_masks): Remove -mvsx-small-integer.
22496 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
22497 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22499 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
22500 test for TARGET_VEXTRACTUB was used, and that uses
22502 (p9 extract splitter): Likewise.
22503 (vsx_extract_<mode>_di_p9): Likewise.
22504 (vsx_extract_<mode>_store_p9): Likewise.
22505 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
22506 for TARGET_P9_VECTOR was used. Delete code that is now dead with
22507 the elimination of TARGET_VSX_SMALL_INTEGER.
22508 (vsx_extract_<mode>_p8): Likewise.
22509 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
22510 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
22511 (vsx_set_<mode>_p9): Likewise.
22512 (vsx_set_v4sf_p9): Likewise.
22513 (vsx_set_v4sf_p9_zero): Likewise.
22514 (vsx_insert_extract_v4sf_p9): Likewise.
22515 (vsx_insert_extract_v4sf_p9_2): Likewise.
22516 * config/rs6000/rs6000.md (sign extend splitter): Change
22517 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
22518 (floatsi<mode>2_lfiwax_mem): Likewise.
22519 (floatunssi<mode>2_lfiwzx_mem): Likewise.
22520 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
22521 since a test for TARGET_P9_VECTOR was used.
22522 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22523 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
22524 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22525 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22526 TARGET_P8_VECTOR test.
22527 (fix_trunc<mode>si2_stfiwx): Likewise.
22528 (fix_trunc<mode>si2_internal): Likewise.
22529 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
22530 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22532 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22533 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22534 TARGET_P8_VECTOR test.
22535 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22536 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
22537 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22539 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22540 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
22541 since a test for TARGET_P9_VECTOR was used.
22542 (splitter for loading small constants): Likewise.
22544 2017-07-26 Andrew Pinski <apinski@cavium.com>
22546 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
22549 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
22552 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
22553 (fp_valid_at): Likewise.
22555 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22557 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
22558 (qdf24xx_addrcost_table): Likewise.
22559 (cortexa57_tunings): Update to use generic_branch_cost.
22560 (cortexa72_tunings): Likewise.
22561 (cortexa73_tunings): Likewise.
22562 (qdf24xx_tunings): Likewise.
22564 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22566 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
22567 (thunderx2t99_branch_cost): Likewise.
22568 (cortexa35_tunings): Update to use generic_branch_cost.
22569 (cortexa53_tunings): Likewise.
22570 (cortexa57_tunings): Likewise.
22571 (cortexa72_tunings): Likewise.
22572 (cortexa73_tunings): Likewise.
22573 (thunderx2t99_tunings): Likewise.
22575 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22577 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22578 (sparc_option_override): Honour MASK_FSMULD.
22579 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22580 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22581 * config/sparc/sparc.opt (mfsmuld): New option.
22582 * doc/invoke.texi (mfsmuld): Document option.
22584 2017-07-26 Marek Polacek <polacek@redhat.com>
22586 PR middle-end/70992
22587 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22589 2017-07-26 Richard Biener <rguenther@suse.de>
22591 * gimple-match-head.c (do_valueize): Return OP if valueize
22593 (get_def): New helper to get at the def stmt of a SSA name
22594 if valueize allows.
22595 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22596 do_valueize to get at the def stmt.
22597 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22599 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22601 PR middle-end/46932
22602 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22604 2017-07-26 Martin Liska <mliska@suse.cz>
22607 * function.c (expand_function_start): Make expansion of
22608 nonlocal_goto_save_area after parm_birth_insn.
22610 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22612 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22613 from all CPU target flags enable members.
22615 2017-07-26 Richard Biener <rguenther@suse.de>
22617 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22618 (decision_tree::gen): Make 'type' const.
22619 (write_predicate): Likewise.
22621 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22623 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22624 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22625 (rs6000_option_override_internal): Likewise.
22626 (rs6000_expand_vector_set): Likewise.
22627 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22628 (TARGET_UPPER_REGS_SF): Likewise.
22629 (TARGET_UPPER_REGS_DI): Likewise.
22630 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22631 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22632 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22633 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22634 (Splitters for DI constants in Altivec registers): Likewise.
22635 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22636 (vsx_set_v4sf_p9): Likewise.
22637 (vsx_set_v4sf_p9_zero): Likewise.
22638 (vsx_insert_extract_v4sf_p9): Likewise.
22639 (vsx_insert_extract_v4sf_p9_2): Likewise.
22641 2017-07-25 Carl Love <cel@us.ibm.com>
22643 * doc/extend.texi: Update the built-in documentation file for the
22644 existing built-in functions
22645 vector signed char vec_cnttz (vector signed char);
22646 vector unsigned char vec_cnttz (vector unsigned char);
22647 vector signed short vec_cnttz (vector signed short);
22648 vector unsigned short vec_cnttz (vector unsigned short);
22649 vector signed int vec_cnttz (vector signed int);
22650 vector unsigned int vec_cnttz (vector unsigned int);
22651 vector signed long long vec_cnttz (vector signed long long);
22652 vector unsigned long long vec_cnttz (vector unsigned long long);
22654 2017-07-25 Andrew Pinski <apinski@cavium.com>
22656 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22657 accesses where the use is for the first operand of a BIT_INSERT.
22659 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22662 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22663 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22665 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22667 * config/i386/gstabs.h: Delete.
22668 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22670 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22672 * config/i386/i386.c (ix86_decompose_address): Do not check for
22673 register RTX when looking at index_reg or base_reg.
22674 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22676 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22678 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22679 to update EH info here.
22681 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22683 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22685 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22687 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22689 2017-07-25 Torsten Duwe <duwe@suse.de>
22691 * common.opt: Introduce -fpatchable-function-entry
22692 command line option, and its variables function_entry_patch_area_size
22693 and function_entry_patch_area_start.
22694 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22695 including a two-value parser.
22696 * target.def (print_patchable_function_entry): New target hook.
22697 * targhooks.h (default_print_patchable_function_entry): New function.
22698 * targhooks.c (default_print_patchable_function_entry): Likewise.
22699 * toplev.c (process_options): Switch off IPA-RA if
22700 patchable function entries are being generated.
22701 * varasm.c (assemble_start_function): Look at the
22702 patchable-function-entry command line switch and current
22703 function attributes and maybe generate NOP instructions by
22704 calling the print_patchable_function_entry hook.
22705 * doc/extend.texi: Document patchable_function_entry attribute.
22706 * doc/invoke.texi: Document -fpatchable_function_entry
22707 command line option.
22708 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22710 * doc/tm.texi: Re-generate.
22712 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22715 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22716 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22718 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22720 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22722 2017-07-25 Richard Biener <rguenther@suse.de>
22724 PR tree-optimization/81455
22725 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22726 not walk in cycles when looking for guards.
22728 2017-07-25 Richard Biener <rguenther@suse.de>
22730 PR tree-optimization/81529
22731 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22732 when optimizing backedge uses.
22734 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22736 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22738 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22739 to dl_section_ref. On AIX, append an expression to subtract
22740 the size of the section length to dl_section_ref.
22742 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22744 * configure.ac: If any of the config.* scripts fail, exit 1.
22745 * configure: Regenerate.
22747 2017-07-25 Richard Biener <rguenther@suse.de>
22749 PR middle-end/81546
22750 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22751 of immediate uses, be more verbose on errors.
22753 2017-07-25 Richard Biener <rguenther@suse.de>
22755 PR tree-optimization/81510
22756 * tree-vect-loop.c (vect_is_simple_reduction): When the
22757 reduction stmt is not inside the loop bail out.
22759 2017-07-25 Richard Biener <rguenther@suse.de>
22761 PR tree-optimization/81303
22762 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22763 profitability check against LOOP_VINFO_NITERSM1.
22765 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22767 * domwalk.c (cmp_bb_postorder): Simplify.
22768 (sort_bbs_postorder): New function. Use it...
22769 (dom_walker::walk): ...here to optimize common cases.
22771 2017-07-25 Martin Liska <mliska@suse.cz>
22774 * ipa-visibility.c (function_and_variable_visibility): Make the
22775 redirection just on target that supports aliasing.
22776 Fix GNU coding style.
22778 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22781 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22783 * config/arm/rtems.h: Likewise.
22784 * config/bfin/rtems.h: Likewise.
22785 * config/i386/rtemself.h: Likewise.
22786 * config/lm32/rtems.h: Likewise.
22787 * config/m32c/rtems.h: Likewise.
22788 * config/m68k/rtemself.h: Likewise.
22789 * config/microblaze/rtems.h: Likewise.
22790 * config/mips/rtems.h: Likewise.
22791 * config/moxie/rtems.h: Likewise.
22792 * config/nios2/rtems.h: Likewise.
22793 * config/powerpcspe/rtems.h: Likewise.
22794 * config/rs6000/rtems.h: Likewise.
22795 * config/rtems.h: Likewise.
22796 * config/sh/rtems.h: Likewise.
22797 * config/sh/rtemself.h: Likewise.
22798 * config/sparc/rtemself.h: Likewise.
22800 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22803 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22804 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22805 * tree-ssa-structalias.c (alias_get_name): Same.
22807 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22810 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22811 instructions if no du chain is found.
22813 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22815 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22817 2017-07-25 Richard Biener <rguenther@suse.de>
22819 PR middle-end/81505
22820 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
22823 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22825 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
22826 upper-regs options.
22827 (ISA_2_7_MASKS_SERVER): Likewise.
22828 (ISA_3_0_MASKS_IEEE): Likewise.
22829 (OTHER_P8_VECTOR_MASKS): Likewise.
22830 (OTHER_VSX_VECTOR_MASKS): Likewise.
22831 (POWERPC_MASKS): Likewise.
22832 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
22833 duplicate list of options.
22834 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
22835 explicit -mupper-regs options.
22836 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
22837 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
22838 alias for -mupper-regs-df.
22839 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
22840 (rs6000_init_hard_regno_mode_ok): Likewise.
22841 (rs6000_option_override_internal): Likewise.
22842 (rs6000_opt_masks): Likewise.
22843 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
22844 options in terms of whether -mvsx or -mpower8-vector was used.
22845 (TARGET_UPPER_REGS_DI): Likewise.
22846 (TARGET_UPPER_REGS_SF): Likewise.
22847 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
22848 -mupper-regs-* options.
22850 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22852 * passes.c (emergency_dump_function): Print some empty lines and a
22853 header before the RTL dump.
22855 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22857 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
22859 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
22862 * config/aarch64/aarch64.c (aarch64_classify_symbol):
22863 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
22865 2017-07-24 Carl Love <cel@us.ibm.com>
22867 * config/rs6000/rs6000-c.c: Add support for built-in functions
22868 vector float vec_extract_fp32_from_shorth (vector unsigned short);
22869 vector float vec_extract_fp32_from_shortl (vector unsigned short);
22870 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
22871 vec_extract_fp_from_shortl): Add defines for the two builtins.
22872 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
22873 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
22875 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
22876 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
22877 * doc/extend.texi: Update the built-in documentation file for the
22878 new built-in function.
22880 2017-07-24 Jakub Jelinek <jakub@redhat.com>
22883 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
22885 * doc/generic.texi: Likewise.
22886 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22887 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22889 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
22891 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
22892 (aarch64_mls_elt_merge<mode>): Likewise.
22894 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
22896 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22897 having __cxa_atexit.
22899 2017-07-23 Michael Collison <michael.collison@arm.com>
22901 * config/arm/arm.c (arm_option_override): Deprecate
22902 use of -mstructure-size-boundary.
22903 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
22904 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
22906 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22909 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22910 Reduce cost estimate for direct moves.
22912 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
22915 * config/i386/i386.c (ix86_option_override_internal): Disable
22916 BMI, BMI2 and TBM instructions for -m16.
22918 2017-07-21 Carl Love <cel@us.ibm.com>
22920 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22921 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22922 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22923 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22924 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22925 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22926 VMULOSW): New enum "unspec" values.
22927 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22928 altivec_vmulosw): New patterns.
22929 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22930 VMULOSW): Add definitions.
22932 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
22934 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
22935 (qdf24xx): Likewise.
22936 * config/aarch64/aarch64-options-extensions.def (rdma); New.
22937 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
22938 (AARCH64_FL_V8_1): Renumber.
22939 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
22940 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
22941 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
22942 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
22943 rdma to feature modifiers list.
22945 2017-07-21 Yury Gribov <tetra2005@gmail.com>
22947 PR middle-end/56727
22948 * ipa-visibility (function_and_variable_visibility): Convert
22949 recursive PLT call to direct call if appropriate.
22951 2017-07-21 Andrew Pinski <apinski@cavium.com>
22953 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
22954 operand 1 to see if the types precision matches.
22955 * fold-const.c (operand_equal_p): Likewise.
22957 2017-07-21 Richard Biener <rguenther@suse.de>
22959 PR tree-optimization/81303
22960 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
22961 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
22962 (vect_peeling_hash_get_lowest_cost): Adjust.
22963 (vect_enhance_data_refs_alignment): Likewise. Use
22964 vect_get_peeling_costs_all_drs to compute the penalty for no
22965 peeling to match up costs.
22967 2017-07-21 Richard Biener <rguenther@suse.de>
22969 PR tree-optimization/81500
22970 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
22971 we didn't identify a reduction path.
22973 2017-07-21 Tom de Vries <tom@codesourcery.com>
22974 Cesar Philippidis <cesar@codesourcery.com>
22976 PR gcov-profile/81442
22977 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
22980 2017-07-21 Tom de Vries <tom@codesourcery.com>
22983 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
22985 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
22986 nvptx_override_options_after_change.
22988 2017-07-21 Ulrich Drepper <drepper@redhat.com>
22990 * dwarf2out.c (output_file_names): Avoid double testing for
22991 dwarf_version >= 5.
22993 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
22995 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
22997 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22999 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23001 (try_crossjump_to_edge): Do not punt on partitioned functions.
23003 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23005 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23006 Put all BBs reachable only via paths crossing cold region to cold
23008 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23010 2016-07-21 Richard Biener <rguenther@suse.de>
23012 PR tree-optimization/81303
23013 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23014 into account prologue and epilogue iterations when raising
23015 min_profitable_iters to sth at least covering one vector iteration.
23017 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23019 * config/arm/arm.c (arm_test_cpu_arch_dat):
23022 2017-07-20 Nathan Sidwell <nathan@acm.org>
23024 Remove TYPE_METHODS.
23025 * tree.h (TYPE_METHODS): Delete.
23026 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23027 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23028 (dbxout_type_methods): Scan TYPE_FIELDS.
23029 (dbxout_type): Don't check TYPE_METHODS here.
23030 * function.c (use_register_for_decl): Always ignore register for
23031 class types when not optimizing.
23032 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23033 * tree.c (free_lang_data_in_type): Stitch out member functions and
23034 templates from TYPE_FIELDS.
23035 (build_distinct_type_copy, verify_type_variant,
23036 verify_type): Member fns are on TYPE_FIELDS.
23037 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23038 * tree-pretty-print.c (dump_generic_node): Likewise.
23040 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23043 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23044 V2TImode and V4TImode.
23045 (ix86_expand_vector_extract): Likewise.
23046 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23047 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23048 (ssescalarmode): Handle V4TImode and V2TImode.
23049 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23050 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23051 (VEXTRACTI128_MODE): New mode iterator.
23052 (splitter for *vec_extractv?ti first element): New.
23053 (VEC_INIT_MODE): New mode iterator.
23054 (vec_init<mode>): Consolidate 3 expanders into one using
23055 VEC_INIT_MODE mode iterator.
23057 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23059 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23060 non_spilled_static_chain_regno_p.
23062 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23064 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
23066 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
23068 * bb-reorder.c (connect_traces): Allow copying of blocks within
23071 2017-07-20 Richard Biener <rguenther@suse.de>
23073 * gimple.h (gimple_phi_result): Add gphi * overload.
23074 (gimple_phi_result_ptr): Likewise.
23075 (gimple_phi_arg): Likewise. Adjust index assert to only
23076 allow actual argument accesses rather than all slots available
23078 (gimple_phi_arg_def): Add gphi * overload.
23079 * tree-phinodes.c (make_phi_node): Initialize only actual
23081 (resize_phi_node): Clear memory not covered by old node,
23082 do not initialize excess argument slots.
23083 (reserve_phi_args_for_new_edge): Initialize new argument slot
23086 2017-07-20 Bin Cheng <bin.cheng@arm.com>
23088 PR tree-optimization/81388
23090 2016-07-21 Bin Cheng <bin.cheng@arm.com>
23092 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
23093 by removing computation of may_be_zero.
23095 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23096 Tom de Vries <tom@codesourcery.com>
23098 PR middle-end/81030
23099 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
23100 when gimple level profile disagrees with what RTL expander did.
23102 2017-07-20 Richard Biener <rguenther@suse.de>
23104 PR tree-optimization/61171
23105 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
23106 (vect_analyze_stmt): Add slp instance parameter.
23107 (vectorizable_reduction): Likewise.
23108 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
23109 (vect_is_simple_reduction): Deal with chains not detected
23110 as SLP reduction chain, specifically not properly associated
23111 chains containing a mix of plus/minus.
23112 (get_reduction_op): Remove.
23113 (get_initial_defs_for_reduction): Simplify, pass in whether
23114 this is a reduction chain, pass in the SLP node for the PHIs.
23115 (vect_create_epilog_for_reduction): Get the SLP instance as
23117 (vectorizable_reduction): Get the SLP instance as arg.
23118 During analysis remember the SLP node with the PHIs in the
23119 instance. Simplify getting at the vectorized reduction PHIs.
23120 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
23121 through SLP instance.
23122 (vect_slp_analyze_operations): Likewise.
23123 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
23124 (vect_transform_stmt): Likewise.
23126 2017-07-20 Tom de Vries <tom@codesourcery.com>
23128 PR tree-optimization/81489
23129 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
23130 read of phi arg location to before loop that modifies phi.
23132 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23134 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
23137 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23139 PR middle-end/81331
23140 * except.c (execute): Fix ordering issue.
23142 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23144 PR rtl-optimization/81423
23145 * combine.c (make_compound_operation_int): Don't try to optimize
23146 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
23148 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23150 PR rtl-optimization/81423
23151 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
23152 with a constant that is -1 in the truncated to mode.
23154 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23156 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
23157 (determine_unlikely_bbs): ... here.
23158 * predict.h (propagate_unlikely_bbs_forward): Declare.
23159 * cfgexpand.c (pass_expand::execute): Use it.
23160 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
23162 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
23163 propagate_unlikely_bbs_forward.
23165 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23167 PR middle-end/81331
23168 * except.c (maybe_add_nop_after_section_switch): New function.
23171 2017-07-19 Tom de Vries <tom@codesourcery.com>
23173 * gimple.h (gimple_phi_set_arg): Make assert more strict.
23175 2017-07-19 Tom de Vries <tom@codesourcery.com>
23177 * gimple.h (gimple_phi_arg): Make assert more strict.
23179 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
23181 * config.gcc (powerpc*-*-*): Add mmintrin.h.
23182 * config/rs6000/mmintrin.h: New file.
23183 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
23185 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23187 PR tree-optimization/81346
23188 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
23190 2017-07-19 Tom de Vries <tom@codesourcery.com>
23192 * config/nvptx/nvptx.md (VECIM): Add V2DI.
23194 2017-07-19 Tom de Vries <tom@codesourcery.com>
23196 * config/nvptx/nvptx-modes.def: Add V2DImode.
23197 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
23198 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
23199 (nvptx_output_mov_insn): Handle lack of mov.b128.
23200 (nvptx_print_operand): Handle 'H' and 'L' codes.
23201 (nvptx_vector_mode_supported): Allow V2DImode.
23202 (nvptx_preferred_simd_mode): New function.
23203 (nvptx_data_alignment): New function.
23204 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
23205 nvptx_preferred_simd_mode.
23206 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
23208 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
23210 2017-07-19 Tom de Vries <tom@codesourcery.com>
23212 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
23213 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
23214 (nvptx_vector_mode_supported): New function. Allow V2SImode.
23215 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
23216 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
23217 (mov<VECIM>_insn): New define_insn.
23218 (define_expand "mov<VECIM>): New define_expand.
23220 2017-07-19 Tom de Vries <tom@codesourcery.com>
23222 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
23224 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23226 PR tree-optimization/81346
23227 * fold-const.h (fold_div_compare, range_check_type): Declare.
23228 * fold-const.c (range_check_type): New function.
23229 (build_range_check): Use range_check_type.
23230 (fold_div_compare): No longer static, rewritten into
23231 a match.pd helper function.
23232 (fold_comparison): Don't call fold_div_compare here.
23233 * match.pd (X / C1 op C2): New optimization using fold_div_compare
23234 as helper function.
23236 2017-07-19 Nathan Sidwell <nathan@acm.org>
23238 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
23239 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
23240 * tree.c (find_decls_types_r, verify_type): Use
23241 TYPE_{MIN,MAX}_VALUE_RAW.
23242 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
23243 (hash_tree): Likewise.
23244 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
23246 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
23249 2017-07-18 Tom de Vries <tom@codesourcery.com>
23251 PR middle-end/81464
23252 * omp-expand.c (expand_omp_for_static_chunk): Handle
23253 equal-argument loop exit phi.
23255 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
23258 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
23259 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
23260 operand 2 predicate.
23261 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
23262 operand 2 predicate.
23263 (ror,rol -> rorx splitters): Use const_int_operand as
23264 operand 2 predicate.
23266 2017-06-18 Richard Biener <rguenther@suse.de>
23268 PR tree-optimization/81410
23269 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
23270 the gap in the ! slp_perm SLP case after each group.
23272 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23274 PR middle-end/81463
23275 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
23278 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23280 PR middle-end/81462
23281 * predict.c (set_even_probabilities): Cleanup; do not affect
23282 probabilities that are already known.
23283 (combine_predictions_for_bb): Call even when count is set.
23285 2017-07-18 Nathan Sidwell <nathan@acm.org>
23287 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
23290 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23293 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
23294 optimization for loop niter analysis.
23296 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
23299 * config/avr/avr.c (avr_optimize_casesi): Don't use
23300 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
23302 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
23304 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
23305 body_cost_vec from _vect_peel_extended_info.
23306 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
23307 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
23310 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23312 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
23314 2017-07-18 Richard Biener <rguenther@suse.de>
23316 PR tree-optimization/80620
23317 PR tree-optimization/81403
23318 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
23319 info when re-using a VN table entry.
23321 2017-07-18 Richard Biener <rguenther@suse.de>
23323 PR tree-optimization/81418
23324 * tree-vect-loop.c (vectorizable_reduction): Properly compute
23325 vectype_in. Verify that with lane-reducing reduction operations
23326 we have a single def-use cycle.
23328 2017-07-17 Carl Love <cel@us.ibm.com>
23330 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
23332 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23333 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23334 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23335 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23336 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23337 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23338 VMULOSW): New enum "unspec" values.
23339 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23340 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23341 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23342 altivec_vmulosw): New patterns.
23343 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23344 VMULOSW): Add definitions.
23346 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
23348 * config/alpha/alpha.c: Include predict.h.
23350 2017-07-17 Yury Gribov <tetra2005@gmail.com>
23352 * tree-vrp.c (compare_assert_loc): Fix comparison function
23353 to return predictable results.
23355 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23357 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
23359 (subdi3): Likewise.
23360 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
23361 * doc/invoke.texi (mexpand-adddi): Update text.
23363 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23365 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
23366 that also clobbers the CC register. The old expand code is moved
23368 (*arc_clzsi2): ... here.
23369 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
23370 the CC register. The old expand code is moved to ...
23371 (arc_ctzsi2): ... here.
23373 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23375 * config/arc/arc.opt (mindexed-loads): Use initial value
23376 TARGET_INDEXED_LOADS_DEFAULT.
23377 (mauto-modify-reg): Use initial value
23378 TARGET_AUTO_MODIFY_REG_DEFAULT.
23379 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23380 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23381 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23382 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23384 2017-07-17 Martin Liska <mliska@suse.cz>
23387 * opts.c (finish_options): Do not allow -fgnu-tm
23388 w/ -fsanitize={kernel-,}address. Say sorry.
23390 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23393 * tree-loop-distribution.c (classify_partition): Only assert on
23394 numer of iterations.
23395 (merge_dep_scc_partitions): Delete prameter. Update function call.
23396 (distribute_loop): Remove code handling loop with unknown niters.
23397 (pass_loop_distribution::execute): Skip loop with unknown niters.
23399 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23402 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
23403 function sort_partitions_by_post_order.
23405 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23407 PR tree-optimization/81374
23408 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
23409 the max index of basic blocks, rather than number of basic blocks.
23411 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23413 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
23415 (arc_legitimate_pic_operand_p): Likewise.
23416 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
23418 (arc_needs_pcl_p): Likewise.
23419 (arc_legitimate_pc_offset_p): Likewise.
23420 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
23421 function is also used in constrains.md.
23422 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
23423 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
23424 PLUS. Only return true/false in known cases, otherwise assert.
23425 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
23426 is already called in arc_legitimate_constant_p.
23427 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
23429 (LEGITIMATE_PIC_OPERAND_P): Use
23430 arc_raw_symbolic_reference_mentioned_p function.
23431 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
23436 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23437 Andrew Burgess <andrew.burgess@embecosm.com>
23439 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
23440 (arc_return_address_register): New function.
23441 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
23442 (arc_handle_fndecl_attribute): Add naked attribute.
23443 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
23444 (TARGET_WARN_FUNC_RETURN): Likewise.
23445 (arc_allocate_stack_slots_for_args): New function.
23446 (arc_warn_func_return): Likewise.
23447 (machine_function): Change type fn_type.
23448 (arc_compute_function_type): Consider new naked function type,
23449 change function return type.
23450 (arc_must_save_register): Adapt to handle new
23451 arc_compute_function_type's return type.
23452 (arc_expand_prologue): Likewise.
23453 (arc_expand_epilogue): Likewise.
23454 (arc_return_address_regs): Delete.
23455 (arc_return_address_register): New function.
23456 (arc_epilogue_uses): Use above function.
23457 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
23458 (arc_function_type): Change encoding, add naked type.
23459 (ARC_INTERRUPT_P): Change to handle the new encoding.
23460 (ARC_FAST_INTERRUPT_P): Likewise.
23461 (ARC_NORMAL_P): Define.
23462 (ARC_NAKED_P): Likewise.
23463 (arc_compute_function_type): Delete prototype.
23464 * config/arc/arc.md (in_ret_delay_slot): Use
23465 arc_return_address_register function.
23466 (simple_return): Likewise.
23467 (p_return_i): Likewise.
23469 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23471 PR tree-optimization/81428
23472 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
23473 can't be built for those types.
23475 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23477 Remove stuff dead since r239246.
23479 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
23480 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
23481 (avr_inform_devices): Remove dead stuff.
23483 2017-07-17 Tamar Christina <tamar.christina@arm.com>
23485 * config/arm/arm_neon.h: Fix softp typo.
23487 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23489 PR tree-optimization/81365
23490 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
23491 aggregate moves onto bb predecessor edges, make sure there are no
23492 loads that could alias the lhs in between the start of bb and the
23495 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23498 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
23499 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
23500 [LSHIFTRT, outer_code = TRUNCATE]: Same.
23502 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23504 PR tree-optimization/81396
23505 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
23506 (init_symbolic_number): Initialize it to 1.
23507 (perform_symbolic_merge): Add n_ops from both operands into the new
23509 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
23510 without base_addr as useless if they need more than one operation.
23511 (bswap_replace): Handle !bswap case for NULL base_addr.
23513 2017-07-17 Tom de Vries <tom@codesourcery.com>
23516 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
23519 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
23521 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
23522 conditional builtin define __FIX_LEON3FT_B2BST.
23524 2017-07-17 Daniel Cederman <cederman@gaisler.com>
23526 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
23527 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
23530 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23532 PR rtl-optimization/81424
23533 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
23534 to remove potential trapping from operands if -fnon-call-exceptions.
23536 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23538 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
23539 profile_proability for scalling.
23540 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
23542 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23544 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
23546 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23548 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
23549 fixpoint arithmetics.
23551 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23553 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
23554 fixpoint arithmetics.
23556 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23558 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
23559 fixpoint arithmetics.
23561 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23563 * profile-count.h (profile_probability::from_reg_br_prob_note,
23564 profile_probability::to_reg_br_prob_note): New functions.
23565 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
23566 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
23567 * predict.c (probability_reliable_p): Update.
23568 (edge_probability_reliable_p): Update.
23569 (br_prob_note_reliable_p): Update.
23570 (invert_br_probabilities): Update.
23571 (add_reg_br_prob_note): New function.
23572 (combine_predictions_for_insn): Update.
23573 * asan.c (asan_clear_shadow): Update.
23574 * cfgbuild.c (compute_outgoing_frequencies): Update.
23575 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23576 (update_br_prob_note): Update.
23577 (rtl_verify_edges): Update.
23578 (purge_dead_edges): Update.
23579 (fixup_reorder_chain): Update.
23580 * emit-rtl.c (try_split): Update.
23581 * ifcvt.c (cond_exec_process_insns): Update.
23582 (cond_exec_process_if_block): Update.
23583 (dead_or_predicable): Update.
23584 * internal-fn.c (expand_addsub_overflow): Update.
23585 (expand_neg_overflow): Update.
23586 (expand_mul_overflow): Update.
23587 * loop-doloop.c (doloop_modify): Update.
23588 * loop-unroll.c (compare_and_jump_seq): Update.
23589 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23590 * predict.h: Update.
23591 * reorg.c (mostly_true_jump): Update.
23593 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23594 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23595 * config/arc/arc.c: (emit_unlikely_jump): Update.
23596 * config/arm/arm.c: (emit_unlikely_jump): Update.
23597 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23598 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23599 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23600 (ix86_print_operand): Update.
23601 (ix86_split_fp_branch): Update.
23602 (predict_jump): Update.
23603 * config/ia64/ia64.c (ia64_print_operand): Update.
23604 * config/mmix/mmix.c (mmix_print_operand): Update.
23605 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23606 (rs6000_expand_split_stack_prologue): Update.
23607 * config/rs6000/rs6000.c: Update.
23608 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23609 (s390_expand_vec_movstr): Update.
23610 (s390_expand_cs_tdsi): Update.
23611 (s390_expand_split_stack_prologue): Update.
23612 * config/sh/sh.c (sh_print_operand): Update.
23613 (expand_cbranchsi4): Update.
23614 (expand_cbranchdi4): Update.
23615 * config/sparc/sparc.c (output_v9branch): Update.
23616 * config/spu/spu.c (get_branch_target): Update.
23617 (ea_load_store_inline): Update.
23618 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23619 * config/tilepro/tilepro.c: Update.
23621 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23623 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23624 (gimplify_save_expr): Likewise.
23626 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23628 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23630 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23632 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23634 * timevar.def (TV_IPA_FNSUMMARY): Define.
23636 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23638 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23639 to back store errata sensitive sequence from being generated.
23640 (sqrtdf2_fix): Likewise.
23642 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23644 * tree-ssa-threadupdate.c (compute_path_counts,
23645 update_joiner_offpath_counts): Use profile_probability.
23647 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23650 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23652 * config/arm/arm-c.c (arm_cpu_builtins): Define
23653 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23655 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23657 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23658 array entries to represent __ieee128 versions of the
23659 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23660 scalar_extract_sig, and scalar_insert_exp built-in functions.
23661 (altivec_resolve_overloaded_builtin): Add special case handling
23662 for the __builtin_scalar_insert_exp function, as represented by
23663 the P9V_BUILTIN_VEC_VSIEDP constant.
23664 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23665 exponent support for __ieee128 argument.
23666 (VSESQP): Add scalar extract signature support for __ieee128
23668 (VSTDCNQP): Add scalar test negative support for __ieee128
23670 (VSIEQP): Add scalar insert exponent support for __int128 argument
23671 with __ieee128 result.
23672 (VSIEQPF): Add scalar insert exponent support for __ieee128
23673 argument with __ieee128 result.
23674 (VSTDCQP): Add scalar test data class support for __ieee128
23676 (VSTDCNQP): Add overload support for scalar test negative with
23677 __ieee128 argument.
23678 (VSTDCQP): Add overload support for scalar test data class
23679 __ieee128 argument.
23680 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23681 UNSPEC_VSX_SXSIGDP.
23682 (UNSPEC_VSX_SIEXPQP): New constant.
23683 (xsxexpqp): New insn for VSX scalar extract exponent quad
23685 (xsxsigqp): New insn for VSX scalar extract significand quad
23687 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23688 precision with floating point argument.
23689 (xststdcqp): New expand for VSX scalar test data class quad
23691 (xststdcnegqp): New expand for VSX scalar test negative quad
23693 (xststdcqp): New insn to match expansions for VSX scalar test data
23694 class quad precision and VSX scalar test negative quad precision.
23695 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23696 special case operand checking to enforce that second operand of
23697 VSX scalar test data class with quad precision argument is a 7-bit
23699 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23700 prototypes and descriptions of __ieee128 versions of
23701 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23702 scalar_test_data_class, and scalar_test_neg built-in functions.
23704 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23706 PR tree-optimization/81162
23707 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23708 replace a negate with an add.
23710 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23712 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23714 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23716 * config/arm/arm-c.c (arm_cpu_builtins): Define
23717 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23719 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23721 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23722 (armv8-r): Set ARM Cortex-R52 as default CPU.
23723 * config/arm/arm-tables.opt: Regenerate.
23724 * config/arm/arm-tune.md: Regenerate.
23725 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23727 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23728 extension for -mcpu=cortex-r52.
23730 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23732 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23733 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23734 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23735 (fp-armv8): Define it as FP_ARMv8 only.
23736 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23737 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23739 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23740 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23741 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23742 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23743 than TARGET_FPU_ARMV8.
23744 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23745 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23746 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23748 * config/arm/neon.md (neon_vrint): Likewise.
23749 (neon_vcvt): Likewise.
23750 (neon_<fmaxmin_op><mode>): Likewise.
23751 (<fmaxmin><mode>3): Likewise.
23752 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23753 * config/arm/predicates.md (arm_cond_move_operator): Check against
23754 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23756 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23758 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23759 to top of function.
23761 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23763 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23764 loop in comment with memset.
23766 2017-07-14 Martin Liska <mliska@suse.cz>
23768 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23769 * dwarf2out.c (is_java): Remove the function.
23770 (output_pubname): Remove usage of the function.
23771 (lower_bound_default): Remove usage of DW_LANG_Java.
23772 (gen_compile_unit_die): Likewise.
23773 * gcc.c: Remove compiler defaults for .java and .zip files.
23774 * gimple-expr.c (remove_suffix): Change as there's no longer
23775 extension than 4-letter one.
23776 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23777 (gimplify_save_expr): Likewise.
23778 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23779 as it's possible even for other languages than Java.
23780 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23781 * lto-opts.c (lto_write_options): Remove reference to Java.
23782 * opts.c (strip_off_ending): Update file extension handling.
23783 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23784 * tree-eh.c (lower_resx): Likewise.
23785 * tree.c (free_lang_data_in_type): Remove dead code.
23786 (find_decls_types_r): Likewise.
23787 (build_common_builtin_nodes): Remove Java from a comment.
23788 (verify_type): Remove dead code.
23789 * varasm.c (assemble_external): Remove Java from a comment.
23791 2017-07-14 Martin Liska <mliska@suse.cz>
23793 * opts.c (finish_options): Add quotes.
23794 (common_handle_option): Likewise.
23796 2017-07-14 Martin Liska <mliska@suse.cz>
23798 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23799 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23800 Remove N_SO_PASCAL.
23801 * dwarf2out.c (lower_bound_default): Do not handle
23803 (gen_compile_unit_die): Likewise.
23804 * gcc.c: Remove default extension binding for GNU Pascal.
23805 * stmt.c: Remove Pascal language from a comment.
23806 * xcoffout.c: Likewise.
23808 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23811 * diagnostic-show-locus.c (fixit_cmp): New function.
23812 (layout::layout): Sort m_fixit_hints.
23813 (column_range::column_range): Assert that the values are valid.
23814 (struct char_span): New struct.
23815 (correction::overwrite): New method.
23816 (struct source_line): New struct.
23817 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23818 calls in terms of classes source_line and char_span, and
23819 correction::overwrite.
23820 (selftest::test_overlapped_fixit_printing_2): New function.
23821 (selftest::diagnostic_show_locus_c_tests): Call it.
23823 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
23825 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
23826 early if there is no lhs.
23828 2017-07-13 Martin Liska <mliska@suse.cz>
23830 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
23831 (gen_reference_type_die): Likewise.
23832 * stor-layout.c: Remove Pascal-related comment.
23834 2017-07-13 Martin Liska <mliska@suse.cz>
23836 * opts.c (finish_options): Add quotes to error messages.
23837 (parse_sanitizer_options): Likewise.
23839 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
23841 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
23843 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
23845 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
23847 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23849 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
23851 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
23853 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23856 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
23857 provides the hardware capability bits, define the macro
23858 __BUILTIN_CPU_SUPPORTS__.
23859 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
23860 if GLIBC does not provide the hardware capability bits. Add a
23861 gcc_unreachable call if the built-in cpu function is neither
23862 __builtin_cpu_is nor __builtin_cpu_supports.
23863 (rs6000_get_function_versions_dispatcher): Change the warning
23864 that an old GLIBC is used which does not export the capability
23865 bits to be an error.
23866 * doc/extend.texi (target_clones attribute): Document the
23867 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
23868 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
23869 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
23870 the macros defined by GCC if the newer GLIBC is available.
23872 2017-07-12 Jeff Law <law@redhat.com>
23874 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
23875 remaining includes slightly.
23876 * config/riscv/riscv-builtins.c: Include profile-count.h.
23878 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23881 * config/avr/avr.c (avr_set_current_function): In diagnostic
23882 messages: Quote keywords and (parts of) identifiers.
23883 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
23886 2017-07-12 Carl Love <cel@us.ibm.com>
23888 * config/rs6000/rs6000-c.c: Add support for built-in functions
23889 vector bool char vec_revb (vector bool char);
23890 vector bool short vec_revb (vector short char);
23891 vector bool int vec_revb (vector bool int);
23892 vector bool long long vec_revb (vector bool long long);
23893 * doc/extend.texi: Update the built-in documentation file for the
23894 new built-in functions.
23896 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23898 * config/s390/s390.md: Remove movcc splitter.
23900 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23902 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
23903 load/store on condition.
23905 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23908 * config/avr/avr.c (avr_encode_section_info)
23909 [progmem && !TREE_READONLY]: Error if progmem object needs
23912 2017-07-11 Michael Collison <michael.collison@arm.com>
23914 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
23917 2017-07-11 Carl Love <cel@us.ibm.com>
23919 * config/rs6000/rs6000-c.c: Add support for builtins
23920 vector unsigned int vec_parity_lsbb (vector signed int);
23921 vector unsigned int vec_parity_lsbb (vector unsigned int);
23922 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
23923 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
23924 vector unsigned long long vec_parity_lsbb (vector signed long long);
23925 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
23926 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
23927 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
23928 * doc/extend.texi: Update the built-in documentation file for the
23929 new built-in functions.
23931 2017-07-11 David Malcolm <dmalcolm@redhat.com>
23933 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
23934 (layout::m_primary_loc): New field.
23935 (layout::layout): Initialize new field. Move location filtering
23936 logic from here to...
23937 (layout::maybe_add_location_range): ...this new method. Add
23938 support for filtering to just the lines already specified by other
23940 (layout::will_show_line_p): New method.
23941 (gcc_rich_location::add_location_if_nearby): New method.
23942 (selftest::test_add_location_if_nearby): New test function.
23943 (selftest::diagnostic_show_locus_c_tests): Call it.
23944 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
23947 2017-07-11 Tom de Vries <tom@codesourcery.com>
23949 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
23950 (bb_first_real_insn): New function.
23951 (nvptx_single): Add extra initialization of broadcasted condition
23954 2017-07-11 Nathan Sidwell <nathan@acm.org>
23956 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
23958 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
23960 * doc/extend.texi (AVR Function Attributes): Remove weblink to
23961 Binutils doc as TEXI will mess them up.
23962 * doc/invoke.texi (AVR Options): Same here.
23964 2017-07-11 Daniel Cederman <cederman@gaisler.com>
23966 * config/sparc/sparc.opt (mfix-ut700): New option.
23967 (mfix-gr712rc): Likewise.
23968 (sparc_fix_b2bst): New variable.
23969 * doc/invoke.texi (SPARC options): Document them.
23970 (ARM options): Fix warnings.
23971 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
23972 instructions to prevent sequences that can trigger the store-store
23973 errata for certain LEON3FT processors.
23974 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
23975 (sparc_option_override): Set sparc_fix_b2bst appropriately.
23976 * config/sparc/sparc.md (fix_b2bst): New attribute.
23977 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
23979 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
23982 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
23984 (*rsqrtsf2_sse): Ditto.
23986 (div<mode>3): Macroize insn from divdf3 and divsf3
23987 using MODEF mode iterator.
23989 2017-07-10 Martin Sebor <msebor@redhat.com>
23991 PR tree-optimization/80397
23992 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
23993 instead of testing for equality to INTEGER_TYPE.
23995 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
23997 * config.gcc: Remove uclibc from arc target spec.
23999 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24001 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24003 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24006 * lto-wrapper.c (remove_option): New function.
24007 (merge_and_complain): Merge PIC/PIE options more realistically.
24009 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24011 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24015 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24016 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24017 * config.in: Regenerate.
24018 * configure: Regenerate.
24019 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24020 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24021 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24022 (TARGET_GASISR_PROLOGUES): ...target mask.
24023 * common/config/avr/avr-common.c
24024 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24025 Set -mgas-isr-prologues.
24026 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24027 INSERT_PASS_BEFORE for it.
24028 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24029 * config/avr/avr.c (avr_option_override)
24030 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24031 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24032 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24033 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24034 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24036 (make_avr_pass_pre_proep): New function.
24037 (emit_push_sfr) <treg>: Add argument to function and use it
24038 instead of TMP_REG.
24039 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24040 and set machine->gasisr.yes.
24041 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24042 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24043 __gcc_isr.n_pushed to .L__stack_usage.
24044 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24045 (avr_asm_final_postscan_insn): ...this new static function.
24046 * config/avr/avr.h (machine_function)
24047 <is_no_gccisr, use_L__stack_usage>: New fields.
24048 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24049 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24050 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24051 (gasisr, *gasisr): New expander and insn.
24052 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24053 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24054 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24056 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24058 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24061 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24063 Move jump-tables out of .text again.
24066 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
24067 (ASM_OUTPUT_ADDR_VEC): New function.
24068 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
24069 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
24070 INSN_ADDRESSes as asm comment.
24071 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
24072 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
24073 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
24074 * config/avr/avr.md (*tablejump): Adjust comment.
24075 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
24076 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
24078 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
24079 (avr_output_addr_vec): New proto.
24080 (avr_log_t) <insn_addresses>: New field.
24082 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
24085 * config/i386/i386.c (ix86_function_arg_advance): Set
24086 outgoing_args_on_stack to true if there are outgoing arguments
24088 (ix86_function_arg): Likewise.
24089 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
24090 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
24091 * config/i386/i386.h (machine_function): Add
24092 outgoing_args_on_stack.
24094 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
24096 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
24097 supporting pthreds.
24098 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
24100 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
24102 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
24103 (REAL_H): Remove $(MACHMODE_H).
24104 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
24106 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
24107 $(MACHMODE_H) and double-int.h.
24108 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
24110 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
24113 2017-07-07 Andrew Pinski <apinski@cavium.com>
24115 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
24116 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
24118 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
24120 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
24121 Add warning if GCC was not configured to link against a GLIBC that
24122 exports the hardware capability bits.
24123 (make_resolver_func): Make resolver function private and not a
24124 COMDAT function. Create the name with clone_function_name instead
24125 of make_unique_name.
24128 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
24129 correct operand in doing the split.
24131 2017-07-07 Carl Love <cel@us.ibm.com>
24133 * config/rs6000/rs6000-c: Add support for built-in function
24134 vector unsigned short vec_pack_to_short_fp32 (vector float,
24136 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
24137 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
24138 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
24139 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
24140 (convert_4f32_8i16): Add define_expand.
24141 * doc/extend.texi: Update the built-in documentation file for the
24142 new built-in function.
24144 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24146 * config/sparc/m8.md: New file.
24147 * config/sparc/sparc.md: Include m8.md.
24149 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24151 * config/sparc/sparc.opt: New option -mvis4b.
24152 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
24153 (sparc_option_override): Handle VIS4B.
24154 (enum sparc_builtins): Define
24155 SPARC_BUILTIN_DICTUNPACK{8,16,32},
24156 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
24157 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
24158 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
24159 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
24160 (check_constant_argument): New function.
24161 (sparc_vis_init_builtins): Define builtins
24162 __builtin_vis_dictunpack{8,16,32},
24163 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
24164 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
24165 __builtin_vis_fpcmpde{8,16,32}shl and
24166 __builtin_vis_fpcmpur{8,16,32}shl.
24167 (sparc_expand_builtin): Check that the constant operands to
24168 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
24169 constant and in range.
24170 * config/sparc/sparc-c.c (sparc_target_macros): Handle
24172 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
24173 (SPARC_IMM5_P): Likewise.
24174 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
24175 (enabled): Handle vis4b.
24176 (UNSPEC_DICTUNPACK): New unspec.
24177 (UNSPEC_FPCMPSHL): Likewise.
24178 (UNSPEC_FPUCMPSHL): Likewise.
24179 (UNSPEC_FPCMPDESHL): Likewise.
24180 (UNSPEC_FPCMPURSHL): Likewise.
24181 (cpu_feature): New CPU feature `vis4b'.
24182 (dictunpack{8,16,32}): New insns.
24183 (FPCSMODE): New mode iterator.
24184 (fpcscond): New code iterator.
24185 (fpcsucond): Likewise.
24186 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
24187 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
24188 (fpcmpde{8,16,32}{si,di}shl): Likewise.
24189 (fpcmpur{8,16,32}{si,di}shl): Likewise.
24190 * config/sparc/constraints.md: Define constraints `q' for unsigned
24191 2-bit integer constants and `t' for unsigned 5-bit integer
24193 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
24195 (imm5_operand_dictunpack16): Likewise.
24196 (imm5_operand_dictunpack32): Likewise.
24197 (imm2_operand): Likewise.
24198 * doc/invoke.texi (SPARC Options): Document -mvis4b.
24199 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
24200 ditunpack* and fpcmp*shl builtins.
24202 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24204 * config.gcc: Handle m8 in --with-{cpu,tune} options.
24205 * config.in: Add HAVE_AS_SPARC6 define.
24206 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
24208 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
24210 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
24211 (CPP_CPU_SPEC): Handle m8.
24212 (ASM_CPU_SPEC): Likewise.
24213 * config/sparc/sparc-opts.h (enum processor_type): Add
24215 * config/sparc/sparc.c (m8_costs): New struct.
24216 (sparc_option_override): Handle TARGET_CPU_m8.
24217 (sparc32_initialize_trampoline): Likewise.
24218 (sparc64_initialize_trampoline): Likewise.
24219 (sparc_issue_rate): Likewise.
24220 (sparc_register_move_cost): Likewise.
24221 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
24222 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
24223 (ASM_CPU64_DEFAULT_SPEC): Likewise.
24224 (CPP_CPU_SPEC): Handle M8.
24225 (ASM_CPU_SPEC): Likewise.
24226 (AS_M8_FLAG): Define.
24227 * config/sparc/sparc.md: Add m8 to the cpu attribute.
24228 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
24229 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
24231 * configure: Regenerate.
24232 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
24235 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24237 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
24239 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
24240 ("*movdi_insn_sp32"): Do not set v3pipe.
24241 ("*movsi_insn"): Likewise.
24242 ("*movdi_insn_sp64"): Likewise.
24243 ("*movsf_insn"): Likewise.
24244 ("*movdf_insn_sp32"): Likewise.
24245 ("*movdf_insn_sp64"): Likewise.
24246 ("*zero_extendsidi2_insn_sp64"): Likewise.
24247 ("*sign_extendsidi2_insn"): Likewise.
24248 ("*mov<VM32:mode>_insn"): Likewise.
24249 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24250 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24251 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24252 ("<vlop:code><VL:mode>3"): Likewise.
24253 ("*not_<vlop:code><VL:mode>3"): Likewise.
24254 ("*nand<VL:mode>_vis"): Likewise.
24255 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
24256 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
24257 ("one_cmpl<VL:mode>2"): Likewise.
24258 ("faligndata<VM64:mode>_vis"): Likewise.
24259 ("alignaddrsi_vis"): Likewise.
24260 ("alignaddrdi_vis"): Likweise.
24261 ("alignaddrlsi_vis"): Likewise.
24262 ("alignaddrldi_vis"): Likewise.
24263 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24264 ("bmaskdi_vis"): Likewise.
24265 ("bmasksi_vis"): Likewise.
24266 ("bshuffle<VM64:mode>_vis"): Likewise.
24267 ("cmask8<P:mode>_vis"): Likewise.
24268 ("cmask16<P:mode>_vis"): Likewise.
24269 ("cmask32<P:mode>_vis"): Likewise.
24270 ("pdistn<P:mode>_vis"): Likewise.
24271 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24273 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24275 * config/sparc/sparc.md ("subtype"): New insn attribute.
24276 ("*wrgsr_sp64"): Set insn subtype.
24277 ("*rdgsr_sp64"): Likewise.
24278 ("alignaddrsi_vis"): Likewise.
24279 ("alignaddrdi_vis"): Likewise.
24280 ("alignaddrlsi_vis"): Likewise.
24281 ("alignaddrldi_vis"): Likewise.
24282 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24283 ("fexpand_vis"): Likewise.
24284 ("fpmerge_vis"): Likewise.
24285 ("faligndata<VM64:mode>_vis"): Likewise.
24286 ("bshuffle<VM64:mode>_vis"): Likewise.
24287 ("cmask8<P:mode>_vis"): Likewise.
24288 ("cmask16<P:mode>_vis"): Likewise.
24289 ("cmask32<P:mode>_vis"): Likewise.
24290 ("fchksm16_vis"): Likewise.
24291 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
24292 ("fmean16_vis"): Likewise.
24293 ("fp<plusminus_insn>64_vis"): Likewise.
24294 ("<plusminus_insn>v8qi3"): Likewise.
24295 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24296 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
24297 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
24298 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
24299 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
24300 ("*movqi_insn"): Likewise.
24301 ("*movhi_insn"): Likewise.
24302 ("*movsi_insn"): Likewise.
24303 ("movsi_pic_gotdata_op"): Likewise.
24304 ("*movdi_insn_sp32"): Likewise.
24305 ("*movdi_insn_sp64"): Likewise.
24306 ("movdi_pic_gotdata_op"): Likewise.
24307 ("*movsf_insn"): Likewise.
24308 ("*movdf_insn_sp32"): Likewise.
24309 ("*movdf_insn_sp64"): Likewise.
24310 ("*zero_extendhisi2_insn"): Likewise.
24311 ("*zero_extendqihi2_insn"): Likewise.
24312 ("*zero_extendqisi2_insn"): Likewise.
24313 ("*zero_extendqidi2_insn"): Likewise.
24314 ("*zero_extendhidi2_insn"): Likewise.
24315 ("*zero_extendsidi2_insn_sp64"): Likewise.
24316 ("ldfsr"): Likewise.
24317 ("prefetch_64"): Likewise.
24318 ("prefetch_32"): Likewise.
24319 ("tie_ld32"): Likewise.
24320 ("tie_ld64"): Likewise.
24321 ("*tldo_ldub_sp32"): Likewise.
24322 ("*tldo_ldub1_sp32"): Likewise.
24323 ("*tldo_ldub2_sp32"): Likewise.
24324 ("*tldo_ldub_sp64"): Likewise.
24325 ("*tldo_ldub1_sp64"): Likewise.
24326 ("*tldo_ldub2_sp64"): Likewise.
24327 ("*tldo_ldub3_sp64"): Likewise.
24328 ("*tldo_lduh_sp32"): Likewise.
24329 ("*tldo_lduh1_sp32"): Likewise.
24330 ("*tldo_lduh_sp64"): Likewise.
24331 ("*tldo_lduh1_sp64"): Likewise.
24332 ("*tldo_lduh2_sp64"): Likewise.
24333 ("*tldo_lduw_sp32"): Likewise.
24334 ("*tldo_lduw_sp64"): Likewise.
24335 ("*tldo_lduw1_sp64"): Likewise.
24336 ("*tldo_ldx_sp64"): Likewise.
24337 ("*mov<VM32:mode>_insn"): Likewise.
24338 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24339 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24341 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24343 * config/sparc/sparc.md ("type"): New insn type viscmp.
24344 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
24346 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
24347 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
24348 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24349 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
24351 ("n7_vis_logical_11cycle"): Likewise.
24352 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
24353 * config/sparc/niagara2.md ("niag3_vis": Likewise.
24354 * config/sparc/niagara.md ("niag_vis"): Likewise.
24355 * config/sparc/ultra3.md ("us3_fga"): Likewise.
24356 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
24358 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24360 * config/sparc/sparc.md: New instruction type `bmask'.
24361 (bmaskdi_vis): Use the `bmask' type.
24362 (bmasksi_vis): Likewise.
24363 * config/sparc/ultra3.md (us3_array): Likewise.
24364 * config/sparc/niagara7.md (n7_array): Likewise.
24365 * config/sparc/niagara4.md (n4_array): Likewise.
24366 * config/sparc/niagara2.md (niag2_vis): Likewise.
24367 (niag3_vis): Likewise.
24368 * config/sparc/niagara.md (niag_vis): Likewise.
24370 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24372 * ipa-comdats.c: Remove optimize check from gate.
24373 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
24374 for functions not optimized.
24375 (ipa_fn_summary_read): Skip optimize check.
24376 (ipa_fn_summary_write): Likewise.
24377 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
24379 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
24381 (can_inline_edge_p): Check flag_pcc_struct_return for match.
24382 (check_callers): Give up on caller which is not optimized.
24383 (inline_small_functions): Likewise.
24384 (ipa_inline): Do not give up when not optimizing.
24385 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
24386 away unoptimizes cdtors.
24387 (whole_program_function_and_variable_visibility): Do
24388 ipa_discover_readonly_nonaddressable_vars in LTO mode.
24389 * ipa.c (process_references): Do not check optimize.
24390 (symbol_table::remove_unreachable_nodes): Update optimize check.
24391 (set_writeonly_bit): Update optimize check.
24392 (pass_ipa_cdtor_merge::gate): Do not check optimize.
24393 (pass_ipa_single_use::gate): Remove.
24395 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24397 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
24398 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
24399 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
24400 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
24401 permute_load, permute_store, adjust_extract, adjust_splat,
24402 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
24403 replace_swap_with_copy, dump_swap_insn_table,
24404 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
24405 recombine_lvx_pattern, recombine_stvx_pattern,
24406 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
24407 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
24408 to file rs6000-p8swap.c.
24409 * config/rs6000/rs6000-p8swap.c: New file.
24410 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
24411 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
24412 and rs6000*-*-* targets.
24414 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24416 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
24418 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24420 * lto-wrapper.c (merge_and_complain): Do not merge
24421 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
24422 fsigned_zeros, ftrapping_math, fwrapv.
24423 (append_compiler_options): Do not track these options.
24424 (append_linker_options): Likewie
24426 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24428 * cgraphunit.c (cgraph_node::finalize_function): When
24429 !flag_toplevel_reorde set no_reorder flag.
24430 (varpool_node::finalize_decl): Likewise.
24431 (symbol_table::compile): Drop no toplevel reorder path.
24433 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24435 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
24436 edges; zero probability is not better than uninitialized.
24438 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24440 * asan.h (asan_sanitize_allocas_p): Declare.
24441 * asan.c (asan_sanitize_allocas_p): New function.
24442 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
24443 (handle_builtin_alloca): Likewise.
24444 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
24445 if !asan_sanitize_allocas_p.
24446 * params.def (asan-instrument-allocas): Add new option.
24447 * params.h (ASAN_PROTECT_ALLOCAS): Define.
24448 * opts.c (common_handle_option): Disable allocas sanitization for
24451 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24453 * asan.c: Include gimple-fold.h.
24454 (get_last_alloca_addr): New function.
24455 (handle_builtin_stackrestore): Likewise.
24456 (handle_builtin_alloca): Likewise.
24457 (asan_emit_allocas_unpoison): Likewise.
24458 (get_mem_refs_of_builtin_call): Add new parameter, remove const
24459 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
24460 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
24461 (instrument_builtin_call): Pass gimple iterator to
24462 get_mem_refs_of_builtin_call.
24463 (last_alloca_addr): New global.
24464 * asan.h (asan_emit_allocas_unpoison): Declare.
24465 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
24466 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
24467 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
24468 if function calls alloca.
24469 * gimple-fold.c (replace_call_with_value): Remove static keyword.
24470 * gimple-fold.h (replace_call_with_value): Declare.
24471 * internal-fn.c: Include asan.h.
24472 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
24473 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
24475 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24477 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
24478 (C_SELFTEST_FLAGS): New.
24479 (CPP_SELFTEST_FLAGS): New.
24480 (SELFTEST_DEPS): New, from deps of s-selftest.
24481 (C_SELFTEST_DEPS): New, from deps of s-selftest.
24482 (CPP_SELFTEST_DEPS): New.
24483 (selftest): Add dependency on s-selftest-c++.
24484 (s-selftest): Rename to...
24485 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
24486 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
24487 than SELFTEST_FLAGS.
24488 (selftest-gdb): Rename to...
24489 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
24491 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
24492 (selftest-valgrind): Rename to...
24493 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
24495 (selftest-valgrind): Reintroduce as an alias for
24496 selftest-c-valgrind.
24497 (s-selftest-c++): New.
24498 (selftest-c++-gdb): New.
24499 (selftest-c++-valgrind): New.
24501 2017-07-06 Olivier Hainque <hainque@adacore.com>
24503 * gcc.c (process_command): When deciding if undefined variables
24504 should be ignored when processing specs, accept "gcc -v" as well.
24506 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24508 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
24509 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
24511 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24513 * config/arm/arm-cpus.in (armv8-r): Add new entry.
24514 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
24515 * config/arm/arm-tables.opt: Regenerate.
24516 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
24518 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
24520 2017-07-06 Carl Love <cel@us.ibm.com>
24522 * ChangeLog: Clean up from mid air collision
24524 2017-07-06 Carl Love <cel@us.ibm.com>
24526 * config/rs6000/rs6000-c.c: Add support for built-in functions
24527 vector signed int vec_subc (vector signed int, vector signed int);
24528 vector signed __int128 vec_subc (vector signed __int128,
24529 vector signed __int128);
24530 vector unsigned __int128 vec_subc (vector unsigned __int128,
24531 vector unsigned __int128);
24532 vector signed int vec_sube (vector signed int, vector signed int,
24533 vector signed int);
24534 vector unsigned int vec_sube (vector unsigned int,
24535 vector unsigned int,
24536 vector unsigned int);
24537 vector signed __int128 vec_sube (vector signed __int128,
24538 vector signed __int128,
24539 vector signed__int128);
24540 vector unsigned __int128 vec_sube (vector unsigned __int128,
24541 vector unsigned __int128,
24542 vector unsigned __int128);
24543 vector signed int vec_subec (vector signed int, vector signed int,
24544 vector signed int);
24545 vector unsigned int vec_subec (vector unsigned int,
24546 vector unsigned int,
24547 vector unsigned int);
24548 vector signed __int128 vec_subec (vector signed __int128,
24549 vector signed __int128,
24550 vector signed__int128);
24551 vector unsigned __int128 vec_subec (vector unsigned __int128,
24552 vector unsigned __int128,
24553 vector unsigned __int128);
24554 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
24555 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
24556 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
24557 BU_ALTIVEC_OVERLOAD_X definitions.
24558 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
24559 * doc/extend.texi: Update the built-in documentation file for the new
24560 built-in functions.
24562 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24565 * diagnostic-show-locus.c (layout::layout): Use start and finish
24566 spelling location for the start and finish of each range.
24567 * genmatch.c (linemap_client_expand_location_to_spelling_point):
24568 Add unused aspect param.
24569 * input.c (expand_location_1): Add "aspect" param, and use it
24570 to access the correct part of the location.
24571 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
24573 (expand_location_to_spelling_point): Likewise.
24574 (linemap_client_expand_location_to_spelling_point): Add "aspect"
24575 param, and pass it to expand_location_1.
24577 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24579 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24580 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24581 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24582 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24583 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24584 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24585 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24586 _mm_maskz_getmant_ss): New intrinsics.
24587 (__builtin_ia32_getexpss128_mask): Changed to ...
24588 __builtin_ia32_getexpss128_round ... this.
24589 (__builtin_ia32_getexpsd128_mask): Changed to ...
24590 __builtin_ia32_getexpsd128_round ... this.
24591 * config/i386/i386-builtin-types.def
24592 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24593 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24594 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24595 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24596 __builtin_ia32_getmantss_mask_round): New builtins.
24597 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24598 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24599 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24600 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24601 * config/i386/sse.md
24602 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24603 avx512f_sgetexp<mode><mask_scalar_name>
24604 <round_saeonly_scalar_name> ... this.
24605 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24606 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24607 vgetexp<ssescalarmodesuffix>
24608 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24609 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24610 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24611 avx512f_vgetmant<mode><mask_scalar_name>
24612 <round_saeonly_scalar_name> ... this.
24613 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24614 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24615 vgetmant<ssescalarmodesuffix>
24616 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24617 %0<mask_scalar_operand4>, %1, %2
24618 <round_saeonly_scalar_mask_op4>, %3} ... this.
24619 * config/i386/subst.md (mask_scalar_operand4,
24620 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24621 round_saeonly_scalar_nimm_predicate): New subst attributes.
24623 2017-07-06 Julia Koval <julia.koval@intel.com>
24625 * config/i386/i386.c (ix86_erase_embedded_rounding):
24626 Remove code for old rounding pattern.
24628 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24630 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24632 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24634 * doc/sourcebuild.texi (Test Directives, Variants of
24635 dg-require-support): Add documentation for dg-require-stack-check.
24637 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24639 * config/i386/subst.md (mask_scalar, round_scalar,
24640 round_saeonly_scalar): New meta-templates.
24641 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24642 round_scalar_mask_operand3, round_scalar_mask_op3,
24643 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24644 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24645 round_saeonly_scalar_constraint,
24646 round_saeonly_scalar_prefix): New subst attribute.
24647 * config/i386/sse.md
24648 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24649 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24650 <round_scalar_name> ... this.
24651 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24652 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24653 <round_scalar_name> ... this.
24654 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24655 <sse>_vm<code><mode>3<mask_scalar_name>
24656 <round_saeonly_scalar_name> ... this.
24657 (v<plusminus_mnemonic><ssescalarmodesuffix>
24658 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24659 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24660 v<plusminus_mnemonic><ssescalarmodesuffix>
24661 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24662 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24663 (v<multdiv_mnemonic><ssescalarmodesuffix>
24664 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24665 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24666 v<multdiv_mnemonic><ssescalarmodesuffix>
24667 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24668 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24669 (v<maxmin_float><ssescalarmodesuffix>
24670 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24671 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24672 v<maxmin_float><ssescalarmodesuffix>
24673 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24674 %0<mask_scalar_operand3>, %1, %<iptr>2
24675 <round_saeonly_scalar_mask_op3>} ... this.
24677 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24679 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24680 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24682 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24683 Alan Hayward <alan.hayward@arm.com>
24684 David Sherwood <david.sherwood@arm.com>
24686 * combine.c (simplify_if_then_else): Remove "enum" before
24688 * compare-elim.c (can_eliminate_compare): Likewise.
24689 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24691 (aarch64_lookup_simd_builtin_type): Likewise.
24692 (aarch64_simd_builtin_type): Likewise.
24693 (aarch64_init_simd_builtin_types): Likewise.
24694 (aarch64_simd_expand_args): Likewise.
24695 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24697 (aarch64_reverse_mask): Likewise.
24698 (aarch64_simd_emit_reg_reg_move): Likewise.
24699 (aarch64_gen_adjusted_ldpstp): Likewise.
24700 (aarch64_ccmp_mode_to_code): Likewise.
24701 (aarch64_operands_ok_for_ldpstp): Likewise.
24702 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24703 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24705 (aarch64_min_divisions_for_recip_mul): Likewise.
24706 (aarch64_reassociation_width): Likewise.
24707 (aarch64_get_condition_code_1): Likewise.
24708 (aarch64_simd_emit_reg_reg_move): Likewise.
24709 (aarch64_simd_attr_length_rglist): Likewise.
24710 (aarch64_reverse_mask): Likewise.
24711 (aarch64_operands_ok_for_ldpstp): Likewise.
24712 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24713 (aarch64_gen_adjusted_ldpstp): Likewise.
24714 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24716 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24717 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24718 (arm_lookup_simd_builtin_type): Likewise.
24719 (arm_simd_builtin_type): Likewise.
24720 (arm_init_simd_builtin_types): Likewise.
24721 (arm_expand_builtin_args): Likewise.
24722 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24723 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24724 (ft32_setup_incoming_varargs): Likewise.
24725 (ft32_function_arg): Likewise.
24726 (ft32_function_arg_advance): Likewise.
24727 (ft32_pass_by_reference): Likewise.
24728 (ft32_arg_partial_bytes): Likewise.
24729 (ft32_valid_pointer_mode): Likewise.
24730 (ft32_addr_space_pointer_mode): Likewise.
24731 (ft32_addr_space_legitimate_address_p): Likewise.
24732 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24734 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24735 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24736 (iamcu_alignment): Likewise.
24737 (canonicalize_vector_int_perm): Likewise.
24738 (ix86_noce_conversion_profitable_p): Likewise.
24739 (ix86_mpx_bound_mode): Likewise.
24740 (ix86_operands_ok_for_move_multiple): Likewise.
24741 * config/microblaze/microblaze-protos.h
24742 (microblaze_expand_conditional_branch_reg): Likewise.
24743 * config/microblaze/microblaze.c
24744 (microblaze_expand_conditional_branch_reg): Likewise.
24745 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24747 (rs6000_reassociation_width): Likewise.
24748 (rs6000_invalid_binary_op): Likewise.
24749 (fusion_p9_p): Likewise.
24750 (emit_fusion_p9_load): Likewise.
24751 (emit_fusion_p9_store): Likewise.
24752 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24754 (riscv_hard_regno_mode_ok_p): Likewise.
24755 (riscv_address_insns): Likewise.
24756 (riscv_split_symbol): Likewise.
24757 (riscv_legitimize_move): Likewise.
24758 (riscv_function_value): Likewise.
24759 (riscv_hard_regno_nregs): Likewise.
24760 (riscv_expand_builtin): Likewise.
24761 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24762 (riscv_build_integer): Likewise.
24763 (riscv_split_integer): Likewise.
24764 (riscv_legitimate_constant_p): Likewise.
24765 (riscv_cannot_force_const_mem): Likewise.
24766 (riscv_regno_mode_ok_for_base_p): Likewise.
24767 (riscv_valid_base_register_p): Likewise.
24768 (riscv_valid_offset_p): Likewise.
24769 (riscv_valid_lo_sum_p): Likewise.
24770 (riscv_classify_address): Likewise.
24771 (riscv_legitimate_address_p): Likewise.
24772 (riscv_address_insns): Likewise.
24773 (riscv_load_store_insns): Likewise.
24774 (riscv_force_binary): Likewise.
24775 (riscv_split_symbol): Likewise.
24776 (riscv_force_address): Likewise.
24777 (riscv_legitimize_address): Likewise.
24778 (riscv_move_integer): Likewise.
24779 (riscv_legitimize_const_move): Likewise.
24780 (riscv_legitimize_move): Likewise.
24781 (riscv_address_cost): Likewise.
24782 (riscv_subword): Likewise.
24783 (riscv_output_move): Likewise.
24784 (riscv_canonicalize_int_order_test): Likewise.
24785 (riscv_emit_int_order_test): Likewise.
24786 (riscv_function_arg_boundary): Likewise.
24787 (riscv_pass_mode_in_fpr_p): Likewise.
24788 (riscv_pass_fpr_single): Likewise.
24789 (riscv_pass_fpr_pair): Likewise.
24790 (riscv_get_arg_info): Likewise.
24791 (riscv_function_arg): Likewise.
24792 (riscv_function_arg_advance): Likewise.
24793 (riscv_arg_partial_bytes): Likewise.
24794 (riscv_function_value): Likewise.
24795 (riscv_pass_by_reference): Likewise.
24796 (riscv_setup_incoming_varargs): Likewise.
24797 (riscv_print_operand): Likewise.
24798 (riscv_elf_select_rtx_section): Likewise.
24799 (riscv_save_restore_reg): Likewise.
24800 (riscv_for_each_saved_reg): Likewise.
24801 (riscv_register_move_cost): Likewise.
24802 (riscv_hard_regno_mode_ok_p): Likewise.
24803 (riscv_hard_regno_nregs): Likewise.
24804 (riscv_class_max_nregs): Likewise.
24805 (riscv_memory_move_cost): Likewise.
24806 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24807 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24808 (rl78_addr_space_address_mode): Likewise.
24809 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24811 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24812 (rs6000_reassociation_width): Likewise.
24813 (rs6000_invalid_binary_op): Likewise.
24814 (fusion_p9_p): Likewise.
24815 (emit_fusion_p9_load): Likewise.
24816 (emit_fusion_p9_store): Likewise.
24817 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24818 (ok_for_simple_move_operands): Likewise.
24819 (ok_for_simple_move_strict_operands): Likewise.
24820 (ok_for_simple_arith_logic_operands): Likewise.
24821 (visium_legitimize_reload_address): Likewise.
24822 (visium_select_cc_mode): Likewise.
24823 (output_cbranch): Likewise.
24824 (visium_split_double_move): Likewise.
24825 (visium_expand_copysign): Likewise.
24826 (visium_expand_int_cstore): Likewise.
24827 (visium_expand_fp_cstore): Likewise.
24828 * config/visium/visium.c (visium_pass_by_reference): Likewise.
24829 (visium_function_arg): Likewise.
24830 (visium_function_arg_advance): Likewise.
24831 (visium_libcall_value): Likewise.
24832 (visium_setup_incoming_varargs): Likewise.
24833 (visium_legitimate_constant_p): Likewise.
24834 (visium_legitimate_address_p): Likewise.
24835 (visium_legitimize_address): Likewise.
24836 (visium_secondary_reload): Likewise.
24837 (visium_register_move_cost): Likewise.
24838 (visium_memory_move_cost): Likewise.
24839 (prepare_move_operands): Likewise.
24840 (ok_for_simple_move_operands): Likewise.
24841 (ok_for_simple_move_strict_operands): Likewise.
24842 (ok_for_simple_arith_logic_operands): Likewise.
24843 (visium_function_value_1): Likewise.
24844 (rtx_ok_for_offset_p): Likewise.
24845 (visium_legitimize_reload_address): Likewise.
24846 (visium_split_double_move): Likewise.
24847 (visium_expand_copysign): Likewise.
24848 (visium_expand_int_cstore): Likewise.
24849 (visium_expand_fp_cstore): Likewise.
24850 (visium_split_cstore): Likewise.
24851 (visium_select_cc_mode): Likewise.
24852 (visium_split_cbranch): Likewise.
24853 (output_cbranch): Likewise.
24854 (visium_print_operand_address): Likewise.
24855 * expmed.c (flip_storage_order): Likewise.
24856 * expmed.h (emit_cstore): Likewise.
24857 (flip_storage_order): Likewise.
24858 * genrecog.c (validate_pattern): Likewise.
24859 * hsa-gen.c (gen_hsa_addr): Likewise.
24860 * internal-fn.c (expand_arith_overflow): Likewise.
24861 * ira-color.c (allocno_copy_cost_saving): Likewise.
24862 * lra-assigns.c (find_hard_regno_for_1): Likewise.
24863 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
24864 (process_invariant_for_inheritance): Likewise.
24865 * lra-eliminations.c (move_plus_up): Likewise.
24866 * omp-low.c (lower_oacc_reductions): Likewise.
24867 * simplify-rtx.c (simplify_subreg): Likewise.
24868 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
24869 (TARGET_CHKP_BOUND_MODE): Likewise..
24870 * targhooks.c (default_chkp_bound_mode): Likewise.
24871 (default_setup_incoming_vararg_bounds): Likewise.
24872 * targhooks.h (default_chkp_bound_mode): Likewise.
24873 (default_setup_incoming_vararg_bounds): Likewise.
24874 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
24875 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
24876 (have_whole_vector_shift): Likewise.
24877 * tree-vect-stmts.c (vectorizable_load): Likewise.
24878 * doc/tm.texi: Regenerate.
24880 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24882 Graceful degrade if Binutils PR21472 is not available.
24885 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
24886 .rodata in flash test fails.
24887 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
24888 * confgure: Regenerate.
24889 * config.in: Regenerate.
24890 * config/avr/avr.c (avr_asm_named_section)
24891 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
24892 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
24893 (avr_asm_init_sections): Same.
24895 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24897 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
24898 (fma<VH:mode>4_intrinsic): Likewise.
24899 (*fmsub<VCVTF:mode>4): Likewise.
24900 (*fmsub<VH:mode>4_intrinsic): Likewise.
24902 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24905 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
24906 Don't depend on "optimize > 0".
24907 (out_movhi_r_mr, out_movqi_mr_r): Same.
24908 (out_movhi_mr_r, out_movqi_r_mr): Same.
24909 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
24910 io_address_operand on "optimize > 0".
24912 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24914 * tree-loop-distribution.c: Add general explanantion on the pass.
24915 (generate_loops_for_partition): Mark distributed loop.
24916 (pg_add_dependence_edges): New parameter. Handle alias data
24917 dependence specially and record it in the parameter if asked.
24918 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
24919 (init_partition_graph_vertices, add_partition_graph_edge): New.
24920 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
24921 (free_partition_graph_vdata, build_partition_graph): New.
24922 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
24923 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
24924 (data_ref_segment_size, latch_dominated_by_data_ref): New.
24925 (compute_alias_check_pairs, version_loop_by_alias_check): New.
24926 (version_for_distribution_p, finalize_partitions): New.
24927 (distribute_loop): Handle alias data dependence specially. Factor
24928 out loop fusion code as functions and call these functions.
24930 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24932 * tree-loop-distribution.c (classify_partition): New parameter and
24933 better handle reduction statement.
24934 (rdg_build_partitions): Revise comment.
24935 (distribute_loop): Compute statements in all partitions and pass it
24936 to classify_partition.
24938 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24940 * tree-loop-distribution.c (enum partition_type): New.
24941 (struct partition): New field type.
24942 (partition_merge_into): Add parameter. Update partition type.
24943 (data_dep_in_cycle_p, update_type_for_merge): New functions.
24944 (build_rdg_partition_for_vertex): Compute partition type.
24945 (rdg_build_partitions): Dump partition type.
24946 (distribute_loop): Update calls to partition_merge_into.
24948 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24950 * tree-loop-distribution.c (struct ddr_hasher): New.
24951 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
24953 (classify_partition): Call get_data_dependence.
24954 (pg_add_dependence_edges): Ditto.
24955 (distribute_loop): Release data dependence hash table.
24957 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24959 * tree-loop-distribution.c (ref_base_address): Delete.
24960 (similar_memory_accesses): Rename ...
24961 (share_memory_accesses): ... to this. Check if partitions access
24962 the same memory reference.
24963 (distribute_loop): Call share_memory_accesses.
24965 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24967 * tree-loop-distribution.c (struct partition): New field recording
24968 its data reference.
24969 (partition_alloc, partition_free): Init and release data refs.
24970 (partition_merge_into): Merge data refs.
24971 (build_rdg_partition_for_vertex): Collect data refs for partition.
24972 (pg_add_dependence_edges): Change parameters from vector to bitmap.
24974 (distribute_loop): Remve data refs from vertice data of partition
24977 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24979 * tree-loop-distribution.c (params.h): Include header file.
24980 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
24981 (datarefs_vec): New global var.
24982 (create_rdg_vertices): Use datarefs_vec directly.
24983 (free_rdg): Don't free data references.
24984 (build_rdg): Update use. Don't free data references.
24985 (distribute_loop): Compute global variable for data references.
24986 Bail out if there are too many data references.
24988 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24990 * tree-loop-distribution.c (loop_nest): New global var.
24991 (build_rdg): Use loop directly, rather than loop nest.
24992 (pg_add_dependence_edges): Remove loop nest parameter. Use global
24994 (distribute_loop): Compute global variable loop nest. Update use.
24996 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24998 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
24999 (partition_merge_into): New parameter. Dump reason for fusion.
25000 (distribute_loop): Update use of partition_merge_into.
25002 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25004 * tree-loop-distribution.c (bb_top_order_index): New.
25005 (bb_top_order_index_size, bb_top_order_cmp): New.
25006 (stmts_from_loop): Use topological order.
25007 (pass_loop_distribution::execute): Compute and release topological
25008 order for basic blocks.
25010 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25012 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25015 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25017 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25018 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25019 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25020 * internal-fn.def (LOOP_DIST_ALIAS): New.
25021 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25022 (fold_loop_internal_call): ... this.
25023 (vect_loop_dist_alias_call): New function.
25024 (set_uid_loop_bbs): Call fold_loop_internal_call.
25025 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25028 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25031 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25032 Require dead FLAGS_REG at the beginning of a peephole.
25034 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25037 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25038 arguments in the call to __builtin_ia32_sbb_u32.
25039 (_subborrow_u64): Swap _X and _Y arguments in the call to
25040 __builtin_ia32_sbb_u64.
25042 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25045 * tree-vrp.c (compare_assert_loc): Turn into a function template
25046 with stable template parameter. Only test if a->e is NULL,
25047 !a->e == !b->e has been verified already. Use e == NULL or
25048 e != NULL instead of e or ! e tests. If stable is true, don't use
25049 iterative_hash_expr, on the other side allow a or b or both NULL
25050 and sort the NULLs last.
25051 (process_assert_insertions): Sort using compare_assert_loc<false>
25052 instead of compare_assert_loc, later sort using
25053 compare_assert_loc<true> before calling process_assert_insertions_for
25054 in a loop. Use break instead of continue once seen NULL pointer.
25056 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25058 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25059 Cortex-R7 and Cortex-R8 processors.
25061 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25063 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25064 uninitialized while src is not.
25066 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
25068 * common/config/arm/arm-common.c: Adjust include path for
25070 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
25071 (arm-cpu.h): Create in build directory. Adjust dependency rules.
25072 (arm-cpu-data.h): Likewise.
25073 (arm-cpu-cdata.h): Likewise.
25074 * config/arm/arm-cpu.h: Delete.
25075 * config/arm/arm-cpu-cdata.h: Delete.
25076 * config/arm/arm-cpu-data.h: Delete.
25078 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
25080 * config/arm/arm-cpus.in (cortex-a55): New.
25081 (cortex-a75): Likewise.
25082 (cortex-a75.cortex-a55): Likewise.
25083 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
25085 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
25086 * config/arm/arm-cpu-cdata.h: Regenerate.
25087 * config/arm/arm-cpu-data.h: Regenerate.
25088 * config/arm/arm-cpu.h: Regenerate.
25089 * config/arm/arm-tables.opt: Regenerate.
25090 * config/arm/arm-tune.md: Regenerate.
25092 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25094 * haifa-sched.c (sched_create_recovery_edges): Update profile.
25096 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25098 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
25101 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
25103 PR tree-optimization/81292
25104 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
25105 full_string_p, also call adjust_related_strinfos if the adjustment
25106 is simple, otherwise invalidate related strinfos.
25108 2017-07-04 Martin Liska <mliska@suse.cz>
25111 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
25112 newly created variable as DECL_IGNORED_P.
25114 2017-07-04 Martin Liska <mliska@suse.cz>
25117 * ipa-inline.c (inline_small_functions):
25118 Use xstrdup_for_dump.
25120 2017-07-04 Tom de Vries <tom@codesourcery.com>
25122 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
25124 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25127 * config/darwin.c (darwin_function_switched_text_sections):
25130 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
25132 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
25134 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25136 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
25138 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25140 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
25141 min_profitable_iters, and th as inclusive lower bounds.
25142 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
25143 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
25144 for min_profitable_iters and min_profitable_estimate.
25145 (vect_transform_loop): Treat th as an inclusive lower bound.
25146 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25148 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25151 * config/darwin.c (darwin_function_switched_text_sections):
25152 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
25153 in two pieces, and suppress the use of buf.
25155 2017-07-03 Nathan Sidwell <nathan@acm.org>
25157 * hash-table.h (hash_table_mod1): Fix indentation.
25159 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25161 PR middle-end/81290
25162 * predict.c (force_edge_cold): Be more careful about propagation
25164 * profile-count.h (profile_probability::guessed,
25165 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
25167 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
25169 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
25171 * doc/invoke.texi (rcpc architecture extension): Document it.
25173 2017-07-03 Richard Biener <rguenther@suse.de>
25175 PR tree-optimization/60510
25176 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
25177 the scalar reduction PHI and use it.
25178 (vectorizable_reduction): Properly guard the single_defuse_cycle
25179 path for non-SLP reduction chains where we cannot use it.
25180 Rework reduc_def/index and vector type deduction. Rework
25181 vector operand gathering during reduction op code-gen.
25182 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
25183 chains dissolve the chain and leave it to non-SLP reduction
25186 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25188 * tree-data-ref.h (dr_alignment): Declare.
25189 * tree-data-ref.c (dr_alignment): New function.
25190 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
25191 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
25193 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
25195 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25197 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
25198 and base_misalignment fields.
25199 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
25200 * tree-data-ref.c: Include builtins.h.
25201 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
25202 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
25203 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
25204 * tree-vect-data-refs.c: Include tree-cfg.h.
25205 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
25206 fields instead of calculating an alignment here.
25207 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
25208 innermost_loop_behavior fields.
25210 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25212 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
25214 (DR_STEP_ALIGNMENT): New macro.
25215 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
25216 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
25217 (create_data_ref): Print it.
25218 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
25219 to tell whether the step preserves vector (mis)alignment.
25220 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25221 Move the check for an integer step and generalise to all INTEGER_CST.
25222 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
25223 Print the outer step alignment.
25225 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25227 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
25228 with offset_alignment.
25229 (DR_ALIGNED_TO): Delete.
25230 (DR_OFFSET_ALIGNMENT): New macro.
25231 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
25232 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
25233 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
25234 (create_data_ref): Likewise.
25235 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25236 (vect_analyze_data_refs): Likewise.
25237 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
25238 creating dummy innermost behavior.
25240 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25242 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
25243 with a "innermost_loop_behavior *" and refeence tree.
25244 * tree-data-ref.c (dr_analyze_innermost): Likewise.
25245 (create_data_ref): Update call accordingly.
25246 * tree-predcom.c (find_looparound_phi): Likewise.
25248 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25250 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
25251 fields with dr_wrt_vec_loop.
25252 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
25253 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
25254 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
25255 (vect_dr_behavior): New function.
25256 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25257 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
25258 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
25259 track whether the step preserves the misalignment.
25260 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25261 Use vect_dr_behavior.
25262 (vect_setup_realignment): Update call accordingly.
25263 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
25264 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
25265 call to vect_create_addr_base_for_vector_ref.
25266 (vect_create_cond_for_align_checks): Likewise.
25267 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
25268 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
25269 (vect_recog_mask_conversion_pattern): Likewise.
25270 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
25271 (new_stmt_vec_info): Remove redundant zeroing.
25273 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25275 * common/config/arm/arm-common.c (arm_be8_option): New function.
25276 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
25277 (ISA_ARMv6): Add isa_bit_be8.
25278 * config/arm/arm.h (arm_be8_option): Add prototype.
25279 (BE8_SPEC_FUNCTION): New define.
25280 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
25281 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
25282 (mlittle-endian): Similarly.
25283 (mbe8, mbe32): New options.
25284 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
25285 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
25287 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25289 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
25291 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25293 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
25294 (cleanup_tree_cfg_bb): Use it.
25295 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
25297 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
25299 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25302 * loop-doloop.c (add_test): Update profile.
25304 2017-07-03 Martin Liska <mliska@suse.cz>
25307 * sanopt.c (rewrite_usage_of_param): New function.
25308 (sanitize_rewrite_addressable_params): Likewise.
25309 (pass_sanopt::execute): Call rewrite_usage_of_param.
25311 2017-07-03 Richard Biener <rguenther@suse.de>
25313 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
25314 back to using VIEW_CONVERT_EXPR.
25316 2017-07-03 Martin Liska <mliska@suse.cz>
25319 * doc/extend.texi: Document when a resolver function is
25320 generated for target_clones.
25322 2017-07-03 Martin Liska <mliska@suse.cz>
25324 * asan.c (asan_emit_stack_protection): Unpoison just red zones
25325 and shadow memory of auto variables which are subject of
25326 use-after-scope sanitization.
25327 (asan_expand_mark_ifn): Add do set only when is_poison.
25329 2016-07-03 Richard Biener <rguenther@suse.de>
25331 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
25333 (vect_force_simple_reduction): Record reduction def -> phi mapping.
25334 (vectorizable_reduction): Perform reduction PHI creation when
25335 visiting a reduction PHI and adjust and simplify code generation
25336 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
25337 (vect_transform_loop): Visit reduction PHIs.
25338 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
25339 defs into the SLP tree.
25340 (vect_build_slp_tree): Reduction defs terminate the recursion.
25341 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
25343 (vect_get_vec_defs_for_stmt_copy): Export.
25344 (vect_get_vec_defs): Likewise.
25345 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
25347 (vect_get_vec_defs_for_stmt_copy): Declare.
25348 (vect_get_vec_defs): Likewise.
25350 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25352 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
25353 parameter with a "loop" parameter and use it instead of the
25354 loop containing DR_STMT. Don't check simple_iv when doing
25355 BB analysis. Describe the two analysis modes in the comment.
25357 2017-07-03 Tom de Vries <tom@codesourcery.com>
25359 PR tree-optimization/69468
25360 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
25361 (find_same_succ_bb): Handle ignore_edge_flags.
25363 2017-07-03 Tom de Vries <tom@codesourcery.com>
25365 PR tree-optimization/81192
25366 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
25368 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
25370 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
25372 2017-07-03 Tom de Vries <tom@codesourcery.com>
25374 PR tree-optimization/81192
25375 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
25376 BB_SAME_SUCC (bb) == NULL.
25378 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25380 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
25383 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25385 * dumpfile.c: Include profile-count.h
25386 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
25388 (insert_cond_bb): Update profile.
25389 * tree-cfg.h (insert_cond_bb): Update prototype.
25390 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
25391 * tree-dump.c: Do not include tree-cfg.
25393 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25395 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
25397 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25399 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
25402 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25404 * tree-complex.c (expand_complex_div_wide): update profile.
25406 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25407 Alan Hayward <alan.hayward@arm.com>
25408 David Sherwood <david.sherwood@arm.com>
25410 * Makefile.in (MACHMODE_H): Remove insn-modes.h
25411 (CORETYPES_H): New define.
25412 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
25413 (insn-modes-inline.h, s-modes-inline-h): New rules.
25414 (generated_files): Add insn-modes-inline.h.
25415 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
25416 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
25417 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
25418 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
25419 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
25420 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
25421 (build/gencodes.o, build/genconditions.o): Likewise.
25422 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
25423 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
25424 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
25425 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
25426 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
25427 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
25428 * coretypes.h: Include everything up to real.h for generators.
25429 Include insn-modes.h first. Include wide-int-print.h after
25430 wide-int.h. Include insn-modes-inline.h and then machmode.h.
25431 * machmode.h: Don't include insn-modes.h here.
25432 * function-tests.c: Remove includes of signop.h, machmode.h,
25433 double-int.h and wide-int.h.
25435 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
25437 * optc-save-gen.awk: Likewise.
25438 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
25439 * godump.c: Remove include of wide-int-print.h.
25440 * pretty-print.h: Likewise.
25441 * wide-int-print.cc: Likewise.
25442 * wide-int.cc: Likewise.
25443 * hash-map-tests.c: Remove include of signop.h.
25444 * hash-set-tests.c: Likewise.
25445 * rtl-tests.c: Likewise.
25446 * mkconfig.sh: Remove include of machmode.h.
25447 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
25449 (emit_insn_modes_inline_h): ...this new function. Emit the code
25450 into an insn-modes-inline.h header file, adding appropriate
25451 include guards and end comments.
25452 (emit_insn_modes_c_header): Remove include of machmode.h.
25453 (emit_min_insn_modes_c_header): Include coretypes.h rather than
25455 (main): Handle -i flag and call emit_insn_modes_inline_h when
25458 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25460 * tree-ssa-strlen.c (strinfo): Rename the length field to
25461 nonzero_chars. Add a full_string_p field.
25462 (compare_nonzero_chars, zero_length_string_p): New functions.
25463 (get_addr_stridx): Add an offset_out parameter.
25464 Use compare_nonzero_chars.
25465 (get_stridx): Update accordingly. Use compare_nonzero_chars.
25466 (new_strinfo): Update after above changes to strinfo.
25467 (set_endptr_and_length): Set full_string_p.
25468 (get_string_length): Update after above changes to strinfo.
25469 (unshare_strinfo): Update call to new_strinfo.
25470 (maybe_invalidate): Likewise.
25471 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
25472 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
25473 as a uhwi instead of an shwi. Update after above changes to
25474 strinfo and new_strinfo.
25475 (zero_length_string): Assert that chainsi contains full strings.
25476 Use zero_length_string_p. Update call to new_strinfo.
25477 (adjust_related_strinfos): Update after above changes to strinfo.
25478 Copy full_string_p from origsi.
25479 (adjust_last_stmt): Use zero_length_string_p.
25480 (handle_builtin_strlen): Update after above changes to strinfo and
25481 new_strinfo. Install the lhs as the string length if the previous
25482 entry didn't describe a full string.
25483 (handle_builtin_strchr): Update after above changes to strinfo
25485 (handle_builtin_strcpy): Likewise.
25486 (handle_builtin_strcat): Likewise.
25487 (handle_builtin_malloc): Likewise.
25488 (handle_pointer_plus): Likewise.
25489 (handle_builtin_memcpy): Likewise. Track nonzero characters
25490 that aren't necessarily followed by a nul terminator.
25491 (handle_char_store): Likewise.
25493 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25495 PR tree-optimization/80769
25496 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
25497 for malloc and calloc. Document the new invariant that all related
25498 strinfos have delayed lengths or none do.
25499 (verify_related_strinfos): Move earlier in file.
25500 (set_endptr_and_length): New function, split out from...
25501 (get_string_length): ...here. Also set the lengths of related
25503 (zero_length_string): Assert that chainsi has known (rather than
25505 (adjust_related_strinfos): Likewise.
25507 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25509 PR tree-optimization/81136
25510 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
25511 assert that two references with the same misalignment have the same
25512 compile-time misalignment if those compile-time misalignments
25515 2017-07-01 Andi Kleen <ak@linux.intel.com>
25517 * print-tree.c (print_node): Print all attributes.
25519 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25521 * cfg.c (scale_bbs_frequencies): New function.
25522 * cfg.h (scale_bbs_frequencies): Declare it.
25523 * cfgloopanal.c (single_likely_exit): Cleanup.
25524 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
25526 (scale_loop_profile): Likewise.
25527 (loop_version): Likewise.
25528 (create_empty_loop_on_edge): Update.
25529 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
25530 scale_loop_frequencies, scale_loop_profile, loopify,
25531 loop_version): Update prototypes.
25532 * modulo-sched.c (sms_schedule): Update.
25533 * predict.c (unlikely_executed_edge_p): Also check probability.
25534 (probably_never_executed_edge_p): Fix typo.
25535 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25536 * tree-parloops.c (gen_parallel_loop): Update.
25537 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
25538 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25539 * tree-ssa-loop-split.c (split_loop): Update.
25540 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25541 * tree-vect-loop-manip.c (vect_do_peeling): Update.
25542 (vect_loop_versioning): Update.
25543 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25545 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25547 * trans-mem.c (split_bb_make_tm_edge): Update profile.
25549 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25551 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
25552 to keep profile consistent.
25554 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25556 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
25557 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
25558 * profile-count.h (max_safe_multiplier): Make unsigned.
25559 (profile_count::guessed_zero): New.
25561 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25563 * bb-reorder.c (fix_up_crossing_landing_pad,
25564 fix_crossing_conditional_branches): Use make_single_succ_edge
25565 to keep profile consistent.
25567 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25569 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
25572 2017-07-01 Jakub Jelinek <jakub@redhat.com>
25575 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
25576 the right scopes, make sure cond_jump isn't preserved between multiple
25577 iterations. Search for fallthru edge whenever there are 3+ edges and
25578 use find_fallthru_edge for it.
25580 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25582 Patch by Alexander Monakov <amonakov@ispras.ru>
25583 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25584 probabilities consistently.
25586 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25588 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25589 emit_cmp_and_jump_insns.
25591 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25594 * tree-inline.c (expand_call_inline): Combine profile statuses.
25596 2017-06-30 Andrew Pinski <apinski@cavium.com>
25598 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25599 fold_stmt returned true.
25601 2017-06-30 Nathan Sidwell <nathan@acm.org>
25603 * ggc.h (empty_string): Delete.
25604 * cfgexpand.c (expand_asm_stmt): Use plain "".
25605 * optabs.c (expand_asm_memory_barrier): Likewise.
25606 * stringpool.c (empty_string): Delete.
25607 (digit_vector, digit_string): Delete.
25608 (ggc_alloc_string): Use plain "", don't optimize single digit
25609 strings. Use ggc_alloc_atomic.
25611 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25613 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25614 comparison set and one other set, use the cost of the non-comparison
25617 2017-06-30 Nathan Sidwell <nathan@acm.org>
25619 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25622 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25624 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25625 loops. Remove now unneeded calls to gimple_switch_set_label() that
25626 just set removed labels to NULL_TREE.
25628 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25630 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25631 (set_range_info): ...here. Only call set_range_info_raw if domain
25633 (set_nonzero_bits): Call set_range_info_raw.
25634 * tree-ssanames.h (set_range_info_raw): New.
25636 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25639 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25640 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25641 of nonimmediate_operand and <store_mask_constraint> instead of m
25642 for the input operand. For V8FI iterator, always split if input
25643 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25644 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25645 <store_mask_predicate> instead of register_operand and
25646 <store_mask_constraint> instead of v for the input operand. Make
25647 sure both operands aren't MEMs for if not <mask_applied>.
25649 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25651 * lto-wrapper.c (copy_file) Close both file descriptors before
25654 2017-06-30 Martin Liska <mliska@suse.cz>
25657 * multiple_target.c (create_dispatcher_calls): Make ifunc
25658 also for function that don't have calls or are not referenced.
25660 2017-06-30 Richard Biener <rguenther@suse.de>
25662 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25663 analyze the first scalar stmt. Move vector type computation
25664 for the BB case here from ...
25665 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25666 live operation processing in the SLP case properly.
25668 2017-06-30 Richard Biener <rguenther@suse.de>
25670 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25672 2017-06-30 Martin Liska <mliska@suse.cz>
25675 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25676 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25678 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25680 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25683 2017-06-30 Richard Biener <rguenther@suse.de>
25685 PR tree-optimization/81249
25686 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25687 condition reduction result to original scalar type.
25689 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25691 * profile-count.h (enum profile_quality): Fix typos and whitespace
25694 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25696 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25697 type for branch probabilities.
25699 2017-06-29 Julian Brown <julian@codesourcery.com>
25700 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25702 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25703 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25704 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25705 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25707 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25709 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25710 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25711 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25712 CC usage from generic code to here.
25713 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25714 CC usage into the target macros.
25716 2017-06-29 Maya Rashish <coypu@sdf.org>
25718 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25721 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25723 * arm/arm-builtins.c: Include profile-count.h
25724 * except.c (sjlj_emit_function_enter): Use
25725 profile_probability::unlikely.
25727 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25729 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25730 and tocrel_offset be pointer args rather than implicitly using
25732 (legitimate_constant_pool_address_p, rs6000_emit_move,
25733 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25734 tocrel_offset and use in toc_relative_expr_p call.
25735 (print_operand, print_operand_address): Use static tocrel_base_oac
25736 and tocrel_offset_oac.
25737 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25740 2017-06-29 Maya Rashish <coypu@sdf.org>
25742 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25744 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25746 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25747 objects, take into account only the alignment of 'op0' and 'mode1' if
25750 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25752 * ccmp.c (ccmp_tree_comparison_p): New function.
25753 (ccmp_candidate_p): Update to use above function.
25754 (get_compare_parts): New function.
25755 (expand_ccmp_next): Update to use new functions.
25756 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25758 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25759 take mode as argument.
25760 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25761 * expr.c (expand_expr_real_1): Pass mode as argument.
25763 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25765 * combine.c (combine_instructions): Print insns to dump_file, together
25768 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25770 * asan.c (asan_emit_stack_protection): Update.
25771 (create_cond_insert_point): Update.
25772 * auto-profile.c (afdo_propagate_circuit): Update.
25773 * basic-block.h (struct edge_def): Turn probability to
25774 profile_probability.
25775 (EDGE_FREQUENCY): Update.
25776 * bb-reorder.c (find_traces_1_round): Update.
25777 (better_edge_p): Update.
25778 (sanitize_hot_paths): Update.
25779 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25780 (make_single_succ_edge): Update.
25781 (check_bb_profile): Update.
25782 (dump_edge_info): Update.
25783 (update_bb_profile_for_threading): Update.
25784 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25785 probabilitycount to 0.
25786 * cfgbuild.c (compute_outgoing_frequencies): Update.
25787 * cfgcleanup.c (try_forward_edges): Update.
25788 (outgoing_edges_match): Update.
25789 (try_crossjump_to_edge): Update.
25790 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25791 (expand_gimple_tailcall): Update.
25792 (construct_init_block): Use make_single_succ_edge.
25793 (construct_exit_block): Use make_single_succ_edge.
25794 * cfghooks.c (verify_flow_info): Update.
25795 (redirect_edge_succ_nodup): Update.
25796 (split_edge): Update.
25797 (account_profile_record): Update.
25798 * cfgloopanal.c (single_likely_exit): Update.
25799 * cfgloopmanip.c (scale_loop_profile): Update.
25800 (set_zero_probability): Remove.
25801 (duplicate_loop_to_header_edge): Update.
25802 * cfgloopmanip.h (loop_version): Update prototype.
25803 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25804 (force_nonfallthru_and_redirect): Update.
25805 (update_br_prob_note): Update.
25806 (rtl_verify_edges): Update.
25807 (purge_dead_edges): Update.
25808 (rtl_lv_add_condition_to_bb): Update.
25809 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25810 * cgraphunit.c (init_lowered_empty_function): Update.
25811 (cgraph_node::expand_thunk): Update.
25812 * cilk-common.c: Include profile-count.h
25813 * dojump.c (inv): Remove.
25814 (jumpifnot): Update.
25815 (jumpifnot_1): Update.
25816 (do_jump_1): Update.
25818 (do_jump_by_parts_greater_rtx): Update.
25819 (do_compare_rtx_and_jump): Update.
25820 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25821 do_jump_1. do_compare_rtx_and_jump): Update prototype.
25822 * dwarf2cfi.c: Include profile-count.h
25823 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
25824 (sjlj_emit_dispatch_table): Likewise.
25825 * explow.c: Include profile-count.h
25826 * expmed.c (emit_store_flag_force): Update.
25827 (do_cmp_and_jump): Update.
25828 * expr.c (compare_by_pieces_d::generate): Update.
25829 (compare_by_pieces_d::finish_mode): Update.
25830 (emit_block_move_via_loop): Update.
25831 (store_expr_with_bounds): Update.
25832 (store_constructor): Update.
25833 (expand_expr_real_2): Update.
25834 (expand_expr_real_1): Update.
25835 * expr.h (try_casesi, try_tablejump): Update prototypes.
25836 * gimple-pretty-print.c (dump_probability): Update.
25837 (dump_profile): New.
25838 (dump_gimple_label): Update.
25839 (dump_gimple_bb_header): Update.
25840 * graph.c (draw_cfg_node_succ_edges): Update.
25841 * hsa-gen.c (convert_switch_statements): Update.
25842 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
25843 (find_if_case_1): Update.
25844 (find_if_case_2): Update.
25845 * internal-fn.c (expand_arith_overflow_result_store): Update.
25846 (expand_addsub_overflow): Update.
25847 (expand_neg_overflow): Update.
25848 (expand_mul_overflow): Update.
25849 (expand_vector_ubsan_overflow): Update.
25850 * ipa-cp.c (good_cloning_opportunity_p): Update.
25851 * ipa-split.c (split_function): Use make_single_succ_edge.
25852 * ipa-utils.c (ipa_merge_profiles): Update.
25853 * loop-doloop.c (add_test): Update.
25854 (doloop_modify): Update.
25855 * loop-unroll.c (compare_and_jump_seq): Update.
25856 (unroll_loop_runtime_iterations): Update.
25857 * lra-constraints.c (lra_inheritance): Update.
25858 * lto-streamer-in.c (input_cfg): Update.
25859 * lto-streamer-out.c (output_cfg): Update.
25860 * mcf.c (adjust_cfg_counts): Update.
25861 * modulo-sched.c (sms_schedule): Update.
25862 * omp-expand.c (expand_omp_for_init_counts): Update.
25863 (extract_omp_for_update_vars): Update.
25864 (expand_omp_ordered_sink): Update.
25865 (expand_omp_for_ordered_loops): Update.
25866 (expand_omp_for_generic): Update.
25867 (expand_omp_for_static_nochunk): Update.
25868 (expand_omp_for_static_chunk): Update.
25869 (expand_cilk_for): Update.
25870 (expand_omp_simd): Update.
25871 (expand_omp_taskloop_for_outer): Update.
25872 (expand_omp_taskloop_for_inner): Update.
25873 * omp-simd-clone.c (simd_clone_adjust): Update.
25874 * optabs.c (expand_doubleword_shift): Update.
25875 (expand_abs): Update.
25876 (emit_cmp_and_jump_insn_1): Update.
25877 (expand_compare_and_swap_loop): Update.
25878 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
25879 * predict.c (predictable_edge_p): Update.
25880 (edge_probability_reliable_p): Update.
25881 (set_even_probabilities): Update.
25882 (combine_predictions_for_insn): Update.
25883 (combine_predictions_for_bb): Update.
25884 (propagate_freq): Update.
25885 (estimate_bb_frequencies): Update.
25886 (force_edge_cold): Update.
25887 * profile-count.c (profile_count::dump): Add missing space into dump.
25888 (profile_count::debug): Add newline.
25889 (profile_count::differs_from_p): Explicitly convert to unsigned.
25890 (profile_count::stream_in): Update.
25891 (profile_probability::dump): New member function.
25892 (profile_probability::debug): New member function.
25893 (profile_probability::differs_from_p): New member function.
25894 (profile_probability::differs_lot_from_p): New member function.
25895 (profile_probability::stream_in): New member function.
25896 (profile_probability::stream_out): New member function.
25897 * profile-count.h (profile_count_quality): Rename to ...
25898 (profile_quality): ... this one.
25899 (profile_probability): New.
25900 (profile_count): Update.
25901 * profile.c (compute_branch_probabilities): Update.
25902 * recog.c (peep2_attempt): Update.
25903 * sched-ebb.c (schedule_ebbs): Update.
25904 * sched-rgn.c (find_single_block_region): Update.
25905 (compute_dom_prob_ps): Update.
25906 (schedule_region): Update.
25907 * sel-sched-ir.c (compute_succs_info): Update.
25908 * stmt.c (struct case_node): Update.
25909 (do_jump_if_equal): Update.
25910 (get_outgoing_edge_probs): Update.
25911 (conditional_probability): Update.
25912 (emit_case_dispatch_table): Update.
25913 (expand_case): Update.
25914 (expand_sjlj_dispatch_table): Update.
25915 (emit_case_nodes): Update.
25916 * targhooks.c: Update.
25917 * tracer.c (better_p): Update.
25918 (find_best_successor): Update.
25919 * trans-mem.c (expand_transaction): Update.
25920 * tree-call-cdce.c: Update.
25921 * tree-cfg.c (gimple_split_edge): Upate.
25922 (move_sese_region_to_fn): Upate.
25923 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
25924 * tree-eh.c (lower_resx): Upate.
25925 (cleanup_empty_eh_move_lp): Upate.
25926 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25927 * tree-inline.c (copy_edges_for_bb): Update.
25928 (copy_cfg_body): Update.
25929 * tree-parloops.c (gen_parallel_loop): Update.
25930 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
25931 (gimple_gen_time_profiler): Update.
25932 * tree-ssa-dce.c (remove_dead_stmt): Update.
25933 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25934 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
25935 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
25936 (unloop_loops): Update.
25937 (try_peel_loop): Update.
25938 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25939 * tree-ssa-loop-split.c (connect_loops): Update.
25940 (split_loop): Update.
25941 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25942 (hoist_guard): Update.
25943 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
25944 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
25945 (value_replacement): Update.
25946 * tree-ssa-reassoc.c (branch_fixup): Update.
25947 * tree-ssa-tail-merge.c (replace_block_by): Update.
25948 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
25949 (create_edge_and_update_destination_phis): Update.
25950 (compute_path_counts): Update.
25951 (recompute_probabilities): Update.
25952 (update_joiner_offpath_counts): Update.
25953 (freqs_to_counts_path): Update.
25954 (duplicate_thread_path): Update.
25955 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
25956 (struct switch_conv_info): Update.
25957 (gen_inbound_check): Update.
25958 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25959 (vect_do_peeling): Update.
25960 (vect_loop_versioning): Update.
25961 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25962 (optimize_mask_stores): Update.
25963 * ubsan.c (ubsan_expand_null_ifn): Update.
25964 * value-prof.c (gimple_divmod_fixed_value): Update.
25965 (gimple_divmod_fixed_value_transform): Update.
25966 (gimple_mod_pow2): Update.
25967 (gimple_mod_pow2_value_transform): Update.
25968 (gimple_mod_subtract): Update.
25969 (gimple_mod_subtract_transform): Update.
25970 (gimple_ic): Update.
25971 (gimple_stringop_fixed_value): Update.
25972 (gimple_stringops_transform): Update.
25973 * value-prof.h: Update.
25975 2017-06-29 Carl Love <cel@us.ibm.com>
25977 * config/rs6000/rs6000-c.c: Add support for built-in functions
25978 vector signed int vec_signed (vector float);
25979 vector signed long long vec_signed (vector double);
25980 vector signed int vec_signed2 (vector double, vector double);
25981 vector signed int vec_signede (vector double);
25982 vector signed int vec_signedo (vector double);
25983 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
25984 instruction generator.
25985 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25986 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
25987 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
25989 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
25990 vunsignede_v2df): Add define_expands.
25991 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
25992 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
25993 VEC_UNSIGNEDO): Add definitions.
25994 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25995 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
25996 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
25997 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
25998 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
25999 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26000 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26001 vec_unsignede, vec_unsignedo): Add builtin defines.
26002 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26004 * doc/extend.texi: Update the built-in documentation file for the
26005 new built-in functions.
26007 2017-06-29 Richard Biener <rguenther@suse.de>
26009 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26010 reduction chains to LOOP_VINFO_REDUCTIONS.
26011 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26012 SLP reductions after processing reduction chains.
26014 2017-06-29 Nathan Sidwell <nathan@acm.org>
26016 * builtins.c (fold_builtin_FUNCTION): Use
26017 lang_hooks.decl_printable_name.
26019 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26021 PR middle-end/81194
26022 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26023 with only one label.
26024 * stmt.c (expand_case): Assert NCASES is greater than one.
26026 2017-06-29 Richard Biener <rguenther@suse.de>
26028 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26030 (group_case_labels): Likewise.
26031 (find_taken_edge): Push sanity checking on val to workers...
26032 (find_taken_edge_cond_expr): ... here
26033 (find_taken_edge_switch_expr): ... and here, handle cases
26034 with just a default label.
26035 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26036 (group_case_labels): Likewise.
26037 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26038 group_case_labels does anything cleanup the CFG again.
26040 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26042 PR tree-optimization/81196
26043 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26044 exit condition comparing two IVs.
26046 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26048 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26049 profile to the dummy entry at the end of the list of architectures.
26050 * config/arm/arm-cpu-cdata.h: Regenerated.
26052 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26053 Michael Collison <michael.collison@arm.com>
26056 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26058 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26059 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26060 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26061 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26062 of shift when the shift amount is masked with constant equal to
26063 the size of the mode.
26064 * config/aarch64/predicates.md (subreg_lowpart_operator): New
26067 2017-06-29 Martin Liska <mliska@suse.cz>
26069 * config/i386/i386.opt: Change range from [1,5] to [0,5].
26071 2017-06-29 Yury Gribov <tetra2005@gmail.com>
26074 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
26077 (edge_growth_cache_entry::edge_growth_cache_entry): New
26079 (reset_edge_growth_cache): Update to use constructor.
26081 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26083 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
26084 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26085 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26087 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
26089 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
26090 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
26092 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
26094 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
26095 (*-linux-uclibc*): Add t-uclibc tmake_file.
26096 * config/t-musl: New.
26097 * config/t-uclibc: New.
26099 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
26101 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
26103 (gen_comm_data): Emit architectural setting of arch_prof.
26104 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
26106 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
26107 (armv8-m.base, armv8-m.main): Likewise.
26108 * arm-protos.h (arm_build_target): Add profile field.
26109 (arch_option): Likewise.
26110 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
26112 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
26113 arm_active_target.profile.
26115 2017-06-28 Richard Biener <rguenther@suse.de>
26117 PR middle-end/81227
26118 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
26119 TYPE_OVERFLOW_WRAPS.
26120 * match.pd (negate_expr_p): Likewise.
26121 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
26122 fold_build2, not fold_binary.
26124 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26126 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
26127 Convert memory address to Pmode.
26128 (aarch64_print_operand): Assert MEM operands are always Pmode.
26130 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26133 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
26134 Remove redundant if.
26135 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
26136 * config/arm/aarch-common-protos.h
26137 (aarch_forward_to_shift_is_not_shifted_re): Remove.
26138 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
26140 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
26143 * multiple_target.c (create_dispatcher_calls): Set the default
26144 clone to be static, not public.
26146 2017-06-28 Richard Biener <rguenther@suse.de>
26148 * tree-vect-loop.c (vectorizable_reduction): Move special
26149 cond reduction IV var creation ...
26150 (vect_create_epilog_for_reduction): ... here. Remove induction_index
26151 parameter. Use STMT_VINFO_VECTYPE.
26152 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
26155 2017-06-28 Martin Liska <mliska@suse.cz>
26158 * ipa-visibility.c (non_local_p): Handle visibility.
26160 2017-06-28 Martin Liska <mliska@suse.cz>
26163 * common.opt: Add IntegerRange to various options.
26164 * opt-functions.awk (integer_range_info): New function.
26165 * optc-gen.awk: Add integer_range_info to cl_options struct.
26166 * opts-common.c (decode_cmdline_option): Handle
26167 CL_ERR_INT_RANGE_ARG.
26168 (cmdline_handle_error): Likewise.
26169 * opts.c (print_filtered_help): Show valid interval in
26170 when --help is provided.
26171 * opts.h (struct cl_option): Add range_min and range_max fields.
26172 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
26174 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
26176 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
26177 (x * C EQ/NE y * C): New transformation.
26179 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
26181 * genmultilib (combination_space): Accept '+' in option names.
26183 2017-06-28 Martin Liska <mliska@suse.cz>
26186 * asan.c (instrument_derefs): Bail out inner references
26187 that are hard register variables.
26189 2017-06-28 Jakub Jelinek <jakub@redhat.com>
26192 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
26193 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
26195 2017-06-28 Richard Biener <rguenther@suse.de>
26197 * tree-vectorizer.h (vect_get_vec_defs): Remove.
26198 (vect_get_slp_defs): Adjust.
26199 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
26201 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
26203 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
26204 get_initial_defs_for_reduction instead of vect_get_vec_defs.
26205 (vectorizable_reduction): Adjust.
26206 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
26208 (vect_get_slp_defs): Likewise.
26209 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
26210 (vectorizable_bswap): Adjust.
26211 (vectorizable_call): Likewise.
26212 (vectorizable_conversion): Likewise.
26213 (vectorizable_assignment): Likewise.
26214 (vectorizable_shift): Likewise.
26215 (vectorizable_operation): Likewise.
26216 (vectorizable_store): Likewise.
26217 (vectorizable_condition): Likewise.
26218 (vectorizable_comparison): Likewise.
26220 2017-06-28 Michael Collison <michael.collison@arm.com>
26223 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
26225 (arm_gen_movmemqi): Removed unused variable 'i'.
26226 Convert 'for' loop into 'while' loop.
26227 (arm_expand_prologue): Remove last unnecessary set of insn.
26228 (thumb_pop): Remove unused variable 'pushed_words'.
26229 (thumb_exit): Remove last unnecessary set of regs_to_pop.
26231 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26233 * config/s390/predicates.md: Use s390_rel_address_ok_p.
26234 * config/s390/s390-protos.h: Add prototype of
26235 s390_rel_address_ok_p.
26236 * config/s390/s390.c (s390_got_symbol): New function.
26237 (s390_rel_address_ok_p): New function.
26238 (legitimize_pic_address): Use s390_rel_address_ok_p.
26239 (s390_load_got): Use s390_got_symbol.
26240 (s390_option_override): Issue error if
26241 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
26242 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
26244 * config/s390/s390.opt: New option mpic-data-is-text-relative.
26246 2017-06-27 Andrew Pinski <apinski@cavium.com>
26248 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
26249 (X * copysign (1.0, X)): New pattern.
26250 (X * copysign (1.0, -X)): New pattern.
26251 (copysign (-1.0, CST)): New pattern.
26253 2017-06-27 Joseph Myers <joseph@codesourcery.com>
26255 * genmultilib (combination_space): Remove variable.
26256 Validate reuse rules against regular expression for any sequence
26257 of multilib options in any order.
26259 2017-06-27 Michael Collison <michael.collison@arm.com>
26261 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
26262 call aarch64_split_simd_combine.
26263 * (aarch64_combine_internal<mode>): Delete pattern.
26264 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
26265 Allow register and subreg operands.
26267 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26269 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
26270 specific need, just fallback on defaults.
26271 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
26273 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26274 Olivier Hainque <hainque@adacore.com>
26276 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
26278 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
26279 targets. Pick a default if no particular attempt applied.
26280 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
26283 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26285 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
26286 (x86_64-wrs-vxworks7): Likewise.
26288 2017-06-27 Marek Polacek <polacek@redhat.com>
26291 * ubsan.c (instrument_null): Check get_base_address's result for null.
26293 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26295 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
26297 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26299 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
26300 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
26301 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
26302 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
26303 New function types.
26304 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
26305 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
26306 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
26307 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
26308 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
26309 BUILT_IN_FEUPDATEENV): New builtins.
26310 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
26311 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
26312 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
26313 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
26315 (builtin_structptr_types): Adjust size.
26316 * tree.c (builtin_structptr_types): Add four entries.
26318 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26319 Olivier Hainque <hainque@adacore.com>
26321 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
26322 (TLS_SYM): New local macro, forcing reference to __tls__ on
26323 link command lines for VxWorks 7 RTPs, triggering initialization
26325 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
26326 OS features TLS support, true for RTPs on VxWorks 7.
26327 * config/vxworks.c (vxworks_override_options): Setup emutls
26330 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26332 * predict.c (test_prediction_value_range): Use -1U instead of -1
26333 to avoid narrowing conversion warning.
26334 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
26335 to avoid narrowing conversion warning.
26336 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
26338 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
26340 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26342 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
26343 64bit configurations.
26344 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
26345 (SIZE_TYPE): Likewise.
26346 * config/vxworks.c (vxworks_emutls_var_fields): Use
26347 long_unsigned_type_node instead of unsigned_type_node as the offset
26348 field type, which is "pointer" mode in emutls.c.
26350 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26353 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
26355 PR middle-end/81207
26356 * gimple-fold.c (replace_call_with_call_and_fold): Handle
26357 gimple_vuse copying separately from gimple_vdef copying.
26359 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26361 * value-prof.c (free_hist): Remove call to memset and the enclosing if
26364 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26365 Olivier Hainque <hainque@adacore.com>
26367 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
26368 for all vxworks7 targets.
26369 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
26370 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
26371 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
26372 variations for VX6/VX7 and 32/64bits later on in ...
26373 (VXWORKS_LIB_SPEC): Leverage new macros.
26374 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
26375 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
26377 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26379 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
26380 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
26382 2017-06-26 Carl Love <cel@us.ibm.com>
26384 * config/rs6000/rs6000-c.c: Add support for built-in functions
26385 vector bool char vec_reve (vector bool char);
26386 vector signed char vec_reve (vector signed char);
26387 vector unsigned char vec_reve (vector unsigned char);
26388 vector bool int vec_reve (vector bool int);
26389 vector signed int vec_reve (vector signed int);
26390 vector unsigned int vec_reve (vector unsigned int);
26391 vector bool long long vec_reve (vector bool long long);
26392 vector signed long long vec_reve (vector signed long long);
26393 vector unsigned long long vec_reve (vector unsigned long long);
26394 vector bool short vec_reve (vector bool short);
26395 vector signed short vec_reve (vector signed short);
26396 vector double vec_reve (vector double);
26397 vector float vec_reve (vector float);
26398 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
26399 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
26400 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
26401 (altivec_vreve): New pattern.
26402 * config/rs6000/altivec.h (vec_reve): New define.
26403 * doc/extend.texi (vec_rev): Update the built-in documentation file
26404 for the new built-in functions.
26406 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26408 PR tree-optimization/71815
26409 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
26411 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
26413 (slsr_process_phi): Likewise.
26414 (replace_uncond_cands_and_profitable_phis): Don't replace a
26415 multiply candidate with a stride of 1 (copy or cast).
26416 (phi_incr_cost): Call uses_consumed_by_stmt rather than
26418 (lowest_cost_path): Likewise.
26419 (total_savings): Likewise.
26421 2017-06-26 Richard Biener <rguenther@suse.de>
26424 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
26425 Use def_builtin_pure for all gather builtins.
26427 2017-06-26 Richard Biener <rguenther@suse.de>
26429 PR tree-optimization/81203
26430 * tree-tailcall.c (find_tail_calls): Do not move stmts into
26431 non-dominating BBs.
26433 2017-06-26 Marek Polacek <polacek@redhat.com>
26436 * doc/invoke.texi: Document -Wmultistatement-macros.
26438 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
26440 * doc/sourcebuild.texi (ARM-specific attributes): Document new
26441 arm_neon_ok_no_float_abi effective target.
26443 2017-06-26 Richard Biener <rguenther@suse.de>
26445 PR tree-optimization/80928
26446 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
26447 (copy_bbs): Set BB_DUPLICATED flag early.
26448 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
26450 (execute_on_shrinking_pred): Likewise.
26451 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
26452 BB_DUPLICATED blocks.
26453 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
26454 iterate over all PHIs considering removal of *gsi.
26456 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
26458 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
26461 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26463 * config/rs6000/rs6000-string.c: (expand_block_clear,
26464 do_load_for_compare, select_block_compare_mode,
26465 compute_current_alignment, expand_block_compare,
26466 expand_strncmp_align_check, expand_strn_compare,
26467 expand_block_move, rs6000_output_load_multiple)
26468 Move functions related to string/block move/compare
26469 to a separate file.
26470 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
26471 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
26472 for this function which is now used in two files.
26473 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
26474 * config.gcc: Add rs6000-string.o to extra_objs for
26475 targets powerpc*-*-* and rs6000*-*-*.
26477 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
26480 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
26481 32-bit, since indexed is not valid for DImode.
26482 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
26483 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
26484 (define_peephole2 for Altivec d-form load): Add 32-bit support.
26485 (define_peephole2 for Altivec d-form store): Likewise.
26488 * multiple_target.c (create_dispatcher_calls): Only create the
26489 dispatcher call if the function is the default clone of a
26490 versioned function.
26492 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
26494 PR middle-end/80902
26495 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
26496 a call, force the call to not be a tail call.
26498 2017-06-23 Jeff Law <law@redhat.com>
26500 * doc/contrib.texi: Add entry for Steven Pemberton's work on
26503 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
26505 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
26506 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
26507 handling for early expansion of vector shifts (sl,sr,sra,rl).
26508 (builtin_function_type): Add vector shift right instructions
26509 to the unsigned argument list.
26511 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
26513 rtl-optimizatoin/79286
26514 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
26515 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
26516 trap. PIC register plus a const unspec without offset can never trap.
26518 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
26520 * tree.h (builtin_structptr_type): New type.
26521 (builtin_structptr_types): Declare new array.
26522 * tree.c (builtin_structptr_types): New array.
26523 (free_lang_data, build_common_tree_nodes): Use it.
26525 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
26528 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
26531 2017-06-22 Matt Turner <mattst88@gmail.com>
26533 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
26534 Lake models to skylake case. Assume skylake for unknown
26535 models with clflushopt.
26537 2017-06-22 Jeff Law <law@redhat.com>
26539 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
26540 frame sizes that do not satisfy aarch64_uimm12_shift.
26542 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
26544 * profile-count.h (apply_probability,
26545 apply_scale, probability_in): Fix checks for zero.
26547 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26549 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
26550 * doc/cppdiropts.texi (-I @var{dir}): Document it.
26552 2016-06-22 Richard Biener <rguenther@suse.de>
26554 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26555 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
26556 REDUC_MAX_EXPR support.
26557 (vectorizable_reduction): Likewise.
26558 (vect_create_epilog_for_reduction): Likewise.
26560 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
26562 * match.pd (A / (1 << B) -> A >> B): New.
26563 * generic-match-head.c: Include optabs-tree.h.
26564 * gimple-match-head.c: Likewise.
26565 * optabs-tree.h (target_supports_op_p): New.
26566 * optabs-tree.c (target_supports_op_p): New.
26568 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26570 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
26571 $gcc_cv_ld --help output.
26572 (gcc_cv_ld_demangle): Likewise.
26573 (gcc_cv_ld_eh_frame_hdr): Likewise.
26574 (gcc_cv_ld_pie): Likewise.
26575 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
26576 (gcc_cv_ld_buildid): Likewise.
26577 (gcc_cv_ld_sysroot): Likewise.
26578 (ld_bndplt_support): Likewise.
26579 (ld_pushpopstate_support): Likewise.
26580 * configure: Regenerate.
26581 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26583 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26586 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26587 operands indexes to avoid gap between operands and match_dups.
26589 2017-06-21 Andrew Pinski <apinski@cavium.com>
26591 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26592 Increment Arith_shift and Arith_shift_reg by 1.
26593 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26595 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26596 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26597 (aarch64_strip_extend): Add new argument and test for it.
26598 (aarch64_cheap_mult_shift_p): New function.
26599 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26600 add a cost if it is true.
26601 Update calls to aarch64_strip_extend.
26602 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26604 2017-06-21 Andrew Pinski <apinski@cavium.com>
26606 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26608 (thunderxt88): Likewise.
26609 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26610 (thunderx_prefetch_tune): New variable.
26611 (thunderx2t99_prefetch_tune): Update for the correct values.
26612 (thunderxt88_tunings): New variable.
26613 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26614 generic_prefetch_tune.
26615 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26617 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26619 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26620 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26621 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26622 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26623 (aarch64_atomic_cas<mode>, GPI): Likewise.
26625 2017-06-21 Martin Liska <mliska@suse.cz>
26627 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26628 statements on cold and hot labels.
26629 * predict.c (tree_estimate_probability_bb): Remove the
26630 prediction from this place.
26632 2017-06-21 Martin Liska <mliska@suse.cz>
26634 PR tree-optimization/79489
26635 * gimplify.c (maybe_add_early_return_predict_stmt): New
26637 (gimplify_return_expr): Call the function.
26638 * predict.c (tree_estimate_probability_bb): Remove handling
26640 * predict.def: Update comment about early return predictor.
26641 * gimple-predict.h (is_gimple_predict): New function.
26642 * predict.def: Change default value of early return to 66.
26643 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26645 * passes.def: Put pass_strip_predict_hints to the beginning of
26648 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26650 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26651 FUNCTION_DECL declarations.
26652 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26654 (dwaf2out_decl): Likewise.
26655 * godump.c (go_early_global_decl): Skip call to the real debug hook
26656 for FUNCTION_DECL declarations.
26657 * passes.c (rest_of_decl_compilation): Skip call to the
26658 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26659 -fdump-go-spec is passed.
26661 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26663 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26664 Reorder for compactness.
26665 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26666 (def_builtin_pure, def_builtin_pure2): New functions.
26667 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26669 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26671 * match.pd (nop_convert): New predicate.
26672 ((A +- CST1) +- CST2): Allow some NOP conversions.
26674 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26677 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26678 with ctors/dtors if GOVD_SHARED is set.
26680 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26682 * config/aarch64/aarch64.md (movti_aarch64):
26683 Emit mov rather than orr.
26684 (movtf_aarch64): Likewise.
26685 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26686 Emit mov rather than orr.
26688 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26690 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26691 Swap alternatives, make integer dup more expensive.
26693 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26695 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26696 Return true for non-tls symbols.
26698 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26700 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26701 (cortex-a75): Likewise.
26702 (cortex-a75.cortex-a55): Likewise.
26703 * config/aarch64/aarch64-tune.md: Regenerate.
26704 * doc/invoke.texi (-mtune): Document new values for -mtune.
26706 2017-06-21 Tom de Vries <tom@codesourcery.com>
26708 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26709 stack_size feature.
26710 (Effective-Target Keywords, Other attributes): Suggest using
26711 dg-add-options stack_size feature to get stack limit in stack_size
26712 effective target documentation.
26714 2017-06-21 Julian Brown <julian@codesourcery.com>
26715 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26717 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26718 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26719 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26721 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26722 attribute type list for neon_multiply.
26723 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26724 attribute type list for neon_multiply.
26725 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26726 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26727 attribute type list for neon_multiply.
26728 * config/arm/types.md (crypto_pmull): Add.
26729 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26730 attribute type list.
26732 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26734 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26737 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26739 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26740 to make sure not to dereference a NULL cost_classes_ptr pointer.
26742 2017-06-20 Carl Love <cel@us.ibm.com>
26744 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26745 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26746 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26747 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26748 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26749 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26750 VMULOSW): New enum "unspec" values.
26751 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26752 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26753 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26754 altivec_vmulosw): New patterns.
26755 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26756 VMULOSW): Add definitions.
26758 2017-06-20 Julia Koval <julia.koval@intel.com>
26760 * config/i386/i386.c: Fix rounding expand for new pattern.
26761 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26763 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26765 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26766 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26768 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26770 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26773 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26775 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26776 architecture, then by implementer ID.
26777 * config/aarch64/aarch64-tune.md: Regenerate.
26779 2017-06-20 Richard Biener <rguenther@suse.de>
26781 PR middle-end/81097
26782 * fold-const.c (split_tree): Fold to type before negating.
26784 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26786 * diagnostic-show-locus.c
26787 (selftest::test_fixit_deletion_affecting_newline): New function.
26788 (selftest::diagnostic_show_locus_c_tests): Call it.
26790 2017-06-20 Andreas Schwab <schwab@suse.de>
26793 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26796 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26798 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26799 __ARM_FEATURE_COPROC according to support.
26801 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26803 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26804 Rewritten to avoid overflow for > 32-bit pointers.
26807 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26808 by removing enum keyword.
26809 (ubsan_type_descriptor): Likewise. Formatting fix.
26812 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26813 splitter): Require TARGET_SSE2 in the condition.
26815 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26818 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26819 for doing vector set of SFmode on ISA 3.0.
26820 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26821 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
26823 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
26824 SFmode value into a V4SF variable that was extracted from another
26825 V4SF variable without converting the element to double precision
26826 and back to single precision vector format.
26827 (vsx_insert_extract_v4sf_p9_2): Likewise.
26829 2017-06-19 Jakub Jelinek <jakub@redhat.com>
26831 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
26832 in UWHI to avoid undefined overflow.
26835 * ubsan.h (enum ubsan_encode_value_phase): New.
26836 (ubsan_encode_value): Change second argument to
26837 enum ubsan_encode_value_phase with default value of
26838 UBSAN_ENCODE_VALUE_GENERIC.
26839 * ubsan.c (ubsan_encode_value): Change second argument to
26840 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
26841 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
26842 create_tmp_var_raw instead of create_tmp_var and use a
26844 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
26845 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
26846 ubsan_encode_value callers.
26849 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
26850 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
26851 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
26853 2017-06-19 Richard Biener <rguenther@suse.de>
26855 PR middle-end/81118
26856 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
26857 estimates if we changed anything.
26859 2017-06-19 Richard Biener <rguenther@suse.de>
26861 PR tree-optimization/80887
26862 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
26863 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
26864 simplified lookups, then reset mprts_hook.
26865 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
26867 (try_to_simplify): Likewise.
26869 2017-06-19 Martin Liska <mliska@suse.cz>
26872 * gimplify.c (gimplify_switch_expr):
26873 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
26875 2017-06-19 Martin Liska <mliska@suse.cz>
26877 * doc/install.texi: Document that PGO runs in 4 stages.
26879 2017-06-19 Martin Liska <mliska@suse.cz>
26882 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
26883 to dispatcher function name.
26884 * multiple_target.c (replace_function_decl): New function.
26885 (create_dispatcher_calls): Redirect both edges and references.
26887 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
26889 * profile-count.c (profile_count::dump): Dump quality.
26890 (profile_count::differs_from_p): Update for unsigned val.
26891 * profile-count.h (profile_count_quality): New enum.
26892 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
26894 2017-06-19 Richard Biener <rguenther@suse.de>
26896 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
26897 struct function as arg.
26898 (estimate_numbers_of_iterations): Export overload with loop arg.
26899 (free_numbers_of_iterations_estimates_loop): Use an overload of
26900 free_numbers_of_iterations_estimates instead.
26901 * tree-cfg.c (remove_bb): Adjust.
26902 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
26903 * tree-parloops.c (gen_parallel_loop): Likewise.
26904 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
26906 (tree_unroll_loops_completely): Likewise.
26907 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
26908 Use an overload instead and export.
26909 (estimated_loop_iterations): Adjust.
26910 (max_loop_iterations): Likewise.
26911 (likely_max_loop_iterations): Likewise.
26912 (estimate_numbers_of_iterations): Take struct function as arg
26914 (loop_exits_before_overflow): Adjust.
26915 (free_numbers_of_iterations_estimates_loop): Use an overload.
26916 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
26917 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
26919 2017-06-19 Richard Biener <rguenther@suse.de>
26922 * ipa-prop.c (find_constructor_constant_at_offset): Handle
26923 RANGE_EXPR conservatively.
26925 2017-06-16 Carl Love <cel@us.ibm.com>
26927 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26928 definitions for vec_float, vec_float2, vec_floato,
26929 vec_floate built-ins.
26930 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
26931 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
26933 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
26934 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
26935 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
26936 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
26937 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
26938 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
26939 vec_floato): Add builtin defines.
26940 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
26941 Update the built-in documentation file for the new built-in
26944 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26946 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
26947 (mthumb): Mark as the negative of -marm.
26949 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26951 * doc/invoke.texi (ARM Options, -mcpu): Document supported
26953 (ARM Options, -mtune): Document that this accepts the same
26954 extension options as -mcpu.
26955 (ARM Options, -mfpu): Document addition of -mfpu=auto.
26957 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26959 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
26960 permitted extensions.
26962 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26964 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
26965 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
26966 (armv8-m.main): Add option +nodsp.
26967 * config/arm/arm-cpu-cdata.h: Regenerated.
26969 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26971 * config/arm/t-fuchsia: New file.
26972 * config.gcc (arm*-*-fuchsia*): Use it.
26974 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26976 * config/arm/t-symbian: Rewrite for new option infrastructure.
26978 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26980 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
26981 (MULTILIB_REQUIRED): Likewise.
26983 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26985 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
26986 (MULTILIB_RESUE): Likewise.
26987 (MULTILIB_MATCHES): Likewise.
26988 (MULTLIB_REQUIRED): Likewise.
26990 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26992 * config/arm/t-rtems: Rewrite for new option framework.
26994 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26996 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
26997 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
26998 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
26999 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27000 * config/arm/t-multilib: ... here.
27001 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27002 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27003 armv7-a and armv8*-a when A-profile libraries have not been built.
27004 * config/arm/t-rmprofile: Rewrite.
27006 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27008 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27009 with a backslash. Remove the backslash after substituting unescaped
27011 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27013 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27015 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27016 the driver to pass through the default setting of -mfloat-abi.
27017 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27019 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27020 (all_feat_combs): New rule.
27021 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27023 * config/arm/t-aprofile: Rewrite.
27025 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27027 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27028 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27029 fpu is not specified by the user/command-line.
27030 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27031 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27032 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27033 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27034 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27035 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27037 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27039 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27040 * config/arm/t-arm-elf: Rewritten.
27042 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27044 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27045 have some floating-point instructions.
27046 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27047 (TARGET_MAYBE_HARD_FLOAT): New macro.
27048 * config/arm/arm-builtins.c (arm_init_builtins): Use
27049 TARGET_MAYBE_HARD_FLOAT.
27050 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27052 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27054 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27055 (configargs.h): Include it.
27056 (arm_print_hint_for_fpu_option): New function.
27057 (arm_parse_fpu_option): New function.
27058 (candidate_extension): New class.
27059 (arm_canon_for_multilib): New function.
27060 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27061 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27062 (ARCH_CANONICAL_SPECS): New macro.
27063 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
27065 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27067 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
27068 are set after handling multilib fragments. Set target_cpu_default2
27071 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27073 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
27075 (arm*-*-*): Set target_cpu_default2 to a quoted string.
27076 * config/arm/parsecpu.awk (check_cpu): Validate any extension
27078 (check_arch): Likewise.
27079 * config/arm/arm.c (arm_configure_build_target): Handle
27080 TARGET_CPU_DEFAULT being a string constant. Scan any feature
27081 options in the default.
27083 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27085 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
27086 when an option is an alias of another.
27087 * config/arm/parsecpu.awk (optalias): New parser token.
27088 (gen_comm_data): Mark non-alias options as such. Emit entries
27089 for extension aliases.
27090 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
27091 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
27092 (armv6kz, armv6zk, armv6t2): Likewise.
27093 (armv7): Make vfpv3-d16 an alias.
27094 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
27096 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
27097 Sort in canonical order.
27098 (armv8-a): Sort in canonical order.
27099 (armv8.1-a, armv8.2-a): Likewise.
27100 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
27102 (cortex-a9): Sort in canonical order.
27103 * config/arm/arm.c (selftests.h): Include it.
27104 (arm_test_cpu_arch_data): New function.
27105 (arm_run_self_tests): New function.
27106 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
27107 (targetm): Move declaration to the end of the file.
27108 * arm-cpu-cdata.h: Regenerated.
27110 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27112 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
27113 call to target_mode_check describing the type of option passed.
27114 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
27115 (arm_target_thumb_only): Use arm_parse_arch_option_name or
27116 arm_parse_cpu_option_name to match parameters against list of
27118 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
27119 arm_arch_core_flags data structure.
27120 * config/arm/arm-cpu_cdata.h: Regenerated.
27122 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27124 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
27126 (arm_print_hint_for_cpu_option): Likewise.
27127 (arm_print_hint_for_arch_option): Likewise.
27128 (arm_parse_cpu_option_name): Likewise.
27129 (arm_parse_arch_option_name): Likewise.
27130 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
27131 of entries in the all_fpus list.
27132 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
27133 (arm_parse_cpu_option_name): Declare.
27134 (arm_parse_arch_option_name): Declare.
27135 (arm_parse_option_features): Declare.
27136 (arm_intialize_isa): Declare.
27137 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
27139 (gen_comm_data): ... here. Make definitions non-static.
27140 * config/arm/arm-cpu-data.h: Regenerated.
27141 * config/arm/arm-cpu-cdata.h: Regenerated.
27143 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27145 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
27146 (cpu_arch_extension): New structure.
27147 (cpu_arch_option, arch_option, cpu_option): New structures.
27148 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
27149 architecture types.
27150 (gen_data): Generate new format data tables.
27151 * config/arm/arm.c (cpu_tune): New structure.
27152 (cpu_option, processors): Delete.
27153 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
27154 (arm_print_hint_for_cpu_option): ... this and ...
27155 (arm_print_hint_for_arch_option): ... this.
27156 (arm_parse_arch_cpu_name): Delete. Replace with ...
27157 (arm_parse_cpu_option_name): ... this and ...
27158 (arm_parse_arch_option_name): ... this.
27159 (arm_unrecognized_feature): Change type of target parameter to
27161 (arm_parse_arch_cpu_features): Delete. Replace with ...
27162 (arm_parse_option_features): ... this.
27163 (arm_configure_build_target): Rework to use new configuration data
27165 (arm_print_tune_info): Rework for new configuration data tables.
27166 * config/arm/arm-cpu-data.h: Regenerated.
27167 * config/arm/arm-cpu.h: Regenerated.
27169 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27171 * Makefile.in (OBJS): Move sbitmap.o from here ...
27172 (OBJS-libcommon): ... to here.
27174 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27176 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
27177 (ISA_ALL_CRYPTO): New macro.
27178 (ISA_ALL_SIMD): New macro
27179 (ISA_ALL_FP): New macro.
27180 * config/arm/arm.c (fpu_bitlist): Update initializer.
27181 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
27183 (arm9e): Add fpu. Add option for nofp
27184 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
27185 (arm926ej-s, arm1026ej-s): Likewise.
27186 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
27187 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
27188 neon-fp16, neon-vfpv4, nofp and nosimd.
27189 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
27190 (cortex-a8): Add fpu. Add option for nofp.
27191 (cortex-a9): Add fpu. Add options for nosimd and nofp.
27192 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
27193 (cortex-r4f): Add fpu.
27194 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
27195 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
27197 (cortex-r8): Likewise.
27198 (cortex-m4): Add fpu. Add option for nofp.
27199 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
27200 (cortex-a17.cortex-a7): Likewise.
27201 (cortex-a32): Add fpu. Add options for crypto and nofp.
27202 (cortex-a35, cortex-a53): Likewise.
27203 (cortex-a57): Add fpu. Add option for crypto.
27204 (cortex-a72, cortex-a73): Likewise.
27205 (exynos-m1): Likewise.
27206 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
27207 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
27208 (cortex-m33): Add fpu. Add option for nofp.
27209 * config/arm/arm-cpu-cdata.h: Regenerated
27210 * config/arm/arm-cpu-data.h: Regenerated.
27212 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27214 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
27215 (armv5te, armv5tej): Likewise.
27216 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
27217 (armv7): Add options fp and vfpv3-d16.
27218 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
27219 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
27221 (armv7ve): Likewise.
27222 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
27223 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
27224 (armv8-a): Add nocrypto option.
27225 (armv8.1-a, armv8.2-a): Likewise.
27226 (armv8-m.main): add options fp, fp.dp and nofp.
27228 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27230 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
27232 (armv8-a+crc): Delete.
27233 (armv8.1-a): Add options simd, crypto and nofp.
27234 (armv8.2-a): Add options fp16, simd, crypto and nofp.
27235 (armv8.2-a+fp16): Delete.
27236 (armv8-m.main): Add option dsp.
27237 (armv8-m.main+dsp): Delete.
27238 (cortex-a8): Add fpu. Add nofp option.
27239 (cortex-a9): Add fpu. Add nofp and nosimd options.
27240 * config/arm/parsecpu.awk (gen_data): Generate option tables and
27241 link to main cpu and architecture data structures.
27242 (gen_comm_data): Only put isa attributes from the main architecture
27244 (option): New statement for architecture and CPU entries.
27245 * arm.c (struct cpu_option): New structure.
27246 (struct processors): Add entry for options.
27247 (arm_unrecognized_feature): New function.
27248 (arm_parse_arch_cpu_name): Ignore any characters after the first
27250 (arm_parse_arch_cpu_feature): New function.
27251 (arm_configure_build_target): Separate out any CPU and architecture
27252 features and parse separately. Don't error out if -mfpu=auto is
27253 used with only an architecture string.
27254 (arm_print_asm_arch_directives): New function.
27255 (arm_file_start): Call it.
27256 * config/arm/arm-cpu-cdata.h: Regenerated.
27257 * config/arm/arm-cpu-data.h: Likewise.
27258 * config/arm/arm-tables.opt: Likewise.
27260 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27262 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
27263 assembler when it is not -mfpu=auto.
27265 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27267 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
27268 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
27269 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
27270 (ASM_CPU_SPEC): Rewrite.
27271 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
27272 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
27273 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
27274 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
27275 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
27276 copied string is NUL-terminated. Also strip any characters prefixed
27278 (arm_rewrite_selected_arch): New function.
27279 (arm_rewrite_march): New function.
27281 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27283 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
27284 (x_arm_cpu_string, x_arm_tune_string): Likewise.
27285 (march, mcpu, mtune): Convert to string-based options.
27286 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
27287 (arm_parse_arch_cpu_name): New function.
27288 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
27289 identify selected architecture or CPU.
27290 (arm_option_save): New function.
27291 (TARGET_OPTION_SAVE): Redefine.
27292 (arm_option_restore): Restore string options.
27293 (arm_option_print): Print string options.
27295 2017-06-16 Martin Sebor <msebor@redhat.com>
27297 PR tree-optimization/80933
27298 PR tree-optimization/80934
27299 * builtins.c (fold_builtin_3): Do not handle bcmp here.
27300 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
27301 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
27302 (gimple_fold_builtin): Call them.
27304 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27306 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
27307 as unlikely; update profile.
27309 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27311 * predict.c (force_edge_cold): Handle declaring edges impossible
27314 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27316 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
27318 (try_unroll_loop_completely): Fix reporting.
27320 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27322 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
27324 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
27327 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
27328 if given a non-constant argument for an intrinsic which requires a
27331 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27333 * profile.c (compare_freqs): New function.
27334 (branch_prob): Sort edge list.
27335 (find_spanning_tree): Assume that the list is priority sorted.
27337 2017-06-16 Richard Biener <rguenther@suse.de>
27339 PR tree-optimization/81090
27340 * passes.def (pass_record_bounds): Remove.
27341 * tree-pass.h (make_pass_record_bounds): Likewise.
27342 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
27343 make_pass_record_bounds): Likewise.
27344 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
27345 not free niter estimates at the beginning but at the end.
27346 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
27348 2017-06-16 Richard Biener <rguenther@suse.de>
27350 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
27351 initializer to workaround ICE in host GCC 4.8.
27353 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27355 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
27357 (clone_inlined_nodes): Update.
27359 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27361 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
27362 prefetch settings, and enable prefetching by default at -O3.
27364 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27366 * config/aarch64/aarch64.c (aarch64_override_options_internal):
27367 Set flag_prefetch_loop_arrays according to tuning data.
27369 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27371 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
27372 New tune structure.
27373 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
27374 [Unrelated to main purpose of the patch] Place the pointer field last
27375 to enable type checking errors when tune structure are wrongly merged.
27376 * config/aarch64/aarch64.c (generic_prefetch_tune,)
27377 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
27378 (thunderx2t99_prefetch_tune): New tune constants.
27379 (tune_params *_tunings): Update all tunings (no functional change).
27380 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
27381 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
27382 from tunings structures.
27384 2017-06-16 Jakub Jelinek <jakub@redhat.com>
27387 * ubsan.c (instrument_null): Add T argument, use it instead
27388 of computing it based on IS_LHS.
27389 (instrument_object_size): Likewise.
27390 (pass_ubsan::execute): Adjust instrument_null and
27391 instrument_object_size callers to pass gimple_get_lhs or
27392 gimple_assign_rhs1 result to it. Use instrument_null instead of
27393 calling get_base_address and instrument_mem_ref. Handle
27394 aggregate call arguments for object-size sanitization.
27396 2017-06-16 Yury Gribov <tetra2005@gmail.com>
27398 PR tree-optimization/81089
27399 * tree-vrp.c (is_masked_range_test): Validate operands of
27402 2017-06-15 Martin Sebor <msebor@redhat.com>
27405 * dumpfile.c (dump_register): Avoid calling memset to initialize
27406 a class with a default ctor.
27407 * gcc.c (struct compiler): Remove const qualification.
27408 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
27409 * hash-table.h: Ditto.
27410 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
27412 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
27413 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
27415 * params.h (struct param_info): Make struct members non-const.
27416 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
27417 with default initialization.
27418 * vec.h (vec_copy_construct, vec_default_construct): New helper
27420 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
27421 with vec_copy_construct.
27422 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
27423 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
27424 * doc/invoke.texi (-Wclass-memaccess): Document.
27426 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27428 * emit-rtl.h (is_leaf): Update comment about local
27429 register allocator.
27431 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27434 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
27435 for a variable to have a section before checking if the section has a
27437 Set section to.persistent if persistent attribute is set.
27438 Warn if .persistent attribute is used on an automatic variable.
27440 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
27442 PR rtl-optimization/80474
27443 * reorg.c (update_block): Do not ignore instructions in a delay slot.
27445 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
27447 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
27450 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
27452 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
27453 (casesi): Emit bounds checking as RTL.
27454 (casesi_internal_mips16_<mode>): Remove bounds checking.
27456 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
27458 * config/xtensa/xtensa.c (xtensa_option_override): Append
27459 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
27460 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
27461 xtensa_doloop_hooks): Define unconditionally.
27462 (xtensa_reorg_loops): Only call reorg_loops in the presence of
27464 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
27465 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
27466 for it in xtensa_option_override.
27467 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
27468 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
27470 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
27472 * doc/cppopts.texi: Document '-' special value to -MF.
27474 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
27476 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
27477 (cortex_a53_fconst): Likewise.
27478 (cortex_a53_fpmul): Likewise.
27479 (cortex_a53_f_load_64): Likewise.
27480 (cortex_a53_f_load_many): Likewise.
27481 (cortex_a53_advsimd_alu): Likewise.
27482 (cortex_a53_advsimd_alu_q): Likewise.
27483 (cortex_a53_advsimd_mul): Likewise.
27484 (cortex_a53_advsimd_mul_q): Likewise.
27485 (fpmac bypass): Add new bypass for fpmac-fpmac case.
27486 Add missing fmul, r2f_cvt and fconst cases.
27488 2017-06-14 Richard Biener <rguenther@suse.de>
27490 PR middle-end/81088
27491 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
27493 (fold_binary_loc): When associating do not treat pre-existing
27494 TREE_OVERFLOW on literal constants as a reason to allow
27495 TREE_OVERFLOW on associated literal constants.
27497 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27499 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
27500 (MASK_FEATURES): New macro.
27501 * config/sparc/sparc.c (sparc_option_override): Remove the special
27502 handling of -mfpu and generalize it to all MASK_FEATURES switches.
27504 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27506 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
27507 a division of 0 if non-call exceptions are enabled.
27509 2017-06-14 Andrew Pinski <apinski@cavium.com>
27510 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27513 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
27514 Improve vector initialization code gen for only variable case.
27516 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27518 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
27520 2017-06-14 Richard Biener <rguenther@suse.de>
27522 PR tree-optimization/81083
27523 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
27526 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27528 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
27529 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
27530 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
27531 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
27532 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
27533 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
27535 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27537 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
27538 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
27540 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27542 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
27544 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27546 * config/rs6000/t-rtems: Don't handle SPE.
27548 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27550 * config/rs6000/t-linux: Don't handle SPE.
27552 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27554 * config/rs6000/eabispe.h: Delete file.
27556 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27558 * config/rs6000/t-spe: Delete file.
27560 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27562 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
27563 (rs6000_legitimate_offset_address_p): Return false for anything in
27564 V2SImode or V2SFmode.
27566 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27568 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
27569 except V2SF and V2SI. Rearrange the vector modes, and add comments.
27570 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
27572 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
27573 (rs6000_legitimate_offset_address_p): Ditto.
27574 (rs6000_emit_move): Ditto.
27575 (rs6000_init_builtins): Remove V4HI_type_node.
27577 2017-06-13 Martin Liska <mliska@suse.cz>
27580 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27581 (gate_asan): Likewise.
27582 * asan.h (asan_no_sanitize_address_p): Remove the function.
27583 (sanitize_flags_p): New function.
27584 * builtins.def: Fix coding style.
27585 * common.opt: Use renamed enum value.
27586 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27587 * doc/extend.texi: Document no_sanitize attribute.
27588 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27589 to SANITIZE_UNDEFINED_NONDEFAULT.
27590 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27591 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27592 Use sanitize_flags_p.
27593 * gimplify.c (gimplify_function_tree): Likewise.
27594 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27595 * opts.c (parse_no_sanitize_attribute): New function.
27596 (common_handle_option): Use renamed enum value.
27597 * opts.h (parse_no_sanitize_attribute): Declare.
27598 * tree.c (sanitize_flags_p): New function.
27599 * tree.h: Declared here.
27600 * tsan.c: Use sanitize_flags_p.
27601 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27602 (instrument_mem_ref): Likewise.
27603 (instrument_bool_enum_load): Likewise.
27604 (do_ubsan_in_current_function): Remove the function.
27605 (pass_ubsan::execute): Use sanitize_flags_p.
27606 * ubsan.h: Remove do_ubsan_in_current_function
27607 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27608 (dump_function_to_file): Use it here.
27610 2017-06-13 Martin Jambor <mjambor@suse.cz>
27612 PR tree-optimization/80803
27613 PR tree-optimization/81063
27614 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27615 (propagate_subaccesses_across_link): Enqueue subtree whenever
27616 necessary instead of relying on the caller.
27618 2017-06-13 Martin Jambor <mjambor@suse.cz>
27620 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27621 that have a first_link.
27622 (sort_and_splice_var_accesses): Do not check first_link before
27624 (subtree_mark_written_and_enqueue): Likewise.
27625 (propagate_all_subaccesses): Likewise and do not stop at first
27626 parent with a first_link.
27628 2017-06-13 Martin Jambor <mjambor@suse.cz>
27630 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27633 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27635 * match.pd: New pattern.
27637 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27639 * tree-vrp.c (is_masked_range_test): New function.
27640 (register_edge_assert_for): Determine ranges for
27643 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27645 PR tree-optimization/67328
27646 * fold-const.c (maskable_range_p): New function.
27647 (build_range_check): Generate bittests if possible.
27649 2017-06-13 Martin Liska <mliska@suse.cz>
27651 * gimple-pretty-print.c (dump_probability): Add new argument.
27652 (dump_edge_probability): Dump both probability and count.
27653 (dump_gimple_label): Likewise.
27654 (dump_gimple_bb_header): Likewise.
27656 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27659 * config/avr/avr-devices.c: Fix indentation.
27660 * config/avr/gen-avr-mmcu-specs.c: Dito.
27662 2017-06-13 Richard Biener <rguenther@suse.de>
27664 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27665 instead get vector type from stmt_info.
27666 (vectorizable_reduction): Adjust. Remove dead code.
27668 2017-06-13 Richard Biener <rguenther@suse.de>
27670 PR middle-end/81065
27671 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27672 case of C * (x * C2 + C3).
27673 (fold_addr_of_array_ref_difference): Properly fold index difference.
27675 2017-06-12 David S. Miller <davem@davemloft.net>
27678 * config/sparc/sparc.md (return expander): Emit frame blockage if
27679 function uses alloca.
27681 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27683 * combine.c (make_field_assignment): Check len rather than the mode
27684 precision when calling force_to_mode.
27686 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27688 Support multilibs and devices that see flash in RAM address range.
27691 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27692 (avr_mcu_t) <flash_pm_offset>: New field.
27693 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27694 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27695 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27696 (AVR_TINY_PM_OFFSET): Remove macro.
27697 * config/avr/avr.opt (-mshort-calls): New option.
27698 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27699 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27700 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27701 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27702 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27703 instead of avr_arch->have_jmp_call.
27704 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27705 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27706 avr_arch->flash_pm_offset to define.
27707 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27708 new field flash_pm_offset. Add entry for avrxmega3.
27709 (avr_texinfo): Add entry for avrxmega3.
27710 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27711 attiny212, attiny214,
27712 attiny412, attiny414, attiny416, attiny417,
27713 attiny814, attiny816, attiny817,
27714 attiny1614, attiny1616, attiny1617,
27715 attiny3214, attiny3216, attiny3217.
27716 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27717 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27718 (avr_print_operand_address) [AVR_TINY]: Same.
27719 (avr_asm_init_sections) <readonly_data_section>: Only patch
27720 callback if avr_arch->flash_pm_offset = 0.
27721 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27722 for rodata if avr_arch->flash_pm_offset != 0.
27723 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27724 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27725 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27726 (m_options): Append opt_rcall.
27727 (m_dirnames): Append dir_rcall.
27728 * config/avr/t-multilib: Regenerate.
27730 * configure.ac [target=avr]: Check whether avrxmega3 default
27731 linker description file works as needed.
27732 * configure: Regenerate.
27733 * doc/avr-mmcu.texi: Regenerate.
27734 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27735 <__AVR_ARCH__>: Document avrxmega3 and 103.
27736 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27737 <__AVR_SHORT_CALLS__>: Document it.
27738 <__AVR_PM_BASE_ADDRESS__>: Document it.
27739 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27740 (AVR Variable Attributes) <progmem>: Document this is
27741 not needed for avrxmega3.
27742 (AVR Named Address Spaces) <__flash>: Dito.
27744 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27746 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27748 2017-06-12 Doug Rupp <rupp@adacore.com>
27750 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27751 Append vxworks-stdint.h to the tm_file list.
27752 * config/vxworks-stdint.h: New file.
27754 2017-06-12 Martin Liska <mliska@suse.cz>
27756 PR tree-optimization/81041
27757 * tree-profile.c (gimple_gen_ic_func_profiler):
27758 Create an extra BB in profile-generate
27759 (gimple_gen_time_profiler): Likewise.
27761 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27763 PR tree-optimization/81003
27764 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27765 (update_range_test): Use it instead of force_gimple_operand_gsi.
27767 2017-06-12 Richard Biener <rguenther@suse.de>
27769 PR tree-optimization/81053
27770 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27771 with backedge value not defined in loop. Simplify def stmt
27774 2017-06-11 Tom de Vries <tom@codesourcery.com>
27777 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27779 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27780 nvptx_cannot_force_const_mem.
27782 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27784 * opts.c (finish_options): Move test for flag_split_stack after
27785 it has been initialized.
27787 2017-06-11 Jason Merrill <jason@redhat.com>
27789 * tree.h (id_equal): New.
27790 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27791 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27792 instead of strcmp of IDENTIFIER_POINTER.
27794 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27796 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27797 (mark_all_inlined_calls_cdtor): Fix formating.
27798 (inline_transform): Rescale profile before inlining.
27800 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27802 * cgraph.h (cgraph_edge::clone): Update prototype.
27803 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27804 (cgraph_node::create_clone): Update.
27805 (cgraph_node::create_version_clone): Update.
27806 * tree-inline.c (copy_bb): Update.
27807 (expand_call_inline): Update.
27809 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27811 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27812 factored out from ...
27813 (rs6000_emit_prologue): ... here.
27815 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27817 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27818 factored out from ...
27819 (rs6000_emit_prologue): ... here.
27821 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27823 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
27825 (handle_missing_profiles): Fix computation of tp_first_run.
27826 (counts_to_freqs): Do not touch freqs when count is 0.
27828 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27830 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
27833 2017-06-10 Tom de Vries <tom@codesourcery.com>
27835 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27836 attributes): Document signal effective target.
27838 2017-06-10 Tom de Vries <tom@codesourcery.com>
27840 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27841 Document effective target stack_size.
27843 2017-06-09 David Malcolm <dmalcolm@redhat.com>
27845 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
27846 to the edit_context if they can be auto-applied.
27848 2017-06-9 Ian Lance Taylor <iant@golang.org>
27850 * opts.c (finish_options): If -fsplit-stack, disable implicit
27851 -forder-blocks-and-partition.
27852 * doc/invoke.texi (Optimize Options): Document that when using
27853 -fsplit-stack -forder-blocks-and-partition is not implicitly
27856 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27858 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
27859 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
27860 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
27861 * builtins.def (abort, trap, unreachable): Declare cold.
27862 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
27863 * tree-core.h (ECF_COLD): New.
27864 * tree.c (set_call_expr_flags): Handle ECF_COLD.
27865 (build_common_builtin_nodes): Mark unreachable and abort as cold.
27867 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27869 * predict.c (unlikely_executed_stmt_p): Cleanup.
27871 2017-06-09 Richard Biener <rguenther@suse.de>
27873 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
27874 model if the ref is always written to.
27876 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27878 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
27880 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27882 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
27885 2017-06-09 Tom de Vries <tom@codesourcery.com>
27888 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
27889 "target cannot support label values" when encountering LABEL_REF.
27891 2017-06-09 Martin Liska <mliska@suse.cz>
27893 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
27894 (gimple_gen_ic_func_profiler): Emit direct comparison
27895 of __gcov_indirect_call_callee with NULL.
27896 (gimple_gen_time_profiler): Change probability from
27897 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
27899 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27901 * profile.c (edge_gcov_counts): Turn to pointer.
27902 (compute_branch_probabilities, compute_branch_probabilities): Update.
27903 (branch_prob): Do not clear edge_gcov_count.
27904 * profile.h (edge_gcov_counts): Turn to pointer.
27905 (edge_gcov_count): Update.
27907 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27909 * gimple.h (gimple_check_failed): Mark cold.
27911 2017-06-09 Richard Biener <rguenther@suse.de>
27913 PR tree-optimization/66623
27914 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
27915 refactor check_reduction into two parts, properly computing
27916 whether we have to check reduction validity for outer loop
27919 2017-06-09 Richard Biener <rguenther@suse.de>
27921 PR tree-optimization/79483
27922 * graphite-scop-detection.c (order): New global.
27923 (get_order): Compute bb to order mapping that satisfies code
27924 generation constraints.
27925 (cmp_pbbs): New helper.
27926 (build_scops): Start domwalk at entry block, sort generated
27929 2017-06-09 Richard Biener <rguenther@suse.de>
27931 PR middle-end/81007
27932 * ipa-polymorphic-call.c
27933 (ipa_polymorphic_call_context::restrict_to_inner_class):
27934 Skip FIELD_DECLs with error_mark_node type.
27935 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
27938 2017-06-09 Martin Liska <mliska@suse.cz>
27940 * predict.c (struct branch_predictor): New struct.
27941 (test_prediction_value_range): New test.
27942 (predict_c_tests): New function.
27943 * selftest-run-tests.c (selftest::run_tests): Run the function.
27944 * selftest.h: Declare new tests.
27946 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
27949 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
27950 gen_add3_insn did not fail.
27951 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
27952 r0, construct that number in a temporary reg and add that reg to r0.
27953 If asked to put the result in r0 as well, fail.
27955 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
27957 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27958 for early expansion of vec_eqv.
27960 2017-06-08 Jakub Jelinek <jakub@redhat.com>
27962 PR middle-end/81005
27963 * ubsan.c (instrument_null): Avoid pointless code temporary.
27964 (pass_ubsan::execute): Instrument aggregate arguments of calls.
27966 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
27970 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
27973 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
27976 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27978 * predict.c (unlikely_executed_edge_p): Move ahead.
27979 (probably_never_executed_edge_p): Use it.
27981 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27983 PR middle-end/79988
27984 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
27985 gimple_call_builtin_p call.
27987 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27989 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
27990 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
27991 rtl_check_failed_type2, rtl_check_failed_code1,
27992 rtl_check_failed_code2, rtl_check_failed_code_mode,
27993 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
27994 rtvec_check_failed_bounds, rtl_check_failed_flag,
27995 _fatal_insn_not_found, _fatal_insn): Likewise.
27996 * tree.h (tree_contains_struct_check_failed,
27997 tree_check_failed, tree_not_check_failed,
27998 tree_class_check_failed, tree_range_check_failed,
27999 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28000 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28001 tree_operand_check_failed, omp_clause_check_failed,
28002 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28005 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28007 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28008 flag_branch_probabilities.
28009 * ipa-inline.c (edge_badness): Likewise.
28010 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28011 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28012 * predict.c (maybe_hot_frequency_p): Likewise.
28013 (probably_never_executed): Likewise.
28014 * sched-ebb.c (schedule_ebbs): Likewise.
28015 * sched-rgn.c (find_single_block_region): Likewise.
28016 * tracer.c (tail_duplicate): Likewise.
28018 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28020 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28021 longer requires x_flag_profile_use.
28023 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28025 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28026 instead of flag_reorder_blocks_and_partition.
28027 * dbxout.c (dbxout_function_end): Likewise.
28028 * dwarf2out.c (gen_subprogram_die): Likewise.
28029 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28030 * hw-doloop.c (reorg_loops): Likewise.
28031 * varasm.c (assemble_start_function,
28032 assemble_end_function): Likewise.
28033 (decide_function_section): Do not check for
28034 flag_reorder_blocks_and_partition.
28036 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28038 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28040 (chkp_get_hard_register_fake_addr_expr): Ditto.
28041 (chkp_build_addr_expr): Add check for hard reg case.
28042 (chkp_parse_array_and_component_ref): Ditto.
28043 (chkp_find_bounds_1): Ditto.
28044 (chkp_process_stmt): Don't generate bounds store for
28047 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28049 * predict.c (maybe_hot_bb_p): Do not check profile status.
28050 (maybe_hot_edge_p): Likewise.
28051 (probably_never_executed): Check for zero counts even if profile
28053 (unlikely_executed_edge_p): New function.
28054 (unlikely_executed_stmt_p): New function.
28055 (unlikely_executed_bb_p): New function.
28056 (set_even_probabilities): Use unlikely predicates.
28057 (combine_predictions_for_bb): Likewise.
28058 (predict_paths_for_bb): Likewise.
28059 (predict_paths_leading_to_edge): Likewise.
28060 (determine_unlikely_bbs): New function.
28061 (estimate_bb_frequencies): Use it.
28062 (compute_function_frequency): Use zero counts even if profile is
28064 * profile-count.h: Fix typo.
28066 2017-08-08 Julia Koval <julia.koval@intel.com>
28068 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
28069 _mm512_mask_cvtsepi16_storeu_epi8,
28070 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
28071 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
28072 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
28073 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
28074 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
28075 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
28076 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
28077 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
28078 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
28079 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
28080 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
28081 __builtin_ia32_pmovuswb256mem_mask,
28082 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
28083 __builtin_ia32_pmovwb512mem_mask): New builtins.
28085 2017-08-08 Julia Koval <julia.koval@intel.com>
28087 PR target/73350,80862
28088 * config/i386/subst.md (round): Fix round pattern.
28089 * config/i386/i386.c (ix86_erase_embedded_rounding):
28090 Fix erasing rounding for the fixed pattern.
28092 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28094 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
28096 2017-06-08 Martin Liska <mliska@suse.cz>
28098 PR gcov-profile/80911
28099 * gcov.c (block_info::block_info): New constructor.
28101 2017-06-07 Carl Love <cel@us.ibm.com>
28103 * config/rs6000/rs6000-c: The return type of the following
28104 built-in functions was implemented as int not long long. Fix sign
28105 of return value for the unsigned version of vec_mulo and vec_mule.
28106 vector unsigned long long vec_bperm (vector unsigned long long,
28107 vector unsigned char)
28108 vector signed long long vec_mule (vector signed int,
28110 vector unsigned long long vec_mule (vector unsigned int,
28111 vector unsigned int)
28112 vector signed long long vec_mulo (vector signed int,
28114 vector unsigned long long vec_mulo (vector unsigned int,
28115 vector unsigned int)
28116 * doc/extend.texi: Fix the documentation for the built-in
28119 2017-06-07 Carl Love <cel@us.ibm.com>
28122 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
28125 2017-06-07 Carl Love <cel@us.ibm.com>
28127 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
28128 support, Generate doublehv for signed int/float for BE case only.
28130 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
28132 * doc/invoke.texi (mcx16): Rewrite.
28134 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28136 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
28137 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
28138 *mov<mode>_softfloat, and an anonymous splitter): Use
28139 nonimmediate_operand instead of rs6000_nonimmediate_operand.
28141 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28143 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
28145 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28146 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
28147 (rs6000_debug_reg_global): Adjust.
28148 (rs6000_init_hard_regno_mode_ok): Adjust.
28149 (rs6000_dbx_register_number): Adjust.
28150 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
28151 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28152 Remove SPE_ACC and SPEFSCR.
28153 (REG_ALLOC_ORDER): Ditto.
28154 (FRAME_POINTER_REGNUM): Change to 111.
28155 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
28156 (REG_CLASS_NAMES): Ditto.
28157 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
28158 (REGISTER_NAMES): Ditto.
28159 (ADDITIONAL_REG_NAMES): Ditto.
28160 (rs6000_reg_names): Ditto.
28161 * config/rs6000/rs6000.md: Renumber some register number
28164 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28166 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
28168 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28169 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
28171 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
28172 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28173 Delete the SPE high registers.
28174 (REG_ALLOC_ORDER): Ditto.
28175 (enum reg_class): Remove SPE_HIGH_REGS.
28176 (REG_CLASS_NAMES): Ditto.
28177 (REG_CLASS_CONTENTS): Delete the SPE high registers.
28178 (REGISTER_NAMES): Ditto.
28179 (rs6000_reg_names): Ditto.
28180 * doc/tm.texi.in: Remove SPE as example.
28181 * doc/tm.texi: Regenerate.
28183 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28185 * config/rs6000/8540.md (ppc8540_brinc): Delete.
28186 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
28187 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
28188 * config/rs6000/rs6000.md (type): Remove "brinc".
28190 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28192 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
28193 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
28194 * config/rs6000/linuxspe.h: Delete file.
28195 * config/rs6000/rs6000.md: Don't include spe.md.
28196 * config/rs6000/spe.h: Delete file.
28197 * config/rs6000/spe.md: Delete file.
28198 * config/rs6000/t-rs6000: Remove spe.md.
28200 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28202 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
28203 (reg_or_none500mem_operand): Delete.
28204 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
28205 instead of reg_or_none500mem_operand.
28207 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28209 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28210 handling of SPE flags.
28211 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
28213 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28215 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
28217 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
28218 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
28219 paired_divv2sf3): Similar.
28220 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
28221 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
28222 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
28224 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
28225 Rename the paired_* instruction patterns.
28226 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
28228 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
28229 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
28230 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
28231 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
28232 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
28233 PAIRED_VECTOR_MODE.
28234 (struct machine_function): Delete field spe_insn_chain_scanned_p.
28235 (spe_func_has_64bit_regs_p): Delete.
28236 (spe_expand_predicate_builtin): Delete.
28237 (spe_expand_evsel_builtin): Delete.
28238 (TARGET_DWARF_REGISTER_SPAN): Do not define.
28239 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
28240 (invalid_e500_subreg): Delete.
28241 (rs6000_legitimize_address): Always force_reg op2 as well, for
28242 paired single memory accesses.
28243 (rs6000_member_type_forces_blk): Delete.
28244 (rs6000_spe_function_arg): Delete.
28245 (rs6000_expand_unop_builtin): Delete SPE handling.
28246 (rs6000_expand_binop_builtin): Ditto.
28247 (spe_expand_stv_builtin): Delete.
28248 (bdesc_2arg_spe): Delete.
28249 (spe_expand_builtin): Delete.
28250 (spe_expand_predicate_builtin): Delete.
28251 (spe_expand_evsel_builtin): Delete.
28252 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
28253 (spe_init_builtins): Delete.
28254 (spe_func_has_64bit_regs_p): Delete.
28255 (savres_routine_name): Delete "info" parameter. Adjust callers.
28256 (rs6000_emit_stack_reset): Ditto.
28257 (rs6000_dwarf_register_span): Delete.
28258 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
28259 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
28260 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
28262 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
28264 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
28265 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
28266 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
28267 mulv2sf3, divv2sf3): Delete expanders.
28269 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28271 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
28273 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28275 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
28276 * config/rs6000/rs6000.c: Ditto.
28278 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28280 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
28281 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
28282 comparison_operator.
28284 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28286 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
28287 * config/rs6000/rs6000.opt: Ditto.
28288 * config/rs6000/t-rtems: Ditto.
28290 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28292 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
28293 TARGET_E500_SINGLE by 0, simplify.
28294 * config/rs6000/rs6000.c: Ditto.
28295 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
28296 (spe_build_register_parallel): Delete.
28297 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
28298 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
28299 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
28300 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
28301 (E500_CONVERT): Delete.
28302 * config/rs6000/spe.md: Remove many patterns and all define_constants.
28304 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28306 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
28307 * config/rs6000/dfp.md: Ditto.
28308 (negdd2, *negdd2_fpr): Merge.
28309 (absdd2, *absdd2_fpr): Merge.
28310 (negtd2, *negtd2_fpr): Merge.
28311 (abstd2, *abstd2_fpr): Merge.
28312 * config/rs6000/e500.h: Delete file.
28313 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
28314 TARGET_FPRS by 1 and simplify.
28315 * config/rs6000/rs6000-c.c: Ditto.
28316 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
28317 TARGET_DF_SPE by 0.
28318 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
28320 * config/rs6000/rs6000.md: Ditto.
28321 (floatdidf2, *floatdidf2_fpr): Merge.
28322 (move_from_CR_gt_bit): Delete.
28323 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
28324 (E500_CR_IOR_COMPARE): Delete.
28325 (All patterns that require !TARGET_FPRS): Delete.
28326 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
28328 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28330 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
28332 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28334 * graphds.c (add_edge): Intitialize edge's attached data.
28335 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
28336 pointer parameter. Call pointed function on each edge during
28337 graph traversing. Skip traversing the edge when the function
28339 (graphds_dfs, graphds_scc): Ditto.
28340 (for_each_edge): New parameter. Pass the new parameter to callback
28342 * graphds.h (skip_edge_callback): New function pointer type.
28343 (graphds_dfs, graphds_scc): New function pointer parameter.
28344 (graphds_edge_callback, for_each_edge): New parameter.
28346 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28348 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
28349 out code checking if runtime alias check is possible to below ...
28350 Call the new function.
28351 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
28352 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
28354 2017-06-07 Marek Polacek <polacek@redhat.com>
28357 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
28358 TYPE_OVERFLOW_WRAPS check.
28360 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28362 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
28363 if versioning is required.
28364 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
28365 peeling with the check for versioning.
28367 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28369 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
28370 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
28371 Set true to new parameter if new ssa variable is defined.
28372 (vect_gen_vector_loop_niters): Refactor. Set range information
28373 for the new vector loop bound variable.
28374 (vect_do_peeling): Ditto.
28376 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28378 * tree-affine.c (ssa.h): Include header file.
28379 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
28380 has wrapping overflow behavior.
28382 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28384 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
28386 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28388 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
28389 (tree_to_aff_combination): ... here.
28391 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28393 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
28394 reg_pressure model function.
28395 (ivopts_global_cost_for_size): Delete.
28396 (determine_set_costs, iv_ca_recount_cost): Call new model function
28397 ivopts_estimate_reg_pressure.
28399 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28401 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
28402 expensive than udiv. Remove floating point cases from mod.
28404 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28406 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
28407 Increase idiv cost.
28409 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28411 * config/aarch64/aarch64.md
28412 (copysignsf3): Fix mask generation.
28414 2017-06-07 Jakub Jelinek <jakub@redhat.com>
28416 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
28418 (class dump_manager): Add register_dumps method.
28419 * dumpfile.c: Include langhooks.h.
28420 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
28421 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
28422 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
28423 (dump_manager::dump_register): Start with 512 entries instead of 32.
28424 (dump_manager::register_dumps): New method.
28425 * toplev.c (general_init): Instead of invoking register_dumps
28426 langhook, invoke register_dumps method on the dump manager.
28427 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
28430 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
28432 * doc/md.texi: Clarify the restrictions on a define_insn condition.
28433 Say that # requires an associated define_split to exist, and that
28434 the define_split must be suitable for use after register allocation.
28436 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28438 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
28439 (compute_outgoing_frequencies): Also initialize zero counts.
28440 (find_many_sub_basic_blocks): Do not produce uninitialized profile
28441 around loops; preserve more of profile when nothing changes.
28443 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
28445 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
28447 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
28448 * config/arm/arm-cpu-cdata.h: Regenerate.
28449 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
28450 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
28451 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
28452 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
28453 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
28455 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
28457 * config/arm/t-rmprofile: Likewise.
28458 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
28460 2017-06-06 David S. Miller <davem@davemloft.net>
28463 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
28464 blockage if function uses alloca.
28466 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28468 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
28469 New "uid" fields to hold pretty-print IDs of group and ref.
28470 Memory references are now identified as <group_id>:<ref_id>
28471 instead of using [random] addresses.
28472 (dump_mem_details): Simplify, no functional change.
28473 (dump_mem_ref): Simplify and make output more concise.
28474 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
28475 (find_or_create_group): Initialize group uid.
28476 (record_ref): Initialize ref uid. Improve debug output.
28477 (prune_group_by_reuse, should_issue_prefetch_p,)
28478 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
28479 (mark_nontemporal_store, determine_loop_nest_reuse):
28480 Improve debug output.
28482 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28484 * dbgcnt.def (prefetch): New debug counter.
28485 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
28486 (schedule_prefetches): Stop issueing prefetches if debug counter
28489 2017-06-06 Tom de Vries <tom@codesourcery.com>
28491 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
28492 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
28494 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28496 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
28497 Use aarch64_reg_or_zero predicate for operand 4.
28498 (aarch64_compare_and_swap<mode> define_insn_and_split):
28499 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
28500 (aarch64_store_exclusive<mode>): Likewise for operand 2.
28502 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
28504 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
28505 (arm_compute_save_core_reg_mask): This.
28506 (thumb1_compute_save_reg_mask): Rename into ...
28507 (thumb1_compute_save_core_reg_mask): This.
28508 (arm_compute_save_reg0_reg12_mask): Adapt comment.
28509 (arm_compute_frame_layout): Likewise.
28511 2017-06-06 Richard Biener <rguenther@suse.de>
28513 PR tree-optimization/80974
28514 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
28515 keep or clear leaders SSA info.
28517 2017-06-06 Tom de Vries <tom@codesourcery.com>
28519 * config/nvptx/nvptx.c (split_mode_p): New function.
28520 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
28522 2017-06-06 Tom de Vries <tom@codesourcery.com>
28524 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
28526 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28529 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
28532 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28534 * shrink-wrap.c (handle_simple_exit): Update profile.
28535 (try_shrink_wrapping): Upate profile.
28537 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28539 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
28540 (tree_guess_outgoing_edge_probabilities): New.
28541 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
28542 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
28544 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28546 * ipa-split.c (split_function): Initialize return bb profile.
28548 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28550 * profile.c (compute_branch_probabilities): Also initialize
28551 EXIT_BLOCK profile.
28553 2017-06-06 Richard Biener <rguenther@suse.de>
28555 PR tree-optimization/80928
28556 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
28557 (vect_analyze_loop_operations): Properly guard analysis for
28559 (vect_transform_loop): Likewise.
28560 (vect_analyze_loop_2): Also reset SLP type on PHIs.
28561 (vect_model_induction_cost): Do not cost for pure SLP.
28562 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
28563 of induction in inner loop vectorization.
28564 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
28565 (vect_get_and_check_slp_defs): Handle vect_induction_def.
28566 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
28568 (vect_analyze_slp_cost_1): Cost induction.
28569 (vect_detect_hybrid_slp_stmts): Handle PHIs.
28570 (vect_get_slp_vect_defs): Likewise.
28571 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
28572 (vect_transform_stmt): Handle SLP reductions.
28573 * tree-vectorizer.h (vectorizable_induction): Adjust.
28575 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28577 * config/rs6000/rs6000.c (make_resolver_func): Update
28578 init_lowered_empty_function call.
28580 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28582 * doc/invoke.texi: Document the -fprofile-abs-path option.
28583 * common.opt (fprofile-abs-path): New option.
28584 * gcov-io.h (gcov_write_filename): Declare.
28585 * gcov-io.c (gcov_write_filename): New function.
28586 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28587 * profile.c (output_location): Likewise.
28589 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28591 * shring-wrap.c: Revert accidental commit.
28593 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28595 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28597 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28599 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28601 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28602 profile in callgraph edge.
28603 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28604 (apply_scale): Likewise.
28605 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28606 Also scale profile when inlining function with zero profile.
28607 (initialize_cfun): Update exit block profile even when it is zero.
28608 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28609 when profile is read.
28611 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28613 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28614 (CLONE_*): New constants to define the processors we can generate
28615 code for with the target_clone attribute.
28616 (rs6000_clone_map): New array to identify which clone processors
28617 the current program is running on.
28618 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28619 target_clone attribute.
28620 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28621 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28622 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28623 (cpu_expand_builtin): Add support for target_clone attribute.
28624 (rs6000_valid_attribute_p): Allow "default" attribute.
28625 (get_decl_name): New debug function to simplify printing the
28626 current function name in debugging statements.
28627 (rs6000_clone_priority): New functions to support the target_clone
28628 attribute, and be able to generate code to switch between ISA 2.05
28629 through ISA 3.0 (power6 through power9).
28630 (rs6000_compare_version_priority): Likewise.
28631 (rs6000_get_function_versions_dispatcher): Likewise.
28632 (make_resolver_func): Likewise.
28633 (add_condition_to_bb): Likewise.
28634 (dispatch_function_versions): Likewise.
28635 (rs6000_generate_version_dispatcher_body): Likewise.
28636 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28637 (fusion_gpr_load_p): Fix a spacing issue.
28638 * doc/extend.texi (Common Function Attributes): Document that the
28639 PowerPC supports the target_clone attribute.
28641 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28643 * config/arm/arm.h: explain F symbol found in description of ARM
28644 register allocation in its legend.
28646 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28648 * config/mips/frame-header-opt.c: Include profile-count.h.
28649 * config/riscv/riscv.c: Include profile-count.h
28651 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28653 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28655 (sm_set_flag_if_changed): Add bbs field.
28656 (execute_sm_if_changed_flag_set): Pass BBS.
28657 (execute_sm): Update.
28659 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28661 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28664 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28666 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28667 (peephole2): New peephole2 to emit the above.
28668 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28670 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28672 * config/aarch64/aarch64.c (define_peephole2 above
28673 *sub_<shift>_<mode>): New peephole.
28675 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28677 * config/i386/i386.c (make_resolver_func): Update.
28678 * Makefile.in: Add profile-count.h and profile-count.o
28679 * auto-profile.c (afdo_indirect_call): Update to new API.
28680 (afdo_set_bb_count): Update.
28681 (afdo_propagate_edge): Update.
28682 (afdo_propagate_circuit): Update.
28683 (afdo_calculate_branch_prob): Update.
28684 (afdo_annotate_cfg): Update.
28685 * basic-block.h: Include profile-count.h
28686 (struct edge_def): Turn count to profile_count.
28687 (struct basic_block_def): Likewie.
28688 (REG_BR_PROB_BASE): Move to profile-count.h
28689 (RDIV): Move to profile-count.h
28690 * bb-reorder.c (max_entry_count): Turn to profile_count.
28691 (find_traces): Update.
28692 (rotate_loop):Update.
28693 (connect_traces):Update.
28694 (sanitize_hot_paths):Update.
28695 * bt-load.c (migrate_btr_defs): Update.
28696 * cfg.c (RDIV): Remove.
28697 (init_flow): Use alloc_block.
28698 (alloc_block): Uninitialize count.
28699 (unchecked_make_edge): Uninitialize count.
28700 (check_bb_profile): Update.
28701 (dump_edge_info): Update.
28702 (dump_bb_info): Update.
28703 (update_bb_profile_for_threading): Update.
28704 (scale_bbs_frequencies_int): Update.
28705 (scale_bbs_frequencies_gcov_type): Update.
28706 (scale_bbs_frequencies_profile_count): New.
28707 * cfg.h (update_bb_profile_for_threading): Update.
28708 (scale_bbs_frequencies_profile_count): Declare.
28709 * cfgbuild.c (compute_outgoing_frequencies): Update.
28710 (find_many_sub_basic_blocks): Update.
28711 * cfgcleanup.c (try_forward_edges): Update.
28712 (try_crossjump_to_edge): Update.
28713 * cfgexpand.c (expand_gimple_tailcall): Update.
28714 (construct_exit_block): Update.
28715 * cfghooks.c (verify_flow_info): Update.
28716 (dump_bb_for_graph): Update.
28717 (split_edge): Update.
28718 (make_forwarder_block): Update.
28719 (duplicate_block): Update.
28720 (account_profile_record): Update.
28721 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28722 (get_estimated_loop_iterations): Update.
28723 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28724 (single_likely_exit): Update.
28725 * cfgloopmanip.c (scale_loop_profile): Update.
28727 (set_zero_probability): Update.
28728 (lv_adjust_loop_entry_edge): Update.
28729 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28730 (purge_dead_edges): Update.
28731 (rtl_account_profile_record): Update.
28732 * cgraph.c (cgraph_node::create): Uninitialize count.
28733 (symbol_table::create_edge): Uninitialize count.
28734 (cgraph_update_edges_for_call_stmt_node): Update.
28735 (cgraph_edge::dump_edge_flags): Update.
28736 (cgraph_node::dump): Update.
28737 (cgraph_edge::maybe_hot_p): Update.
28738 * cgraph.h: Include profile-count.h
28739 (create_clone), create_edge, create_indirect_edge): Update.
28740 (cgraph_node): Turn count to profile_count.
28741 (cgraph_edge0: Likewise.
28742 (make_speculative, clone): Update.
28743 (create_edge): Update.
28744 (init_lowered_empty_function): Update.
28745 * cgraphclones.c (cgraph_edge::clone): Update.
28746 (duplicate_thunk_for_node): Update.
28747 (cgraph_node::create_clone): Update.
28748 * cgraphunit.c (cgraph_node::analyze): Update.
28749 (cgraph_node::expand_thunk): Update.
28750 * final.c (dump_basic_block_info): Update.
28751 * gimple-streamer-in.c (input_bb): Update.
28752 * gimple-streamer-out.c (output_bb): Update.
28753 * graphite.c (print_global_statistics): Update.
28754 (print_graphite_scop_statistics): Update.
28755 * hsa-brig.c: Include basic-block.h.
28756 * hsa-dump.c: Include basic-block.h.
28757 * hsa-gen.c (T sum_slice): Update.
28758 (convert_switch_statements):Update.
28759 * hsa-regalloc.c: Include basic-block.h.
28760 * ipa-chkp.c (chkp_produce_thunks): Update.
28761 * ipa-cp.c (struct caller_statistics): Update.
28762 (init_caller_stats): Update.
28763 (gather_caller_stats): Update.
28764 (ipcp_cloning_candidate_p): Update.
28765 (good_cloning_opportunity_p): Update.
28766 (get_info_about_necessary_edges): Update.
28767 (dump_profile_updates): Update.
28768 (update_profiling_info): Update.
28769 (update_specialized_profile): Update.
28770 (perhaps_add_new_callers): Update.
28771 (decide_about_value): Update.
28772 (ipa_cp_c_finalize): Update.
28773 * ipa-devirt.c (struct odr_type_warn_count): Update.
28774 (struct decl_warn_count): Update.
28775 (struct final_warning_record): Update.
28776 (possible_polymorphic_call_targets): Update.
28777 (ipa_devirt): Update.
28778 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28779 * ipa-icf.c (sem_function::merge): Update.
28780 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28781 * ipa-inline.c (compute_uninlined_call_time): Update.
28782 (compute_inlined_call_time): Update.
28783 (want_inline_small_function_p): Update.
28784 (want_inline_self_recursive_call_p): Update.
28785 (edge_badness): Update.
28786 (lookup_recursive_calls): Update.
28787 (recursive_inlining): Update.
28788 (inline_small_functions): Update.
28789 (dump_overall_stats): Update.
28790 (dump_inline_stats): Update.
28791 * ipa-profile.c (ipa_profile_generate_summary): Update.
28792 (ipa_propagate_frequency): Update.
28793 (ipa_profile): Update.
28794 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28795 * ipa-utils.c (ipa_merge_profiles): Update.
28796 * loop-doloop.c (doloop_modify): Update.
28797 * loop-unroll.c (report_unroll): Update.
28798 (unroll_loop_runtime_iterations): Update.
28799 * lto-cgraph.c (lto_output_edge): Update.
28800 (lto_output_node): Update.
28801 (input_node): Update.
28802 (input_edge): Update.
28803 (merge_profile_summaries): Update.
28804 * lto-streamer-in.c (input_cfg): Update.
28805 * lto-streamer-out.c (output_cfg): Update.
28806 * mcf.c (create_fixup_graph): Update.
28807 (adjust_cfg_counts): Update.
28808 (sum_edge_counts): Update.
28809 * modulo-sched.c (sms_schedule): Update.
28810 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28811 * predict.c (maybe_hot_count_p): Update.
28812 (probably_never_executed): Update.
28813 (dump_prediction): Update.
28814 (combine_predictions_for_bb): Update.
28815 (propagate_freq): Update.
28816 (handle_missing_profiles): Update.
28817 (counts_to_freqs): Update.
28818 (rebuild_frequencies): Update.
28819 (force_edge_cold): Update.
28820 * predict.h: Include profile-count.h
28821 (maybe_hot_count_p, counts_to_freqs): UPdate.
28822 * print-rtl-function.c: Do not include cfg.h
28823 * print-rtl.c: Include basic-block.h
28824 * profile-count.c: New file.
28825 * profile-count.h: New file.
28826 * profile.c (is_edge_inconsistent): Update.
28827 (correct_negative_edge_counts): Update.
28828 (is_inconsistent): Update.
28829 (set_bb_counts): Update.
28830 (read_profile_edge_counts): Update.
28831 (compute_frequency_overlap): Update.
28832 (compute_branch_probabilities): Update; Initialize and deinitialize
28834 (branch_prob): Update.
28835 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
28836 (edge_gcov_count): New.
28837 (bb_gcov_count): New.
28838 * shrink-wrap.c (try_shrink_wrapping): Update.
28839 * tracer.c (better_p): Update.
28840 * trans-mem.c (expand_transaction): Update.
28841 (ipa_tm_insert_irr_call): Update.
28842 (ipa_tm_insert_gettmclone_call): Update.
28843 * tree-call-cdce.c: Update.
28844 * tree-cfg.c (gimple_duplicate_sese_region): Update.
28845 (gimple_duplicate_sese_tail): Update.
28846 (gimple_account_profile_record): Update.
28847 (execute_fixup_cfg): Update.
28848 * tree-inline.c (copy_bb): Update.
28849 (copy_edges_for_bb): Update.
28850 (initialize_cfun): Update.
28851 (freqs_to_counts): Update.
28852 (copy_cfg_body): Update.
28853 (expand_call_inline): Update.
28854 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
28855 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
28856 (try_unroll_loop_completely): Update.
28857 (try_peel_loop): Update.
28858 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
28859 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
28860 * tree-ssa-loop-split.c (connect_loops): Update.
28861 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
28862 * tree-ssa-reassoc.c (branch_fixup): Update.
28863 * tree-ssa-tail-merge.c (replace_block_by): Update.
28864 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
28865 (compute_path_counts): Update.
28866 (update_profile): Update.
28867 (recompute_probabilities): Update.
28868 (update_joiner_offpath_counts): Update.
28869 (estimated_freqs_path): Update.
28870 (freqs_to_counts_path): Update.
28871 (clear_counts_path): Update.
28872 (ssa_fix_duplicate_block_edges): Update.
28873 (duplicate_thread_path): Update.
28874 * tree-switch-conversion.c (case_bit_test_cmp): Update.
28875 (struct switch_conv_info): Update.
28876 * tree-tailcall.c (decrease_profile): Update.
28877 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
28878 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
28879 * value-prof.c (check_counter): Update.
28880 (gimple_divmod_fixed_value): Update.
28881 (gimple_mod_pow2): Update.
28882 (gimple_mod_subtract): Update.
28883 (gimple_ic_transform): Update.
28884 (gimple_stringop_fixed_value): Update.
28885 * value-prof.h (gimple_ic): Update.
28887 2017-06-02 Carl Love <cel@us.ibm.com>
28889 * config/rs6000/rs6000-c: Add support for built-in functions
28890 vector double vec_doublee (vector signed int);
28891 vector double vec_doublee (vector unsigned int);
28892 vector double vec_doublee (vector float);
28893 vector double vec_doubleh (vector signed int);
28894 vector double vec_doubleh (vector unsigned int);
28895 vector double vec_doubleh (vector float);
28896 vector double vec_doublel (vector signed int);
28897 vector double vec_doublel (vector unsigned int);
28898 vector double vec_doublel (vector float);
28899 vector double vec_doubleo (vector signed int);
28900 vector double vec_doubleo (vector unsigned int);
28901 vector double vec_doubleo (vector float);.
28902 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
28903 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
28905 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
28906 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
28907 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
28909 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
28910 vec_doublel, vec_doubleh.
28911 * doc/extend.texi: Update the built-in documentation file for the
28912 new built-in functions.
28914 2017-06-02 David Malcolm <dmalcolm@redhat.com>
28917 * ipa-inline-analysis.c (free_growth_caches): Set
28918 edge_removal_hook_holder to NULL after removing it.
28920 2017-06-02 Sudakshina Das <sudi.das@arm.com>
28922 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
28923 comparision with zero.
28925 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28926 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28927 for early expansion of vec_min and vec_max builtins.
28928 (builtin_function_type): Add min/max unsigned variants to those
28929 identified as having unsigned arguments.
28931 2017-06-02 Olivier Hainque <hainque@adacore.com>
28933 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
28935 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28937 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
28938 Use VALL_F16 iterator rather than VALL.
28940 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28942 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
28943 Emit CBNZ inside loop when doing a strong exchange and comparing
28944 against zero. Generate the CC flags after the loop.
28946 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
28948 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
28949 (dl_section_ref): New.
28950 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
28951 On AIX, append an expression to subtract the size of the
28952 section length to dl_section_ref.
28954 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28956 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28957 for early expansion of vector absolute builtins.
28959 2017-06-02 Richard Biener <rguenther@suse.de>
28961 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
28962 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
28964 2017-06-02 Richard Biener <rguenther@suse.de>
28966 PR tree-optimization/80948
28967 * tree-tailcall.c (find_tail_calls): Track stmts to move in
28968 stmt order as well.
28970 2017-06-02 Richard Biener <rguenther@suse.de>
28972 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
28974 * tree-vect-stmts.c (process_use): Do not mark backedge defs
28975 for inductions as relevant.
28977 2017-06-02 Richard Biener <rguenther@suse.de>
28979 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
28980 (vectorizable_induction): ... this. Remove dead code.
28982 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
28984 * builtins. (expand_builtin_alloca): Remove second parameter and
28985 infer its value from the first parameter instead.
28986 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
28988 2017-06-02 Jakub Jelinek <jakub@redhat.com>
28990 PR rtl-optimization/80903
28991 * loop-doloop.c (add_test): Unshare sequence.
28993 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
28995 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
28997 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
28999 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29001 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29002 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29003 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29004 (xlogue_layout::get_stub_rtx): Make static.
29005 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29006 (xlogue_layout::compute_stub_managed_regs): Rename to...
29007 (xlogue_layout::count_stub_managed_regs): ...this.
29008 (xlogue_layout::is_stub_managed_reg): New function.
29009 (xlogue_layout::m_stub_names): Rename to...
29010 (xlogue_layout::s_stub_names): ...this, make static.
29011 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29012 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29013 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29014 xlogue_layout::s_stub_names): Instantiate statics.
29015 (stub_managed_regs): Remove.
29016 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29017 (disable_call_ms2sysv_xlogues): Rename to...
29018 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29019 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29021 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29022 change after reload_completed.
29023 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29025 (ix86_expand_prologue): Likewise.
29026 (ix86_expand_epilogue): Likewise.
29027 (ix86_expand_split_stack_prologue): Likewise.
29028 (ix86_compute_frame_layout): Remove frame parameter ...
29029 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29030 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29032 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29033 (ix86_frame): Move from here ...
29034 * config/i386/i386.h (ix86_frame): ... to here.
29035 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29036 complete ix86_frame data structure instead. Remove some_ld_name.
29038 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29040 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29041 symbols that hold a DECL_VALUE_EXPR.
29043 2017-06-01 Martin Jambor <mjambor@suse.cz>
29045 PR tree-optimization/80898
29046 * tree-sra.c (process_subtree_disqualification): Removed.
29047 (disqualify_candidate): Do not acll
29048 process_subtree_disqualification.
29049 (subtree_mark_written_and_enqueue): New function.
29050 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29051 RHS has been disqualified and re-queue LHS if necessary. Apart
29052 from that, ignore disqualified RHS.
29054 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29056 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29057 address fetch for z10 or later.
29059 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29061 * config/arc/arc.md (tst_movb): Add guard when splitting.
29063 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29065 * config/arc/arc.c (arc_can_eliminate): Test against
29066 arc_frame_pointer_needed.
29068 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29070 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
29071 to prevent store reordering.
29072 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
29073 (type): Add block type.
29074 (stack_tie): Define special instruction to be used in
29077 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29079 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
29080 constraint. It is not valid for the pattern.
29081 (noncommutative_binary_comparison): Likewise.
29083 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29085 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
29088 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29090 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
29091 be used by the reg-alloc.
29093 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29095 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
29096 reg-alloc when having mul64 or mul32x16 instructions.
29097 (mulsidi3): Likewise.
29098 (umulsidi3): Likewise.
29099 (mulsi32x16): New pattern.
29100 (mulsi64): Likewise.
29101 (mulsidi64): Likewise.
29102 (umulsidi64): Likewise.
29103 (MUL32x16_REG): Define.
29104 (mul64_600): Use MUL32x16_REG.
29105 (mac64_600): Likewise.
29106 (umul64_600): Likewise.
29107 (umac64_600): Likewise.
29109 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29111 * config/arc/arc.md (mulsi3_700): Make it commutative.
29113 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
29115 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
29117 (*sign_extendsidi2_insn): Likewise for movstosw.
29119 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29121 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
29122 the type of the input discriminant value. Convert the
29123 discriminant value of signedness vary.
29125 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
29127 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
29128 Add to -Wall section.
29130 2017-06-01 Richard Biener <rguenther@suse.de>
29132 PR middle-end/66313
29133 * fold-const.c (fold_plusminus_mult_expr): If the factored
29134 factor may be zero use a wrapping type for the inner operation.
29135 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
29136 and handle moved defs.
29137 (process_assignment): Properly guard the unary op case. Return a
29138 tri-state indicating that moving the stmt before the call may allow
29139 to continue. Pass through to_move.
29140 (find_tail_calls): Handle moving unrelated defs before
29143 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
29146 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
29147 splitter result in the canonical way.
29149 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
29151 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
29152 also for 32bit target. Update insn attributes.
29153 (zero-extendsidi2 splitter): Allow all registers for operand 1.
29155 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
29157 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
29158 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
29159 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
29160 (_mm_maskz_min_ss): New intrinsics.
29162 2017-05-31 Martin Liska <mliska@suse.cz>
29164 * tree-vect-loop.c (vect_create_epilog_for_reduction):
29165 Change comment style to one we normally use.
29166 (vectorizable_reduction): Likewise.
29167 (vectorizable_induction): Likewise.
29168 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
29169 (vectorizable_call): Likewise.
29170 (vectorizable_simd_clone_call): Likewise.
29171 (vectorizable_conversion): Likewise.
29172 (vectorizable_assignment): Likewise.
29173 (vectorizable_shift): Likewise.
29174 (vectorizable_operation): Likewise.
29175 (vectorizable_store): Likewise.
29176 (vectorizable_load): Likewise.
29177 * tree-vectorizer.h: Likewise.
29179 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
29181 * passes.c (emergency_dump_function): New.
29182 * tree-pass.h (emergency_dump_function): Declare.
29183 * plugin.c (plugins_internal_error_function): Remove.
29184 * plugin.h (plugins_internal_error_function): Remove declaration.
29185 * toplev.c (internal_error_function): New static function. Use it...
29186 (general_init): ...here.
29188 2017-05-31 Graham Markall <graham.markall@embecosm.com>
29190 * config/arc/arc.c (arc_print_operand): Handle constant operands.
29191 (arc_rtx_costs): Add costs for new patterns.
29192 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
29193 * config/arc/predicates.md: Add _1_2_3_operand predicate.
29195 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29197 * tree-ssa-strlen.c (get_next_strinfo): New function.
29198 (get_stridx_plus_constant): Use it.
29199 (zero_length_string): Likewise.
29200 (adjust_related_strinfos): Likewise.
29201 (adjust_last_stmt): Likewise.
29203 2017-05-31 Richard Biener <rguenther@suse.de>
29206 * config/i386/i386.c (ix86_expand_builtin): Remove assert
29207 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
29209 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29211 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
29212 loop_vinfo argument and use of dependence distance vectors.
29213 Check instead whether the two references differ only in their
29214 initial value and assume that they have the same alignment if the
29215 difference is a multiple of the vector alignment.
29216 (vect_analyze_data_refs_alignment): Update call accordingly.
29218 2017-05-31 Martin Liska <mliska@suse.cz>
29221 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
29223 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29225 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
29226 (create_intersect_range_checks): Move from ...
29227 * tree-data-ref.c (create_intersect_range_checks_index)
29228 (create_intersect_range_checks): ... to here.
29229 (create_runtime_alias_checks): New function factored from ...
29230 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
29231 here. Call above function.
29232 * tree-data-ref.h (create_runtime_alias_checks): New function.
29234 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29236 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
29237 segment length for dr_b and compute it in wide_int.
29239 2017-05-31 Richard Biener <rguenther@suse.de>
29241 PR tree-optimization/80906
29242 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
29243 and pass through iv_map.
29244 (copy_bb_and_scalar_dependences): Adjust.
29245 (translate_pending_phi_nodes): Likewise.
29246 (copy_loop_close_phi_args): Handle code-generating IVs instead
29249 2017-05-30 David Malcolm <dmalcolm@redhat.com>
29251 * diagnostic-color.c (color_dict): Add "type-diff".
29252 (parse_gcc_colors): Update comment.
29253 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
29254 -fdiagnostics-show-template-tree and -fno-elide-type.
29255 (GCC_COLORS): Add type-diff to example.
29257 (-fdiagnostics-show-template-tree): New.
29258 (-fno-elide-type): New.
29259 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
29260 the pp_format_decoder callback. Call any m_format_postprocessor's
29262 (pretty_printer::pretty_printer): Initialize
29263 m_format_postprocessor.
29264 (pretty_printer::~pretty_printer): Delete any
29265 m_format_postprocessor.
29266 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
29267 (class format_postprocessor): New class.
29268 (struct pretty_printer::format_decoder): Document the new parameters.
29269 (struct pretty_printer::m_format_postprocessor): New field.
29270 * tree-diagnostic.c (default_tree_printer): Update for new
29271 bool and const char ** params.
29272 * tree-diagnostic.h (default_tree_printer): Likewise.
29274 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
29276 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
29277 (lwa_operand): Delete rs6000_gen_cell_microcode test.
29278 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29279 rs6000_gen_cell_microcode code.
29280 (rs6000_final_prescan_insn): Delete.
29281 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
29282 "warn-cell-microcode" entries.
29283 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
29284 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
29285 throughout. Change cc_reg_not_micro_cr0_operand to
29286 cc_reg_not_cr0_operand throughout.
29287 (*extendhi<mode>2_noload): Delete.
29288 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
29289 (mwarn-cell-microcode): Delete.
29290 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
29291 -mgen-cell-microcode and -mwarn-cell-microcode.
29293 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
29296 * config/i386/constraints.md (Yd): New constraint.
29298 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
29299 and (?Yd, r) alternatives. Update insn attributes.
29300 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
29301 and (?*Yd, r) alternatives. Update insn attributes.
29302 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
29304 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
29306 * gimplify.c (gimplify_modify_expr): Don't create a
29307 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
29310 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
29312 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
29314 2017-05-30 Richard Biener <rguenther@suse.de>
29316 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
29317 and reduc_def fields.
29318 (STMT_VINFO_REDUC_TYPE): New define.
29319 (STMT_VINFO_REDUC_DEF): Likewise.
29320 (vect_force_simple_reduction): Adjust prototype.
29321 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
29322 (vect_is_simple_reduction): Remove check_reduction argument.
29323 (vect_force_simple_reduction): Adjust and set
29324 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29325 (vectorizable_reduction): Do not re-do reduction analysis
29326 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29327 * tree-parloops.c (gather_scalar_reductions): Adjust.
29329 2017-05-30 Richard Biener <rguenther@suse.de>
29331 PR middle-end/80901
29332 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
29335 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
29337 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
29338 Introduce unknown_misalignment parameter and remove vf.
29339 (vect_peeling_hash_get_lowest_cost):
29340 Pass unknown_misalignment parameter.
29341 (vect_enhance_data_refs_alignment):
29342 Fix unsupportable data ref treatment.
29344 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29346 * tree-vect-data-refs.c (vect_get_data_access_cost):
29347 Workaround for SLP handling.
29348 (vect_enhance_data_refs_alignment):
29349 Compute costs for doing no peeling at all, compare to the best
29350 peeling costs so far and avoid peeling if cheaper.
29352 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29354 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
29355 Return peeling info and set costs to zero for unlimited cost
29357 (vect_enhance_data_refs_alignment): Also inspect all datarefs
29358 with unknown misalignment. Compute and costs for unknown
29359 misalignment, compare them to the costs for known misalignment
29360 and choose the cheapest for peeling.
29362 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29364 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
29365 (vect_get_peeling_costs_all_drs): Create function.
29366 (vect_peeling_hash_get_lowest_cost):
29367 Use vect_get_peeling_costs_all_drs.
29368 (vect_peeling_supportable): Create function.
29369 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
29371 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29373 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
29374 DR_HAS_NEGATIVE_STEP.
29375 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
29376 (vect_enhance_data_refs_alignment): Use.
29377 (vect_duplicate_ssa_name_ptr_info): Use.
29378 * tree-vectorizer.h (dr_misalignment): Use.
29379 (known_alignment_for_access_p): Use.
29381 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
29384 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
29386 (has_section_name): New function.
29388 2017-05-30 Martin Liska <mliska@suse.cz>
29391 * auto-profile.c (get_function_decl_from_block): Fix
29394 2017-05-30 Richard Biener <rguenther@suse.de>
29396 PR middle-end/80876
29397 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
29399 2017-05-30 Martin Liska <mliska@suse.cz>
29401 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
29402 * dumpfile.h (struct dump_file_info): Remove ctors.
29404 2017-05-30 Martin Liska <mliska@suse.cz>
29406 * predict.def: Fix GNU coding style.
29408 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29410 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29411 Mark 'to' argument with ATTRIBUTE_UNUSED.
29413 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29415 * config/xtensa/xtensa.c (xtensa_emit_call): Use
29416 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
29417 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
29420 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
29422 * doc/install.texi (Options specification): Restore entry of
29423 --enable-sjlj-exceptions.
29425 2017-05-27 Michael Eager <eager@eagercon.com>
29428 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
29430 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
29432 * config/microblaze/microblaze.h
29433 (FIXED_REGISTERS): Update in macro.
29434 (CALL_USED_REGISTERS): Update in macro.
29436 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
29438 * doc/install.texi: Add links to macOS binary distributions.
29440 2017-05-27 Jakub Jelinek <jakub@redhat.com>
29444 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29446 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29448 2017-05-26 Martin Liska <mliska@suse.cz>
29450 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
29452 2017-05-26 Martin Liska <mliska@suse.cz>
29454 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
29455 always leading ';; '.
29456 (dump_bb_info): Likewise.
29457 (brief_dump_cfg): Likewise.
29458 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
29459 * dumpfile.c: Remove usage of TDF_VERBOSE.
29460 * dumpfile.h (enum dump_kind): Likewise.
29461 (dump_gimple_bb_header): Do not use TDF_COMMENT.
29462 * print-tree.c (debug_verbose): Remove.
29463 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
29464 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
29465 * tree-diagnostic.c (default_tree_printer): Replace
29466 TDF_DIAGNOSTIC with TDF_SLIM.
29468 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29470 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
29471 in parameter loop, rather than loop_vinfo.
29472 (create_intersect_range_checks): Ditto.
29473 (vect_create_cond_for_alias_checks): Update call to above functions.
29475 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29477 PR tree-optimization/80815
29478 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
29479 for merging runtime alias checks. Handle negative DR_STEPs.
29481 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29483 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
29485 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
29486 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
29487 out code pruning runtime alias checks.
29488 * tree-data-ref.c (prune_runtime_alias_test_list): New function
29489 factored out from above.
29490 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29492 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29494 (prune_runtime_alias_test_list): New decalaration.
29496 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29498 * tree-vect-data-refs.c (compare_tree): Rename and move ...
29499 * tree-data-ref.c (data_ref_compare_tree): ... to here.
29500 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
29501 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
29502 (operator==, comp_dr_with_seg_len_pair): Ditto.
29503 (vect_prune_runtime_alias_test_list): Ditto.
29505 2017-05-26 Martin Liska <mliska@suse.cz>
29508 * params.def: Bound partial-inlining-entry-probability param.
29510 2017-05-26 Marek Polacek <polacek@redhat.com>
29513 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
29516 2017-05-26 Richard Biener <rguenther@suse.de>
29518 PR tree-optimization/80842
29519 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
29522 2017-05-26 Richard Biener <rguenther@suse.de>
29524 PR tree-optimization/80844
29525 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
29527 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
29529 * doc/md.texi (Machine Constraints): Update x86 family
29530 machine constraints section to match 'config/i386/constraints.md'.
29532 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
29534 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
29536 2017-05-25 Nathan Sidwell <nathan@acm.org>
29538 * doc/invoke.texi (--enable-languages): Update documentation.
29540 2017-05-25 Martin Liska <mliska@suse.cz>
29542 * dumpfile.c: Add TDF_FOLDING.
29543 * dumpfile.h (enum dump_kind): Likewise.
29544 * genmatch.c (dt_simplify::gen_1): Use it.
29546 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29548 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
29550 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29552 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29553 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
29555 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29557 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
29558 * match.pd (X == C): Rewrite it here.
29559 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
29560 with_certain_nonzero_bits2): New predicates.
29561 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
29563 2017-05-24 Nathan Sidwell <nathan@acm.org>
29565 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
29568 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
29571 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29573 * config/powerpcspe: New port. Files are copied from the rs6000
29574 port, with "rs6000" in filenames replaced by "powerpcspe".
29576 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29578 PR rtl-optimization/80754
29579 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29581 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29583 * config/sparc/sparc.md (length): Return the correct value for -mflat
29584 sibcalls to match output_sibcall.
29586 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29590 * config/rs6000/rs6000.c (struct machine_function): Add new field
29592 (rs6000_get_separate_components): Init that field, use it.
29593 (rs6000_components_for_bb): Use the field.
29595 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29597 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29599 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29601 PR middle-end/80823
29602 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29604 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29607 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29608 address against address_operand predicate.
29609 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29611 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29613 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29614 parameters passed indirectly.
29616 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29618 * config/i386/i386.md (*movdi_internal): Remove SSE4
29619 alternative 18 (?r, *v). Update insn attributes.
29620 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29621 Update insn attributes.
29622 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29623 Update insn attributes.
29624 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29625 alternative 1 (r, v). Remove isa attribute.
29626 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29627 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29628 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29630 2017-05-23 Tom de Vries <tom@codesourcery.com>
29632 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29635 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29637 * cgraphunit.c (symbol_table::process_new_functions): Update.
29638 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29639 (inline_generate_summary): Rename to ...
29640 (ipa_fn_summary_generate): ... this one.
29641 (inline_read_summary): Rename to ...
29642 (ipa_fn_summary_read): ... this one.
29643 (inline_write_summary): Rename to ...
29644 (ipa_fn_summary_write): ... this one.
29645 (inline_free_summary): Rename to ...
29646 (ipa_free_fn_summary): ... this one.
29647 (pass_data_local_fn_summary, pass_local_fn_summary,
29648 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29649 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29650 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29651 make_pass_ipa_fn_summary): New.
29652 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29653 inline_write_summary, inline_free_summary): Remove.
29654 (ipa_free_fn_summary) : New.
29655 * ipa-inline.c (ipa_inline): Update.
29656 (pass_ipa_inline): Do not generate summaries.
29657 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29659 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29660 and add pass_ipa_fn_summary.
29661 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29663 (make_pass_inline_parameters): Remove.
29665 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29667 * omp-low.c (struct omp_context): Remove "default_kind" member.
29670 * omp-offload.c (execute_oacc_device_lower): Remove the
29671 parallelism dimensions function attributes for unparallelized
29672 OpenACC kernels constructs.
29674 2017-05-23 Martin Liska <mliska@suse.cz>
29676 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29678 (cgraph_edge::make_speculative): Likewise.
29679 (cgraph_edge::resolve_speculation): Likewise.
29680 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29681 (cgraph_node::dump): Likewise.
29682 * cgraph.h: Likewise.
29683 * cgraphunit.c (analyze_functions): Likewise.
29684 (symbol_table::compile): Likewise.
29685 * ipa-cp.c (print_all_lattices): Likewise.
29686 (determine_versionability): Likewise.
29687 (initialize_node_lattices): Likewise.
29688 (ipcp_verify_propagated_values): Likewise.
29689 (estimate_local_effects): Likewise.
29690 (update_profiling_info): Likewise.
29691 (create_specialized_node): Likewise.
29692 (perhaps_add_new_callers): Likewise.
29693 (decide_about_value): Likewise.
29694 (decide_whether_version_node): Likewise.
29695 (identify_dead_nodes): Likewise.
29696 (ipcp_store_bits_results): Likewise.
29697 * ipa-devirt.c (dump_targets): Likewise.
29698 (ipa_devirt): Likewise.
29699 * ipa-icf.c (sem_item::dump): Likewise.
29700 (sem_function::equals): Likewise.
29701 (sem_variable::equals): Likewise.
29702 (sem_item_optimizer::read_section): Likewise.
29703 (sem_item_optimizer::execute): Likewise.
29704 (congruence_class::dump): Likewise.
29705 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29706 (dump_inline_summary): Likewise.
29707 (estimate_node_size_and_time): Likewise.
29708 (inline_analyze_function): Likewise.
29709 * ipa-inline-transform.c (inline_call): Likewise.
29710 * ipa-inline.c (report_inline_failed_reason): Likewise.
29711 (want_early_inline_function_p): Likewise.
29712 (edge_badness): Likewise.
29713 (update_edge_key): Likewise.
29714 (inline_small_functions): Likewise.
29715 * ipa-profile.c (ipa_profile): Likewise.
29716 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29717 (ipa_make_edge_direct_to_target): Likewise.
29718 (remove_described_reference): Likewise.
29719 (ipa_impossible_devirt_target): Likewise.
29720 (propagate_controlled_uses): Likewise.
29721 (ipa_print_node_params): Likewise.
29722 (ipcp_transform_function): Likewise.
29723 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29724 (propagate_pure_const): Likewise.
29725 * ipa-reference.c (generate_summary): Likewise.
29726 (read_write_all_from_decl): Likewise.
29727 (propagate): Likewise.
29728 (ipa_reference_read_optimization_summary): Likewise.
29729 * ipa-utils.c (ipa_merge_profiles): Likewise.
29730 * ipa.c (walk_polymorphic_call_targets): Likewise.
29731 (symbol_table::remove_unreachable_nodes): Likewise.
29732 (ipa_single_use): Likewise.
29733 * passes.c (execute_todo): Likewise.
29734 * predict.c (drop_profile): Likewise.
29735 * symtab.c (symtab_node::get_dump_name): New function.
29736 (symtab_node::dump_name): Likewise.
29737 (symtab_node::dump_asm_name): Likewise.
29738 (symtab_node::dump_references): Likewise.
29739 (symtab_node::dump_referring): Likewise.
29740 (symtab_node::dump_base): Likewise.
29741 (symtab_node::debug_symtab): Likewise.
29742 * tree-sra.c (convert_callers_for_node): Likewise.
29743 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29744 * value-prof.c (init_node_map): Likewise.
29746 2017-05-23 Martin Liska <mliska@suse.cz>
29748 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29749 and symtab_node::debug_symtab to symbol_table::debug.
29750 * cgraphunit.c (analyze_functions): Use the renamed function.
29751 (symbol_table::compile): Likewise.
29752 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29753 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29754 * passes.c (execute_todo): Likewise.
29755 * symtab.c (symbol_table::dump): New function.
29756 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29758 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29760 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29761 that nonconst implies exec.
29763 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29765 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29766 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29767 (inline_edge_summary_vec): Turn into ...
29768 (ipa_call_summaries): ... this one.
29769 (redirect_to_unreachable, edge_set_predicate,
29770 evaluate_properties_for_edge, inline_summary_alloc,
29771 reset_ipa_call_summary, reset_inline_summary,
29772 inline_summary_t::duplicate): Update.
29773 (inline_edge_duplication_hook): Turn to ...
29774 (ipa_call_summary_t::duplicate): ... this one.
29775 (inline_edge_removal_hook): Turn to ...
29776 (ipa_call_summary_t::remove): ... this one.
29777 (dump_inline_edge_summary): Turn to ...
29778 (dump_ipa_call_summary): ... this one.
29779 (estimate_function_body_sizes): Update.
29780 (inline_update_callee_summaries): Update.
29781 (remap_edge_change_prob): Update.
29782 (remap_edge_summaries): Update.
29783 (inline_merge_summary): Update.
29784 (do_estimate_edge_time): Update.
29785 (inline_generate_summary): Update.
29786 (inline_read_section): Update.
29787 (inline_read_summary): Update.
29788 (inline_free_summary): Update.
29789 * ipa-inline.c (can_inline_edge_p): Update.
29790 (compute_inlined_call_time): Update.
29791 (want_inline_small_function_p): Update.
29792 (edge_badness): Update.
29793 (early_inliner): Update.
29794 * ipa-inline.h (inline_edge_summary): Turn to ...
29795 (ipa_call_summary): ... this one.
29796 (ipa_call_summary_t): New class.
29797 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29798 (ipa_call_summaries): New.
29799 (inline_edge_summary): Remove.
29800 (estimate_edge_growth): Update.
29801 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29802 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29803 * ipa-split.c (execute_split_functions): Update.
29804 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29806 2017-05-23 Tom de Vries <tom@codesourcery.com>
29808 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29809 attributes): Document rdrand effective target.
29811 2017-05-23 Tom de Vries <tom@codesourcery.com>
29813 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29814 attributes): Sort alphabetically.
29816 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29818 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29820 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29823 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
29824 V2DF/V2DI splat into two separate patterns, one that handles
29825 registers, and the other that only handles memory. Drop support
29826 for splatting from a GPR on ISA 2.07 and then splitting the
29827 splat into direct move and splat.
29828 (vsx_splat_<mode>_reg): Likewise.
29829 (vsx_splat_<mode>_mem): Likewise.
29831 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
29833 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
29835 2017-05-22 Jakub Jelinek <jakub@redhat.com>
29837 PR middle-end/80809
29838 * omp-low.c (finish_taskreg_remap): New function.
29839 (finish_taskreg_scan): If unit size of ctx->record_type
29840 is non-constant, unshare the size expression and replace
29841 decls in it with possible outer var refs.
29843 PR middle-end/80809
29844 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
29845 GOVD_SHARED rather than GOVD_PRIVATE with it.
29846 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
29847 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
29849 PR middle-end/80853
29850 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
29851 as last argument to build_outer_var_ref for pointer bases of array
29852 section reductions.
29854 2017-05-19 Martin Sebor <msebor@redhat.com>
29856 * print-tree.c (print_node): Print DECL_READ_P flag.
29858 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29860 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
29861 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
29862 * cgraph.c: Likewise.
29863 * cgraphunit.c: Likewise.
29864 * gengtype.c: Likewise.
29865 * ipa-cp.c: Likewise.
29866 * ipa-devirt.c: Likewise.
29867 * ipa-icf.c: Likewise.
29868 * ipa-predicate.c: Likewise.
29869 * ipa-profile.c: Likewise.
29870 * ipa-prop.c: Likewise.
29871 * ipa-split.c: Likewise.
29873 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
29874 edge_predicate_pool, dump_inline_hints,
29875 inline_summary::account_size_time, redirect_to_unreachable,
29876 edge_set_predicate, set_hint_predicate,
29877 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
29878 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
29879 inline_summary_t::remove, remap_hint_predicate_after_duplication,
29880 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
29881 ipa_call_summary_t::remove, initialize_growth_caches,
29882 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
29883 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
29884 mark_modified, unmodified_parm_1, unmodified_parm,
29885 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
29886 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
29887 compute_bb_predicates, will_be_nonconstant_expr_predicate,
29888 will_be_nonconstant_predicate, record_modified_bb_info,
29889 get_minimal_bb, record_modified, param_change_prob,
29890 phi_result_unknown_predicate, predicate_for_phi_result,
29891 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
29892 estimate_function_body_sizes, compute_inline_parameters,
29893 compute_inline_parameters_for_curren, pass_data_inline_parameters,
29894 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
29895 inline_update_callee_summaries, remap_edge_change_prob,
29896 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
29897 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
29898 inline_analyze_function, inline_summary_t::insert,
29899 inline_generate_summary, read_ipa_call_summary, inline_read_section,
29900 inline_read_summary, write_ipa_call_summary, inline_write_summary,
29901 inline_free_summary): Move to ipa-fnsummary.h
29902 (predicate_t): Remove.
29903 * ipa-fnsummary.c: New file.
29904 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
29905 (enum inline_hints_vals, inline_hints, agg_position_info,
29906 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
29907 inline_summaries, ipa_call_summary, ipa_call_summary_t,
29908 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
29909 dump_inline_summary, dump_inline_hints, inline_generate_summary,
29910 inline_read_summary, inline_write_summary, inline_free_summary,
29911 inline_analyze_function, initialize_inline_failed,
29912 inline_merge_summary, inline_update_overall_summary,
29913 compute_inline_parameters): Move to ipa-fnsummary.h
29914 * ipa-fnsummary.h: New file.
29915 * ipa-inline-transform.h: Include ipa-inline.h.
29916 * ipa-inline.c: LIkewise.
29918 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29920 * ipa-inline.c (edge_badness): Use inlined_time instead of
29921 inline_summaries->get.
29923 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29925 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
29927 2017-05-22 Nathan Sidwell <nathan@acm.org>
29929 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
29930 (fdump-lang): Document 'raw' option.
29931 * dumpfile.h (TDI_tu): Delete.
29932 * dumpfile.c (dump_files): Remove translation-unit.
29933 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
29935 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
29937 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
29938 command option from $(AWK) call.
29939 * config/avr/genmultilib.awk: Simplify and rewrite so that it
29940 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
29941 [FORMAT]: Remove handling of variable.
29942 * config/avr/t-multilib: Regenerate.
29944 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29946 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
29948 (dump_inline_summary): Do not print self_time.
29949 (estimate_function_body_sizes): Do not set self_time.
29950 (compute_inline_parameters): Likewise.
29951 (inline_read_section, inline_write_summary): Do not stream self_time.
29952 * ipa-inline.h (inline_summary): Drop self_time.
29954 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29956 * ipa-inline-analysis.c (account_size_time): Rename to ...
29957 (inline_summary::account_size_time): ... this one.
29958 (reset_ipa_call_summary): Turn to ...
29959 (ipa_call_summary::reset): ... this one.
29960 (reset_inline_summary): Turn to ...
29961 (inline_summary::reset): ... this one.
29962 (inline_summary_t::remove): Update.
29963 (inline_summary_t::duplicate): Update.
29964 (ipa_call_summary_t::remove): Update.
29965 (dump_inline_summary): Update.
29966 (estimate_function_body_sizes): Update.
29967 (compute_inline_parameters): Update.
29968 (estimate_node_size_and_time): Update.
29969 (inline_merge_summary): Update.
29970 (inline_update_overall_summary): Update.
29971 (inline_read_section): Update.
29972 (inline_write_summary): Update.
29973 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
29974 add account_size_time and reset member functions.
29975 (ipa_call_summary): Add reset function.
29976 * ipa-predicate.h (predicate::operator &): Constify.
29978 2017-05-22 Richard Biener <rguenther@suse.de>
29980 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
29982 2017-05-19 Jason Merrill <jason@redhat.com>
29984 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
29986 2017-05-19 Marek Polacek <polacek@redhat.com>
29989 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
29990 TYPE_OVERFLOW_WRAPS checks.
29992 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
29994 * tree-core.h (enum omp_clause_default_kind): Add
29995 "OMP_CLAUSE_DEFAULT_PRESENT".
29996 * tree-pretty-print.c (dump_omp_clause): Handle it.
29997 * gimplify.c (enum gimplify_omp_var_data): Add
29998 "GOVD_MAP_FORCE_PRESENT".
29999 (gimplify_adjust_omp_clauses_1): Map it to
30000 "GOMP_MAP_FORCE_PRESENT".
30001 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30003 * gimplify.c (oacc_default_clause): Clarify.
30005 2017-05-19 Nathan Sidwell <nathan@acm.org>
30007 LANG_HOOK_REGISTER_DUMPS
30008 * toplev.c (general_init): Call register dump lang hook.
30009 * doc/invoke.texi: Document -fdump-lang option family.
30010 * dumpfile.c (dump_files): Remove class dump here.
30011 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30012 * dumpfile.h (tree_dump_index): Remove TDI_class.
30013 * langhooks-def.h (lhd_register_dumps): Declare.
30014 (LANG_HOOKS_REGISTER_DUMPS): Define.
30015 (LANG_HOOKS_INITIALIZER): Add it.
30016 * langhooks.c (lhd_register_dumps): Define.
30017 * langhooks.h (struct lang_hooks): Add register_dumps.
30019 2017-05-19 Nathan Sidwell <nathan@acm.org>
30021 * context.h (context::set_passes): New.
30022 * context.c (context::context): Do not create pass manager.
30023 * toplev.c (general_init): Create pass manager here.
30025 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30027 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30028 use this splitter if two add or or instructions would also work for
30029 the constant we want to generate.
30031 2017-05-19 Richard Biener <rguenther@suse.de>
30034 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30035 predicate evaluation.
30037 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30039 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30041 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30044 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30046 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30047 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30048 (inline_edge_summary_vec): Turn into ...
30049 (ipa_call_summaries): ... this one.
30050 (redirect_to_unreachable, edge_set_predicate,
30051 evaluate_properties_for_edge, inline_summary_alloc,
30052 reset_ipa_call_summary, reset_inline_summary,
30053 inline_summary_t::duplicate): Update.
30054 (inline_edge_duplication_hook): Turn to ...
30055 (ipa_call_summary_t::duplicate): ... this one.
30056 (inline_edge_removal_hook): Turn to ...
30057 (ipa_call_summary_t::remove): ... this one.
30058 (dump_inline_edge_summary): Turn to ...
30059 (dump_ipa_call_summary): ... this one.
30060 (estimate_function_body_sizes): Update.
30061 (inline_update_callee_summaries): Update.
30062 (remap_edge_change_prob): Update.
30063 (remap_edge_summaries): Update.
30064 (inline_merge_summary): Update.
30065 (do_estimate_edge_time): Update.
30066 (inline_generate_summary): Update.
30067 (inline_read_section): Update.
30068 (inline_read_summary): Update.
30069 (inline_free_summary): Update.
30070 * ipa-inline.c (can_inline_edge_p): Update.
30071 (compute_inlined_call_time): Update.
30072 (want_inline_small_function_p): Update.
30073 (edge_badness): Update.
30074 (early_inliner): Update.
30075 * ipa-inline.h (inline_edge_summary): Turn to ...
30076 (ipa_call_summary): ... this one.
30077 (ipa_call_summary_t): New class.
30078 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30079 (ipa_call_summaries): New.
30080 (inline_edge_summary): Remove.
30081 (estimate_edge_growth): Update.
30082 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30083 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30084 * ipa-split.c (execute_split_functions): Update.
30085 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30087 2017-05-19 Richard Biener <rguenther@suse.de>
30089 PR middle-end/80764
30090 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
30092 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
30094 * config/rs6000/rs6000.c (struct machine_function): Add field
30095 fpr_is_wrapped_separately.
30096 (rs6000_get_separate_components): Use 64 components. Handle the
30097 new FPR components.
30098 (rs6000_components_for_bb): Handle the FPR components.
30099 (rs6000_emit_prologue_components): Handle the FPR components.
30100 (rs6000_emit_epilogue_components): Handle the FPR components.
30101 (rs6000_set_handled_components): Handle the FPR components.
30102 (rs6000_emit_prologue): Don't output prologue code for those FPRs
30103 that are already separately shrink-wrapped.
30104 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
30105 that are already separately shrink-wrapped.
30107 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30110 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
30113 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
30114 (define_peephole2 for Altivec d-form load): Add peepholes to catch
30115 cases where the register allocator uses a move and an offsettable
30116 memory operation to/from a FPR register on ISA 2.06/2.07.
30117 (define_peephole2 for Altivec d-form store): Likewise.
30119 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
30122 * config/i386/mmx.md (*mov<mode>_internal): Enable
30123 alternatives 11, 12, 13 and 14 also for 32bit targets.
30124 Remove alternatives 15, 16, 17 and 18.
30125 * config/i386/sse.md (vec_concatv2di): Change
30126 alternative (!x, *y) to (x, ?!*Yn).
30128 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
30130 * dumpfile.h (enum dump_kind): Remove stray comma.
30132 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30134 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
30135 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
30136 predicate::num_conditions
30137 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
30138 (CHANGED): turn into predicate::changed.
30139 (agg_position_info): Move to ipa-predicate.h
30140 (add_condition, predicate::add_clause, predicate::operator &=,
30141 predicate::or_with, predicate::evaluate, predicate::probability,
30142 dump_condition, dump_clause, predicate::dump,
30143 predicate::remap_after_duplication, predicate::remap_after_inlining,
30144 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
30145 (evaluate_conditions_for_known_args): Update.
30146 (set_cond_stmt_execution_predicate): Update.
30147 * ipa-inline.h: Include ipa-predicate.h
30148 (condition, inline_param_summary, conditions, agg_position_info,
30149 predicate): Move to ipa-predicate.h
30150 * ipa-predicate.c: New file.
30151 * ipa-predicate.h: New file.
30153 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
30155 * final.c (leaf_function_p): Check we are not in a sequence.
30157 2017-05-18 Martin Liska <mliska@suse.cz>
30159 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
30160 * dumpfile.c (dump_register): Use new enum dump_kind.
30161 (get_dump_file_name): Likewise.
30162 (dump_enable_all): Likewise.
30163 (dump_switch_p_1): Likewise.
30164 (enable_rtl_dump_file): Remove usage of TDF_RTL.
30165 * dumpfile.h (enum dump_kind): New enum type.
30166 (struct dump_file_info): Create constructor and
30167 format fields and comments.
30168 * passes.c (pass_manager::register_one_dump_file):
30170 * statistics.c (statistics_early_init): Likewise.
30171 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
30172 TDF_TREE with TDF_SLIM.
30173 (gather_memory_references_ref): Likewise.
30175 2017-05-18 Martin Liska <mliska@suse.cz>
30177 * vec.h (struct vnull): Use it.
30179 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30181 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
30182 (true_predicate, false_predicate, true_predicate_p,
30183 false_predicate_p): Remove.
30184 (single_cond_predicate, not_inlined_predicate): Turn to member function
30186 (add_condition): Update.
30187 (add_clause): Turn to...
30188 (predicate::add_clause): ... this one; update; allow passing NULL
30190 (and_predicates): Turn to ...
30191 (predicate::operator &=): ... this one.
30192 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
30193 (or_predicates): Turn to ...
30194 (predicate::or_with): ... this one.
30195 (evaluate_predicate): Turn to ...
30196 (predicate::evaluate): ... this one.
30197 (predicate_probability): Turn to ...
30198 (predicate::probability): ... this one.
30199 (dump_condition): Update.
30200 (dump_predicate): Turn to ...
30201 (predicate::dump): ... this one.
30202 (account_size_time): Update.
30203 (edge_set_predicate): Update.
30204 (set_hint_predicate): UPdate.
30205 (evaluate_conditions_for_known_args): Update.
30206 (evaluate_properties_for_edge): Update.
30207 (remap_predicate_after_duplication): Turn to...
30208 (predicate::remap_after_duplication): ... this one.
30209 (remap_hint_predicate_after_duplication): Update.
30210 (inline_summary_t::duplicate): UPdate.
30211 (dump_inline_edge_summary): Update.
30212 (dump_inline_summary): Update.
30213 (set_cond_stmt_execution_predicate): Update.
30214 (set_switch_stmt_execution_predicate): Update.
30215 (compute_bb_predicates): Update.
30216 (will_be_nonconstant_expr_predicate): Update.
30217 (will_be_nonconstant_predicate): Update.
30218 (phi_result_unknown_predicate): Update.
30219 (predicate_for_phi_result): Update.
30220 (array_index_predicate): Update.
30221 (estimate_function_body_sizes): Update.
30222 (estimate_node_size_and_time): Update.
30223 (estimate_ipcp_clone_size_and_time): Update.
30224 (remap_predicate): Rename to ...
30225 (predicate::remap_after_inlining): ... this one.
30226 (remap_hint_predicate): Update.
30227 (inline_merge_summary): Update.
30228 (inline_update_overall_summary): Update.
30229 (estimate_size_after_inlining): Update.
30230 (read_predicate): Rename to ...
30231 (predicate::stream_in): ... this one.
30232 (read_inline_edge_summary): Update.
30233 (write_predicate): Rename to ...
30234 (predicate::stream_out): ... this one.
30235 (write_inline_edge_summary): Update.
30236 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
30237 (clause_t): Turn to uint32_t
30238 (predicate): Turn to class; implement constructor and operators
30240 (size_time_entry): Update.
30241 (inline_summary): Update.
30242 (inline_edge_summary): Update.
30244 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
30246 * fold-const.c (fold_binary_loc): Move transformation...
30247 * match.pd (C - X CMP X): ... here.
30249 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
30251 * config/sparc/sparc.c (sparc_option_override): Set function
30252 alignment for -mcpu=niagara7 to 64 to match the I$ line.
30253 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
30255 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
30257 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
30259 2017-05-18 Marek Polacek <polacek@redhat.com>
30262 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
30263 (pass_ubsan::execute): Call gimple_assign_single_p instead of
30264 gimple_assign_load_p.
30266 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
30268 PR middle-end/80692
30269 * real.c (do_compare): Give decimal_do_compare preference over
30270 comparing just the signs.
30272 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
30274 * doc/md.texi (Canonicalization of Instructions): Describe the
30275 canonical form of instructions that inherently set a condition
30278 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
30280 PR middle-end/80775
30281 * tree-cfg.c: Move deletion of unreachable case statements to after
30282 the merging of consecutive case labels.
30284 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
30286 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
30287 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
30288 restoring of callee-saved registers.
30290 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
30292 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
30293 * config/visium/visium.c (single_set_and_flags): Likewise.
30294 * config/visium/visium.md (Substitutions): Likewise.
30296 2017-05-17 Martin Liska <mliska@suse.cz>
30298 * cfg.c: Introduce dump_flags_t type and
30299 use it instead of int type.
30301 * cfghooks.c: Likewise.
30302 * cfghooks.h (struct cfg_hooks): Likewise.
30303 * cfgrtl.c: Likewise.
30304 * cfgrtl.h: Likewise.
30305 * cgraph.c (cgraph_node::get_body): Likewise.
30306 * coretypes.h: Likewise.
30307 * domwalk.c: Likewise.
30308 * domwalk.h: Likewise.
30309 * dumpfile.c (struct dump_option_value_info): Likewise.
30310 (dump_enable_all): Likewise.
30311 (dump_switch_p_1): Likewise.
30312 (opt_info_switch_p): Likewise.
30313 * dumpfile.h (enum tree_dump_index): Likewise.
30314 (struct dump_file_info): Likewise.
30315 * genemit.c: Likewise.
30316 * generic-match-head.c: Likewise.
30317 * gengtype.c (open_base_files): Likewise.
30318 * gimple-pretty-print.c: Likewise.
30319 * gimple-pretty-print.h: Likewise.
30320 * graph.c (print_graph_cfg): Likewise.
30321 * graphite-scop-detection.c (dot_all_sese): Likewise.
30322 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
30323 * loop-unroll.c (report_unroll): Likewise.
30324 * passes.c (pass_manager::register_one_dump_file): Likewise.
30325 * print-tree.c: Likewise.
30326 * statistics.c: Likewise.
30327 * tree-cfg.c: Likewise.
30328 * tree-cfg.h: Likewise.
30329 * tree-dfa.c: Likewise.
30330 * tree-dfa.h: Likewise.
30331 * tree-dump.c (dump_function): Likewise.
30332 * tree-dump.h (struct dump_info): Likewise.
30333 * tree-pretty-print.c: Likewise.
30334 * tree-pretty-print.h: Likewise.
30335 * tree-ssa-live.c: Likewise.
30336 * tree-ssa-live.h: Likewise.
30337 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
30338 * tree-vect-loop.c: Likewise.
30339 * tree-vect-slp.c: Likewise.
30341 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
30342 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30344 PR tree-optimization/80457
30345 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
30346 of all arguments to a statement as scalar_to_vec operations.
30347 (vectorizable_call): Adjust call to vect_model_simple_cost for
30349 (vectorizable_conversion): Likewise.
30350 (vectorizable_assignment): Likewise.
30351 (vectorizable_shift): Likewise.
30352 (vectorizable_operation): Likewise.
30353 (vectorizable_comparison): Likewise.
30354 (vect_is_simple_cond): Record the def types for operands.
30355 (vectorizable_condition): Likewise, call vect_model_simple_cost.
30356 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
30357 for statement argument count.
30359 2017-05-16 Carl Love <cel@us.ibm.com>
30361 * config/rs6000/rs6000-c: Add support for built-in functions
30362 vector unsigned long long vec_bperm (vector unsigned long long,
30363 vector unsigned char)
30364 vector signed long long vec_mule (vector signed int,
30366 vector unsigned long long vec_mule (vector unsigned int,
30367 vector unsigned int)
30368 vector signed long long vec_mulo (vector signed int,
30370 vector unsigned long long vec_mulo (vector unsigned int,
30371 vector unsigned int)
30372 vector signed char vec_sldw (vector signed char,
30373 vector signed char,
30375 vector unsigned char vec_sldw (vector unsigned char,
30376 vector unsigned char,
30378 vector signed short vec_sldw (vector signed short,
30379 vector signed short,
30381 vector unsigned short vec_sldw (vector unsigned short,
30382 vector unsigned short,
30384 vector signed int vec_sldw (vector signed int,
30387 vector unsigned int vec_sldw (vector unsigned int,
30388 vector unsigned int,
30390 vector signed long long vec_sldw (vector signed long long,
30391 vector signed long long,
30393 vector unsigned long long vec_sldw (vector unsigned long long,
30394 vector unsigned long long,
30396 * config/rs6000/rs6000-c: Add support for built-in functions
30397 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
30398 * config/rs6000/altivec.h: Add defintion for vec_sldw.
30399 * doc/extend.texi: Update the built-in documentation for the
30400 new built-in functions.
30402 2017-05-16 Marek Polacek <polacek@redhat.com>
30406 * tree.c (save_expr): Don't fold the expression.
30408 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
30410 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
30411 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
30412 and (?*y,m). Update insn attributes.
30414 2017-05-16 Martin Liska <mliska@suse.cz>
30416 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
30417 flags argument of print_gimple_stmt, print_gimple_expr,
30418 print_generic_stmt and print_generic_expr.
30419 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
30420 * coretypes.h: Likewise.
30421 * except.c (dump_eh_tree): Likewise.
30422 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
30423 * gimple-pretty-print.h: Likewise.
30424 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
30425 (backprop::push_to_worklist): Likewise.
30426 (backprop::pop_from_worklist): Likewise.
30427 (backprop::process_use): Likewise.
30428 (backprop::intersect_uses): Likewise.
30429 (note_replacement): Likewise.
30430 * gimple-ssa-store-merging.c
30431 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
30432 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
30433 (pass_store_merging::execute): Likewise.
30434 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
30435 (ssa_base_cand_dump_callback): Likewise.
30436 (dump_incr_vec): Likewise.
30437 (replace_refs): Likewise.
30438 (replace_mult_candidate): Likewise.
30439 (create_add_on_incoming_edge): Likewise.
30440 (create_phi_basis): Likewise.
30441 (insert_initializers): Likewise.
30442 (all_phi_incrs_profitable): Likewise.
30443 (introduce_cast_before_cand): Likewise.
30444 (replace_one_candidate): Likewise.
30445 * gimplify.c (gimplify_expr): Likewise.
30446 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
30447 (set_rename): Likewise.
30448 (rename_uses): Likewise.
30449 (copy_loop_phi_nodes): Likewise.
30450 (add_close_phis_to_merge_points): Likewise.
30451 (copy_loop_close_phi_args): Likewise.
30452 (copy_cond_phi_args): Likewise.
30453 (graphite_copy_stmts_from_block): Likewise.
30454 (translate_pending_phi_nodes): Likewise.
30455 * graphite-poly.c (print_pdr): Likewise.
30456 (dump_gbb_cases): Likewise.
30457 (dump_gbb_conditions): Likewise.
30458 (print_scop_params): Likewise.
30459 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
30460 (build_cross_bb_scalars_use): Likewise.
30461 (gather_bbs::before_dom_children): Likewise.
30462 * hsa-dump.c (dump_hsa_immed): Likewise.
30463 * ipa-cp.c (print_ipcp_constant_value): Likewise.
30464 (get_replacement_map): Likewise.
30465 * ipa-inline-analysis.c (dump_condition): Likewise.
30466 (estimate_function_body_sizes): Likewise.
30467 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
30468 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
30469 * ipa-prop.c (ipa_dump_param): Likewise.
30470 (ipa_print_node_jump_functions_for_edge): Likewise.
30471 (ipa_modify_call_arguments): Likewise.
30472 (ipa_modify_expr): Likewise.
30473 (ipa_dump_param_adjustments): Likewise.
30474 (ipa_dump_agg_replacement_values): Likewise.
30475 (ipcp_modif_dom_walker::before_dom_children): Likewise.
30476 * ipa-pure-const.c (check_stmt): Likewise.
30477 (pass_nothrow::execute): Likewise.
30478 * ipa-split.c (execute_split_functions): Likewise.
30479 * omp-offload.c (dump_oacc_loop_part): Likewise.
30480 (dump_oacc_loop): Likewise.
30481 * trans-mem.c (tm_log_emit): Likewise.
30482 (tm_memopt_accumulate_memops): Likewise.
30483 (dump_tm_memopt_set): Likewise.
30484 (dump_tm_memopt_transform): Likewise.
30485 * tree-cfg.c (gimple_verify_flow_info): Likewise.
30486 (print_loop): Likewise.
30487 * tree-chkp-opt.c (chkp_print_addr): Likewise.
30488 (chkp_gather_checks_info): Likewise.
30489 (chkp_get_check_result): Likewise.
30490 (chkp_remove_check_if_pass): Likewise.
30491 (chkp_use_outer_bounds_if_possible): Likewise.
30492 (chkp_reduce_bounds_lifetime): Likewise.
30493 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
30494 (chkp_mark_completed_bounds): Likewise.
30495 (chkp_register_incomplete_bounds): Likewise.
30496 (chkp_mark_invalid_bounds): Likewise.
30497 (chkp_maybe_copy_and_register_bounds): Likewise.
30498 (chkp_build_returned_bound): Likewise.
30499 (chkp_get_bound_for_parm): Likewise.
30500 (chkp_build_bndldx): Likewise.
30501 (chkp_get_bounds_by_definition): Likewise.
30502 (chkp_generate_extern_var_bounds): Likewise.
30503 (chkp_get_bounds_for_decl_addr): Likewise.
30504 * tree-chrec.c (chrec_apply): Likewise.
30505 * tree-data-ref.c (dump_data_reference): Likewise.
30506 (dump_subscript): Likewise.
30507 (dump_data_dependence_relation): Likewise.
30508 (analyze_overlapping_iterations): Likewise.
30509 * tree-inline.c (expand_call_inline): Likewise.
30510 (tree_function_versioning): Likewise.
30511 * tree-into-ssa.c (dump_defs_stack): Likewise.
30512 (dump_currdefs): Likewise.
30513 (dump_names_replaced_by): Likewise.
30514 (dump_update_ssa): Likewise.
30515 (update_ssa): Likewise.
30516 * tree-object-size.c (pass_object_sizes::execute): Likewise.
30517 * tree-parloops.c (build_new_reduction): Likewise.
30518 (try_create_reduction_list): Likewise.
30519 (ref_conflicts_with_region): Likewise.
30520 (oacc_entry_exit_ok_1): Likewise.
30521 (oacc_entry_exit_single_gang): Likewise.
30522 * tree-pretty-print.h: Likewise.
30523 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
30524 (get_scalar_evolution): Likewise.
30525 (add_to_evolution): Likewise.
30526 (get_loop_exit_condition): Likewise.
30527 (analyze_evolution_in_loop): Likewise.
30528 (analyze_initial_condition): Likewise.
30529 (analyze_scalar_evolution): Likewise.
30530 (instantiate_scev): Likewise.
30531 (number_of_latch_executions): Likewise.
30532 (gather_chrec_stats): Likewise.
30533 (final_value_replacement_loop): Likewise.
30534 (scev_const_prop): Likewise.
30535 * tree-sra.c (dump_access): Likewise.
30536 (disqualify_candidate): Likewise.
30537 (create_access): Likewise.
30538 (reject): Likewise.
30539 (maybe_add_sra_candidate): Likewise.
30540 (create_access_replacement): Likewise.
30541 (analyze_access_subtree): Likewise.
30542 (analyze_all_variable_accesses): Likewise.
30543 (sra_modify_assign): Likewise.
30544 (initialize_constant_pool_replacements): Likewise.
30545 (find_param_candidates): Likewise.
30546 (decide_one_param_reduction): Likewise.
30547 (replace_removed_params_ssa_names): Likewise.
30548 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
30549 * tree-ssa-copy.c (dump_copy_of): Likewise.
30550 (copy_prop_visit_cond_stmt): Likewise.
30551 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
30552 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
30553 (record_equivalences_from_stmt): Likewise.
30554 * tree-ssa-dse.c (compute_trims): Likewise.
30555 (delete_dead_call): Likewise.
30556 (delete_dead_assignment): Likewise.
30557 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
30558 (forward_propagate_into_cond): Likewise.
30559 (pass_forwprop::execute): Likewise.
30560 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
30561 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
30563 (move_computations_worker): Likewise.
30564 (execute_sm): Likewise.
30565 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
30566 (remove_exits_and_undefined_stmts): Likewise.
30567 (remove_redundant_iv_tests): Likewise.
30568 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
30569 (adjust_iv_update_pos): Likewise.
30570 * tree-ssa-math-opts.c (bswap_replace): Likewise.
30571 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
30572 (value_replacement): Likewise.
30573 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
30574 * tree-ssa-pre.c (print_pre_expr): Likewise.
30575 (get_representative_for): Likewise.
30576 (create_expression_by_pieces): Likewise.
30577 (insert_into_preds_of_block): Likewise.
30578 (eliminate_insert): Likewise.
30579 (eliminate_dom_walker::before_dom_children): Likewise.
30580 (eliminate): Likewise.
30581 (remove_dead_inserted_code): Likewise.
30582 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30583 * tree-ssa-reassoc.c (get_rank): Likewise.
30584 (eliminate_duplicate_pair): Likewise.
30585 (eliminate_plus_minus_pair): Likewise.
30586 (eliminate_not_pairs): Likewise.
30587 (undistribute_ops_list): Likewise.
30588 (eliminate_redundant_comparison): Likewise.
30589 (update_range_test): Likewise.
30590 (optimize_range_tests_var_bound): Likewise.
30591 (optimize_vec_cond_expr): Likewise.
30592 (rewrite_expr_tree): Likewise.
30593 (rewrite_expr_tree_parallel): Likewise.
30594 (linearize_expr): Likewise.
30595 (break_up_subtract): Likewise.
30596 (linearize_expr_tree): Likewise.
30597 (attempt_builtin_powi): Likewise.
30598 (attempt_builtin_copysign): Likewise.
30599 (transform_stmt_to_copy): Likewise.
30600 (transform_stmt_to_multiply): Likewise.
30601 (dump_ops_vector): Likewise.
30602 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30603 (print_scc): Likewise.
30604 (set_ssa_val_to): Likewise.
30605 (visit_reference_op_store): Likewise.
30606 (visit_use): Likewise.
30607 (sccvn_dom_walker::before_dom_children): Likewise.
30608 (run_scc_vn): Likewise.
30609 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30611 (expr_hash_elt::print): Likewise.
30612 (const_and_copies::pop_to_marker): Likewise.
30613 (const_and_copies::record_const_or_copy_raw): Likewise.
30614 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30615 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30616 (dump_predicates): Likewise.
30617 (find_uninit_use): Likewise.
30618 (warn_uninitialized_phi): Likewise.
30619 (pass_late_warn_uninitialized::execute): Likewise.
30620 * tree-ssa.c (verify_vssa): Likewise.
30621 (verify_ssa): Likewise.
30622 (maybe_optimize_var): Likewise.
30623 * tree-vrp.c (dump_value_range): Likewise.
30624 (dump_all_value_ranges): Likewise.
30625 (dump_asserts_for): Likewise.
30626 (register_edge_assert_for_2): Likewise.
30627 (vrp_visit_cond_stmt): Likewise.
30628 (vrp_visit_switch_stmt): Likewise.
30629 (vrp_visit_stmt): Likewise.
30630 (vrp_visit_phi_node): Likewise.
30631 (simplify_cond_using_ranges_1): Likewise.
30632 (fold_predicate_in): Likewise.
30633 (evrp_dom_walker::before_dom_children): Likewise.
30634 (evrp_dom_walker::push_value_range): Likewise.
30635 (evrp_dom_walker::pop_value_range): Likewise.
30636 (execute_early_vrp): Likewise.
30638 2017-05-16 Richard Biener <rguenther@suse.de>
30640 * dwarf2out.c (loc_list_from_tree_1): Do not create
30641 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30643 2017-05-16 Richard Biener <rguenther@suse.de>
30645 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30647 (note_variable_value_in_expr): If we resolved the decl ref
30648 do not push to the stack.
30650 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30652 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30653 operations in fast-math mode.
30654 (vaddq_f16): Likewise.
30655 (vmul_f16): Likewise.
30656 (vmulq_f16): Likewise.
30657 (vsub_f16): Likewise.
30658 (vsubq_f16): Likewise.
30659 * config/arm/neon.md (add<mode>3): New.
30661 (fma:<VH:mode>3): New. Also remove outdated comment.
30664 2017-05-16 Martin Liska <mliska@suse.cz>
30668 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30669 (odr_types_equivalent_p): Likewise.
30671 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30673 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30675 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30678 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30679 non-interunit SSE move alternatives with '?'.
30680 (zero-extendsidi peephole2): New peephole to skip intermediate
30681 general register in SSE zero-extend sequence.
30683 2017-05-15 Jeff Law <law@redhat.com>
30685 * reorg.c (relax_delay_slots): Create a new variable to hold
30686 the temporary target rather than clobbering TARGET_LABEL.
30688 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30689 missing argument to extract_bit_field call.
30690 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30692 2017-05-15 Martin Liska <mliska@suse.cz>
30695 * gcc.c (process_command): Do not allow empty argument of -o option.
30697 2017-05-15 Renlin Li <renlin.li@arm.com>
30699 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30700 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30701 * config/aarch64/constraints.md (Usf): Add long call check.
30702 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30703 (call_value): Likewise.
30704 (sibcall): Likewise.
30705 (sibcall_value): Likewise.
30707 (call_value_insn): New.
30708 (sibcall_insn): Update rtx pattern.
30709 (sibcall_value_insn): Likewise.
30710 (call_internal): Remove.
30711 (call_value_internal): Likewise.
30712 (sibcall_internal): Likewise.
30713 (sibcall_value_internal): Likewise.
30714 (call_reg): Likewise.
30715 (call_symbol): Likewise.
30716 (call_value_reg): Likewise.
30717 (call_value_symbol): Likewise.
30719 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30722 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30724 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30726 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30727 compatible with CCGOCmode and with CCZmode.
30729 2017-05-14 Martin Sebor <msebor@redhat.com>
30731 PR middle-end/77671
30732 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30733 (gimple_fold_builtin_snprintf): Same.
30734 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30735 (gimple_fold_builtin_snprintf): Same.
30736 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30737 of character types.
30738 (is_call_safe): New function.
30739 (try_substitute_return_value): Call it.
30740 (try_simplify_call): New function.
30741 (pass_sprintf_length::handle_gimple_call): Call it.
30743 2017-05-14 Martin Sebor <msebor@redhat.com>
30745 PR middle-end/80669
30746 * builtins.c (expand_builtin_stpncpy): Simplify.
30748 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30750 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30751 * config/i386/i386.h
30752 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30753 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30754 (struct machine_function): Add new members call_ms2sysv,
30755 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30756 (struct machine_frame_state): New fields sp_realigned and
30757 sp_realigned_offset.
30758 * config/i386/i386.c
30759 (enum xlogue_stub): New enum.
30760 (enum xlogue_stub_sets): New enum.
30761 (class xlogue_layout): New class.
30762 (struct ix86_frame): New fields stack_realign_allocate_offset,
30763 stack_realign_offset and outlined_save_offset. Modify comments to
30764 detail stack layout when using out-of-line stubs.
30765 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30766 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30767 -mcall-ms2sysv-xlogues.
30768 (stub_managed_regs): New static variable.
30769 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30770 registers managed by out-of-line stub.
30771 (disable_call_ms2sysv_xlogues): New function.
30772 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30773 m->call_ms2sysv when appropriate and compute frame layout for
30775 (sp_valid_at, fp_valid_at): New inline functions.
30776 (choose_basereg): New function.
30777 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30779 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30780 Use align parameter of choose_baseaddr to generated aligned SSE movs
30782 (pro_epilogue_adjust_stack): Modify to track
30783 machine_frame_state::sp_realigned.
30784 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30785 (ix86_nsaved_sseregs): Likewise.
30786 (ix86_emit_save_regs): Likewise.
30787 (ix86_emit_save_regs_using_mov): Likewise.
30788 (ix86_emit_save_sse_regs_using_mov): Likewise.
30789 (get_scratch_register_on_entry): Likewise.
30790 (gen_frame_set): New function.
30791 (gen_frame_load): Likewise.
30792 (gen_frame_store): Likewise.
30793 (emit_outlined_ms2sysv_save): Likewise.
30794 (emit_outlined_ms2sysv_restore): Likewise.
30795 (ix86_expand_prologue): Modify stack re-alignment code and call
30796 emit_outlined_ms2sysv_save when appropriate.
30797 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30798 parameter rtx_insn *insn, which allows the function to be used to only
30799 generate the notes.
30800 (ix86_expand_epilogue): Modify validity checks of frame and stack
30801 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30802 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30803 * config/i386/predicates.md
30804 (save_multiple): New predicate.
30805 (restore_multiple): Likewise.
30806 * config/i386/sse.md
30807 (save_multiple<mode>): New pattern.
30808 (save_multiple_realign<mode>): Likewise.
30809 (restore_multiple<mode>): Likewise.
30810 (restore_multiple_and_return<mode>): Likewise.
30811 (restore_multiple_leave_return<mode>): Likewise.
30812 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30814 2017-05-14 Julia Koval <julia.koval@intel.com>
30816 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30817 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30818 (__builtin_ia32_xsetbv): New builtins.
30819 * config/i386/i386.c (ix86_expand_special_args_builtin):
30821 (ix86_expand_builtin): Special expand for new intrinsics.
30822 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
30823 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
30824 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
30826 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30828 * cfganal.c (inverted_post_order_compute): Change argument type
30830 * cfganal.h (inverted_post_order_compute): Adjust prototype.
30831 * df-core.c (rest_of_handle_df_initialize): Adjust.
30832 (rest_of_handle_df_finish): Likewise.
30833 (df_analyze_1): Likewise.
30834 (df_analyze): Likewise.
30835 (loop_inverted_post_order_compute): Change argument to be a vec *.
30836 (df_analyze_loop): Adjust.
30837 (df_get_n_blocks): Likewise.
30838 (df_get_postorder): Likewise.
30839 * df.h (struct df_d): Change field to be a vec.
30840 * lcm.c (compute_laterin): Adjust.
30841 (compute_available): Likewise.
30842 * lra-lives.c (lra_create_live_ranges_1): Likewise.
30843 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
30844 * tree-ssa-pre.c (compute_antic): Likewise.
30846 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30848 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
30849 (depth_first_search::depth_first_search): Change structure init
30850 function to this constructor.
30851 (depth_first_search::add_bb): Rename function to this member.
30852 (depth_first_search::execute): Likewise.
30853 (flow_dfs_compute_reverse_finish): Adjust.
30855 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30857 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
30858 (longest_simple_path): Likewise.
30859 * shrink-wrap.c (spread_components): Likewise.
30860 (disqualify_problematic_components): Likewise.
30861 (emit_common_heads_for_components): Likewise.
30862 (emit_common_tails_for_components): Likewise.
30863 (insert_prologue_epilogue_for_components): Likewise.
30865 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30867 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
30870 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30872 * df-core.c (df_set_blocks): Start using auto_bitmap.
30873 (df_compact_blocks): Likewise.
30874 * df-problems.c (df_rd_confluence_n): Likewise.
30875 * df-scan.c (df_insn_rescan_all): Likewise.
30876 (df_process_deferred_rescans): Likewise.
30877 (df_update_entry_block_defs): Likewise.
30878 (df_update_exit_block_uses): Likewise.
30879 (df_entry_block_bitmap_verify): Likewise.
30880 (df_exit_block_bitmap_verify): Likewise.
30881 (df_scan_verify): Likewise.
30882 * lra-constraints.c (lra_constraints): Likewise.
30883 (undo_optional_reloads): Likewise.
30884 (lra_undo_inheritance): Likewise.
30885 * lra-remat.c (calculate_gen_cands): Likewise.
30886 (do_remat): Likewise.
30887 * lra-spills.c (assign_spill_hard_regs): Likewise.
30888 (spill_pseudos): Likewise.
30889 * tree-ssa-pre.c (bitmap_set_and): Likewise.
30890 (bitmap_set_subtract_values): Likewise.
30892 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30894 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
30895 management with auto_bitmap.
30896 (fix_inter_tick): Likewise.
30897 (fix_recovery_deps): Likewise.
30898 * ira.c (add_store_equivs): Likewise.
30899 (find_moveable_pseudos): Likewise.
30900 (split_live_ranges_for_shrink_wrap): Likewise.
30901 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
30902 (rtx_reuse_manager::seen_def_p): Likewise.
30903 (rtx_reuse_manager::set_seen_def): Likewise.
30904 * print-rtl.h (class rtx_reuse_manager): Likewise.
30906 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30908 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
30910 (migrate_btr_def): Likewise.
30911 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
30912 * df-core.c (loop_post_order_compute): Likewise.
30913 (loop_inverted_post_order_compute): Likewise.
30914 * hsa-common.h: Likewise.
30915 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
30916 * init-regs.c (initialize_uninitialized_regs): Likewise.
30917 * ipa-inline.c (resolve_noninline_speculation): Likewise.
30918 (inline_small_functions): Likewise.
30919 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
30920 * ira.c (combine_and_move_insns): Likewise.
30921 (build_insn_chain): Likewise.
30922 * loop-invariant.c (find_invariants): Likewise.
30923 * lower-subreg.c (propagate_pseudo_copies): Likewise.
30924 * predict.c (tree_predict_by_opcode): Likewise.
30925 (predict_paths_leading_to): Likewise.
30926 (predict_paths_leading_to_edge): Likewise.
30927 (estimate_loops_at_level): Likewise.
30928 (estimate_loops): Likewise.
30929 * shrink-wrap.c (try_shrink_wrapping): Likewise.
30930 (spread_components): Likewise.
30931 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
30932 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
30933 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
30934 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
30935 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
30936 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
30937 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
30938 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
30939 (mark_threaded_blocks): Likewise.
30940 (thread_through_all_blocks): Likewise.
30941 * tree-ssa.c (verify_ssa): Likewise.
30942 (execute_update_addresses_taken): Likewise.
30943 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
30945 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30947 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
30949 (post_order_compute): Likewise.
30950 (inverted_post_order_compute): Likewise.
30951 (pre_and_rev_post_order_compute_fn): Likewise.
30953 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30955 * genrecog.c (int_set::int_set): Explicitly construct our
30956 auto_vec base class.
30957 * vec.h (auto_vec::auto_vec): New constructor.
30959 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30961 * bitmap.h (class auto_bitmap): New constructor taking
30962 bitmap_obstack * argument.
30964 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30966 * bitmap.h (class auto_bitmap): Change type of m_bits to
30967 bitmap_head, and adjust ctor / dtor and member operators.
30969 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
30971 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
30972 when returned register mode doesn't match original mode.
30974 2017-05-12 Jeff Law <law@redhat.com>
30975 Jakub Jelinek <jakub@redhat.com>
30977 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
30978 we look for cc setter after the compare-elim changes.
30979 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
30980 within the vector to match what compare-elim now expects.
30981 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
30982 (xorsi3_flags, one_cmplsi2_flags): Likewise.
30984 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
30985 after the compare-elim changes.
30986 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
30987 the vector to match what compare-elim now expects.
30988 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
30989 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
30990 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
30991 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
30992 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
30994 * config/visium/visium.c (single_set_and_flags): Fix where
30995 we look for cc setter after the compare-elim changes.
30996 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
30997 with the vector to match what compare-elim now expects.
30998 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
30999 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31000 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31001 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31002 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31003 (neg<mode>2_insn_set_overflow): Likewise.
31005 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31007 PR middle-end/79794
31008 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31009 maybe_expand_insn call, set ops[0].target. If still set after call,
31010 set alt_rtl. Add extra arg to recursive calls.
31011 (extract_bit_field): Add alt_rtl argument. Pass to
31013 * expmed.h (extract_bit_field): Fix prototype.
31014 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31015 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31016 to extract_bit_field_calls.
31017 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31018 Pass alt_rtl to extract_bit_field calls.
31019 * calls.c (store_unaligned_arguments_into_psuedos)
31020 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31021 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31023 * optabs.h (struct expand_operand): Add target bitfield.
31025 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31027 * compare-elim.c (try_eliminate_compare): Canonicalize
31028 operation with embedded compare to
31029 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31030 (set (reg) (operation)].
31032 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31034 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31037 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31038 cost of adding a carry flag for ADC instruction.
31039 [case MINUS]: Ignore the cost of subtracting a carry flag
31040 for SBB instruction.
31042 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31044 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31046 * config/rs6000/bmiintrin.h: New file.
31047 * config/rs6000/bmi2intrin.h: New file.
31048 * config/rs6000/x86intrin.h: New file.
31050 2017-05-12 Jeff Law <law@redhat.com>
31052 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31055 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31057 PR middle-end/80707
31058 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31060 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31062 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31063 early expansion of vector divide builtins.
31064 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
31065 builtins identified as having unsigned arguments.
31067 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31069 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
31070 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
31071 expansion of vector logical operations (and, andc, or, xor,
31074 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31076 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
31077 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
31079 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31081 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31082 early GIMPLE expansion of vector multiplies.
31084 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31086 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
31087 TARGET_HAVE_MOVT conditional.
31088 (movt splitter): Likewise.
31090 2017-05-12 Richard Biener <rguenther@suse.de>
31092 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
31093 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31094 Fold all stmts not inplace.
31096 2017-05-12 Richard Biener <rguenther@suse.de>
31098 PR tree-optimization/80713
31099 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
31100 inserted_exprs bit for not removed stmts.
31102 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
31104 PR middle-end/69921
31105 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
31106 parallelized" attribute for parallelized OpenACC kernels.
31107 * omp-offload.c (execute_oacc_device_lower): Use it.
31109 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
31110 Set "oacc kernels" attribute.
31111 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
31112 parameter. Adjust all users.
31113 (oacc_fn_attrib_kernels_p): Remove function.
31114 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
31115 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
31116 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
31117 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
31118 assert "oacc kernels" attribute is set.
31120 2017-05-11 Carl Love <cel@us.ibm.com>
31122 * config/rs6000/rs6000-c: Add support for built-in functions
31123 vector unsigned char vec_popcnt (vector signed char)
31124 vector unsigned char vec_popcnt (vector unsigned char)
31125 vector unsigned short vec_popcnt (vector signed short)
31126 vector unsigned short vec_popcnt (vector unsigned short)
31127 vector unsigned int vec_popcnt (vector signed int)
31128 vector unsigned int vec_popcnt (vector unsigned int)
31129 vector unsigned long long vec_popcnt (vector signed long long)
31130 vector unsigned long long vec_popcnt (vector unsigned long long)
31131 vector signed long long vec_slo (vector signed long long,
31132 vector signed char)
31133 vector signed long long vec_slo (vector signed long long,
31134 vector unsigned char)
31135 vector unsigned long long vec_slo (vector unsigned long long,
31136 vector signed char)
31137 vector unsigned long long vec_slo (vector unsigned long long,
31138 vector unsigned char)
31139 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
31140 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
31141 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
31142 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
31143 * doc/extend.texi: Update the built-in documentation file for the
31144 new built-in functions.
31146 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
31148 * attribs.h (sorted_attr_string): Move machine independent
31149 functions for target clone support from the i386 port to common
31150 code. Rename ix86_function_versions to common_function_versions.
31151 Rename make_name to make_unique_name.
31152 (common_function_versions): Likewise.
31153 (make_unique_name): Likewise.
31154 (make_dispatcher_decl): Likewise.
31155 (is_function_default_version): Likewise.
31156 * attribs.c (attr_strcmp): Likewise.
31157 (sorted_attr_string): Likewise.
31158 (common_function_versions): Likewise.
31159 (make_unique_name): Likewise.
31160 (make_dispatcher_decl): Likewise.
31161 (is_function_default_version): Likewise.
31162 * config/i386/i386.c (attr_strcmp): Likewise.
31163 (sorted_attr_string): Likewise.
31164 (ix86_function_versions): Likewise.
31165 (make_name): Likewise.
31166 (make_dispatcher_decl): Likewise.
31167 (is_function_default_version): Likewise.
31168 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
31170 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31173 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
31174 Account for direct move costs for vec_construct of integer
31177 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
31180 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
31181 (UNSPEC_STX_ATOMIC): Ditto.
31182 (loaddi_via_sse): New insn.
31183 (storedi_via_sse): Ditto.
31184 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
31185 Update corresponding peephole2 patterns.
31186 (atomic_storedi_fpu): Ditto.
31188 2017-05-11 Julia Koval <julia.koval@intel.com>
31190 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
31191 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
31193 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
31194 (__builtin_ia32_rsqrt14ss_mask): New builtins.
31195 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
31197 2017-05-11 Nathan Sidwell <nathan@acm.org>
31199 * graphite-poly.c: Include dumpfile.h.
31201 * dumpfle.h (dump_function): Declare here ...
31202 * tree-dump.h (dump_function): ... not here.
31203 * dumpfile.c: #include tree-cfg.h.
31204 (dump_function): Move here from ...
31205 * tree-dump.c (dump_function): ... here.
31206 * gimplify.c: #include splay-tree.h, not tree-dump.h.
31207 * graphite-poly.c: Don't include tree-dump.h.
31208 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
31209 * print-tree.c: Likewise.
31210 * stor-layout.c: Likewise.
31211 * tree-nested.c: Likewise.
31213 * dumpfile.c (dump_start): Use TDF_FLAGS.
31214 (dump_enable_all): Fix TDF_KIND check thinko.
31216 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
31218 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31219 array entries to represent two legal parameterizations of the
31220 overloaded __builtin_cmpb function, as represented by the
31221 P6_OV_BUILTIN_CMPB constant.
31222 (altivec_resolve_overloaded_builtin): Add special case handling
31223 for the __builtin_cmpb function, as represented by the
31224 P6_OV_BUILTIN_CMPB constant.
31225 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
31226 (BU_P6_64BIT_2): New macro.
31227 (BU_P6_OVERLOAD_2): New macro
31228 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
31229 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
31230 (CMPB): Add overload support to represent both 32-bit and 64-bit
31231 compare-bytes function.
31232 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
31233 support for TARGET_CMPB.
31234 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
31235 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31236 documentation of the __builtin_cmpb overloaded built-in function.
31238 2017-05-11 Richard Biener <rguenther@suse.de>
31240 PR tree-optimization/80705
31241 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
31242 bases are not vectorizable.
31244 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31246 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
31247 when counting register pressure.
31249 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31251 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
31252 (iv_ca_more_deps): Renamed to ...
31253 (iv_ca_compare_deps): ... this.
31254 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
31256 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31258 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
31260 (determine_group_iv_costs): ... here.
31261 (find_inv_vars_cb): Record inv var if it's not recorded before.
31263 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31265 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
31266 (get_shiftadd_cost): Ditto.
31268 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31270 * tree-ssa-address.c: Include header file.
31271 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
31273 (add_to_parts): Refactor.
31274 (addr_to_parts): New parameter. Update use of move_hint_to_base.
31275 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
31278 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31280 PR tree-optimization/53090
31281 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
31283 (extract_cond_operands): Detect condition with IV on both sides
31284 and return COMP_IV_EXPR_2.
31285 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
31286 (rewrite_use_compare): Simplify by removing call to function
31287 extract_cond_operands.
31289 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31291 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
31292 (extract_cond_operands): Detect condition comparing against non-
31293 invariant bound and return appropriate enum value.
31294 (find_interesting_uses_cond): Update use of extract_cond_operands.
31295 Handle its return value accordingly.
31296 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
31298 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31300 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
31301 nonlinear iv_use computation in loop invariant sensitive way.
31303 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31305 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
31306 (find_iv_candidates): Call relate_compare_use_with_all_cands.
31308 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31310 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
31311 (dump_cand): Support iv_cand.inv_exprs.
31312 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
31314 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
31317 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31319 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
31321 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
31322 as local function. Include necessary header files.
31323 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
31325 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31327 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
31329 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31331 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
31332 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
31333 RSHIFT_EXPR and BIT_NOT_EXPR.
31335 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31337 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
31338 (adjust_setup_cost): New parameter supporting round up adjustment.
31339 (struct address_cost_data): Delete.
31340 (force_expr_to_var_cost): Don't bound cost with spill_cost.
31341 (split_address_cost, ptr_difference_cost): Delete.
31342 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
31343 (struct ainc_cost_data): New struct.
31344 (get_address_cost_ainc): New function.
31345 (get_address_cost, get_computation_cost): Reimplement.
31346 (determine_group_iv_cost_address): Record inv_expr for all uses of
31348 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
31349 (iv_ca_has_deps): Reimplemented to ...
31350 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
31352 (iv_ca_extend): Call iv_ca_more_deps.
31354 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31356 * tree-ssa-address.c (struct mem_address): Move to header file.
31357 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
31358 * tree-ssa-address.h (struct mem_address): Move from C file.
31359 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
31361 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31363 * tree-affine.h (aff_combination_type): New interface.
31364 (aff_combination_zero_p): Remove static.
31365 (aff_combination_const_p): New interface.
31366 (aff_combination_singleton_var_p): New interfaces.
31368 2017-05-11 Richard Biener <rguenther@suse.de>
31370 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31371 Skip unreachable blocks and destinations.
31372 (eliminate): Move stmt removal and fixup ...
31373 (fini_eliminate): ... here. Skip inserted exprs.
31374 (pass_pre::execute): Move fini_pre after fini_eliminate.
31375 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
31376 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
31377 PRE to get rid of dead code that has invalid SSA form and
31378 split critical edges again.
31380 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31382 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31384 2017-05-11 Richard Biener <rguenther@suse.de>
31386 * passes.c (execute_function_todo): Verify loops if they are
31387 said to be up-to-date.
31388 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
31389 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
31391 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
31394 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
31395 handle calling assemble_external ourself.
31398 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
31399 modes with zero size. Enhance comment.
31401 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31403 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
31404 built-ins for vec_xl and vec_xst with short and char pointer
31407 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31409 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
31410 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
31411 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
31412 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
31413 (_mm_maskz_min_round_ss): New intrinsics.
31414 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
31415 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
31416 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
31417 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
31418 (__builtin_ia32_minss_mask_round): New builtins.
31419 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31420 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31421 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
31423 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
31424 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
31426 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
31429 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31431 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
31432 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
31433 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
31434 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
31435 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
31436 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
31437 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
31438 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31439 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31440 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
31441 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
31442 (__builtin_ia32_mulss_mask_round): New builtins.
31443 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31444 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31445 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
31447 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
31448 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31450 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31453 2017-05-10 Julia Koval <julia.koval@intel.com>
31455 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
31456 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
31457 (_mm256_setr_m128i): New intrinsics.
31459 2017-05-10 Julia Koval <julia.koval@intel.com>
31461 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
31462 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
31463 (_mm_maskz_rcp14_ss): New intrinsics.
31464 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
31465 (__builtin_ia32_rcp14ss_mask): New builtins.
31466 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
31468 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
31470 PR tree-optimization/51513
31471 * tree-cfg.c (gimple_seq_unreachable_p): New function.
31472 (assert_unreachable_fallthru_edge_p): Use it.
31473 (group_case_labels_stmt): Likewise.
31474 * tree-cfg.h: Prototype it.
31475 * stmt.c: Include cfghooks.h and tree-cfg.h.
31476 (emit_case_dispatch_table) <gap_label>: New local variable.
31477 Use it to fill dispatch table gaps.
31478 Test for default_label before updating probabilities.
31479 (expand_case) <default_label>: Remove unneeded initialization.
31480 Test for unreachable default case statement and remove its edge.
31481 Set default_label accordingly.
31482 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
31484 2017-05-10 Carl Love <cel@us.ibm.com>
31486 * config/rs6000/rs6000-c: Add support for built-in functions
31487 vector signed char vec_neg (vector signed char)
31488 vector signed short int vec_neg (vector short int)
31489 vector signed int vec_neg (vector signed int)
31490 vector signed long long vec_neg (vector signed long long)
31491 vector float vec_neg (vector float)
31492 vector double vec_neg (vector double)
31493 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
31495 * config/rs6000/altivec.h: Add define for vec_neg
31496 * doc/extend.texi: Update the built-in documentation for the
31497 new built-in functions.
31499 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31501 PR tree-optimization/77644
31502 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
31504 2017-05-10 Nathan Sidwell <nathan@acm.org>
31506 * dumpfile.h (TDI_lang_all): New.
31507 (TDF_KIND): New. Renumber others
31508 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
31510 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
31512 (get_dump_file_name): Adjust suffix generation.
31513 (dump_enable_all): Use TDF_KIND.
31514 * doc/invoke.texi (-fdump-lang-all): Document.
31516 * dumpfile.h: Tabify.
31518 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
31521 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
31522 Move member access before delete.
31524 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
31526 * tree-inline.c (expand_call_inline): Split block at stmt
31529 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
31532 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
31533 are now unused after splitting mov{sf,sd}_hardfloat.
31534 (f32_lr2): Likewise.
31535 (f32_lm): Likewise.
31536 (f32_lm2): Likewise.
31537 (f32_li): Likewise.
31538 (f32_li2): Likewise.
31539 (f32_lv): Likewise.
31540 (f32_sr): Likewise.
31541 (f32_sr2): Likewise.
31542 (f32_sm): Likewise.
31543 (f32_sm2): Likewise.
31544 (f32_si): Likewise.
31545 (f32_si2): Likewise.
31546 (f32_sv): Likewise.
31547 (f32_dm): Likewise.
31548 (f32_vsx): Likewise.
31549 (f32_av): Likewise.
31550 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
31551 For movsf, order stores so the VSX stores occur before the GPR
31552 store which encourages the register allocator to use a traditional
31553 FPR instead of a GPR. For movsd, order the stores so that the GPR
31554 store comes before the VSX stores to allow the power6 to work.
31555 This is due to the power6 not having a 32-bit integer store
31556 instruction from a FPR.
31557 (movsf_hardfloat): Likewise.
31558 (movsd_hardfloat): Likewise.
31560 2017-05-09 Martin Sebor <msebor@redhat.com>
31562 PR translation/80280
31563 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
31566 PR translation/80280
31567 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
31568 data member added in r247778.
31569 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
31571 2017-05-09 Nathan Sidwell <nathan@acm.org>
31573 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
31575 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31578 2017-05-09 Marek Polacek <polacek@redhat.com>
31580 * doc/invoke.texi: Fix typo.
31582 2017-05-09 Richard Biener <rguenther@suse.de>
31584 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31585 (vrp_val_is_min): Likewise.
31586 (set_value_range_to_value): Likewise.
31587 (set_value_range_to_nonnegative): Likewise.
31588 (gimple_assign_nonzero_p): Likewise.
31589 (gimple_stmt_nonzero_p): Likewise.
31590 (vrp_int_const_binop): Likewise. Remove unreachable case.
31591 (adjust_range_with_scev): Adjust comments.
31592 (compare_range_with_value): Likewise.
31593 (extract_range_from_phi_node): Likewise.
31594 (test_for_singularity): Likewise.
31596 2017-05-09 Richard Biener <rguenther@suse.de>
31598 * tree-vrp.c (get_single_symbol): Add assert that we don't
31599 get overflowed constants as invariant part.
31600 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31601 checks. Use wi::cmp instead of recursing for integer constants.
31602 (compare_values): Just ignore whether we assumed undefined
31603 overflow instead of failing the compare.
31604 (extract_range_for_var_from_comparison_expr): Add comment before the
31605 TREE_NO_WARNING sets.
31606 (test_for_singularity): Likewise.
31607 (extract_range_from_comparison): Do not disable optimization
31608 when we assumed undefined overflow.
31609 (extract_range_basic): Remove init of unused var.
31611 2017-05-09 Richard Biener <rguenther@suse.de>
31613 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31614 (extract_range_from_multiplicative_op_1): Adjust.
31615 (extract_range_from_binary_expr_1): Use int_const_binop.
31617 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31620 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31621 rs6000_store_data_bypass_p in seven define_bypass directives and
31622 in several comments.
31623 * config/rs6000/rs6000-protos.h: Add prototype for
31624 rs6000_store_data_bypass_p function.
31625 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31626 function implements slightly different (rs6000-specific) semantics
31627 than store_data_bypass_p, returning false rather than aborting
31628 with assertion error when arguments do not satisfy the
31629 requirements of store data bypass.
31630 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31631 rs6000_store_data_bypass_p.
31633 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31635 * config/xtensa/xtensa-protos.h
31636 (xtensa_initial_elimination_offset): New declaration.
31637 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31638 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31639 macro definition, add case for FRAME_POINTER_REGNUM when
31640 FRAME_GROWS_DOWNWARD.
31641 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31642 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31643 xtensa_initial_elimination_offset.
31645 2017-05-08 Nathan Sidwell <nathan@acm.org>
31647 * doc/invoke.texi: Alphabetize -fdump options.
31649 2017-05-08 Martin Sebor <msebor@redhat.com>
31651 PR translation/80280
31652 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31654 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31656 * target.def (compute_frame_layout): New optional target hook.
31657 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31658 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31659 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31661 * reload1.c (verify_initial_elim_offsets): Likewise.
31662 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31663 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31664 (arm_get_frame_offsets): Split up into this ...
31665 (arm_compute_frame_layout): ... and this function.
31667 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31669 * config/aarch64/constraints.md (Usa): New constraint.
31670 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31672 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31674 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31675 with_multilib_list after it has been checked.
31677 2017-05-08 Richard Biener <rguenther@suse.de>
31679 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31680 (bitmap_set_subtract_values): Likewise.
31682 2017-05-08 Richard Biener <rguenther@suse.de>
31684 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31685 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31687 (gimple_stmt_nonzero_warnv_p): Rename to ...
31688 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31690 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31691 (extract_range_basic): Adjust, do not disable propagation on
31692 strict overflow sensitive simplification.
31693 (vrp_visit_cond_stmt): Likewise.
31695 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31697 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31698 body size unconditionally.
31700 2017-05-07 Jeff Law <law@redhat.com>
31703 2017-05-06 Jeff Law <law@redhat.com>
31704 PR tree-optimization/78496
31705 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31708 PR tree-optimization/78496
31709 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31710 (simplify_stmt_using_ranges): Call it.
31711 (vrp_dom_walker::before_dom_children): Extract equivalences
31712 from an ASSERT_EXPR with an equality comparison against a
31715 2017-05-06 Jeff Law <law@redhat.com>
31717 PR tree-optimization/78496
31718 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31721 PR tree-optimization/78496
31722 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31723 (simplify_stmt_using_ranges): Call it.
31724 (vrp_dom_walker::before_dom_children): Extract equivalences
31725 from an ASSERT_EXPR with an equality comparison against a
31728 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31730 * lra-constraints.c (lra_copy_reg_equiv): New function.
31731 (split_reg): Use it to copy equivalence information from the
31732 original register to the spill register.
31734 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31736 PR rtl-optimization/75964
31737 * simplify-rtx.c (simplify_const_relational_operation): Remove
31738 invalid handling of comparisons of integer ABS.
31740 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31742 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31743 initialize to zero.
31744 (init_regs): Remove declaration.
31745 (function_arg_advance_32): Initialize error_p as boolean variable.
31747 2017-05-05 Nathan Sidwell <nathan@acm.org>
31749 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31750 lines. Use for (;;).
31752 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31754 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31755 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31756 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31757 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31758 VF=2 that require versioning.
31760 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31762 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31765 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31767 * diagnostic.h (diagnostic_override_option_index): Convert from
31768 macro to inline function.
31770 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31772 * diagnostic.c (last_module_changed_p): New function.
31773 (set_last_module): New function.
31774 (diagnostic_report_current_module): Convert macro usage to
31775 the above functions.
31776 * diagnostic.h (diagnostic_context::last_module): Strengthen
31777 from const line_map * to const line_map_ordinary *.
31778 (diagnostic_last_module_changed): Delete macro.
31779 (diagnostic_set_last_module): Delete macro.
31781 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31783 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31784 with diagnostic_report_diagnostic.
31785 (diagnostic_n_impl_richloc): Likewise.
31786 * diagnostic.h (report_diagnostic): Delete macro.
31787 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31788 with diagnostic_report_diagnostic.
31789 * substring-locations.c (format_warning_va): Likewise.
31791 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31793 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31794 save/restor of format_spec. Move option-printing code to...
31795 (print_option_information): ...this new function, and
31796 reimplement by simply printing to the pretty_printer,
31797 rather than appending to the format string.
31799 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31801 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31802 handling logic into...
31803 (update_effective_level_from_pragmas): ...this new function.
31805 2017-05-04 Andrew Waterman <andrew@sifive.com>
31807 * config/riscv/riscv.opt (mstrict-align): New option.
31808 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31809 (SLOW_UNALIGNED_ACCESS): Define.
31810 (riscv_slow_unaligned_access): Declare.
31811 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31813 (riscv_slow_unaligned_access): New variable.
31814 (rocket_tune_info): Set slow_unaligned_access to true.
31815 (optimize_size_tune_info): Set slow_unaligned_access to false.
31816 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31817 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31818 (riscv_option_override): Set riscv_slow_unaligned_access.
31819 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31821 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
31823 * config/riscv/riscv.md: Unify indentation.
31825 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
31830 * config/rs6000/rs6000.md (u code attribute): Add FIX and
31832 (extendsi<mode>2): Add support for doing sign extension via
31833 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
31834 don't have ISA 3.0 instructions.
31835 (extendsi<mode>2 splitter): Likewise.
31836 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
31837 generate the normal insns since SImode can now go in vector
31838 registers. Disallow the special UNSPECs needed for previous
31839 machines to hide SImode being used. Add new insns
31840 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
31841 (fix_trunc<mode>si2_stfiwx): Likewise.
31842 (fix_trunc<mode>si2_internal): Likewise.
31843 (fixuns_trunc<mode>si2): Likewise.
31844 (fixuns_trunc<mode>si2_stfiwx): Likewise.
31845 (fctiw<u>z_<mode>_smallint): Likewise.
31846 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
31847 of floating point to 32-bit integer from doing a direct move to
31848 the GPR registers to do a store.
31849 (fctiwz_<mode>): Break long line.
31851 2017-05-05 Bin Cheng <bin.cheng@arm.com>
31853 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
31854 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
31855 (addr_list, addr_offset_valid_p): New.
31856 (split_address_groups): Check offset validity with above function.
31857 (gt-tree-ssa-loop-ivopts.h): Include header file.
31859 2017-05-05 Nathan Sidwell <nathan@acm.org>
31861 * config.gcc (arm*-*-*): Add missing 'fi'.
31863 2017-05-05 Steve Ellcey <sellcey@cavium.com>
31865 * doc/invoke.texi (-fopt-info): Explicitly say order of options
31866 included in -fopt-info does not matter.
31867 * doc/optinfo.texi (-fopt-info): Fix description of default
31868 behavour. Explicitly say order of options included in -fopt-info
31871 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
31873 * config.gcc: Allow combinations of aprofile and rmprofile values for
31874 --with-multilib-list.
31875 * config/arm/t-multilib: New file.
31876 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
31877 variables. Remove setting of ISA and floating-point ABI in
31878 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
31879 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
31880 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
31881 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
31883 * config/arm/t-rmprofile: Likewise except for the matches changes.
31884 * doc/install.texi (--with-multilib-list): Document the combination of
31885 aprofile and rmprofile values and warn about pitfalls in doing that.
31887 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31889 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
31890 (movdi_aarch64): Likewise.
31892 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31894 PR tree-optimization/80632
31895 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
31897 (build_arrays): Initialize it for virtual phis.
31898 (fix_phi_nodes): Use it for virtual phis.
31900 PR tree-optimization/80558
31901 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
31902 [x, y] op z into [x op, y op z] for op & or | if conditions
31905 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
31906 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31909 * config/arm/arm.md (use_literal_pool): Remove.
31910 (64-bit immediate split): No longer takes cost into consideration
31911 if arm_disable_literal_pool is enabled.
31912 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
31913 used when arm_disable_literal_pool is enabled.
31914 (arm_max_const_double_inline_cost): Remove use of
31915 arm_disable_literal_pool.
31916 (push_minipool_fix): Add assert.
31917 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
31918 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
31919 (no_literal_pool_sf_immediate): New.
31921 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31923 PR tree-optimization/80613
31924 * tree-ssa-dce.c (propagate_necessity): Remove cases for
31925 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
31927 2017-05-05 Richard Biener <rguenther@suse.de>
31929 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
31931 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
31933 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
31934 of this flag from insn conditions due to removal from r247495.
31936 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31938 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
31940 (arm_early_store_addr_dep_ptr): Likewise.
31941 * config/arm/aarch-common-protos.h
31942 (arm_early_load_addr_dep_ptr): Add prototype.
31943 (arm_early_store_addr_dep_ptr): Likewise.
31944 * config/arm/cortex-a53.md: Add new bypasses.
31946 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31948 * tree.c (next_type_uid): Change type to unsigned.
31949 (type_hash_canon): Decrement back next_type_uid if
31950 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
31951 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
31954 2017-05-04 Martin Sebor <msebor@redhat.com>
31956 * builtins.c: Fix a trivial typo in a comment.
31958 PR middle-end/79234
31959 * builtins.c (check_sizes): Adjust to handle reading past the end.
31960 Avoid printing excessive upper bound of ranges. Use %E to print
31961 tree nodes instead of converting them to %wu.
31962 (expand_builtin_memchr): New function.
31963 (compute_dest_size): Rename...
31964 (compute_objsize): ...to this.
31965 (expand_builtin_memcpy): Adjust.
31966 (expand_builtin_mempcpy): Adjust.
31967 (expand_builtin_strcat): Adjust.
31968 (expand_builtin_strcpy): Adjust.
31969 (check_strncat_sizes): Adjust.
31970 (expand_builtin_strncat): Adjust.
31971 (expand_builtin_strncpy): Adjust and simplify.
31972 (expand_builtin_memset): Adjust.
31973 (expand_builtin_bzero): Adjust.
31974 (expand_builtin_memcmp): Adjust.
31975 (expand_builtin): Handle memcmp.
31976 (maybe_emit_chk_warning): Check strncat just once.
31978 2017-05-04 Martin Sebor <msebor@redhat.com>
31980 PR preprocessor/79214
31981 PR middle-end/79222
31982 PR middle-end/79223
31983 * builtins.c (check_sizes): Add inlining context and issue
31984 warnings even when -Wno-system-headers is set.
31985 (check_strncat_sizes): Same.
31986 (expand_builtin_strncat): Same.
31987 (expand_builtin_memmove): New function.
31988 (expand_builtin_stpncpy): Same.
31989 (expand_builtin): Handle memmove and stpncpy.
31991 2017-05-04 Bin Cheng <bin.cheng@arm.com>
31993 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
31994 which is not used any more.
31996 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31998 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32000 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32002 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32003 (cortexa53_tunings): Likewise.
32004 (cortexa57_tunings): Likewise.
32005 (cortexa72_tunings): Likewise.
32006 (cortexa73_tunings): Likewise.
32008 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32010 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32011 Set loop alignment to 8.
32013 2017-05-04 Martin Sebor <msebor@redhat.com>
32015 PR translation/80280
32016 * builtins.c (expand_builtin_object_size): Add missing quoting to
32017 %D and like directives.
32018 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32019 (hsa_type_for_tree_type): Same.
32020 (verify_function_arguments): Same.
32021 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32022 * varasm.c (get_section): Same.
32025 2017-05-04 Martin Sebor <msebor@redhat.com>
32027 PR translation/80280
32028 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32030 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32032 * config/aarch64/aarch64.c (generic_addrcost_table):
32033 Change HI/TI mode setting.
32035 2017-05-04 Martin Jambor <mjambor@suse.cz>
32037 PR tree-optimization/80622
32038 * tree-sra.c (comes_initialized_p): New function.
32039 (build_accesses_from_assign): Only set write lazily when
32040 comes_initialized_p is false.
32041 (analyze_access_subtree): Use comes_initialized_p.
32042 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32043 instead of testing for PARM_DECL.
32045 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32047 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32048 constraint on operand 0 to allow more general addressing modes.
32049 Adjust output template.
32050 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32052 * config/aarch64/aarch64-protos.h
32053 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32054 * config/aarch64/constraints.md (Dp): New address constraint.
32055 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32058 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32060 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32061 update use of estimate_ipcp_clone_size_and_time.
32062 (estimate_local_effects): Update use of
32063 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32064 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
32065 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
32066 Return nonspecialized time.
32068 2017-05-04 Richard Biener <rguenther@suse.de>
32070 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
32071 for the last VUSE which def dominates the PHI. Directly call
32073 (get_continuation_for_phi_1): Remove.
32075 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
32077 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
32078 to explain the use of truncating division. Cap the number of
32079 iterations to the maximum given by nb_iterations_upper_bound,
32082 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32084 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
32085 * configure: Regenerate.
32086 * config.in: Regenerate.
32087 * config/i386/driver-mingw32.c: new file.
32088 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
32089 * config.host: Link driver-mingw32.o on MinGW host.
32090 * doc/install.texi: Document new --enable-mingw-wildcard configure
32093 2017-05-04 Marek Polacek <polacek@redhat.com>
32095 PR tree-optimization/80612
32096 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
32098 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32099 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
32101 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
32102 (movt splitter): Likewise.
32103 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
32104 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
32105 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
32106 block for Thumb-1 with MOVT.
32107 (thumb2_legitimate_address_p): Move code block ...
32108 (can_avoid_literal_pool_for_label_p): ... into this new function.
32109 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
32111 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
32112 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
32113 "M-profile targets with the MOVT instruction".
32115 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32117 * config/arm/arm-builtins.c (arm_init_builtins): Rename
32118 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
32119 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
32121 2017-05-04 Martin Liska <mliska@suse.cz>
32123 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
32124 variable cond_code.
32126 2017-05-04 Richard Biener <rguenther@suse.de>
32128 * tree.c (array_at_struct_end_p): Handle arrays at struct
32129 end with flexarrays more conservatively. Refactor and treat
32130 arrays of arrays or aggregates more strict. Fix
32131 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
32132 * tree.c (array_at_struct_end_p): Adjust prototype.
32133 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
32134 * gimple-fold.c (get_range_strlen): Likewise.
32135 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
32137 2017-05-04 Richard Biener <rguenther@suse.de>
32139 PR tree-optimization/31130
32140 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
32142 (supports_overflow_infinity): Likewise.
32143 (is_negative_overflow_infinity): Likewise.
32144 (is_positive_overflow_infinity): Likewise.
32145 (is_overflow_infinity): Likewise.
32146 (stmt_overflow_infinity): Likewise.
32147 (overflow_infinity_range_p): Likewise.
32148 (usable_range_p): Remove as always returning true.
32149 (make_overflow_infinity): Remove.
32150 (negative_overflow_infinity): Likewise.
32151 (positive_overflow_infinity): Likewise.
32152 (avoid_overflow_infinity): Likewise.
32153 (set_value_range): Adjust accordingly.
32154 (set_value_range_to_nonnegative): Likewise, remove now unused
32155 overflow_infinity arg.
32156 (vrp_operand_equal_p): Adjust.
32157 (update_value_range): Likewise.
32158 (range_int_cst_singleton_p): Likewise.
32159 (operand_less_p): Likewise.
32160 (compare_values_warnv): Likewise.
32161 (extract_range_for_var_from_comparison_expr): Likewise.
32162 (vrp_int_const_binop): Likewise.
32163 (zero_nonzero_bits_from_vr): Likewise.
32164 (extract_range_from_multiplicative_op_1): Likewise.
32165 (extract_range_from_binary_expr_1): Likewise.
32166 (extract_range_from_unary_expr): Likewise.
32167 (extract_range_from_comparison): Likewise.
32168 (extract_range_basic): Likewise.
32169 (adjust_range_with_scev): Likewise.
32170 (compare_ranges): Likewise.
32171 (compare_range_with_value): Likewise.
32172 (dump_value_range): Likewise.
32173 (test_for_singularity): Likewise, remove strict_overflow_p parameter
32175 (simplify_cond_using_ranges): Adjust.
32177 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32179 * brig-builtins.def: Added a builtin for class_f64.
32180 * builtin-types.def: Added a builtin type needed by class_f64.
32182 2017-05-03 Jason Merrill <jason@redhat.com>
32184 * timevar.def: Add TV_CONSTEXPR.
32186 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32188 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
32190 2017-05-03 Martin Jambor <mjambor@suse.cz>
32192 * ipa-prop.c (ipa_update_after_lto_read): Removed.
32193 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
32194 * ipa-cp.c (ipcp_propagate_stage): Do not call
32195 ipa_update_after_lto_read.
32196 * ipa-inline.c (ipa_inline): Likewise.
32198 2017-05-03 Martin Jambor <mjambor@suse.cz>
32200 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
32201 tag. Added a default constructor and a destructor.
32202 (ipa_edge_args_sum_t): New class;
32203 (ipa_edge_args_sum): Declare.
32204 (ipa_edge_args_vector): Remove declaration.
32205 (IPA_EDGE_REF): Use ipa_edge_args_sum.
32206 (ipa_free_edge_args_substructures): Remove declaration.
32207 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
32208 (ipa_edge_args_info_available_for_edge_p): Likewise.
32209 * ipa-prop.c (ipa_edge_args_vector): Removed.
32210 (edge_removal_hook_holder): Likewise.
32211 (edge_duplication_hook_holder): Likewise.
32212 (ipa_edge_args_sum): New variable.
32213 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
32214 ipa_edge_args_vector.
32215 (ipa_free_edge_args_substructures): Likewise.
32216 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
32217 ipa_edge_args_vector.
32218 (ipa_edge_removal_hook): Turned into method
32219 ipa_edge_args_sum_t::remove.
32220 (ipa_edge_duplication_hook): Turned into method
32221 ipa_edge_args_sum_t::duplicate.
32222 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
32223 registering edge hooks.
32224 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
32225 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
32226 ipa_edge_args_sum instead of ipa_edge_args_vector.
32227 * ipa-profile.c (ipa_profile): Likewise.
32229 2017-05-03 Martin Jambor <mjambor@suse.cz>
32231 * symbol-summary.h (function_summary): New method exists.
32232 (function_summary::symtab_removal): Deallocate through release.
32233 (call_summary): New class.
32234 (gt_ggc_mx): New overload.
32235 (gt_pch_nx): Likewise.
32236 (gt_pch_nx): Likewise.
32238 2017-05-03 Jeff Law <law@redhat.com>
32240 PR tree-optimization/78496
32241 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
32242 from simplify_cond_using_ranges. Split off code to walk
32243 backwards through casts into ...
32244 (simplify_cond_using_ranges_2): New function.
32245 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
32246 (execute_vrp): After identifying jump threads, call
32247 simplify_cond_using_ranges_2.
32249 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
32252 * ipa-inline.h (inline_summary): Add ctor.
32253 (create_ggc): Do not use ggc_cleared_alloc.
32255 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
32256 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32258 * gcc.c (handle_braces): Support escaping in switch matching
32260 * doc/invoke.texi (Spec Files): Document it.
32261 Remove superfluous @code markup in items.
32263 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32265 * diagnostic-show-locus.c (struct column_range): New struct.
32266 (get_affected_columns): New function.
32267 (get_printed_columns): New function.
32268 (struct correction): New struct.
32269 (correction::ensure_capacity): New function.
32270 (correction::ensure_terminated): New function.
32271 (struct line_corrections): New struct.
32272 (line_corrections::~line_corrections): New dtor.
32273 (line_corrections::add_hint): New function.
32274 (layout::print_trailing_fixits): Reimplement in terms of the new
32276 (selftest::test_overlapped_fixit_printing): New function.
32277 (selftest::diagnostic_show_locus_c_tests): Call it.
32279 2017-05-03 Nathan Sidwell <nathan@acm.org>
32281 Canonicalize canonical type hashing
32282 * tree.h (type_hash_canon_hash): Declare.
32283 * tree.c (type_hash_list, attribute_hash_list): Move into
32284 type_hash_canon_hash.
32285 (build_type_attribute_qual_variant): Break out hash code calc into
32286 type_hash_canon_hash.
32287 (type_hash_canon_hash): New. Generic type hash computation.
32288 (build_range_type_1, build_array_type_1, build_function_type,
32289 build_method_type_directly, build_offset_type, build_complex_type,
32290 make_vector_type): Call it.
32292 2017-05-03 Richard Biener <rguenther@suse.de>
32294 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
32295 When all DRs have unknown misaligned do not always peel
32296 when there is a store but apply the same costing model as if
32297 there were only loads.
32299 2017-05-03 Richard Biener <rguenther@suse.de>
32302 PR tree-optimization/80492
32303 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32304 compare_base_decls returning dont-know properly.
32306 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
32308 * config/arm/iterators.md (CCSI): New mode iterator.
32309 (arch): New mode attribute.
32310 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
32311 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
32312 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
32313 code iterator for success result mode.
32314 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
32315 the corresponding new insn generators.
32317 2017-05-03 Bin Cheng <bin.cheng@arm.com>
32320 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32321 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32323 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
32325 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
32326 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
32327 (DDR_A): Wrap DDR argument in brackets.
32328 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
32329 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
32330 (DDR_REVERSED_P): Likewise.
32332 2017-05-03 Jakub Jelinek <jakub@redhat.com>
32334 PR tree-optimization/79472
32335 * tree-switch-conversion.c (struct switch_conv_info): Add
32336 contiguous_range and default_case_nonstandard fields.
32337 (collect_switch_conv_info): Compute contiguous_range and
32338 default_case_nonstandard fields, don't clear final_bb if
32339 contiguous_range and only the default case doesn't have the required
32341 (check_all_empty_except_final): Set default_case_nonstandard instead
32342 of failing if contiguous_range and the default case doesn't have empty
32344 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
32345 and only the default case doesn't have the required constants. Skip
32347 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
32348 if default_case_nonstandard.
32349 (build_constructors): Build constant 1 just once. Assert that default
32350 values aren't inserted in between cases if contiguous_range. Skip
32352 (build_arrays): Skip virtual phis.
32353 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
32354 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
32355 Handle virtual phis.
32356 (gen_inbound_check): Handle default_case_nonstandard case.
32357 (process_switch): Adjust check_final_bb caller. Call
32358 gather_default_values with the first non-default case instead of
32359 default case if default_case_nonstandard.
32361 2017-05-02 Nathan Sidwell <nathan@acm.org>
32363 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
32364 check. Fix formatting.
32366 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
32368 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
32369 errors when comparing specialized and unspecialized times.
32371 2017-05-02 David Malcolm <dmalcolm@redhat.com>
32373 * diagnostic-show-locus.c
32374 (layout::should_print_annotation_line_p): Make private.
32375 (layout::print_annotation_line): Make private.
32376 (layout::annotation_line_showed_range_p): Make private.
32377 (layout::show_ruler): Make private.
32378 (layout::print_source_line): Make private. Pass in line and
32379 line_width, rather than calling location_get_source_line. Drop
32381 (layout::print_leading_fixits): New method.
32382 (layout::print_any_fixits): Rename to...
32383 (layout::print_trailing_fixits): ...this, and make private.
32384 Don't print newline fixits.
32385 (diagnostic_show_locus): Move logic for printing one row into...
32386 (layout::print_line): ...this new function. Move the
32387 location_get_source_line call and error-handling from
32388 print_source_line to here. Call print_leading_fixits, and rename
32389 print_any_fixits to print_trailing_fixits.
32390 (selftest::test_fixit_insert_containing_newline): Update now that
32391 newlines are partially supported.
32392 (selftest::test_fixit_insert_containing_newline_2): New test.
32393 (selftest::test_fixit_replace_containing_newline): Update comments.
32394 (selftest::diagnostic_show_locus_c_tests): Call the new test.
32395 * edit-context.c (class added_line): New class.
32396 (class edited_line): Describe newline handling in comment.
32397 (edited_line::actually_edited_p): New method.
32398 (edited_line::print_content): Delete redundant decl.
32399 (edited_line::m_predecessors): New field.
32400 (edited_file::print_content): Call edited_line::print_content.
32401 (edited_file::print_diff): Update to support newlines.
32402 (edited_file::print_diff_hunk): Likewise.
32403 (edited_file::print_run_of_changed_lines): New function.
32404 (edited_file::print_diff_line): Convert to...
32405 (print_diff_line): ...this.
32406 (edited_file::get_effective_line_count): New function.
32407 (edited_line::edited_line): Initialize new field m_predecessors.
32408 (edited_line::~edited_line): Clean up m_predecessors.
32409 (edited_line::apply_fixit): Handle newlines.
32410 (edited_line::get_effective_line_count): New function.
32411 (edited_line::print_content): New function.
32412 (edited_line::print_diff_lines): New function.
32413 (selftest::test_applying_fixits_insert_containing_newline): New
32415 (selftest::test_applying_fixits_replace_containing_newline): New
32417 (selftest::insert_line): New function.
32418 (selftest::test_applying_fixits_multiple_lines): Add example of
32420 (selftest::edit_context_c_tests): Call the new tests.
32422 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32424 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
32425 parameter cand. Update dump information.
32426 (get_computation_cost): Update uses.
32428 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32430 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
32431 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
32432 (get_computation_at, rewrite_use_address): Update use of
32433 get_computation_aff.
32435 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32437 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
32438 (get_computation): Delete.
32439 (get_computation_cost): Implement like get_computation_cost_at.
32440 Use get_computation_at.
32441 (get_computation_cost_at): Delete.
32442 (rewrite_use_nonlinear_expr): Use get_computation_at.
32443 (rewrite_use_compare, remove_unused_ivs): Ditto.
32445 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32447 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
32449 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32451 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
32452 (ivopts_global_cost_for_size): Rename parameter and update uses.
32453 (iv_ca_recount_cost): Update uses.
32454 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
32455 candidates seperately in n_invs and n_cands.
32456 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
32458 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32460 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
32461 (find_inv_vars_cb): New.
32462 (find_depends): Renamed to ...
32463 (find_inv_vars): ... this.
32464 (add_candidate_1, force_var_cost): Call find_inv_vars.
32465 (split_address_cost, determine_group_iv_cost_cond): Ditto.
32467 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32469 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
32470 inv_vars. Add inv_exprs.
32471 (struct iv_cand): Rename depends_on to inv_vars.
32472 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
32473 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
32474 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
32475 (dump_cand): Dump inv_vars.
32476 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
32477 (record_invariant, find_depends, add_candidate_1): Ditto.
32478 (set_group_iv_cost, force_var_cost): Ditto.
32479 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
32480 (get_computation_cost_at, get_computation_cost): Ditto.
32481 (determine_group_iv_cost_generic): Ditto.
32482 (determine_group_iv_cost_address): Ditto.
32483 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
32484 (determine_group_iv_costs): Ditto.
32485 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
32486 (iv_ca_set_remove_invariants): Renamed to ...
32487 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
32488 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
32489 (iv_ca_set_add_invariants): Renamed to ...
32490 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
32491 (iv_ca_set_cp): Use iv_ca_set_add_invs.
32492 (iv_ca_has_deps): Support inv_vars and inv_exprs.
32493 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
32494 (create_new_ivs): Remove useless dump.
32496 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32498 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
32500 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
32501 (iv_ca_set_no_cp, create_new_iv): Ditto.
32503 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32505 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32507 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32509 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
32510 function tree_check2.
32512 2017-05-02 Martin Liska <mliska@suse.cz>
32514 * doc/gcov.texi: Add missing preposition.
32515 * gcov.c (function_info::function_info): Properly fill up
32516 all member variables.
32518 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32520 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
32522 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32524 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
32526 2017-05-02 Martin Liska <mliska@suse.cz>
32529 * lto-streamer-in.c (lto_read_tree_1): Remove
32530 LTO_STREAMER_DEBUG.
32531 * lto-streamer.c (struct tree_hash_entry): Likewise.
32532 (struct tree_entry_hasher): Likewise.
32533 (tree_entry_hasher::hash): Likewise.
32534 (tree_entry_hasher::equal): Likewise.
32535 (lto_streamer_init): Likewise.
32536 (lto_orig_address_map): Likewise.
32537 (lto_orig_address_get): Likewise.
32538 (lto_orig_address_remove): Likewise.
32539 * lto-streamer.h: Likewise.
32540 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
32541 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
32543 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
32545 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
32546 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
32547 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
32548 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
32549 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
32550 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
32551 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
32552 (mm_maskz_sub_ss): New intrinsics.
32553 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32554 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32555 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
32556 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
32557 (__builtin_ia32_subss_mask_round): New builtins.
32558 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32559 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32560 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
32562 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
32563 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32565 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32568 2017-05-02 Martin Jambor <mjambor@suse.cz>
32570 PR tree-optimization/78687
32571 * tree-sra.c (access): New field parent.
32572 (process_subtree_disqualification): New function.
32573 (disqualify_candidate): Call it.
32574 (build_accesses_from_assign): Reset write flag if creating an
32576 (build_access_subtree): Fill in parent field and also prpagate
32577 down grp_write flag.
32578 (create_artificial_child_access): New parameter set_grp_write, set
32579 grp_write to its value.
32580 (propagate_subaccesses_across_link): Also propagate grp_write flag
32582 (propagate_all_subaccesses): Push the closest parent back to work
32583 queue if add_access_to_work_queue returned true.
32585 2017-05-02 Richard Biener <rguenther@suse.de>
32587 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32588 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32589 -fstrict-overflow documentation.
32590 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32591 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32592 flag_strict_overflow.
32593 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32594 * lto-opts.c (lto_write_options): Do not stream it.
32595 * lto-wrapper.c (merge_and_complain): Do not handle it.
32596 * opts.c (default_options_table): Do not set -fstrict-overflow.
32597 (finish_options): Likewise do not clear it when sanitizing.
32598 * simplify-rtx.c (simplify_const_relational_operation): Do not
32599 test flag_strict_overflow.
32601 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32603 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32604 using enabled attribute.
32605 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32606 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32607 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32608 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32609 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32610 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32611 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32612 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32613 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32614 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32616 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32618 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32620 2017-05-02 Richard Biener <rguenther@suse.de>
32622 PR tree-optimization/80591
32624 2017-04-10 Richard Biener <rguenther@suse.de>
32626 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32629 2017-05-02 Richard Biener <rguenther@suse.de>
32631 PR tree-optimization/80549
32632 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32633 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32634 headers if they do not have a preheader.
32636 2017-05-02 Martin Liska <mliska@suse.cz>
32639 * common.opt: Fix typo.
32640 * doc/invoke.texi: Likewise.
32642 2017-05-01 Jan Beulich <jbeulich@suse.com>
32644 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32645 swapping, add (x,x,m,x,n) alternative.
32647 2017-05-01 Nathan Sidwell <nathan@acm.org>
32649 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32650 unnecessary unadjusted_alignment check.
32652 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32655 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32657 * gimplify.c (gimplify_cilk_detach): New function.
32658 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32659 * tree-core.h: Document EXPR_CILK_SPAWN.
32660 * tree.h (EXPR_CILK_SPAWN): Define.
32662 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32664 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32665 to use new fixit_hint representation, using the "replace" logic.
32666 (get_line_span_for_fixit_hint): Likewise.
32667 (layout::print_any_fixits): Likewise.
32668 (selftest::test_one_liner_many_fixits): Rename to...
32669 (selftest::test_one_liner_many_fixits_1): ...this, and update
32670 comment and expected output to reflect that the multiple fix-it
32671 hints are now consolidated into one insertion.
32672 (selftest::test_one_liner_many_fixits_2): New test.
32673 (selftest::test_diagnostic_show_locus_one_liner): Update for
32675 (selftest::test_fixit_consolidation): Update for fix-it API
32677 * diagnostic.c (print_parseable_fixits): Likewise.
32678 * edit-context.c (edited_line::m_line_events): Convert from
32679 auto_vec <line_event *> to auto_vec <line_event>.
32680 (class line_event): Convert from abstract base class to a concrete
32681 class, taking over the role of replace_event.
32682 (class insert_event): Delete.
32683 (class replace_event): Rename to class line_event. Convert to
32685 (edit_context::add_fixits): Reimplement.
32686 (edit_context::apply_insert): Delete.
32687 (edit_context::apply_replace): Rename to...
32688 (edit_context::apply_fixit): ...this. Convert to half-open range.
32689 (edited_file::apply_insert): Delete.
32690 (edited_file::apply_replace): Rename to...
32691 (edited_file::apply_fixit): ...this.
32692 (edited_line::~edited_line): Drop deletion of events.
32693 (edited_line::apply_insert): Delete.
32694 (edited_line::apply_replace): Rename to...
32695 (edited_line::apply_fixit): ...this. Convert to half-open range.
32696 Update for change to type of m_line_events.
32697 * edit-context.h (edit_context::apply_insert): Delete.
32698 (edit_context::apply_replace): Rename to...
32699 (edit_context::apply_fixit): ...this.
32701 2017-05-01 Martin Sebor <msebor@redhat.com>
32703 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32706 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32709 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32710 __get_cpuid_max returns 0.
32711 (__get_cpuid_count): Ditto.
32713 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32715 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32716 replacement expression is another instance of one of its arguments.
32718 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32721 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32722 check for stack push/pop autoinc.
32723 * config/i386/i386.c (ix86_agi_dependent): Return false
32724 if the only reason why modified_in_p returned true is that
32725 addr is SP based and set_insn is a push or pop.
32727 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32729 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32732 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32735 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32736 (account_size_time): Use two predicates - exec_pred and
32738 (evaluate_conditions_for_known_args): Compute both clause and
32740 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32741 (inline_summary_t::duplicate): Update.
32742 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32744 (compute_inline_parameters): Likewise.
32745 (estimate_edge_size_and_time): Update caluclation of time.
32746 (estimate_node_size_and_time): Compute both time and nonspecialized
32748 (estimate_ipcp_clone_size_and_time): Update.
32749 (inline_merge_summary): Update.
32750 (do_estimate_edge_time): Update.
32751 (do_estimate_edge_size): Update.
32752 (do_estimate_edge_hints): Update.
32753 (inline_read_section, inline_write_summary): Stream both new predicates.
32754 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32756 (compute_inlined_call_time): Cleanup.
32757 (big_speedup_p): Update.
32758 (edge_badness): Update.
32759 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32760 (size_time_entry): Replace predicate by exec_predicate and
32761 nonconst_predicate.
32762 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32763 (estimate_edge_time): Return also nonspec_time.
32764 (reset_edge_growth_cache): Update.
32766 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32768 PR rtl-optimization/80491
32769 * ifcvt.c (noce_process_if_block): When looking for x setter
32770 with missing else_bb, don't check only the insn right before
32771 cond_earliest, but look for the last insn that x is modified in
32772 within the same bb.
32774 PR rtl-optimization/80491
32775 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32777 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32779 PR tree-optimization/80487
32780 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32782 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32784 PR tree-optimization/79697
32785 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32786 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32787 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32789 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32790 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32792 2017-04-28 Martin Sebor <msebor@redhat.com>
32794 PR tree-optimization/80523
32795 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32796 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32798 (maybe_warn, format_directive, parse_directive): Use new functions.
32799 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32801 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32803 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32805 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32807 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32808 target_header_dir): Set correctly.
32809 * configure: Regenerated.
32810 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32811 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32812 instead of SYSTEM_HEADER_DIR.
32814 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32816 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32817 (estimate_local_effects): Likewise.
32818 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32819 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32820 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32821 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
32822 do_estimate_edge_time, estimate_edge_time): Likewise.
32823 * ipa-inline-analysis.c (estimate_node_size_and_time,
32824 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
32825 (estimate_time_after_inlining): Remove.
32827 2017-04-28 Martin Liska <mliska@suse.cz>
32829 * doc/gcov.texi: Enhance documentation of gcov.
32831 2017-04-28 Martin Liska <mliska@suse.cz>
32833 * doc/gcov.texi: Sort options in alphabetic order.
32834 * doc/gcov-dump.texi: Likewise.
32835 * doc/gcov-tool.texi: Likewise.
32836 * gcov.c (print_usage): Likewise.
32837 * gcov-dump.c (print_usage): Likewise.
32838 * gcov-tool.c (print_merge_usage_message): Likewise.
32839 (print_rewrite_usage_message): Likewise.
32840 (print_overlap_usage_message): Likewise.
32842 2017-04-28 Martin Liska <mliska@suse.cz>
32844 PR gcov-profile/53915
32845 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
32847 2017-04-28 Martin Liska <mliska@suse.cz>
32849 PR gcov-profile/79891
32850 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
32851 is marked by compiler as living on a line.
32852 (get_cycles_count): Remove usage of the union.
32853 (output_intermediate_file): Likewise.
32854 (find_source): Fix GNU coding style.
32855 (accumulate_line_counts): Remove old non-all block mode.
32856 (output_lines): Remove usage of the union.
32857 * profile.c (output_location): Include all BBs, even if
32858 belonging to a same line (and file) as a previous BB.
32860 2017-04-28 Martin Liska <mliska@suse.cz>
32862 * gcov.c (process_args): Handle new argument 'w'.
32863 (read_graph_file): Assign ID to BBs.
32864 (output_branch_count): Display BB # if verbose flag is set.
32865 (output_lines): Likewise for arcs.
32866 (print_usage): Add '--verbose' option help.
32867 * doc/gcov.texi: Document --verbose (-w) option.
32869 2017-04-28 Martin Liska <mliska@suse.cz>
32871 * gcov.c (struct block_location_info): New struct.
32872 (process_file): Fill up the new structure.
32873 (read_graph_file): Replace usage of encoding by the newly added
32875 (add_line_counts): Likewise.
32876 (accumulate_line_counts): Remove usage of the union.
32877 (function_info::function_info): New function.
32878 (function_info::~function_info): Likewise.
32879 (process_file): Call delete instead of release_function.
32880 (release_function): Release the function.
32881 (release_structures): Call delete instead of release_function.
32882 (solve_flow_graph): Replace usage of num_blocks.
32883 (find_exception_blocks): Likewise.
32884 (output_lines): Fix GNU coding style.
32886 2017-04-28 Martin Liska <mliska@suse.cz>
32889 * coverage.c (coverage_remove_note_file): New function.
32890 * coverage.h: Declare the function.
32891 * toplev.c (finalize): Clean if an error has been seen.
32893 2017-04-28 Martin Liska <mliska@suse.cz>
32895 PR gcov-profile/80031
32896 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
32897 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
32898 * gcov.c (read_graph_file): Read just number of blocks.
32899 * profile.c (branch_prob): Do not stream 0 flags per a basic
32902 2017-04-28 Martin Liska <mliska@suse.cz>
32904 * gcov-dump.c (tag_*): Add new argument to declarations.
32905 (dump_gcov_file): Likewise.
32906 (tag_blocks): Add and use new argument depth.
32907 (tag_arcs): Likewise.
32908 (tag_lines): Likewise.
32909 (tag_counters): Likewise.
32910 (tag_summary): Likewise.
32911 (dump_working_sets): Use depth to do a proper indentation.
32913 2017-04-28 Jakub Jelinek <jakub@redhat.com>
32916 * cgraph.h (symtab_node::debug_symtab): No longer inline.
32917 * symtab.c (symtab_node::debug_symtab): Move definition here.
32919 2017-04-28 Richard Biener <rguenther@suse.de>
32921 * lto-streamer.h (LTO_major_version): Bump to 7.
32923 2017-04-28 Richard Biener <rguenther@suse.de>
32925 * tree-vrp.c (assert_info): New struct.
32926 (add_assert_info): New helper.
32927 (register_edge_assert_for_2): Refactor to add asserts to a vector
32929 (register_edge_assert_for_1): Likewise.
32930 (register_edge_assert_for): Likewise.
32931 (finish_register_edge_assert_for): New helper actually registering
32932 asserts where live on edge.
32933 (find_conditional_asserts): Adjust.
32934 (find_switch_asserts): Likewise.
32935 (evrp_dom_walker::try_find_new_range): Generalize.
32936 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
32938 2017-04-27 Marek Polacek <polacek@redhat.com>
32941 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
32942 arg10 and arg11 to itype.
32944 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
32946 * doc/extend.texi (Object Size Checking): Improve grammar.
32948 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
32951 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
32952 that the logic for permitting reciprocal estimates matches that
32955 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32958 * tree.c (type_cache_hasher::equal): Only compare
32959 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
32960 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
32961 non-aggregate element types.
32962 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
32963 about the flag on ARRAY_TYPEs in the comment, formatting fix.
32965 2017-04-27 Richard Biener <rguenther@suse.de>
32967 PR middle-end/80533
32968 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
32969 stripping ARRAY_REFs from MEM_EXPR make sure we're not
32970 keeping a reference to a trailing array.
32972 2017-04-27 Richard Biener <rguenther@suse.de>
32974 PR middle-end/80539
32975 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
32976 being in loop-closed SSA form conservatively.
32977 (chrec_fold_multiply_poly_poly): Likewise.
32979 2017-04-27 Tamar Christina <tamar.christina@arm.com>
32981 PR middle-end/79665
32982 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
32983 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
32985 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32988 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
32989 (aarch64_function_arg_alignment): Return unsigned int again, but still
32990 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
32991 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32992 Don't emit -Wpsabi note.
32993 (aarch64_function_arg_boundary): Likewise.
32994 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32997 2017-04-26 Nathan Sidwell <nathan@acm.org>
32999 * tree.h (crc32_unsigned_n): Declare.
33000 (crc32_unsigned, crc32_unsigned): Make inline.
33001 * tree.c (crc32_unsigned_bits): Replace with ...
33002 (crc32_unsigned_n): ... this.
33003 (crc32_unsigned, crc32_byte): Remove.
33004 (crc32_string): Remove unnecessary braces.
33006 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33008 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33009 * ipa-inline-analysis.c (MAX_TIME): Remove.
33010 (account_size_time): Use sreal for time.
33011 (dump_inline_summary): Update.
33012 (estimate_function_body_sizes): Update.
33013 (estimate_edge_size_and_time): Update.
33014 (estimate_calls_size_and_time): Update.
33015 (estimate_node_size_and_time): Update.
33016 (inline_merge_summary): Update.
33017 (inline_update_overall_summary): Update.
33018 (estimate_time_after_inlining): Update.
33019 (inline_read_section): Update.
33020 (inline_write_summary): Update.
33021 * ipa-inline.c (compute_uninlined_call_time): Update.
33022 (compute_inlined_call_time): Update.
33023 (recursive_inlining): Update.
33024 (inline_small_functions): Update.
33025 (dump_overall_stats): Update.
33026 * ipa-inline.h: Include sreal.h.
33027 (size_time_entry): Turn time to sreal.
33028 (inline_summary): Turn self_time nad time to sreal.
33030 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33032 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33034 (sreal::stream_out, sreal::stream_in): New.
33035 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33037 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33039 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33042 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33045 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33047 (dimode_scalar_chain::compute_convert_gain): Ditto.
33048 (dimode_scalar_chain::make_vector_copies): Ditto.
33049 (dimode_scalar_chain::convert_reg): Ditto.
33050 (dimode_scalar_chain::convert_insn): Ditto.
33051 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33052 (VI248_AVX512BW_1): New mode iterator.
33053 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33054 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33057 2017-04-25 Martin Sebor <msebor@redhat.com>
33059 PR tree-optimization/80497
33060 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33061 constants are representable in HOST_WIDE_INT.
33062 (parse_directive): Ditto.
33064 2017-04-25 Martin Sebor <msebor@redhat.com>
33067 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
33068 (new_zero_array): Adjust signature.
33069 (dom_info::dom_init): Used unsigned rather that size_t.
33070 (dom_info::dom_info): Same.
33072 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33073 Jakub Jelinek <jakub@redhat.com>
33076 * config/arm/arm.c: Include gimple.h.
33077 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33078 returns negative, increment ncrn only if it returned positive.
33079 (arm_needs_doubleword_align): Return int instead of bool,
33080 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
33081 members, but if there is any such non-FIELD_DECL
33082 > PARM_BOUNDARY aligned decl, return -1 instead of false.
33083 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33084 returns negative, increment nregs only if it returned positive.
33085 (arm_setup_incoming_varargs): Likewise.
33086 (arm_function_arg_boundary): Emit -Wpsabi note if
33087 arm_needs_doubleword_align returns negative, return
33088 DOUBLEWORD_ALIGNMENT only if it returned positive.
33090 2017-04-25 Marek Polacek <polacek@redhat.com>
33093 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33094 first argument to type.
33096 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
33099 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
33100 type checks to test for compatibility instead of equality.
33102 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33103 Jakub Jelinek <jakub@redhat.com>
33106 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
33108 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
33109 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
33110 the alignment computation, but return their maximum in warn_alignment.
33111 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33112 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
33114 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
33115 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33118 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33120 * config/arc/simdext.md (dmpyh): Fix typo.
33122 2017-04-25 Richard Biener <rguenther@suse.de>
33124 PR tree-optimization/80492
33125 * alias.c (compare_base_decls): Handle registers with asm
33126 specification conservatively.
33127 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33128 compare_base_decls returning dont-know properly.
33130 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33132 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
33133 (legitimate_offset_address_p): New function.
33134 (arc_legitimate_address_p): Use above function.
33136 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33138 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
33140 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33142 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
33143 ACCH registers whenever they are available.
33145 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33147 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
33148 double regs fix when not used.
33150 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33152 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
33154 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
33155 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
33157 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33159 * config/arc/arc.c (arc_output_addsi): Check for h-register class
33160 when emitting short ADD instructions.
33162 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33164 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
33166 (cmpsi_cc_c_insn): Likewise.
33167 (cbranchsi4_scratch): Compute proper instruction length using
33168 compact_hreg_operand.
33169 * config/arc/predicates.md (compact_hreg_operand): New predicate.
33171 2017-04-25 Richard Biener <rguenther@suse.de>
33173 PR middle-end/80509
33174 * passes.c (pass_manager::pass_manager): Initialize
33175 m_name_to_pass_map.
33177 2017-04-25 Richard Biener <rguenther@suse.de>
33179 PR tree-optimization/79201
33180 * tree-ssa-sink.c (statement_sink_location): Handle calls.
33182 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33185 * config/s390/vector.md: Split MEM->GPR vector moves for
33186 non-s_operand addresses.
33188 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33191 * config/s390/predicates.md (reload_const_wide_int_operand): New
33193 * config/s390/s390.md ("movti"): Remove d/P alternative.
33194 ("movti_bigconst"): New pattern definition.
33196 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33199 * s390-protos.h (s390_expand_cs_hqi): Removed.
33200 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
33201 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
33202 modes as well as CCZ1mode and CCZmode.
33203 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
33204 signature of s390_emit_compare_and_swap.
33205 (s390_expand_cs_hqi): Likewise, make static.
33206 (s390_expand_cs_tdsi): Generate an explicit compare before trying
33207 compare-and-swap, in some cases.
33208 (s390_expand_cs): Wrapper function.
33209 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
33211 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
33212 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
33213 patterns for small and large integers. Forbid symref memory operands.
33214 Move expander to s390.c. Require cc register.
33215 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
33216 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
33217 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
33218 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
33219 symref memory operands. Remove CC mode and call s390_match_ccmode
33221 ("atomic_exchange<mode>"): Allow and implement all integer modes.
33223 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33225 * config/s390/s390.md (define_peephole2): New peephole to help
33226 combining the load-and-test pattern with volatile memory.
33228 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33230 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
33231 with CCZmode for TARGET_Z196.
33233 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33235 PR rtl-optimization/80501
33236 * combine.c (make_compound_operation_int): Set subreg_code to SET
33237 even for AND with mask of the sign bit of mode.
33239 PR rtl-optimization/80500
33240 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
33241 sum's initial value.
33243 2017-04-25 Julian Brown <julian@codesourcery.com>
33244 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33246 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
33248 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
33250 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
33252 2017-04-25 Julian Brown <julian@codesourcery.com>
33253 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33255 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
33256 (thunderx2t99_sha): New Reservation.
33258 2017-04-25 Julian Brown <julian@codesourcery.com>
33259 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33261 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
33262 type for 1-element load.
33264 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
33266 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
33268 2017-04-24 Martin Jambor <mjambor@suse.cz>
33270 PR tree-optimization/80293
33271 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
33272 char arrays not totally scalarizable if it is false.
33273 (analyze_all_variable_accesses): Pass correct value in the new
33274 parameter. Add a statistics counter.
33276 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
33278 PR middle-end/79931
33279 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
33281 2017-04-24 Richard Biener <rguenther@suse.de>
33283 PR tree-optimization/80494
33284 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
33285 out for complex types.
33287 2017-04-24 Richard Biener <rguenther@suse.de>
33289 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
33290 * tree-ssa-sccvn.c (print_scc): Print SCC size.
33291 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
33292 (DFS): Adjust and never fail.
33293 (sccvn_dom_walker::fail): Remove.
33294 (sccvn_dom_walker::before_dom_children): Adjust.
33295 (run_scc_vn): Likewise and never fail.
33296 * tree-ssa-pre.c (pass_pre::execute): Adjust.
33297 (pass_fre::execute): Likewise.
33299 2017-04-24 Richard Biener <rguenther@suse.de>
33301 PR tree-optimization/79725
33302 * tree-ssa-sink.c (statement_sink_location): Return whether
33303 failure reason was zero uses. Move that check later.
33304 (sink_code_in_bb): Deal with zero uses by removing the stmt
33307 2017-04-24 Richard Biener <rguenther@suse.de>
33310 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
33311 pointer-based references.
33313 2017-04-24 Richard Biener <rguenther@suse.de>
33316 * pass_manager.h (pass_manager::operator new): Remove.
33317 (pass_manager::operator delete): Likewise.
33318 * passes.c (pass_manager::operator new): Remove.
33319 (pass_manager::operator delete): Likewise.
33320 (pass_manager::pass_manager): Zero individual pass members.
33322 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
33325 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
33326 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
33327 Check "XEXP (src, 1)" operand here.
33328 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
33329 Check "XEXP (src, 1)" operand here.
33330 (dimode_scalar_chain::make_vector_copies): Detect count register
33331 of a shift instruction. Zero extend count register from QImode
33332 to DImode to satisfy vector shift pattern count operand predicate.
33333 Substitute vector shift count operand with a DImode copy.
33334 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
33337 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
33339 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
33340 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33341 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
33342 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33343 (UNSPEC_NOREX_MEM): Remove definition.
33345 2017-04-21 Richard Biener <rguenther@suse.de>
33347 PR tree-optimization/79547
33348 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
33349 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
33350 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
33351 without any constraints.
33353 2017-04-21 Richard Biener <rguenther@suse.de>
33355 PR tree-optimization/78847
33356 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
33358 2017-04-21 Richard Biener <rguenther@suse.de>
33360 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
33361 (build_distinct_type_copy): Likewise.
33362 (build_variant_type_copy): Likewise.
33363 * tree.c (build_qualified_type): Pass down mem-stat info.
33364 (build_distinct_type_copy): Likewise.
33365 (build_variant_type_copy): Likewise.
33367 2017-04-21 Richard Biener <rguenther@suse.de>
33369 PR tree-optimization/80237
33370 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
33372 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
33373 for a simplified result.
33375 2016-04-21 Richard Biener <rguenther@suse.de>
33377 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
33378 sth as strict as a simple_iv but a chrec without symbols and an
33379 operand defined in the loop we are peeling (and not some subloop).
33380 (propagate_constants_for_unrolling): Propagate all constants.
33382 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33385 * config/i386/i386.c (print_reg): Remove assert for disalowed
33386 regno values, call output_operand_lossage instead.
33388 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33391 * config/i386/constraints.md (Yc): New register constraint.
33392 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
33393 Use Yc constraint for alternative 2 of operand 0. Remove
33394 preferred_for_speed attribute.
33396 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33398 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
33399 lastprivate clauses in SIMT case.
33401 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
33403 * doc/invoke.texi (-Wextra-semi): Document new warning option.
33405 2017-04-20 Richard Biener <rguenther@suse.de>
33407 PR tree-optimization/57796
33408 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
33409 as N scalar stores.
33410 (vect_model_load_cost): Cost gathers as N scalar loads.
33412 2017-04-20 Richard Biener <rguenther@suse.de>
33414 * ggc-page.c (ggc_allocated_p): Rename to ...
33415 (safe_lookup_page_table_entry): ... this and return the lookup
33417 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
33419 2017-04-20 Richard Biener <rguenther@suse.de>
33421 PR tree-optimization/80453
33422 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
33423 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
33424 from the conditions.
33425 (vn_phi_eq): Pass them down.
33426 (vn_phi_lookup): Record them.
33427 (vn_phi_insert): Likewise.
33429 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
33431 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
33432 uninitialized variable warning to avoid buffer overrun.
33434 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33437 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
33438 is suppressed for '{ 0 }' in C.
33440 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33442 * BASE-VER: Set to 8.0.0.
33444 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
33446 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
33447 priority .init_array and .fini_array section with SECTION_NOTYPE
33450 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33452 PR middle-end/80423
33453 * tree.h (build_array_type): Add typeless_storage default argument.
33454 * tree.c (type_cache_hasher::equal): Also compare
33455 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
33456 (build_array_type): Add typeless_storage argument, set
33457 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
33459 (build_nonshared_array_type): Adjust build_array_type_1 caller.
33460 (build_array_type): Likewise. Add typeless_storage argument.
33462 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33463 Jakub Jelinek <jakub@redhat.com>
33465 PR tree-optimization/80426
33466 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
33467 operation on symbolic operands, also compute the overflow for the
33468 invariant part when the operation degenerates into a negation.
33470 2017-04-19 Jakub Jelinek <jakub@redhat.com>
33473 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
33474 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
33477 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
33479 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
33482 * config/avr/avr.c (tree.h): Include it.
33483 (cgraph.h): Include it.
33484 (avr_encode_section_info): Don't warn for uninitialized progmem
33485 variable if it's just an alias.
33487 2017-04-19 Richard Biener <rguenther@suse.de>
33490 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
33491 when needed by AutoPGO.
33493 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
33496 * doc/lto.texi: Remove an extra 'that'.
33498 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
33500 PR rtl-optimization/80429
33501 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
33502 are only used in debug insns.
33504 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33505 Vladimir Makarov <vmakarov@redhat.com>
33507 * config/sparc/predicates.md (input_operand): Add comment. Return
33508 true for any memory operand when LRA is in progress.
33509 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
33511 2017-04-18 Jeff Law <law@redhat.com>
33514 * mips.md ({return,simple_return}_internal): Do not overwrite
33517 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33519 PR tree-optimization/80443
33520 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
33521 instead of adding 1, subtract -1 and similarly instead of subtracting
33524 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
33526 PR rtl-optimization/80357
33527 * haifa-sched.c (tmp_bitmap): New variable.
33528 (model_recompute): Handle duplicate use records.
33529 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
33530 (free_global_sched_pressure_data): Free it.
33532 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
33535 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33536 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33537 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33538 instead of SYSTEM_HEADER_DIR.
33540 2017-04-18 Jeff Law <law@redhat.com>
33542 PR middle-end/80422
33543 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
33544 predecessors after walking up the insn chain.
33546 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33549 * dwarf2out.c (modified_type_die): Try harder not to emit internal
33550 sizetype type into debug info.
33552 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33555 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
33556 unneeded test for TARGET_UPPER_REGS_SF.
33557 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
33559 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33562 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
33563 instead of gsi_after_labels.
33565 2017-04-18 Jeff Law <law@redhat.com>
33567 * regcprop.c (maybe_mode_change): Avoid creating copies of the
33571 2017-04-13 Jeff Law <law@redhat.com>
33572 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33573 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33575 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33578 * config/avr/avr.c (intl.h): Include it.
33579 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33581 2017-04-18 Martin Liska <mliska@suse.cz>
33583 PR gcov-profile/78783
33584 * gcov-tool.c (gcov_output_files): Validate that destination
33585 file is either removed by the tool or by a user.
33587 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33588 Guy Benyei <guybe@mellanox.com>
33590 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33591 block, and do not negate it, the stored id is already negative.
33593 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33595 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33597 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33600 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33601 masks of options that should be turned off if the VSX vector
33602 options are turned off.
33603 (OTHER_P8_VECTOR_MASKS): Likewise.
33604 (OTHER_VSX_VECTOR_MASKS): Likewise.
33605 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33606 rs6000_disable_incompatible_switches to validate no type switches
33608 (rs6000_incompatible_switch): New function to disallow turning on
33609 other vector options if -mno-vsx, -mno-power8-vector, or
33610 -mno-power9-vector are specified.
33612 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33614 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33616 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33618 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33619 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33620 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33621 (ARG_POINTER_CFA_OFFSET): Likewise.
33623 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33625 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33626 conditions to take advantage of various optimizations.
33628 2017-04-13 Jeff Law <law@redhat.com>
33630 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33631 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33632 (zero_extendsidi2_dext): Likewise.
33634 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33637 * fold-const.c (fold_ternary_loc): Revert
33638 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33641 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33643 PR rtl-optimization/80343
33644 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33645 new scratch pseudo.
33647 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33650 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33651 to ubsan_encode_value.
33653 2017-04-13 Jeff Law <law@redhat.com>
33655 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33656 appearing in DEBUG_INSNs.
33658 2017-04-13 Martin Liska <mliska@suse.cz>
33660 PR gcov-profile/80413
33661 * gcov-io.c (gcov_write_string): Copy to buffer just when
33662 allocated size is greater than zero.
33664 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33667 * dwarf2out.c (decls_for_scope): Ignore declarations of
33668 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33670 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33673 * ipa-visibility.c (non_local_p): Fix typos.
33674 (localize_node): When localizing symbol in same comdat group,
33675 dissolve the group only when we know external symbols are going
33677 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33679 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33681 PR tree-optimization/79390
33682 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33683 order does not result in usable sequence, retry with reversed operand
33689 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33690 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33691 op0 instead of fold_convert_loc (loc, type, arg0).
33693 2017-04-12 Jeff Law <law@redhat.com>
33695 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33696 has a delay slot in the generated code.
33698 * config/cris/cris.md (cris_preferred_reload_class): Return
33699 GENNONACR_REGS rather than GENERAL_REGS.
33701 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33704 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33705 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33706 signedness of the result type.
33708 2017-04-12 Richard Biener <rguenther@suse.de>
33709 Jeff Law <law@redhat.com>
33711 PR tree-optimization/80359
33712 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33713 trim stores to TARGET_MEM_REFs.
33715 2017-04-12 Richard Biener <rguenther@suse.de>
33717 PR tree-optimization/79390
33718 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33719 threading case even more.
33721 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33724 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33725 for quad_address_p for TImode, instead of just not indexed_address.
33727 2017-04-12 Richard Biener <rguenther@suse.de>
33728 Bernd Edlinger <bernd.edlinger@hotmail.de>
33730 PR middle-end/79671
33731 * alias.c (component_uses_parent_alias_set_from): Handle
33732 TYPE_TYPELESS_STORAGE.
33733 (get_alias_set): Likewise.
33734 * tree-core.h (tree_type_common): Add typeless_storage flag.
33735 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33736 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33737 for types containing members with TYPE_TYPELESS_STORAGE.
33738 (place_field): Likewise.
33739 (layout_type): Likewise for ARRAY_TYPE.
33740 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33741 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33742 TYPE_TYPELESS_STORAGE.
33743 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33745 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33748 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33749 first argument to type.
33751 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33755 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33756 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33757 (rs6000_expand_binop_builtin): Likewise.
33758 (rs6000_expand_ternop_builtin): Likewise; also add missing
33759 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33760 vshasigma built-ins.
33761 * doc/extend.texi: Document that vec_xxpermdi's third argument
33762 must be a constant.
33764 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33766 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33767 Use shift_const cost parameter when calculating gain of STV shifts.
33769 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33771 PR rtl-optimization/70478
33772 * lra-constraints.c (process_alt_operands): Check memory for
33773 disfavoring memory insn operand.
33775 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33777 PR middle-end/80100
33778 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33779 left shift in unsigned HOST_WIDE_INT type.
33781 PR rtl-optimization/80385
33782 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33783 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33786 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33787 if they have any depend clauses.
33789 2017-04-11 Martin Liska <mliska@suse.cz>
33792 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33793 * ipa-split.c (split_function): Create a local comdat symbol
33794 if caller is in a comdat group.
33796 2017-04-11 Martin Liska <mliska@suse.cz>
33799 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33802 2017-04-11 Martin Sebor <msebor@redhat.com>
33804 PR middle-end/80364
33805 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33806 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33808 (directive::set_width, directive::set_precision, format_character):
33810 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33813 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33816 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33817 conflict, set target->arch_name instead of target->cpu_name.
33819 2017-04-11 Richard Biener <rguenther@suse.de>
33821 PR tree-optimization/80374
33822 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
33823 build_zero_cst, remove fold_convertible_p check again.
33825 2017-04-11 Martin Liska <mliska@suse.cz>
33828 * ubsan.c (instrument_object_size): Do not instrument register
33831 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33834 * config/i386/i386-builtin-types.def
33835 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
33836 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
33837 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
33838 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
33839 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
33840 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
33841 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
33842 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
33843 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
33844 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
33845 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
33846 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
33847 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
33848 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
33849 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
33850 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
33851 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
33852 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
33853 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
33854 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
33855 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
33856 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
33857 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
33858 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
33859 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
33860 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
33861 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
33862 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
33863 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
33864 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
33865 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
33866 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
33867 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
33868 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
33869 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
33870 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
33871 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
33872 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
33873 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
33874 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
33875 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
33876 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
33877 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
33878 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
33879 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
33880 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
33882 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
33883 flag to second_arg_count, handle 4 argument function type _COUNT
33884 aliases, handle second_arg_count on second argument rather than last.
33886 2017-04-10 Jeff Law <law@redhat.com>
33888 PR tree-optimization/80374
33889 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
33890 record anything if we can not convert integer_zero_node to the
33893 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33896 * config/rs6000/rs6000.c (rs6000_option_override_internal):
33897 Enhance special handling given to the TARGET_P9_MINMAX option in
33898 relation to certain other options.
33900 2017-04-10 Bin Cheng <bin.cheng@arm.com>
33902 PR tree-optimization/80153
33903 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
33904 remove POINTER_PLUS_EXPR's base part directly, rather than through
33907 2017-04-10 Richard Biener <rguenther@suse.de>
33908 Bin Cheng <bin.cheng@arm.com>
33910 PR tree-optimization/80153
33911 * tree-affine.c (aff_combination_to_tree): Get base pointer from
33912 the first element of pointer type aff_tree. Build result expr in
33914 (add_elt_to_tree): Convert to type unconditionally. Remove other
33915 fold_convert calls.
33916 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
33917 (rewrite_use_nonlinear_expr): Check invariant using iv information.
33919 2017-04-10 Richard Biener <rguenther@suse.de>
33921 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33924 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
33926 PR rtl-optimization/70478
33927 * lra-constraints.c (curr_small_class_check): New.
33928 (update_and_check_small_class_inputs): New.
33929 (process_alt_operands): Update curr_small_class_check. Disfavor
33930 alternative insn memory operands. Check available regs for small
33933 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33936 * config/mips/mips.opt (-mvirt): Update description.
33937 * doc/invoke.texi (-mvirt): Likewise.
33939 2017-04-10 Richard Biener <rguenther@suse.de>
33941 PR middle-end/80362
33942 * fold-const.c (fold_binary_loc): Look at unstripped ops when
33943 looking for NEGATE_EXPR in -A / -B to A / B folding.
33945 2017-04-10 Martin Liska <mliska@suse.cz>
33947 PR gcov-profile/80224
33948 * gcov.c (print_usage): Fix usage string.
33949 (get_gcov_intermediate_filename): Remove.
33950 (output_gcov_file): Use both for normal and intermediate format.
33951 (generate_results): Do not initialize special file for
33952 intermediate format.
33954 2017-04-10 Richard Biener <rguenther@suse.de>
33956 PR tree-optimization/80304
33957 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
33960 2017-04-10 Nathan Sidwell <nathan@acm.org>
33963 * config/rs6000/rs6000.c (rs6000_vector_type): New.
33964 (rs6000_init_builtins): Use it.
33966 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33968 * config/arm/arm.md (<mrc>): Add mode to SET source.
33969 (<mrrc>): Likewise.
33971 2017-04-10 Richard Biener <rguenther@suse.de>
33973 PR middle-end/80344
33974 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
33976 2017-04-10 Jakub Jelinek <jakub@redhat.com>
33979 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
33980 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
33981 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
33982 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
33983 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
33984 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
33985 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
33986 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
33987 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
33988 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
33989 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
33990 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
33991 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
33992 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
33993 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
33994 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
33995 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
33996 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
33997 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
33998 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
33999 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34000 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34001 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34003 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34005 PR rtl-optimization/70478
34006 * lra-constraints.c: Reverse the last patch.
34008 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34010 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34011 Add comment for WCHAR_T.
34013 2017-04-08 Martin Liska <mliska@suse.cz>
34016 2017-04-07 Martin Liska <mliska@suse.cz>
34019 * ipa-split.c (split_function): Add function part to a same comdat
34022 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34025 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34027 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34029 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34030 (vec_pairing): ...to this.
34031 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34032 (rs6000_sched_init): Adjust for name change.
34033 (struct rs6000_sched_context): Likewise.
34034 (rs6000_init_sched_context): Likewise.
34035 (rs6000_set_sched_context): Likewise.
34037 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34043 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34045 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34046 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34047 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34049 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34051 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34053 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34055 PR rtl-optimization/70703
34056 * ira-color.c (update_conflict_hard_regno_costs): Use
34057 int64_t instead of HOST_WIDE_INT.
34059 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34061 PR rtl-optimization/70478
34062 * lra-constraints.c (process_alt_operands): Disfavor alternative
34063 insn memory operands.
34065 2017-04-07 Jeff Law <law@redhat.com>
34067 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
34068 CALL and NOTE_INSN_CALL_ARG_LOCATION.
34070 2017-04-07 Martin Liska <mliska@suse.cz>
34073 * config/aarch64/aarch64.c (aarch64_process_target_attr):
34074 Show error message instead of an ICE.
34076 2017-04-07 Martin Liska <mliska@suse.cz>
34079 * ipa-split.c (split_function): Add function part to a same comdat
34082 2017-04-07 Richard Biener <rguenther@suse.de>
34084 PR middle-end/80341
34085 * tree.c (get_unwidened): Also handle ! for_type case for
34087 * convert.c (do_narrow): Split out from ...
34088 (convert_to_integer_1): ... here. Do not pass final truncation
34089 type to get_unwidened for TRUNC_DIV_EXPR.
34091 2017-04-07 Richard Biener <rguenther@suse.de>
34093 * tree-affine.c (wide_int_ext_for_comb): Take type rather
34095 (aff_combination_const): Adjust.
34096 (aff_combination_scale): Likewise.
34097 (aff_combination_add_elt): Likewise.
34098 (aff_combination_add_cst): Likewise.
34099 (aff_combination_convert): Likewise.
34100 (add_elt_to_tree): Likewise. Remove unused argument.
34101 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
34103 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
34105 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
34107 * config/arm/arm.c (arm_default_short_enums): Use
34108 ARM_DEFAULT_SHORT_ENUMS.
34109 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
34111 2017-04-06 Jakub Jelinek <jakub@redhat.com>
34114 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
34115 members with redundant out-of-class redeclaration.
34117 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34120 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
34121 * config/i386/i386.md (*zero_extendsidi2):
34122 Add (?*x,*x) and (?*v,*v) alternatives.
34124 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34127 * config/i386/i386.c (ix86_expand_builtin)
34128 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
34129 mode from insn data. Convert operands to insn operand mode.
34130 Copy operands that don't satisfy insn predicate to a register.
34132 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
34134 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
34137 2017-04-06 Richard Biener <rguenther@suse.de>
34139 PR tree-optimization/80334
34140 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
34141 preserve alignment of accesses.
34143 2017-04-06 Richard Biener <rguenther@suse.de>
34145 PR tree-optimization/80262
34146 * tree-sra.c (build_ref_for_offset): Preserve address-space
34148 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
34149 Drop useless address-space information on MEM_REF offsets.
34151 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
34153 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
34155 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34157 PR rtl-optimization/70703
34158 * ira-color.c (update_conflict_hard_regno_costs): Use
34159 HOST_WIDE_INT instead of long.
34161 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
34164 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
34165 not defined for x86_64 target. Add -mmmx target option when __SSE2__
34167 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
34168 for x86_64 target. Handle -m3dnowa option.
34170 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34172 PR rtl-optimization/70703
34173 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
34174 (update_conflict_hard_regno_costs): Use long instead of unsigned
34175 arithmetic for cost calculation.
34177 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34178 Bernd Edlinger <bernd.edlinger@hotmail.de>
34181 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
34184 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
34187 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
34188 ptr_mode with Pmode throughout.
34189 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
34190 into probe_stack_range and use DImode.
34192 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
34195 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
34196 call_eh_return is true.
34198 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34200 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
34201 Initialize last_match_fntype_index.
34203 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34206 * tree-nvr.c: Include internal-fn.h.
34207 (pass_return_slot::execute): Ignore internal calls without
34210 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34211 Richard Biener <rguenther@suse.de>
34214 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
34215 captures used multiple times, except for the last use.
34216 * generic-match-head.c: Include gimplify.h.
34218 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34220 PR tree-optimization/79390
34221 * target.h (struct noce_if_info): Declare.
34222 * targhooks.h (default_noce_conversion_profitable_p): Declare.
34223 * target.def (noce_conversion_profitable_p): New target hook.
34224 * ifcvt.h (struct noce_if_info): New type, moved from ...
34225 * ifcvt.c (struct noce_if_info): ... here.
34226 (noce_conversion_profitable_p): Renamed to ...
34227 (default_noce_conversion_profitable_p): ... this. No longer
34229 (noce_try_store_flag_constants, noce_try_addcc,
34230 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
34231 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
34232 instead of noce_conversion_profitable_p.
34233 * config/i386/i386.c: Include ifcvt.h.
34234 (ix86_option_override_internal): Don't override
34235 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
34236 (ix86_noce_conversion_profitable_p): New function.
34237 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
34238 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
34239 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
34240 * doc/tm.texi: Regenerated.
34242 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34244 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
34247 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
34250 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
34251 instructions for small multiply cores.
34253 2017-04-04 Jeff Law <law@redhat.com>
34255 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
34257 (mips_expand_vec_perm_const): Initialize elements in orig_perm
34258 that are not set by the loop over the elements.
34260 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34263 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
34264 int mode, convert_modes it to mode as unsigned, otherwise use
34265 lowpart_subreg to mode rather than SImode.
34266 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
34267 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
34268 Use DImode instead of SImode for the shift count operand.
34269 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
34272 2017-04-04 Richard Biener <rguenther@suse.de>
34274 PR middle-end/80281
34275 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
34276 arithmetic done for the negate or the plus. Simplify.
34277 (A - (-B) -> A + B): Likewise.
34278 * fold-const.c (split_tree): Make sure to not negate pointers.
34280 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
34282 PR rtl-optimization/60818
34283 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
34284 a compare of comparisons with the thing compared if this results
34285 in a different machine mode.
34287 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
34289 * alias.c (base_alias_check): Fix typo in comment.
34290 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
34291 * cgraphunit.c (symbol_table::compile): Likewise.
34292 * collect2.c (maybe_run_lto_and_relink): Likewise.
34293 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
34294 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
34295 * config/avr/avr.c (avr_map_op_t): Likewise.
34296 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
34297 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
34298 * config/epiphany/epiphany.md (movcc): Likewise.
34299 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
34300 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
34302 * config/mips/mips.c (mips_save_restore_reg): Likewise.
34303 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
34304 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
34305 * config/sh/sh.c (sh_rtx_costs): Likewise.
34306 * fold-const.c (fold_truth_andor): Likewise.
34307 * genautomata.c (collapse_flag): Likewise.
34308 * gengtype.h (struct type::u::s): Likewise.
34309 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
34310 * input.c (FORMAT_AMOUNT): Likewise.
34311 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
34312 (known_aggs_to_agg_replacement_list): Likewise.
34313 * ipa-inline-analysis.c: Likewise.
34314 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
34315 * ipa-polymorphic-call.c
34316 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
34317 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
34318 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
34320 * modulo-sched.c (apply_reg_moves): Likewise.
34321 * omp-expand.c (build_omp_regions_1): Likewise.
34322 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
34323 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
34324 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
34325 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
34326 * value-prof.c: Likewise.
34327 * var-tracking.c (val_reset): Likewise.
34329 2017-04-03 Richard Biener <rguenther@suse.de>
34331 PR tree-optimization/80275
34332 * fold-const.c (split_address_to_core_and_offset): Handle
34335 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
34337 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
34338 descriptors is at least equal to that of functions.
34340 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34342 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
34344 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34347 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
34348 (mov<IMOD4:mode>): New expander.
34349 (*mov<IMOD4:mode>_internal): New insn and split pattern.
34351 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
34353 PR rtl-optimization/79405
34354 * fwprop.c (propagations_left): New variable.
34355 (forward_propagate_into): Decrement it.
34356 (fwprop_init): Initialize it.
34357 (fw_prop): If the variable has reached zero, stop propagating.
34358 (fwprop_addr): Ditto.
34360 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34363 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
34364 a FUNCTION_DECL, pass it as decl instead of origin to
34367 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
34369 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
34372 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
34375 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
34376 TARGET_VSX_SMALL_INTEGER.
34378 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34380 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
34381 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
34383 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34385 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
34386 extraction from odd-numbered MSA register.
34388 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34390 PR middle-end/80173
34391 * expmed.c (store_bit_field_1): Don't attempt to create
34392 a word subreg out of hard registers wider than word if they
34393 have HARD_REGNO_NREGS of 1 for their mode.
34395 PR middle-end/80163
34396 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
34397 conversions to integer types wider than word and pointer.
34400 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
34401 (rtx_equal_for_cselib_p): Pass 0 to it.
34402 * cselib.c (cselib_hasher::equal): Likewise.
34403 (rtx_equal_for_cselib_1): Add depth argument. If depth
34404 is 128, don't look up VALUE locs and punt. Increment
34405 depth in recursive calls when walking VALUE locs.
34407 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
34409 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
34410 (make_gcov_file_name): Use the canonical path name for generating
34412 (read_line): Fix handling of files with ascii null bytes.
34414 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
34416 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
34417 to initialise a vector register instead
34418 of using a const_int.
34420 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34422 PR translation/80189
34423 * gimplify.c (omp_default_clause): Use %qs instead of %s in
34424 diagnostic messages.
34426 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
34429 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
34430 (dfp_diex_<mode>): Update mode of operand 1.
34431 * doc/extend.texi (dxex, dxexq): Document change to return type.
34432 (diex, diexq): Document change to argument type.
34434 2017-03-30 Martin Jambor <mjambor@suse.cz>
34437 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
34438 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
34439 it reflects the signature changes performed at the callee side.
34440 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
34441 to cgraph_build_function_type_skip_args.
34442 (build_function_decl_skip_args): Adjust call to the above function.
34444 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34447 * config/i386/sse.md
34448 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
34449 register as dest whenever it is a MEM not rtx_equal_p to the
34450 corresponding dup operand, and when forcing into reg move the
34451 reg into the memory afterwards.
34452 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
34453 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
34454 for the force_reg mode.
34455 (avx512vl_vextractf128<mode>): Use register as dest either
34456 always when a MEM, or when it is a MEM not rtx_equal_p to the
34457 corresponding dup operand, or even not when it is a CONST_VECTOR
34458 depending on the mode and lo vs. hi.
34459 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
34461 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
34462 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
34463 Likewise. Require that operands[2] is even.
34464 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
34465 Remove extraneous parens. Require that operands[2] is a multiple
34467 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
34468 operands[0] is a MEM if <mask_applied>, the predicates/constraints
34469 disallow memory then.
34471 2017-03-30 Richard Biener <rguenther@suse.de>
34473 PR tree-optimization/77498
34474 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
34475 to non-constants over backedges.
34477 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
34479 PR rtl-optimization/80233
34480 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
34481 as last_combined_insn. Do not test for BARRIER_P separately.
34483 2017-03-29 Andreas Schwab <schwab@suse.de>
34486 * calls.c (prepare_call_address): Convert funexp to Pmode before
34487 copying to temp reg.
34489 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34491 PR tree-optimization/80158
34492 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
34493 Handle possible future case of more than one alternate
34495 (replace_rhs_if_not_dup): Likewise.
34496 (replace_one_candidate): Likewise.
34498 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
34500 PR rtl-optimization/80193
34501 * ira.c (ira): Do not check allocation for LRA.
34503 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
34505 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
34506 (nvptx_output_simt_exit): Declare.
34507 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
34508 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
34509 (init_softstack_frame): Move initialization of crtl->is_leaf to...
34510 (nvptx_declare_function_name): ...here. Emit declaration of local
34511 memory space buffer for omp_simt_enter insn.
34512 (nvptx_output_unisimt_switch): New.
34513 (nvptx_output_softstack_switch): New.
34514 (nvptx_output_simt_enter): New.
34515 (nvptx_output_simt_exit): New.
34516 * config/nvptx/nvptx.h (struct machine_function): New fields
34517 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
34518 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
34519 (UNSPECV_SIMT_EXIT): Ditto.
34520 (omp_simt_enter_insn): New insn.
34521 (omp_simt_enter): New expansion.
34522 (omp_simt_exit): New insn.
34523 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
34525 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
34526 (expand_GOMP_SIMT_ENTER_ALLOC): New.
34527 (expand_GOMP_SIMT_EXIT): New.
34528 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
34529 (GOMP_SIMT_ENTER_ALLOC): Ditto.
34530 (GOMP_SIMT_EXIT): Ditto.
34531 * target-insns.def (omp_simt_enter): New insn.
34532 (omp_simt_exit): Ditto.
34533 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
34535 (lower_rec_simd_input_clauses): Implement SIMT privatization.
34536 (lower_rec_input_clauses): Likewise.
34537 (lower_lastprivate_clauses): Handle SIMT privatization.
34539 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
34540 (ompdevlow_adjust_simt_enter): New.
34541 (find_simtpriv_var_op): New.
34542 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
34543 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
34545 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
34546 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
34547 (copy_decl_for_dup_finish): Ditto.
34549 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
34551 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
34554 * config/i386/i386.c (ix86_option_override_internal): Always
34555 allow -mpreferred-stack-boundary=3 for 64-bit targets.
34557 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34559 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
34561 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34563 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
34564 mark new edge's irreducible flag accordign to it.
34565 (vect_do_peeling): Check loop preheader edge's irreducible flag
34566 and pass it to function slpeel_add_loop_guard.
34568 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
34570 PR tree-optimization/80218
34571 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34572 Update block frequencies and counts.
34574 2017-03-28 Richard Biener <rguenther@suse.de>
34576 PR tree-optimization/78644
34577 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34578 of a simplification result we may not use it at all.
34580 2017-03-28 Richard Biener <rguenther@suse.de>
34583 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34584 without arguments, generate default definition of a SSA name.
34586 2017-03-28 Richard Biener <rguenther@suse.de>
34588 PR middle-end/80222
34589 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34590 TYPE_REF_CAN_ALIAS_ALL references.
34591 * fold-const.c (fold_indirect_ref_1): Likewise.
34593 2017-03-28 Martin Liska <mliska@suse.cz>
34596 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34597 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34599 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34600 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34602 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34603 (EXTRA_SPECS): Define.
34604 (SUBTARGET_EXTRA_SPECS): Likewise.
34605 (SUBTARGET_CPP_SPEC): Likewise.
34606 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34607 SUBTARGET_EXTRA_SPECS.
34608 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34610 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34612 * config/arc/simdext.md (vst64_insn): Update pattern.
34613 (vld32wh_insn): Likewise.
34614 (vld32wl_insn): Likewise.
34615 (vld64_insn): Likewise.
34616 (vld32_insn): Likewise.
34618 2017-03-28 Marek Polacek <polacek@redhat.com>
34621 * fold-const.c (fold_comparison): Use protected_set_expr_location
34622 instead of SET_EXPR_LOCATION.
34624 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34626 * tree.c (add_expr): Avoid name lookup warning.
34628 2017-03-27 Jeff Law <law@redhat.com>
34630 PR tree-optimization/80216
34631 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34632 function name. Limit recursion depth.
34633 (record_temporary_equivalences): Corresponding changes.
34635 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34637 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34640 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34643 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34645 * cfgcleanup.c (reg_note_cfa_p): New array.
34646 (insns_have_identical_cfa_notes): New function.
34647 (old_insns_match_p): Don't cross-jump in between /f
34648 and non-/f instructions. If both i1 and i2 are frame related,
34649 verify all CFA notes, their order and content.
34651 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34654 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34655 HImode and SImode with zero extend to DImode to one insn.
34656 (bswap<mode>2_extenddi): Likewise.
34657 (bswapsi2_extenddi): Likewise.
34658 (bswaphi2_extendsi): Likewise.
34659 (bswaphi2): Combine bswap HImode and SImode into one insn.
34660 Separate memory insns from swapping register.
34661 (bswapsi2): Likewise.
34662 (bswap<mode>2): Likewise.
34663 (bswaphi2_internal): Delete, no longer used.
34664 (bswapsi2_internal): Likewise.
34665 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34666 store, and gpr<-gpr swap insns.
34667 (bswap<mode>2_store): Likewise.
34668 (bswaphi2_reg): Register only splitter, combine with the splitter.
34669 (bswaphi2 splitter): Likewise.
34670 (bswapsi2_reg): Likewise.
34671 (bswapsi2 splitter): Likewise.
34672 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34673 the insns into load, store, and register/register insns.
34674 (bswapdi2_ldbrx): Likewise.
34675 (bswapdi2_load): Likewise.
34676 (bswapdi2_store): Likewise.
34677 (bswapdi2_reg): Likewise.
34679 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34681 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34682 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34684 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34687 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34689 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34690 special handling for target option conflicts between dform
34691 options (-mpower9-dform, -mpower9-dform-vector,
34692 -mpower9-dform-scalar) and -mno-direct-move.
34694 2017-03-27 Richard Biener <rguenther@suse.de>
34696 PR tree-optimization/80181
34697 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34699 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34701 * config/arc/predicates.md (move_double_src_operand): Replace the
34702 call to move_double_src_operand with a call to address_operand.
34704 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34706 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34707 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34708 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34710 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34712 * config/arc/predicates.md (long_immediate_loadstore_operand):
34713 Consider scaled addresses cases.
34715 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34717 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34718 restored when in interrupt.
34719 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34720 doesn't have delay slot.
34722 2017-03-27 Richard Biener <rguenther@suse.de>
34725 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34726 inlined thunk clones.
34728 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34731 * asan.c (instrument_derefs): Copy over last operand from
34732 original COMPONENT_REF to the new COMPONENT_REF with
34733 DECL_BIT_FIELD_REPRESENTATIVE.
34734 * ubsan.c (instrument_object_size): Likewise.
34736 2017-03-27 Richard Biener <rguenther@suse.de>
34738 PR tree-optimization/80170
34739 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34740 sure DR/SCEV didnt fold in constants we do not see when looking
34741 at the reference base alignment.
34743 2017-03-27 Richard Biener <rguenther@suse.de>
34745 PR middle-end/80171
34746 * gimple-fold.c (fold_ctor_reference): Properly guard against
34747 NULL return value from canonicalize_constructor_val.
34749 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34752 * config/i386/i386.c (ix86_expand_builtin)
34753 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34754 flags reg setting and flags reg using instructions.
34755 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34756 clobbering instructions to zero extend op2.
34758 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34760 * doc/install.texi (Configuration) <--with-aix-soname>:
34761 Update link to AIX ld.
34763 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34765 PR rtl-optimization/80160
34766 PR rtl-optimization/80159
34767 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34768 reg_alternate_class into account.
34770 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34773 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34774 to consider in curr_insn_transform.
34776 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34778 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34779 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34780 and emit_mode_inner.
34782 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34784 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34785 argument to the overloaded builtin variants. Use the new flag to
34786 deprecate certain builtin variants.
34787 * config/s390/s390-builtin-types.def: Add new builtin types.
34788 * config/s390/s390-builtins.h: Support new flags field for
34789 overloaded builtins.
34790 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34791 (s390_macro_to_expand): Enable vector float data type.
34792 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34793 builtins by incrementing the __VEC__ version number.
34794 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34796 (s390_resolve_overloaded_builtin): Emit error messages depending
34797 on the builtin flags.
34798 * config/s390/s390.c (s390_expand_builtin): Support additional
34799 flags argument. Change error message to match the messages
34800 emitted in s390-c.c.
34801 * config/s390/s390.md: New UNSPEC_* constants.
34802 (op_type): Add new instruction types.
34803 * config/s390/vecintrin.h: Add new builtins and test data class
34805 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34806 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34807 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34808 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34809 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34810 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34812 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34813 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34814 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34815 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34817 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34818 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34819 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34820 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
34822 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
34823 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
34824 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
34825 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
34826 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
34827 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
34828 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
34830 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
34831 ("vec_scatter_element<V_HW_4:mode>_DI")
34832 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
34833 ("vec_fpint<mode>", "vflls")
34834 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
34835 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
34836 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
34837 ("*vec_cmphe<mode>_cc"): ... these.
34839 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
34840 mode constant instead of magic value.
34842 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34844 * config/s390/s390.c (s390_expand_vec_compare): Support other
34845 vector floating point modes than just V2DF.
34846 (s390_expand_vcond): Likewise.
34847 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
34848 (s390_cannot_change_mode_class): Prevent mode changes between TF
34849 and V1TF in vector registers.
34850 * config/s390/s390.md (DF, SF): New mode attributes.
34851 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
34852 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
34853 SFmode support for VRs.
34854 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
34856 (VFT, VF_HW): New mode iterators.
34857 (vw, sdx): New mode attributes.
34858 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
34859 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
34860 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
34861 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
34862 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
34863 also the new vector floating point modes. Renaming to ...
34865 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
34866 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
34867 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
34868 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
34869 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
34870 ("vec_unordered<mode>"): ... these.
34872 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
34873 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
34874 ("*vec_extendv2df"): New insn definitions.
34876 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34878 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
34879 ("mulditi3_2", "*muldi3_sign"): New patterns.
34880 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
34881 rename the pattern definition.
34883 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34885 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
34887 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
34889 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34891 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
34892 instruction if possible.
34893 * config/s390/vector.md (vec_halfnumelts): New mode
34895 ("*vec_vllezlf<mode>"): New pattern.
34897 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34899 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
34900 ("popcountv4si2", "popcountv2di2"): Rename to ...
34901 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
34902 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
34904 ("popcount<mode>2_vxe"): New pattern.
34906 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34908 * common/config/s390/s390-common.c (processor_flags_table): Add
34910 * config.gcc: Add arch12.
34911 * config/s390/driver-native.c (s390_host_detect_local_cpu):
34912 Default to arch12 for unknown CPU model numbers.
34913 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
34914 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
34915 PROCESSOR_max sanity check.
34916 * config/s390/s390-opts.h (enum processor_type): Add
34918 * config/s390/s390.c (processor_table): Add arch12.
34919 (s390_expand_builtin): Add check for B_VXE flag.
34920 (s390_issue_rate): Add PROCESSOR_ARCH12.
34921 (s390_get_sched_attrmask): Likewise.
34922 (s390_get_unit_mask): Likewise.
34923 (s390_sched_score): Enable z13 scheduling for arch12.
34924 (s390_sched_reorder): Likewise.
34925 (s390_sched_variable_issue): Likewise.
34926 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
34928 (s390_tune_attr): Use z13 scheduling also for arch12.
34929 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
34930 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
34931 (TARGET_VXE_P): New macros.
34932 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
34933 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
34934 * config/s390/s390.opt: Add arch12 as processor_type.
34936 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34938 * config/s390/s390.md
34939 ("fixuns_truncdddi2", "fixuns_trunctddi2")
34940 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
34941 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
34943 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
34944 Rename expanders to ...
34946 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
34947 ("fixuns_truncdddi2_emu"): ... these.
34949 ("fixuns_trunc<mode>si2_emu"): New expander.
34951 ("*fixuns_truncdfdi2_z13"): Rename to ...
34952 ("*fixuns_truncdfdi2_vx"): ... this.
34954 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34956 * config/s390/2964.md: Remove the single element vector compare
34957 instructions which are no longer used.
34958 * config/s390/s390.c (s390_select_ccmode): Remove handling of
34960 (s390_canonicalize_comparison): Remove handling of DFmode
34962 (s390_expand_vec_compare_scalar): Remove function.
34963 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
34964 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
34966 ("*cmp<mode>_ccs"): Add wfcdb instruction.
34968 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34970 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
34972 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
34973 will anyway by matched by mov<mode>_64dfp.
34975 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34977 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
34978 vlef/vstef. Add missing operand to vleif.
34980 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34982 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
34983 pair for all vector types with 64 bit elements.
34984 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
34985 * config/s390/vector.md (V_HW_64): ... here.
34986 (V_128_NOSINGLE): New mode iterator.
34987 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
34988 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
34989 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
34990 ("*vec_load_pairv2di"): Change to ...
34991 ("*vec_load_pair<mode>"): ... this one.
34993 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34995 * config/s390/constraints.md: Add comments.
34996 (jKK): Reject element sizes > 8 bytes.
34997 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
34999 * config/s390/s390.md: Add the s_operand checks formerly in
35000 s390_split_ok_p to various splitters where they are still
35002 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35003 for 128 bit vectors. Plus two splitters.
35005 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35007 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35010 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35013 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35014 error if the boundary argument is not constant.
35016 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35018 PR rtl-optimization/80112
35019 * loop-doloop.c (doloop_condition_get): Don't check condition
35020 if cmp isn't SET with IF_THEN_ELSE src.
35022 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35024 PR tree-optimization/80158
35025 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35026 replacing a candidate statement, also replace it for the
35027 candidate's alternate interpretation.
35028 (replace_rhs_if_not_dup): Likewise.
35029 (replace_one_candidate): Likewise.
35031 2017-03-24 Richard Biener <rguenther@suse.de>
35033 PR tree-optimization/80167
35034 * graphite-isl-ast-to-gimple.c
35035 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35037 (translate_isl_ast_to_gimple::get_rename): Likewise.
35039 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35041 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35042 handling of certain combinations of target options, including the
35043 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35044 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35046 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35049 * config/arm/arm.md (*load_multiple): Add reload_completed to
35050 matching condition.
35052 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35053 Richard Biener <rguenth@suse.de>
35055 PR tree-optimization/79908
35056 PR tree-optimization/80136
35057 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35058 been cast away, gimplify_and_add suffices.
35060 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35062 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35064 2017-03-23 Richard Biener <rguenther@suse.de>
35066 PR tree-optimization/80032
35067 * gimplify.c (gimple_push_cleanup): Forced unconditional
35068 cleanups still have to go to the conditional_cleanups
35071 2017-03-22 Jakub Jelinek <jakub@redhat.com>
35073 PR tree-optimization/80072
35074 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
35076 (next_operand_entry_id): Change type to unsigned int.
35077 (sort_by_operand_rank): Make sure to return the right return value
35078 even if unsigned fields are bigger than INT_MAX.
35079 (struct oecount): Change cnt and id type to unsigned int.
35080 (oecount_hasher::equal): Formatting fix.
35081 (oecount_cmp): Make sure to return the right return value
35082 even if unsigned fields are bigger than INT_MAX.
35083 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
35086 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
35087 TREE_READONLY on result if writing it more than once.
35090 * doc/invoke.texi (-fsanitize=thread): Document that with
35091 -fnon-call-exceptions atomics are not able to throw
35095 * tsan.c: Include tree-eh.h.
35096 (instrument_builtin_call): Call maybe_clean_eh_stmt or
35097 maybe_clean_or_replace_eh_stmt where needed.
35098 (instrument_memory_accesses): Add cfg_changed argument.
35099 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
35100 if it returned true.
35101 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
35103 PR rtl-optimization/63191
35104 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
35105 wrapper function, moved the whole old content into ...
35106 (ix86_delegitimize_address_1): ... this. New inline function.
35107 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
35108 true as last argument instead of ix86_delegitimize_address.
35110 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35112 * config/aarch64/aarch64.c (generic_branch_cost): Copy
35113 cortexa57_branch_cost.
35115 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35117 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
35119 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35122 * doc/md.texi (Constraints): Document wA constraint.
35123 * config/rs6000/constraints.md (wA): New.
35124 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
35125 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
35126 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
35127 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
35129 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
35132 * gimplify.c (is_oacc_declared): New function.
35133 (oacc_default_clause): Use it to set default flags for acc declared
35134 variables inside parallel regions.
35135 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
35136 declared variables.
35137 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
35138 declare attribute to any decl as necessary.
35140 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
35143 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
35144 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
35145 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
35146 (arm_arch_lpae): This.
35147 * config/arm/arm.c (arm_arch7ve): Rename into ...
35148 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
35149 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
35152 2017-03-22 Martin Liska <mliska@suse.cz>
35155 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
35156 error message instead of an ICE.
35158 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35160 * doc/extend.texi (6.11 Additional Floating Types): Revise.
35162 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
35164 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35166 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35169 2017-03-21 Martin Sebor <msebor@redhat.com>
35171 * doc/extend.texi: Use "cannot" instead of "can't."
35172 * doc/hostconfig.texi: Same.
35173 * doc/install.texi: Same.
35174 * doc/invoke.texi: Same.
35175 * doc/loop.texi: Same.
35176 * doc/md.texi: Same.
35177 * doc/objc.texi: Same.
35178 * doc/rtl.texi: Same.
35179 * doc/tm.texi: Same.
35180 * doc/tm.texi.in: Same.
35181 * doc/trouble.texi: Same.
35183 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
35186 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
35187 (collect_checksum_attributes): Set it.
35188 (die_checksum_ordered): Use it.
35190 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35192 PR tree-optimization/79908
35193 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
35194 change: For a VA_ARG whose LHS has been cast away, use
35195 force_gimple_operand to construct the side effects.
35197 2017-03-21 David Malcolm <dmalcolm@redhat.com>
35199 PR translation/80001
35200 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
35201 more amenable to translation.
35202 (oacc_loop_auto_partitions): Likewise.
35204 2017-03-21 Marek Polacek <polacek@redhat.com>
35205 Martin Sebor <msebor@redhat.com>
35207 PR tree-optimization/80109
35208 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
35209 on INTEGRAL_TYPE_P.
35211 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35212 Segher Boessenkool <segher@kernel.crashing.org>
35215 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
35216 check reg_used_between_p between insn and one of succ or succ2
35217 depending on if succ is artificial insn not inserted into insn
35220 2017-03-21 Martin Liska <mliska@suse.cz>
35222 PR gcov-profile/80081
35223 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
35224 * doc/gcc.texi: Include gcov-dump stuff.
35225 * doc/gcov-dump.texi: New file.
35227 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
35229 PR rtl-optimization/79150
35230 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
35231 conditional jump, if the jump is the last insn of the loop.
35233 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35234 Richard Biener <rguenth@suse.de>
35236 PR tree-optimization/79908
35237 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35238 been cast away, use force_gimple_operand to construct the side
35241 2017-03-21 Martin Liska <mliska@suse.cz>
35243 PR libfortran/79956
35244 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
35247 2017-03-21 Brad Spengler <spender@grsecurity.net>
35250 * plugin.c (htab_hash_plugin): New function.
35251 (add_new_plugin): Use it and adjust.
35252 (parse_plugin_arg_opt): Adjust.
35253 (init_one_plugin): Likewise.
35255 2017-03-21 Richard Biener <rguenther@suse.de>
35257 PR tree-optimization/80032
35258 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
35259 if set force the cleanup to happen unconditionally.
35260 (gimplify_target_expr): Push inserted clobbers with force_uncond
35261 to avoid them being removed by control-dependent DCE.
35263 2017-03-21 Richard Biener <rguenther@suse.de>
35265 PR tree-optimization/80122
35266 * tree-inline.c (copy_bb): Do not expans va-arg packs or
35267 va_arg_pack_len when the inlined call stmt requires pack
35269 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
35271 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35274 * tsan.c (instrument_builtin_call): If the memory model argument
35275 is not a constant, assume it is valid.
35278 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
35281 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35283 PR rtl-optimization/79910
35284 * combine.c (can_combine_p): Do not allow combining an I0 or I1
35285 if its dest is used by an insn before I2 (other than the combined
35286 insns themselves, which are properly handled already).
35288 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35291 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35293 * combine.c (record_used_regs): New static function.
35294 (try_combine): Handle situations where there is an additional
35295 instruction between I2 and I3 which needs to have a LOG_LINK
35299 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35301 * combine.c (try_combine): Delete redundant i1 test. Call
35302 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35304 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35307 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
35308 alternatives 13/14.
35310 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35312 PR tree-optimization/80054
35313 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
35314 the optimization if a PHI or any of its arguments is not dominated
35315 by the candidate's basis. Use gphi* rather than gimple* as
35317 (replace_profitable_candidates): Clean up a gimple* variable that
35318 should be a gphi* variable.
35320 2017-03-20 Martin Sebor <msebor@redhat.com>
35323 * doc/extend.texi (attribute constructor): Document present limitation.
35325 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
35328 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
35329 __POWER9_VECTOR__ #ifdef control, change template definition to
35330 use Power9-specific built-in function.
35331 (vec_any_eq): Likewise.
35332 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
35333 to control outcomes from this test.
35334 (vector_ae_<mode>p): For VEC_F modes, likewise.
35336 2017-03-20 Ian Lance Taylor <iant@google.com>
35338 * config/i386/i386.c (ix86_function_regparm): Save an extra
35339 register for -fsplit-stack with DECL_STATIC_CHAIN.
35341 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35344 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
35345 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
35347 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35349 * config/riscv/riscv.c (riscv_print_operand): Use "fence
35351 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
35354 2017-03-20 Marek Polacek <polacek@redhat.com>
35357 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
35359 2017-03-20 Richard Biener <rguenther@suse.de>
35361 PR tree-optimization/80113
35362 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
35363 allocate extra SSA name for PHI def.
35364 (add_close_phis_to_outer_loops): Likewise.
35365 (add_close_phis_to_merge_points): Likewise.
35366 (copy_loop_close_phi_args): Likewise.
35367 (copy_cond_phi_nodes): Likewise.
35369 2017-03-20 Martin Liska <mliska@suse.cz>
35371 PR middle-end/79753
35372 * tree-chkp.c (chkp_build_returned_bound): Do not build
35373 returned bounds for a LHS that's not a BOUNDED_P type.
35375 2017-03-20 Martin Liska <mliska@suse.cz>
35379 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
35380 COMPLEX_CST and VECTOR_CST.
35382 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35385 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
35386 target operand. A new splitter adds the clobber statement in case
35387 the target operand is dead anyway.
35389 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
35391 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
35392 to age-old versions of binutils and glibc.
35394 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
35396 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
35398 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35400 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
35402 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35404 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
35405 requirement for binutils 2.13.
35407 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35409 * combine.c (try_combine): Delete redundant i1 test. Call
35410 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35412 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
35414 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
35415 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
35417 <riscv64-*-elf>: Re-arrange section
35418 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
35419 <riscv32-*-linux>: Likewise.
35420 <riscv64-*-elf>: Likewise
35421 <riscv64-*-linux>: Likewise.
35423 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
35426 * aarch64.opt(verbose-cost-dump): Fix typo.
35428 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
35431 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
35432 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
35434 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35436 * reload.c (find_reloads): When reloading a nonoffsettable address,
35437 use RELOAD_OTHER for it and its address reloads.
35439 PR rtl-optimization/79910
35440 * combine.c (record_used_regs): New static function.
35441 (try_combine): Handle situations where there is an additional
35442 instruction between I2 and I3 which needs to have a LOG_LINK
35445 2017-03-17 Jeff Law <law@redhat.com>
35447 PR tree-optimization/71437
35448 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
35449 conditional in the hash table first.
35450 (vrp_dom_walker::before_dom_children): Extract condition from
35451 ASSERT_EXPR. Record condition, its inverion and any implied
35452 conditions as well.
35454 2017-03-17 Marek Polacek <polacek@redhat.com>
35455 Markus Trippelsdorf <markus@trippelsdorf.de>
35457 PR tree-optimization/80079
35458 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
35461 2017-03-17 Richard Biener <rguenther@suse.de>
35463 PR middle-end/80075
35464 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
35465 Properly verify the LHS before the RHS possibly claims to be
35467 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
35470 2017-03-17 Martin Jambor <mjambor@suse.cz>
35472 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
35473 (List of -O2 options): Likewise.
35474 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
35477 2017-03-17 Tom de Vries <tom@codesourcery.com>
35479 * gcov-dump.c (print_usage): Print bug_report_url.
35481 2017-03-17 Richard Biener <rguenther@suse.de>
35483 PR middle-end/80050
35484 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
35485 (parser::peek): Likewise.
35487 2017-03-17 Richard Biener <rguenther@suse.de>
35489 PR tree-optimization/80048
35490 * sese.c (free_sese_info): Properly release rename_map and
35491 copied_bb_map elements.
35493 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
35495 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
35496 Add linked-list forward and backlinks. Insert on
35497 construction, remove on destruction.
35498 (class pass_store_merging): Add m_stores_head field.
35499 (pass_store_merging::terminate_and_process_all_chains):
35500 Iterate over m_stores_head list.
35501 (pass_store_merging::terminate_all_aliasing_chains):
35503 (pass_store_merging::execute): Check for debug stmts first.
35504 Push new chains onto the m_stores_head stack.
35506 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
35509 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
35510 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
35511 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
35513 2017-03-16 Jeff Law <law@redhat.com>
35515 PR tree-optimization/71437
35516 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
35517 member function. Implementation moved into after_dom_children
35518 member function and into the threader's thread_outgoing_edges
35520 (dom_opt_dom_walker::after_dom_children): Simplify by moving
35521 some code into new thread_outgoing_edges.
35522 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
35523 definition. Simplify marker handling (do it here). Assume we always
35524 have the available expression and the const/copies tables.
35525 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
35527 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
35528 * tree-vrp.c (equiv_stack): No longer file scoped.
35529 (vrp_dom_walker): New class.
35530 (vrp_dom_walker::before_dom_children): New member function.
35531 (vrp_dom_walker::after_dom_children): Likewise.
35532 (identify_jump_threads): Setup domwalker. Use it rather than
35533 walking edges in a random order by hand. Simplify setup/finalization.
35534 (finalize_jump_threads): Remove.
35535 (vrp_finalize): Do not call identify_jump_threads here.
35536 (execute_vrp): Do it here instead and call thread_through_all_blocks
35539 PR tree-optimization/71437
35540 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
35542 (simplify_stmt_for_jump_threading): Add basic_block argument. All
35544 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
35545 (dom_opt_dom_walker::thread_across_edge): Remove
35546 handle_dominating_asserts argument. All callers changed.
35547 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
35548 changes. Remove calls to lhs_of_dominating_assert. Other
35549 uses of handle_dominating_asserts turn into unconditional code
35550 (simplify_control_stmt_condition_1): Likewise.
35551 (simplify_control_stmt_condition): Likewise.
35552 (thread_through_normal_block, thread_across_edge): Likewise.
35553 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
35554 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
35555 object if it is not an SSA_NAME.
35556 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
35557 before calling into the VRP specific simplifiers.
35558 (identify_jump_threads): Remove handle_dominating_asserts
35561 2017-03-16 Jakub Jelinek <jakub@redhat.com>
35564 * tree-diagnostic.c (default_tree_printer): No longer static.
35565 * tree-diagnostic.h (default_tree_printer): New prototype.
35567 2017-03-16 Tamar Christina <tamar.christina@arm.com>
35569 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
35570 Change ins into fmov.
35572 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35574 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
35575 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
35576 Use h_con constraint for operand 1.
35577 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35578 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35580 2017-03-15 Jeff Law <law@redhat.com>
35582 PR tree-optimization/71437
35583 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35584 (record_temporary_equivalences): Use it.
35586 PR tree-optimization/71437
35587 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35588 tree-ssa-scopedtables.
35589 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35590 (record_conditions, record_cond, vuse_eq): Likewise.
35591 (record_edge_info): Adjust to API tweak of record_conditions.
35592 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35593 (record_temporary_equivalences, optimize_stmt): Likewise.
35594 (eliminate_redundant_computations): Likewise.
35595 (record_equivalences_from_stmt): Likewise.
35596 * tree-ssa-scopedtables.c: Include options.h and params.h.
35597 (vuse_eq): New function, moved from tree-ssa-dom.c
35598 (build_and_record_new_cond): Likewise.
35599 (record_conditions): Likewise. Accept vector of conditions rather
35600 than edge_equivalence structure for first argument.
35601 for the first argument.
35602 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35603 from tree-ssa-dom.c.
35604 (avail_exprs_stack::record_cond): Likewise.
35605 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35606 from tree-ssa-dom.c.
35607 (avail_exprs_stack): Add new member functions lookup_avail_expr
35609 (record_conditions): Declare.
35611 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35614 * lra-constraints.c (process_alt_operands): Increase reject for
35615 reloading an input/output operand.
35617 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35620 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35621 insns to convert from signed/unsigned char/short to IEEE 128-bit
35623 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35625 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35628 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35629 subreg of inner mode for values already in registers.
35631 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35633 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35634 iteration reg is used after the loop.
35636 2017-03-14 Martin Sebor <msebor@redhat.com>
35638 PR tree-optimization/79800
35639 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35640 precision in negative-positive range.
35641 (format_floating): Call non-const overload with adjusted precision.
35643 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35646 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35649 2017-03-14 Martin Sebor <msebor@redhat.com>
35651 PR middle-end/80020
35652 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35653 * builtins.def (aligned_alloc): Use it.
35656 * Makefile.in (GTFILES): Add calls.c.
35657 * calls.c: Include "gt-calls.h".
35659 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35661 PR rtl-optimization/79728
35662 * regs.h (struct target_regs): New field
35663 x_contains_allocatable_regs_of_mode.
35664 (contains_allocatable_regs_of_mode): New macro.
35665 * reginfo.c (init_reg_sets_1): Initialize it, and change
35666 contains_reg_of_mode so it includes global regs as well.
35667 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35668 rather than contains_regs_of_mode.
35670 2017-03-14 Martin Liska <mliska@suse.cz>
35672 * doc/invoke.texi: Document options that can't be combined with
35673 -fcheck-pointer-bounds.
35675 2017-03-14 Martin Liska <mliska@suse.cz>
35677 PR middle-end/79831
35678 * doc/invoke.texi (-Wchkp): Document the option.
35680 2017-03-14 Martin Liska <mliska@suse.cz>
35682 * Makefile.in: Install gcov-dump.
35684 2017-03-14 Martin Liska <mliska@suse.cz>
35686 * multiple_target.c (expand_target_clones): Bail out for
35687 an invalid attribute.
35689 2017-03-14 Richard Biener <rguenther@suse.de>
35691 * alias.c (struct alias_set_entry): Pack properly.
35692 * cfgloop.h (struct loop): Likewise.
35693 * cse.c (struct set): Likewise.
35694 * ipa-utils.c (struct searchc_env): Likewise.
35695 * loop-invariant.c (struct invariant): Likewise.
35696 * lra-remat.c (struct cand): Likewise.
35697 * recog.c (struct change_t): Likewise.
35698 * rtl.h (struct address_info): Likewise.
35699 * symbol-summary.h (function_summary): Likewise.
35700 * tree-loop-distribution.c (struct partition): Likewise.
35701 * tree-object-size.c (struct object_size_info): Likewise.
35702 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35703 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35704 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35705 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35706 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35707 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35708 (struct _stmt_vec_info): Likewise.
35710 2017-03-14 Martin Liska <mliska@suse.cz>
35713 * multiple_target.c (create_dispatcher_calls): Check that
35714 a target can create a function dispatcher.
35716 2017-03-14 Martin Liska <mliska@suse.cz>
35719 * multiple_target.c (expand_target_clones): Drop local.local
35720 flag for default implementation.
35722 2017-03-14 Richard Biener <rguenther@suse.de>
35724 PR tree-optimization/80030
35725 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35727 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35729 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35730 gcc_fallthrough() instead of __attribute__((fallthrough));
35732 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35734 * doc/gcc.texi: Remove "up" link to (DIR).
35735 * doc/gccint.texi: Ditto.
35737 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35739 * doc/install.texi (Specific) <avr>: Remove reference to
35742 2017-03-13 Jeff Law <law@redhat.com>
35744 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35745 attribute rather than comments.
35747 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35748 match_scratch operand is highest.
35750 2017-03-13 Martin Liska <mliska@suse.cz>
35752 PR middle-end/78339
35753 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35754 is a CHKP clone, use original declaration.
35756 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35758 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35759 (arc_conditional_register_usage): Use a different allocation order
35760 when optimizing for size.
35761 * common/config/arc/arc-common.c (arc_option_optimization_table):
35762 Section anchors default on when optimizing for size.
35764 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35766 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35768 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35770 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35771 * config/arc/arc.md (cpu_facility): Add cd variant.
35772 (*movqi_insn): Add code density variant.
35773 (*movhi_insn): Likewise.
35774 (*movqi_insn): Likewise.
35775 (*addsi3_mixed): Likewise.
35776 (subsi3_insn): Likewise.
35778 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35780 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35782 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35784 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35785 expressions with MINUS and UNARY ops.
35787 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35790 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35792 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35793 between vec_select and vector argument.
35794 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35795 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35796 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35797 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35798 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35799 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35801 2017-03-13 Richard Biener <rguenther@suse.de>
35804 * params.def (vect-max-peeling-for-alignment): Fix typo.
35806 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35808 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35809 issue that only occurred with binutils below 2.18.
35811 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35813 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35814 refer to binutils 2.11/2.12 minimum.
35816 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35818 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35819 ftp.kernel.org and simplify binutils requirement.
35821 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
35823 * doc/invoke.texi (Warning Options): Fix spelling of link-time
35825 (Optimize Options): Ditto. Also remove redundancy.
35827 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35829 PR translation/79848
35830 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
35832 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
35833 to G_ to avoid double translation.
35835 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35837 PR translation/79923
35838 * auto-profile.c (get_combined_location): Convert leading
35839 character of diagnostics to lower case and remove trailing period.
35840 (read_profile): Likewise for various diagnostics.
35841 * config/arm/arm.c (arm_option_override): Remove trailing period
35842 from various diagnostics.
35843 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
35844 (msp430_expand_delay_cycles): Likewise.
35846 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35849 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
35850 full command-line argument, rather than just "str".
35851 (aarch64_validate_march): Likewise.
35852 (aarch64_validate_mtune): Likewise.
35854 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
35856 PR rtl-optimization/78911
35857 * lra-assigns.c (must_not_spill_p): New function.
35858 (spill_for): Use it.
35860 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35862 PR tree-optimization/79981
35863 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
35864 ATOMIC_COMPARE_EXCHANGE ifn result.
35865 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
35866 IFN_ATOMIC_COMPARE_EXCHANGE.
35868 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35871 * opts.c (parse_sanitizer_options): Add missing question mark to
35872 "did you mean" message.
35874 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35876 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
35878 (VMULEUH_UNS): Likewise.
35879 (VMULOUB_UNS): Likewise.
35880 (VMULOUH_UNS): Likewise.
35881 * config/rs6000/rs6000.c (builtin_function_type): Remove
35882 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
35884 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35887 * read-rtl-function.c (function_reader::read_rtx_operand): Update
35888 x with result of extra_parsing_for_operand_code_0.
35889 (function_reader::extra_parsing_for_operand_code_0): Convert
35890 return type from void to rtx, returning x. When reading
35891 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
35892 larger size containing struct block_symbol.
35894 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
35896 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
35897 -mfloat128-hardware without -m64.
35899 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
35902 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
35903 entries to the case statement that marks unsigned arguments to
35904 overloaded functions.
35906 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
35908 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
35909 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
35911 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
35914 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
35915 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
35917 2017-03-10 Martin Liska <mliska@suse.cz>
35921 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
35922 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
35925 2017-03-10 Olivier Hainque <hainque@adacore.com>
35927 * tree-switch-conversion (array_value_type): Start by resetting
35928 candidate type to it's main variant.
35930 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35932 PR rtl-optimization/79909
35933 * combine.c (try_combine): Use simplify_replace_rtx on individual
35934 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
35935 of the whole CALL_INSN_FUNCTION_USAGE.
35937 PR tree-optimization/79972
35938 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
35939 get_range_info on SSA_NAMEs. Formatting fixes.
35941 2017-03-10 Richard Biener <rguenther@suse.de>
35942 Jakub Jelinek <jakub@redhat.com>
35944 PR tree-optimization/77975
35945 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
35946 edge to be constant.
35947 (get_val_for): For constant x return it. Formatting fix.
35948 (loop_niter_by_eval): Avoid pointless looping if the next iteration
35949 would use the same bases as the current one.
35951 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35953 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
35954 instead of vec_select for V1TImode.
35955 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
35957 (VSX_LE_128): Add V1TI to this mode iterator.
35958 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
35959 (*vsx_le_perm_store_<mode>): Likewise.
35960 (pre-reload splitter for VSX stores): Likewise.
35961 (post-reload splitter for VSX stores): Likewise.
35962 (*vsx_xxpermdi2_le_<mode>): Likewise.
35963 (*vsx_lxvd2x2_le_<mode>): Likewise.
35964 (*vsx_stxvd2x2_le_<mode>): Likewise.
35966 2017-03-09 Michael Eager <eager@eagercon.com>
35968 Correct failures with --enable-checking=yes,rtl.
35970 * config/microblaze/microblaze.c (microblaze_expand_shift):
35971 Replace GET_CODE test with CONST_INT_P and INTVAL test with
35972 test for const0_rtx.
35973 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
35974 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
35976 2017-03-09 Richard Biener <rguenther@suse.de>
35978 PR tree-optimization/79977
35979 * graphite-scop-detection.c (scop_detection::merge_sese):
35980 Handle the case of extra exits to blocks dominating the entry.
35982 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
35984 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
35987 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
35989 PR rtl-optimization/79949
35990 * lra-constraints.c (process_alt_operands): Check memory when
35991 trying to predict a cycle. Print about the overall increase.
35993 2017-03-09 Richard Biener <rguenther@suse.de>
35995 PR middle-end/79971
35996 * gimple-expr.c (useless_type_conversion_p): Preserve
35997 TYPE_SATURATING for fixed-point types.
35999 2017-03-09 Richard Biener <rguenther@suse.de>
36002 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36003 alignment of BLKmode params.
36005 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36008 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36009 (VALL_NO_V2Q): Likewise.
36011 * config/aarch64/aarch64-simd.md
36012 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36014 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36015 VALL_NO_V2Q mode iterator.
36016 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36018 2017-03-09 Martin Liska <mliska@suse.cz>
36020 PR tree-optimization/79631
36021 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36022 tree_int_cst_sign_bit just for INTEGER constants.
36024 2017-03-09 Martin Liska <mliska@suse.cz>
36028 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36031 2017-03-09 Marek Polacek <polacek@redhat.com>
36034 * tree.c (inchash::add_expr): Handle TREE_VEC.
36036 2017-03-09 Martin Liska <mliska@suse.cz>
36039 (chkp_narrow_size_and_offset): New function.
36040 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36041 (void chkp_parse_bit_field_ref): New function.
36042 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36043 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36045 2017-03-09 Martin Liska <mliska@suse.cz>
36048 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36049 (chkp_find_bounds_1): Remove gcc_unreachable.
36051 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36054 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36055 BUILT_IN_SYNC*, determine the access type from the size suffix and
36056 always build a MEM_REF with that type. Handle forgotten
36057 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36060 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36061 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36062 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36063 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36064 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
36065 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
36066 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
36067 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
36068 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
36069 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
36070 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
36071 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
36072 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
36073 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
36074 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
36075 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
36076 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
36077 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
36078 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
36079 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
36080 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
36081 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
36082 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
36083 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
36084 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
36085 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
36086 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
36087 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
36088 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
36089 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
36090 definitions outside of __OPTIMIZE__ guarded section.
36093 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
36094 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
36095 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
36096 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
36099 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36101 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
36102 ("vfenez<mode>"): Add missing constraints.
36104 2017-03-08 Martin Sebor <msebor@redhat.com>
36107 * config/nds32/nds32.c (nds32_option_override):
36108 Fix misspelled diagnostic.
36110 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36113 * gimplify.c (gimplify_omp_for): Replace index var in outer
36114 taskloop statement with an artificial variable and add
36115 OMP_CLAUSE_PRIVATE clause for it.
36117 2017-03-08 Richard Biener <rguenther@suse.de>
36119 PR tree-optimization/79955
36120 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
36121 for accesses that are completely outside of the variable.
36123 2017-03-08 Andrew Haley <aph@redhat.com>
36125 PR tree-optimization/79943
36126 * tree-ssa-loop-split.c (compute_new_first_bound): When
36127 calculating the new upper bound, (END-BEG) should be added, not
36130 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36132 * config/avr/avr.md (setmemhi): Make sure match_dup
36133 operand number comes before match_scratch.
36135 2017-03-08 Richard Biener <rguenther@suse.de>
36137 PR tree-optimization/79920
36138 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
36139 with ncopies == 1 to ...
36140 (vect_transform_slp_perm_load): ... here. Properly compute
36141 all element loads by iterating VF times over the group. Do
36142 not handle ncopies (computed in a broken way) in
36143 vect_create_mask_and_perm.
36145 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36148 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
36149 is a uniform vector, use uniform_vector_p return value instead of
36150 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
36152 2017-03-07 Marek Polacek <polacek@redhat.com>
36154 PR middle-end/79809
36155 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
36156 (alloca_call_type): Likewise.
36158 2017-03-07 Martin Liska <mliska@suse.cz>
36160 * gcov.c (process_args): Put comment to correct location.
36162 2017-03-07 Martin Liska <mliska@suse.cz>
36164 PR middle-end/68270
36165 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
36166 Use array_at_struct_end_p instead of DECL_CHAIN (field).
36167 (chkp_narrow_bounds_for_field): Likewise.
36168 (chkp_parse_array_and_component_ref): Pass one more argument to
36171 2017-03-07 Richard Biener <rguenther@suse.de>
36173 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
36176 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
36178 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
36179 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
36181 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36184 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
36185 to end of description.
36186 (PARAM_MAX_STORES_TO_MERGE): Likewise.
36188 2017-03-07 Jakub Jelinek <jakub@redhat.com>
36190 PR rtl-optimization/79901
36191 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
36193 (*avx512f_<code><mode>3<mask_name>): ... this.
36194 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
36195 iterator instead of VI8_AVX2_AVX512BW.
36197 PR rtl-optimization/79901
36198 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
36199 min/max expander, expand it using expand_vec_cond_expr.
36202 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
36205 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36208 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
36209 to void * for PCH reasons.
36210 * dwarf2out.c (output_loc_operands, output_die): Cast
36211 v.val_vec.array to unsigned char *.
36213 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
36216 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
36219 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
36221 PR rtl-optimization/79571
36222 * lra-constraints.c (process_alt_operands): Calculate static
36223 reject and subtract it from overall when only addresses will be
36226 2017-03-06 Julia Koval <julia.koval@intel.com>
36229 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
36230 incoming stack boundary to 128 for 64-bit targets.
36232 2017-03-06 Richard Biener <rguenther@suse.de>
36234 PR tree-optimization/79894
36235 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
36236 to NULL after folding it.
36238 2017-03-06 Richard Biener <rguenther@suse.de>
36240 PR tree-optimization/79824
36241 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
36242 check disabling peeling for gaps.
36244 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
36246 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
36247 attributes): Document gettimeofday.
36249 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36251 * config/s390/s390.c (s390_option_override_internal): Set
36252 PARAM_MIN_VECT_LOOP_BOUND
36254 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36256 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
36257 * config/s390/s390.md: Likewise.
36259 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36262 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
36263 (<avx2_avx512>_perm<mode>): Rename to ...
36264 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
36266 (<avx512>_perm<mode>_mask): Rename to ...
36267 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
36269 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
36270 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
36271 instead of VI8F_256_512.
36272 (avx512f_perm<mode>): New define_expand.
36273 (avx512f_perm<mode>_mask): Likewise.
36274 (avx512f_perm<mode>_1<mask_name>): New define_insn.
36275 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
36277 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36279 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
36280 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
36282 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
36284 2017-03-06 Martin Liska <mliska@suse.cz>
36287 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
36288 when having a SSA NAME w/o VAR_DECL assigned to it.
36290 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36292 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
36293 msa_dpsub_<su>_d): Fix MODE for vec_select.
36295 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36297 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
36299 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
36301 2017-03-06 Richard Biener <rguenther@suse.de>
36303 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
36304 * plugin.c (register_plugin_info): Likewise.
36305 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
36307 2017-03-05 Jakub Jelinek <jakub@redhat.com>
36309 * config/i386/sse.md (sse_storehps, sse_storelps,
36310 avx_<castmode><avxsizesuffix>_<castmode>,
36311 avx512f_<castmode><avxsizesuffix>_<castmode>,
36312 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
36313 in condition that at least one operand is not a MEM.
36315 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36317 PR middle-end/79805
36318 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
36319 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
36321 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
36322 gimple_call_nothrow_p flag based on whether original builtin can throw.
36323 If it can, emit following stmts on the fallthrough edge.
36324 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
36325 don't create new bb if inserting just debug stmts on the edge, try to
36326 insert them on the fallthru bb or just reset debug stmts.
36328 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
36331 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
36332 restore recog_data (including the operand rtxes inside it) around
36333 the call to get_insn_template.
36335 2017-03-03 Martin Sebor <msebor@redhat.com>
36337 PR tree-optimization/79699
36338 * context.c (context::~context): Free MPFR caches to avoid
36339 a memory leak on program exit.
36341 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36343 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
36344 Use wide_int::ulow () instead of .elt (0).
36346 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36348 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
36349 (*pushxf): Limit oF constraint to 32bit targets and add oC
36350 constraint for 64bit targets.
36351 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
36352 (*pushdf): Change rmF constraint to rmC.
36354 2017-03-03 Martin Liska <mliska@suse.cz>
36356 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
36357 Remove unused variable.
36359 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36362 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
36363 is a memory operand, increase num_memory.
36364 (ix86_expand_args_builtin): Likewise.
36366 2017-03-03 Jan Hubicka <jh@suse.cz>
36369 * ipa-devirt.c (maybe_record_node): Properly handle
36370 __cxa_pure_virtual visibility.
36372 2017-03-03 Martin Liska <mliska@suse.cz>
36374 PR tree-optimization/79803
36375 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
36377 (pass_loop_prefetch::execute): Disabled optimization if an
36378 assumption about L1 cache size is not met.
36380 2017-03-03 Martin Liska <mliska@suse.cz>
36382 PR rtl-optimization/79574
36383 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
36384 (hash_scan_set): Likewise.
36385 (dump_hash_table): Likewise.
36386 (hoist_code): Likewise.
36388 2017-03-03 Richard Biener <rguenther@suse.de>
36390 * fixed-value.c (fixed_from_string): Restore use of elt (1)
36391 in place of uhigh ().
36392 (fixed_convert_from_real): Likewise.
36394 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36397 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
36399 2017-03-03 Richard Biener <rguenther@suse.de>
36401 PR middle-end/79818
36402 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
36403 TYPE_OVERFLOW_UNDEFINED check.
36405 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36407 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
36409 (vector_ae_<mode>_p): Likewise.
36410 (vector_nez_<mode>_p): Likewise.
36411 (vector_ne_v2di_p): Likewise.
36412 (vector_ae_v2di_p): Likewise.
36413 (vector_ne_<mode>_p): Likewise.
36414 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
36416 (vsx_tsqrt<mode>2_fe): Likewise.
36418 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
36421 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
36423 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36425 PR rtl-optimization/79780
36426 * cprop.c (one_cprop_pass): When second and further conditional trap
36427 in a single basic block is turned into an unconditional trap, turn it
36428 into a deleted note to avoid RTL verification failures.
36430 2017-03-02 Richard Biener <rguenther@suse.de>
36432 * fold-const.c (const_binop): Use ulow () instead of elt (0).
36434 2017-03-02 Richard Biener <rguenther@suse.de>
36436 PR tree-optimization/79345
36438 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
36439 param and abort the walk, returning -1 if it is hit.
36440 (walk_aliased_vdefs): Take a limit param and pass it on.
36441 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
36442 defaulting to 0 and return a signed int.
36443 * tree-ssa-uninit.c (struct check_defs_data): New struct.
36444 (check_defs): New helper.
36445 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
36446 about uninitialized memory.
36447 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
36448 bogus uninitialized warning.
36449 (fixed_convert_from_real): Likewise.
36451 2017-03-02 Bin Cheng <bin.cheng@arm.com>
36453 PR tree-optimization/66768
36454 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
36455 iv_use if base object can't be determined.
36457 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36459 PR tree-optimization/79345
36460 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
36461 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
36462 (get_pattern_stats): Initialize it.
36463 * genemit.c (gen_expand): Verify match_scratch numbers come after
36464 match_operand/match_dup numbers.
36465 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
36466 match_scratch numbers.
36467 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
36469 * config/s390/s390.md (trunctdsd2): Likewise.
36471 2017-03-02 Richard Biener <rguenther@suse.de>
36473 * wide-int.h (wide_int_storage::operator=): Implement in terms
36476 2017-03-02 Richard Biener <rguenther@suse.de>
36478 PR tree-optimization/79777
36479 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
36480 the to insert expression to sth existing.
36482 2017-03-01 Martin Sebor <msebor@redhat.com>
36484 PR middle-end/79692
36485 * gimple-ssa-sprintf.c
36486 (directive::known_width_and_precision): New function.
36487 (format_integer): Use it.
36488 (get_mpfr_format_length): Consider the full range of precision
36489 when computing %g output with the # flag. Set the likely byte
36490 count to 3 rather than 1 when precision is indeterminate.
36491 (format_floating): Correct the lower bound of precision.
36493 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36495 * doc/invoke.texi: Document default code model for 64-bit Linux.
36497 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36500 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
36501 udiv rather than div since input pattern is unsigned.
36503 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
36505 * config/i386/i386.c (print_reg): Warn for values of
36506 unsupported size in integer register.
36508 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
36511 * config/rs6000/predicates.md (current_file_function_operand): Do
36512 not allow self calls to be local if the function is replaceable.
36514 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
36517 * config/rs6000/altivec.h (vec_ctz and others): Change the
36518 preprocessor macro that controls conditional compilation from
36519 _ARCH_PWR9 to __POWER9_VECTOR__.
36520 (vec_all_ne): Change parameterization of __altivec_scalar_pred
36521 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
36522 control (instead of _ARCH_PWR9 control) so that template
36523 definition uses power9-specific function.
36524 (vec_any_eq): Likewise.
36525 (vec_all_ne): Change macro definition to use a power9-specific
36526 expansion under #ifdef __POWER9_VECTOR__ control (instead of
36527 _ARCH_PWR9 control).
36528 (vec_any_eq) Likewise.
36529 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
36530 expansion for CMPNEF to remove support for xvcmpnesp instruction.
36531 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
36532 support for xvcmpnedp instruction.
36533 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
36534 macro expansion so that Power9 implementation of vec_all_ne does
36535 not use the AltiVec predicate framework.
36536 (VCMPNEH_P): Likewise.
36537 (VCMPNEW_P): Likewise.
36538 (VCMPNED_P): Likewise.
36539 (VCMPNEFP_P): Likewise.
36540 (VCMPNEDP_P): Likewise.
36541 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
36542 implementation of vec_any_eq to not use AltiVec predicate
36544 (VCMPAEH_P): Likewise.
36545 (VCMPAEW_P): Likewise.
36546 (VCMPAED_P): Likewise.
36547 (VCMPAEFP_P): Likewise.
36548 (VCMPAEDP_P): Likewise.
36549 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
36550 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
36551 not use the AltiVec predicate framework.
36552 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
36553 of vec_any_eq to not use AltiVec predicate framework.
36554 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36555 support for predefined __POWER9_VECTOR__ macro to indicate that
36556 Power9 instruction selection is enabled.
36557 (altivec_overloaded_builtins): Remove extraneous
36558 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
36559 function argument types RS6000_BTI_bool_V16QI and
36560 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
36561 entry for overloaded function argument types RS6000_BTI_bool_V4SI
36562 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
36563 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
36564 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
36565 Power9 for implementations of vec_cmpne. Change the signature for
36566 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
36567 (representing vec_all_ne) to remove the previously described first
36568 argument of type RS6000_BTI_INTSI, as this was an artifact of
36569 reliance on the AltiVec predicate framework, which is no longer
36570 used in the implementation of these functions. Add
36571 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
36572 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
36573 since, unlike the AltiVec predicate framework implementation, we
36574 do not share function descriptors between vec_alle and vec_anyeq.
36575 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
36576 set of modes that receive special treatment even when
36577 TARGET_P9_VECTOR is true. The special treatment emits code that
36578 does not depend on Power9 instructions.
36579 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36580 define_expand to not rely on AltiVec predicate framework.
36581 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36583 (vector_ne_v2di_p): Change this define_expand to not rely on
36584 AltiVec predicate framework.
36585 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36587 (vector_ne_<mode>_p): Change this define_expand to not rely on
36588 AltiVec predicate framework.
36589 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36591 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36592 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36593 define_insn pattern.
36594 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36595 define_insn pattern because the xvcmpne<VSs>. instruction is not
36597 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36598 instruction is not supported.
36600 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36602 * config/nvptx/nvptx.c: Include intl.h.
36604 2017-03-01 Martin Jambor <mjambor@suse.cz>
36607 * ipa-prop.h (ipa_bits): Removed field known.
36608 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36609 to pointers. Adjusted their comments to warn about their sharing.
36610 (ipcp_transformation_summary): Change bits to a vector of pointers.
36611 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36612 (ipa_get_ipa_bits_for_value): Declare.
36613 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36614 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36615 (ipa_bits_hash_table): Likewise.
36616 (ipa_vr_ggc_hash_traits): Likewise.
36617 (ipa_vr_hash_table): Likewise.
36618 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36619 being pointers and vr_known being removed.
36620 (ipa_set_jf_unknown): Likewise.
36621 (ipa_get_ipa_bits_for_value): New function.
36622 (ipa_set_jfunc_bits): Likewise.
36623 (ipa_get_value_range): New overloaded functions.
36624 (ipa_set_jfunc_vr): Likewise.
36625 (ipa_compute_jump_functions_for_edge): Use the above functions to
36626 construct bits and vr parts of jump functions.
36627 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36628 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36630 (ipcp_grow_transformations_if_necessary): Also allocate
36631 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36633 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36634 them. Fix too long lines.
36635 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36636 vr_known being removed.
36637 (ipa_read_jump_function): Use new setter functions to construct bits
36638 and vr parts of jump functions or set them to NULL.
36639 (write_ipcp_transformation_info): Adjust for bits being pointers.
36640 (read_ipcp_transformation_info): Likewise.
36641 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36643 Include gt-ipa-prop.h.
36644 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36646 (ipcp_store_bits_results): Likewise.
36647 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36648 Do not write to existing jump functions but use a temporary instead.
36650 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36653 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36654 attempt to use its first operand as BIT_FIELD_REF base.
36656 2017-03-01 Richard Biener <rguenther@suse.de>
36658 PR middle-end/79721
36659 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36660 interpolating formula in wrapping arithmetic.
36661 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36663 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36665 PR tree-optimization/79734
36666 * tree-vect-generic.c (expand_vector_condition): Optimize
36667 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36668 Handle VEC_COND_EXPR where comparison has different inner width from
36669 type's inner width.
36671 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36673 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36674 markup, and similar issues. Remove @opindex entries for things
36675 that aren't options. Add missing -mmpy-option entries.
36677 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36679 PR tree-optimization/79737
36680 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36681 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36682 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36683 instead of byte_size. Formatting fix.
36684 (shift_bytes_in_array_right): Formatting fix.
36686 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36689 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36690 condition on optimize for the leaf function test.
36692 2017-02-28 Martin Liska <mliska@suse.cz>
36695 * read-rtl-function.c (function_reader::handle_unknown_directive):
36696 Bail out when one uses -flto.
36698 2017-02-28 Martin Liska <mliska@suse.cz>
36700 * common.opt: Replace space with tabular for options of <number>
36702 * config/i386/i386.opt: Show <number> value for
36703 -mlarge-data-threshold.
36704 * opts.c (print_filtered_help): Do not display number in hexadecimal
36707 2017-02-28 Martin Liska <mliska@suse.cz>
36709 * common.opt: Fix --help=option -Q for options which are of
36712 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36714 * config/i386/i386.c (print_reg): Error out for values
36715 of 8-bit size in invalid integer register.
36717 2017-02-28 Martin Sebor <msebor@redhat.com>
36719 PR tree-optimization/79691
36720 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36722 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36725 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36726 gcc_unreachable with output_operand_lossage.
36728 2017-02-28 Richard Biener <rguenther@suse.de>
36730 PR tree-optimization/79740
36731 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36733 (visit_nary_op): Insert the nary into the hashtable if we
36734 pattern-matched sth.
36735 * tree-ssa-pre.c (eliminate_insert): Robustify.
36737 2017-02-28 Richard Biener <rguenther@suse.de>
36739 PR middle-end/79731
36740 * fold-const.c (decode_field_reference): Reject out-of-bound
36743 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36745 * config/i386/i386.c: Include intl.h.
36746 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36747 instead of just cond ? "..." : "...".
36748 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36749 * coverage.c (read_counts_file): Likewise.
36750 * omp-offload.c: Include intl.h.
36751 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36752 of just cond ? "..." : "...".
36753 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36754 of just cond ? "..." : "...".
36756 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36759 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36761 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36762 'tune for' CPU name.
36763 * config/arm/arm-cpu-data.h: Regenerated.
36765 2017-02-28 Richard Biener <rguenther@suse.de>
36767 PR tree-optimization/79732
36768 * tree-inline.c (expand_call_inline): Do not shadow var.
36770 2017-02-28 Richard Biener <rguenther@suse.de>
36772 PR tree-optimization/79723
36773 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36774 address-space properly.
36776 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36778 * doc/optinfo.texi (Optimization groups): Fix option used for
36780 * doc/invoke.texi (-fopt-info): Document "omp".
36781 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36782 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36783 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36784 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36785 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36787 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36789 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36790 OPTGROUP_OMP with "omp".
36792 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36795 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36796 for arithmetic shift of unsigned V2DI.
36798 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36800 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36801 arc/linux.h headers.
36802 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36803 (LINK_SPEC): Likewise.
36804 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36805 (EXTRA_SPECS): Likewise.
36806 (STARTFILE_SPEC): Likewise.
36807 (ENDFILE_SPEC): Likewise.
36808 (LIB_SPEC): Likewise.
36809 (TARGET_SDATA_DEFAULT): Likewise.
36810 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36811 (MULTILIB_DEFAULTS): Likewise.
36812 (DWARF2_UNWIND_INFO): Likewise.
36813 * config/arc/big.h: New file.
36814 * config/arc/elf.h: Likewise.
36815 * config/arc/linux.h: Likewise.
36816 * config/arc/t-uClibc: Remove.
36818 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36820 PR tree-optimization/77536
36821 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
36822 (tree_transform_and_unroll_loop): Use above function to compute the
36823 estimated niter of unrolled loop and use it when scaling profile.
36824 Also use count info rather than frequency if it's non-zero.
36825 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
36826 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
36827 (vect_transform_loop): Call above function.
36829 2017-02-27 Richard Biener <rguenther@suse.de>
36831 PR tree-optimization/45397
36832 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
36833 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
36834 (visit_nary_op): Add pattern matching for CSEing sign-changed
36835 or truncated operations with wider ones.
36837 2017-02-27 Richard Biener <rguenther@suse.de>
36839 PR tree-optimization/79690
36840 * tree-vect-stmts.c (vectorizable_store): Use vector type
36841 built from the DR with address-space.
36843 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
36845 * doc/invoke.texi (Optimize Options): Refine the description
36846 of asan-use-after-return.
36848 2017-02-25 Alan Modra <amodra@gmail.com>
36850 PR rtl-optimization/79584
36851 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
36852 base, not ad->base_term, the reg within base. Remove assertion
36853 that ad->base == ad->base_term. Replace gen_int_mode using
36854 bogus mode with const0_rtx.
36856 2017-02-25 Jakub Jelinek <jakub@redhat.com>
36858 PR middle-end/79396
36859 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
36860 FMA_EXPR like tcc_binary or tcc_unary.
36862 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
36865 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
36867 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
36868 (output_loc_operands): Handle DW_OP_call_ref and
36869 DW_OP_GNU_variable_value.
36870 (struct variable_value_struct): New type.
36871 (struct variable_value_hasher): Likewise.
36872 (variable_value_hash): New variable.
36873 (string_types): Remove.
36874 (copy_loc_descr): New function.
36875 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
36876 (prepend_loc_descr_to_each): New function.
36877 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
36878 instead of add_loc_descr_to_each if the first argument is single
36879 location list and the second has multiple.
36880 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
36881 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
36882 when looking for variable value which doesn't have other location info.
36883 (loc_list_from_tree): Formatting fix.
36884 (gen_array_type_die): Simplify DW_AT_string_length handling.
36885 (adjust_string_types): Remove.
36886 (gen_subprogram_die): Don't call adjust_string_types nor test/set
36887 string_types. Call resolve_variable_values.
36888 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
36889 (resolve_addr_in_expr): Likewise. Add A argument.
36890 (copy_deref_exprloc): Remove deref argument. Adjust for the
36891 original expression being DW_OP_GNU_variable_value with optionally
36892 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
36893 optionally after it.
36894 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
36895 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
36896 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
36897 (variable_value_hasher::hash, variable_value_hasher::equal): New
36899 (resolve_variable_value_in_expr, resolve_variable_value,
36900 resolve_variable_values, note_variable_value_in_expr,
36901 note_variable_value): New functions.
36902 (dwarf2out_early_finish): Call note_variable_value on all toplevel
36905 2017-02-24 Jakub Jelinek <jakub@redhat.com>
36908 * opts.h (handle_generated_option): Add GENERATED_P argument.
36909 * opts-common.c (handle_option): Adjust function comment.
36910 (handle_generated_option): Add GENERATED_P argument, pass it to
36912 (control_warning_option): Pass false to handle_generated_option
36914 * opts.c (maybe_default_option): Pass true to handle_generated_option
36916 * optc-gen.awk: Likewise.
36918 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36920 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
36921 a REG, look at the REG it is a SUBREG of.
36922 (splitter for cmpeqsi_t): Ditto.
36924 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36926 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
36927 the special USEs with the pattern of the insn, not the insn itself.
36929 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
36932 * doc/invoke.texi: Document -mload-store-pairs.
36934 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36935 Sandra Loosemore <sandra@codesourcery.com>
36937 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
36938 argument isn't a CONST_INT.
36939 (nios2_alternate_compare_const): Assert op is a CONST_INT.
36940 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
36941 (nios2_validate_compare): Bypass alternate compare logic if *op2
36942 is not a CONST_INT.
36943 (ldstwm_operation_p): Return false if first_base is not a REG or
36944 if first_offset is not a CONST_INT.
36946 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36948 * config/cris/cris.md: Use correct operand in a define_peephole2.
36950 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36952 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
36954 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36956 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
36957 this_insn if it is an INSN or JUMP_INSN.
36958 (force_offsettable): Look at base, not at addr.
36959 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
36960 on things that aren't necessarily CONST_INTs.
36962 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
36964 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
36965 -mfpmath=sse is the default also for x86-32 targets with SSE2
36966 instruction set when @option{-ffast-math} is enabled
36968 2017-02-24 Jeff Law <law@redhat.com>
36970 PR rtl-optimizatoin/79286
36971 * ira.c (update_equiv_regs): Drop may_trap_p exception to
36974 2017-02-24 Richard Biener <rguenther@suse.de>
36976 PR tree-optimization/79389
36977 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
36980 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
36982 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
36983 function comment to reflect reality.
36984 (loop_exits_before_overflow): Fix typo in function description.
36986 2017-02-24 Richard Biener <rguenther@suse.de>
36988 PR tree-optimization/79389
36989 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
36990 properly that a threading opportunity exists. Detect conditional
36991 copy/constant propagation opportunities.
36993 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
36995 * config/visium/visium.md (type): Add trap.
36996 (b): New mode attribute.
36997 (*btst): Rename into...
36998 (*btst<mode>): ...this and adjust.
36999 (*cbranchsi4_btst_insn): Rename into...
37000 (*cbranch<mode>4_btst_insn): ...this and adjust.
37001 (trap): New define_insn.
37003 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37005 PR tree-optimization/79389
37006 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37007 (noce_reversed_cond_code): New function.
37008 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37009 reversed_comparison_code. Formatting fix.
37010 (noce_try_store_flag): Test rev_cond != NULL in addition to
37011 reversed_comparison_code.
37012 (noce_try_store_flag_constants): Likewise.
37013 (noce_try_store_flag_mask): Likewise.
37014 (noce_try_addcc): Use rev_cond if non-NULL instead of
37015 reversed_comparison_code.
37016 (noce_try_cmove_arith): Likewise. Formatting fixes.
37017 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37018 (noce_find_if_block): Initialize rev_cond.
37019 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37020 instead of false as last argument never attempt to reverse it
37023 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37025 PR tree-optimization/79663
37026 * tree-predcom.c (combine_chains): Process refs in reverse order
37027 only for ZERO length chains, and add explaining comment.
37029 2017-02-23 Jeff Law <law@redhat.com>
37031 PR tree-optimization/79578
37032 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37033 in call to operand_equal_p.
37035 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37038 * config/i386/cpuid.h: Fix another undefined behavior.
37040 2017-02-23 Richard Biener <rguenther@suse.de>
37042 PR tree-optimization/79683
37043 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37044 vector types for data-refs.
37046 2017-02-23 Martin Liska <mliska@suse.cz>
37048 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37050 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37052 PR middle-end/79665
37053 * internal-fn.c (get_range_pos_neg): Moved to ...
37054 * tree.c (get_range_pos_neg): ... here. No longer static.
37055 * tree.h (get_range_pos_neg): New prototype.
37056 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37057 are known to be in between 0 and signed maximum inclusive, try to
37058 expand both unsigned and signed divmod and use the cheaper one from
37061 2017-02-22 Jeff Law <law@redhat.com>
37063 PR tree-optimization/79578
37064 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37065 to compare base operands.
37067 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
37070 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
37071 gpc_reg_operand instead of fpr_reg_operand.
37073 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
37075 * config/mips/mips.c (mips_return_in_memory): Force FP
37076 vector types to be returned in memory for o32 ABI.
37078 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37080 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
37081 instead of DW_TAG_member for static data member declarations and don't
37082 set no_linkage_name for static inline data members.
37083 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
37086 2017-02-22 Martin Liska <mliska@suse.cz>
37088 * doc/invoke.texi: Replace inequality signs with square brackets
37091 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
37094 * lra-constraints.c (simplify_operand_subreg): Handle
37095 WORD_REGISTER_OPERATIONS targets.
37097 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37100 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
37101 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
37102 elimination by swapping fld*.
37104 2017-02-22 Richard Biener <rguenther@suse.de>
37106 PR tree-optimization/79673
37107 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
37108 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
37109 irrelevant address-space qualifiers and avoiding a
37110 ADDR_SPACE_CONVERT_EXPR from fold_convert.
37112 2017-02-22 Richard Biener <rguenther@suse.de>
37114 PR tree-optimization/79666
37115 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
37116 to not symbolically negate if that may introduce undefined
37119 2017-02-22 Martin Liska <mliska@suse.cz>
37122 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
37123 * data-streamer-out.c (streamer_write_gcov_count_stream):
37125 * value-prof.c (stream_out_histogram_value): Make assert more
37126 precise based on type of counter.
37128 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
37131 * config/i386/i386.md (standard_x87sse_constant_load splitter):
37132 Use nonimmediate_operand instead of memory_operand for operand 1.
37133 (float-extend standard_x87sse_constant_load splitter): Ditto.
37135 2017-02-21 Jeff Law <law@redhat.com>
37137 PR tree-optimization/79621
37138 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
37139 blocks with edges to themselves.
37141 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37144 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
37145 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
37146 Use gimple_call_builtin_p.
37149 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
37150 on temporarily removed DEBUG_INSNs.
37152 PR tree-optimization/79649
37153 * tree-loop-distribution.c (classify_partition): Give up on
37154 non-generic address space loads/stores.
37156 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
37158 * doc/loop.texi (Loop manipulation): Remove nonexistent
37159 tree_ssa_loop_version from the documentation.
37160 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
37162 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37165 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
37166 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
37167 * config/rs6000/rs6000.c: Include except.h.
37168 (rs6000_expand_split_stack_prologue): Call
37169 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
37171 2017-02-21 Martin Jambor <mjambor@suse.cz>
37174 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
37175 have been analyzed.
37177 2017-02-21 Martin Jambor <mjambor@suse.cz>
37179 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
37180 for backward compatibility only.
37181 * doc/invoke.texi (Option Summary): Remove all references to
37182 -fipa-cp-alignment.
37184 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
37188 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37190 * lra-constraints.c (curr_insn_transform): Handle
37191 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37193 2017-02-21 Martin Liska <mliska@suse.cz>
37195 * config/i386/i386.opt: Replace -masm-dialect with -masm.
37197 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
37199 PR translation/79638
37200 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
37202 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
37205 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
37206 (arm_function_ok_for_sibcall): Return false for an indirect call by
37207 descriptor if all the argument registers are used.
37208 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
37209 alignment of the function.
37211 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37213 PR tree-optimization/61441
37214 * simplify-rtx.c (simplify_const_unary_operation): For
37215 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
37216 the sNaN unmodified.
37218 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
37220 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
37221 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
37222 instead of SYSTEM_HEADER_DIR.
37224 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
37225 Martin Liška <mliska@suse.cz>
37227 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
37228 Fix typos and grammar, use active voice, and clarify.
37230 2017-02-20 Marek Polacek <polacek@redhat.com>
37232 PR middle-end/79537
37233 * gimplify.c (gimplify_expr): Handle unused *&&L;.
37236 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
37238 2017-02-20 Jakub Jelinek <jakub@redhat.com>
37241 * config/i386/i386.c (ix86_expand_builtin): Handle
37242 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
37243 ix86_builtins_isa[fcode].isa as a requirement of those
37244 flags and any other flag in the bitmask.
37245 (ix86_init_mmx_sse_builtins): Use 0 instead of
37246 ~OPTION_MASK_ISA_64BIT as mask.
37247 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
37248 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
37249 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
37250 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
37252 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37255 * lra-constraints.c (split_reg): Check requested split mode
37256 is supported by the register.
37258 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37260 * lra-constraints.c (simplify_operand_subreg): Remove early
37263 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37266 * lra-constraints.c (curr_insn_transform): Tighten condition
37267 for converting SUBREG reloads from OP_OUT to OP_INOUT.
37269 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37272 * lra-constraints.c (curr_insn_transform): Handle
37273 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37275 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
37278 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
37280 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
37282 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
37285 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
37288 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37290 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
37293 2017-02-18 Jakub Jelinek <jakub@redhat.com>
37295 * final.c (last_columnnum, override_columnnum): New variables.
37296 (final_start_function): Set last_columnnum, pass it to begin_prologue
37297 hook and pass 0 to dwarf2out_begin_prologue.
37298 (final_scan_insn): Update override_columnnum. Pass last_columnnum
37299 to source_line debug hook.
37300 (notice_source_line): Compute last_columnnum and for debug_column_info
37301 return true on column changes.
37302 * debug.h (struct gcc_debug_hooks): Add column argument to
37303 source_line and begin_prologue hooks.
37304 (debug_nothing_int_charstar_int_bool): Remove prototype.
37305 (debug_nothing_int_int_charstar,
37306 debug_nothing_int_int_charstar_int_bool): New prototypes.
37307 (dwarf2out_begin_prologue): Add column argument.
37308 * debug.c (do_nothing_debug_hooks): Adjust source_line and
37309 begin_prologue hooks.
37310 (debug_nothing_int_charstar_int_bool): Remove.
37311 (debug_nothing_int_int_charstar,
37312 debug_nothing_int_int_charstar_int_bool): New functions.
37313 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
37314 through to dwarf2out_source_line.
37315 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
37316 (dwarf2out_source_line): Add column argument, emit it if requested.
37317 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
37319 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37320 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37321 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
37322 through to dwarf2out_begin_prologue.
37323 (vmsdbgout_source_line): Add column argument, pass it through to
37324 dwarf2out_source_line.
37325 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
37326 dbxout_source_line caller.
37327 (dbxout_source_line): Add column argument.
37329 * common.opt (gno-column-info, gcolumn-info): New options.
37330 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
37331 (check_die): Also test for multiple DW_AT_decl_column attributes.
37332 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
37333 DW_AT_decl_column if requested.
37334 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
37336 (gen_variable_die): Likewise.
37337 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
37338 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
37341 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
37342 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
37343 (ix86_handle_option): Handle OPT_m3dnowa.
37344 * doc/invoke.texi (-m3dnowa): Document.
37345 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
37346 -m3dnowa instead of -m3dnow -march=athlon.
37349 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
37350 instead of gcc_assert for K, r and R code checks. Formatting fixes.
37352 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37355 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
37356 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
37357 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
37358 generator for vsx_xxpermdi_<mode>_be.
37359 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
37360 force big-endian semantics.
37361 (vsx_xxpermdi_<mode>_be): New define_expand with same
37362 implementation as previous version of vsx_xxpermdi_<mode>.
37364 2017-02-17 Jakub Jelinek <jakub@redhat.com>
37366 PR tree-optimization/79327
37367 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
37368 variable, its initialization and use.
37370 2017-02-17 Julia Koval <julia.koval@intel.com>
37372 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
37373 (OPTION_MASK_ISA_PKU_UNSET): New.
37374 (ix86_handle_option): Handle -mrdpid.
37375 * config/i386/cpuid.h (bit_RDPID): New.
37376 * config/i386/driver-i386.c (host_detect_local_cpu):
37377 Detect RDPID feature.
37378 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
37379 * config/i386/i386-c.c (ix86_target_macros_internal):
37381 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
37382 (ix86_valid_target_attribute_inner_p): Add "rdpid".
37383 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
37384 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
37385 * config/i386/i386.md (define_insn "rdpid"): New.
37386 * config/i386/i386.opt Add -mrdpid.
37387 * config/i386/immintrin.h (_rdpid_u32): New.
37389 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
37391 PR rtl-optimization/79541
37392 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
37393 instead of transforming it into USE.
37395 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
37397 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
37398 If HONOR_SNANS (SFmode) force the input to a register.
37399 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
37400 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
37401 an frsp or similar insn.
37403 2017-02-17 Martin Liska <mliska@suse.cz>
37405 PR rtl-optimization/79577
37406 * params.def (selsched-max-sched-times): Increase minimum to 1.
37408 2017-02-17 Martin Liska <mliska@suse.cz>
37410 PR rtl-optimization/79574
37411 * gcse.c (want_to_gcse_p): Prevent integer overflow.
37413 2017-02-17 Martin Liska <mliska@suse.cz>
37415 PR tree-optimization/79529
37416 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
37417 ssa_defined_default_def_p to handle cases which are implicitly
37419 * tree-ssa.c (ssa_defined_default_def_p): New function.
37420 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
37421 which are implicitly defined.
37422 * tree-ssa.h (ssa_defined_default_def_p): Declare.
37424 2017-02-17 Richard Biener <rguenther@suse.de>
37426 PR middle-end/79576
37427 * params.def (max-ssa-name-query-depth): Limit to 10.
37429 2017-02-17 Richard Biener <rguenther@suse.de>
37431 PR tree-optimization/79552
37432 * tree-ssa-structalias.c (visit_loadstore): Properly verify
37435 2017-02-17 Richard Biener <rguenther@suse.de>
37438 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
37440 2017-02-17 Marek Polacek <polacek@redhat.com>
37442 PR middle-end/79536
37443 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
37444 (fold_negate_expr): New wrapper.
37446 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
37448 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
37449 Correct terminology and de-emphasize pre-standard behavior.
37451 2017-02-16 Alan Modra <amodra@gmail.com>
37453 PR rtl-optimization/79286
37454 * ira.c (def_dominates_uses): New function.
37455 (update_equiv_regs): Don't create an equivalence for insns that
37456 may trap where the register def does not dominate the use.
37458 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
37460 PR rtl-optimization/78127
37461 * lra.c (lra): Call lra_eliminate before finish the loop after
37464 2017-02-16 Richard Biener <rguenther@suse.de>
37466 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
37468 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
37469 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
37470 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
37471 (isl_val_int_from_wi): New function.
37472 (extract_affine_gmp): Rename to ...
37473 (extract_affine_wi): ... this, take a widest_int.
37474 (extract_affine_int): Just wrap extract_affine_wi.
37475 (add_param_constraints): Use isl_val_int_from_wi.
37476 (add_loop_constraints): Likewise, and extract_affine_wi.
37478 2017-02-15 Jeff Law <law@redhat.com>
37480 PR middle-end/79521
37481 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
37482 ira_init_register_move_cost_if_necessary.
37484 2017-02-15 Martin Sebor <msebor@redhat.com>
37486 PR middle-end/32003
37487 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
37488 removed in a prior commit.
37490 2017-02-15 Bin Cheng <bin.cheng@arm.com>
37492 PR tree-optimization/79347
37493 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
37494 counters during peeling.
37496 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
37498 * Makefile.in (site.exp): Remove "set ISLVER".
37500 2017-02-15 Jakub Jelinek <jakub@redhat.com>
37503 * real.c (real_from_integer): Call real_convert even for decimal.
37505 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
37508 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
37510 2017-02-14 Andrew Pinski <apinski@cavium.com>
37512 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
37513 cores and change the partno/implementer to be correct.
37514 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
37515 the 'B" as the implementer.
37516 * config/aarch64/aarch64-tune.md: Regenerate.
37518 2017-02-14 Carl Love <cel@us.ibm.com>
37520 * config/rs6000/rs6000.c: Add case statement entry to make the
37521 xvcvuxdsp built-in argument unsigned.
37522 * config/rs6000/vsx.md: Fix the source and return operand types so they
37523 match the instruction definitions from the ISA document. Fix typo
37524 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
37527 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
37530 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
37531 member early_clobber_alts.
37532 * lra-lives.c (reg_early_clobber_p): New.
37533 (process_bb_lives): Use it.
37534 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
37535 (debug_operand_data): Initialize early_clobber_alts.
37536 (setup_operand_alternative): Set up early_clobber_alts.
37537 (collect_non_operand_hard_regs): Ditto. Pass early clobber
37538 alternatives to new_insn_reg.
37539 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
37541 (lra_update_insn_regno_info): Pass the new arg.
37543 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37545 PR middle-end/79505
37546 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
37547 (new_oacc_loop_raw): Don't clear already cleared fields.
37550 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
37551 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
37552 _mm512_prefetch_i64gather_ps): New inline functions and macros.
37554 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
37557 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
37559 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
37562 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
37563 the extra instruction to the right place to store 128-bit constant
37566 2017-02-14 Martin Sebor <msebor@redhat.com>
37568 PR middle-end/79448
37569 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
37570 warning for strings of unknown length.
37572 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
37574 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
37576 2017-02-14 Jeff Law <law@redhat.com>
37579 * ira-costs.c (scan_one_insn): Initialize register move costs
37580 for pseudos seen in USE/CLOBBER insns.
37582 PR tree-optimization/79095
37583 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37584 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37585 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37586 if the operands are known to be not equal, then the resulting range
37588 (intersect_ranges): If the new range is ~[0,0] and the old range is
37589 wide, then prefer ~[0,0].
37590 * tree-vrp.c (overflow_comparison_p_1): New function.
37591 (overflow_comparison_p): New function.
37592 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37593 if NAME is used in an overflow test.
37594 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37595 overflow check that can be expressed as an equality test, then adjust
37596 ops to be that equality test.
37598 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37600 * config/s390/s390-builtin-types.def: Remove flags argument.
37601 * config/s390/s390.c (s390_init_builtins): Likewise.
37603 2017-02-14 Martin Liska <mliska@suse.cz>
37605 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37606 vector. Fix trailing white spaces.
37608 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37610 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37613 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37615 PR rtl-optimization/68664
37616 * config/arm/arm.c (arm_sched_can_speculate_insn):
37617 New function. Declare prototype.
37618 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37620 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37622 PR rtl-optimization/68664
37623 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37625 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37627 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37629 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37630 max skip bytes for function, loop and jump.
37632 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37634 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37635 ABS_EXPR for gimple dump.
37637 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37640 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37642 PR tree-optimization/79408
37643 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37644 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37645 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37646 also if rhs1 is INTEGER_CST.
37648 2017-02-14 Richard Biener <rguenther@suse.de>
37650 PR middle-end/79432
37651 * tree-into-ssa.c (insert_phi_nodes): When the function can
37652 have abnormal edges rewrite SSA names with broken use-def
37653 dominance out of SSA and register them for PHI insertion.
37655 2017-02-13 Martin Sebor <msebor@redhat.com>
37657 PR middle-end/79496
37658 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37659 clearing info.nowrite flag when snprintf size argument is a range.
37661 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37663 * cprop.c (cprop_jump): Add missing space in string literal.
37664 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37665 (get_constraint_for_component_ref): Likewise.
37666 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37667 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37668 * lra-constraints.c (process_alt_operands): Likewise.
37669 * ipa-inline.c (inline_small_functions): Likewise.
37670 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37671 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37672 * trans-mem.c (diagnose_tm_1_op): Likewise.
37673 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37674 (grid_parallel_clauses_gridifiable): Likewise.
37676 * config/nvptx/mkoffload.c (process): Add space in between
37679 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37680 "MOD4_SSE_REGS" and "ALL_REGS".
37682 * spellcheck.c (test_data): Add , in between "foo" and "food".
37684 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37687 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37688 boundary crossing check and subsequent code generation agree.
37690 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37692 * config/aarch64/aarch64.c (has_memory_op): Delete.
37693 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37696 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37698 PR rtl-optimization/79388
37699 PR rtl-optimization/79450
37700 * combine.c (distribute_notes): When removing TEM_INSN for which
37701 corresponding dest has last value recorded, invalidate that last
37704 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37706 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37707 of explicit '@'. Add missing assembly comment marker on branch costs
37710 2017-02-13 Nathan Sidwell <nathan@acm.org>
37712 * gengtype-lex.l (<in_struct>): Add '/'.
37714 2017-02-13 Martin Liska <mliska@suse.cz>
37717 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37719 2017-02-13 Richard Biener <rguenther@suse.de>
37721 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37723 * configure: Re-generate.
37724 * config.in: Likewise.
37725 * graphite-dependences.c: Simplify as if
37726 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37727 * graphite-isl-ast-to-gimple.c: Likewise.
37728 * graphite-optimize-isl.c: Likewise.
37729 * graphite-poly.c: Likewise.
37730 * graphite-sese-to-poly.c: Likewise.
37731 * graphite.h: Likewise.
37732 * toplev.c: Include isl/version.h and use isl_version () for
37733 printing the ISL version.
37734 * doc/install.texi: Update ISL requirement.
37736 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37738 * doc/standards.texi (Standards): Update reference to
37741 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37743 * doc/extend.texi (Named Address Spaces): sourceware.org now
37745 * doc/install.texi (Binaries): Ditto.
37748 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37750 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37751 standard terminology "stringize"/"stringizing" throughout.
37752 * doc/cppinternals.texi: Likewise.
37754 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37756 * doc/extend.texi: Fix some spelling mistakes and typos.
37757 * doc/invoke.texi: Likewise.
37759 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37762 * params.def (inline-min-speedup) Change from 10 to 8.
37764 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37766 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37769 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37772 * ipa-inline-analysis.c (get_minimal_bb): New function.
37773 (record_modified): Use it.
37774 (remap_edge_change_prob): Handle also ancestor functions.
37776 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37778 * doc/contrib.texi (Contributors): Remove broken link into
37779 the Mauve CVS repository.
37781 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37783 PR middle-end/79454
37784 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37785 result computation whenever lhs doesn't have vector mode, not
37786 just when it has BLKmode.
37788 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37790 * doc/makefile.texi (profiledbootstrap): Refer to the
37791 installation instructions only in textual form.
37793 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37796 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37798 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37800 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37801 (Specific): Update mingw-w64 reference.
37803 (Specific): Remove broken link to Renesas RX processor.
37805 2017-02-10 Richard Biener <rguenther@suse.de>
37807 * toplev.c (process_options): Do not mention obsolete graphite
37808 options when printing sorry message about missing graphite support.
37809 Mention -floop-nest-optimize.
37811 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37813 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37814 (vtst_p16): Likewise.
37815 (vtstq_p8): Likewise.
37816 (vtstq_p16): Likewise.
37818 (vtstq_p64): Likewise.
37819 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37820 (vset_lane_p64): New.
37821 (vsetq_lane_p64): New.
37823 2017-02-10 Jakub Jelinek <jakub@redhat.com>
37825 PR tree-optimization/79411
37826 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
37827 stmt operands are SSA_NAMEs used in abnormal phis.
37828 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
37831 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37834 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
37837 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37839 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
37841 2017-02-09 Jakub Jelinek <jakub@redhat.com>
37843 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
37847 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
37848 not arbitrary TREE_CONSTANT.
37851 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
37852 "omp declare target link" attribute unless is_global_var.
37853 * omp-offload.c (find_link_var_op): Likewise.
37855 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
37856 Chung-Lin Tang <cltang@codesourcery.com>
37858 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
37860 (gimplify_adjust_omp_clauses): Don't delete TILE.
37861 (gimplify_omp_for): Deal with TILE.
37862 * internal-fn.c (expand_GOACC_TILE): New function.
37863 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
37865 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
37866 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
37868 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
37869 avoid DIV for outermost collapse var.
37870 (expand_oacc_for): Insert tile element loop as needed. Adjust.
37871 Remove out of date comments, fix whitespace.
37872 * omp-general.c (omp_extract_for_data): Deal with tiling.
37873 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
37874 adjust OLF_DIM_BASE value.
37875 (struct omp_for_data): Add tiling field.
37876 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
37877 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
37878 for auto loops. Remove default auto determining, moved to
37879 oacc_loop_fixed_partitions.
37880 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
37881 stmts, add e_mask field.
37882 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
37883 (oacc_thread_numbers): Use oacc_dim_call.
37884 (oacc_xform_tile): New.
37885 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
37886 (finish_oacc_loop): Adjust for ifns vector.
37887 (oacc_loop_discover_walk): Append loop abstraction sites to list,
37888 add case for GOACC_TILE fns.
37889 (oacc_loop_xform_loop): Delete.
37890 (oacc_loop_process): Iterate over call list directly, and add
37891 handling for GOACC_TILE fns.
37892 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
37894 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
37895 vector partitioning to outer loops. Assign 2 partitions to loops
37896 when available. Add TILE handling.
37897 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
37898 (execite_oacc_device_lower): Process GOACC_TILE fns,
37899 ignore unknown specs.
37900 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
37901 * tree.c (omp_clause_num_ops): Adjust TILE ops.
37902 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
37904 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
37906 * configure.ac (ACX_BUGURL): Update.
37907 * configure: Regenerate.
37909 2017-02-09 Richard Biener <rguenther@suse.de>
37911 PR tree-optimization/69823
37912 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
37913 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
37915 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37917 * config/arc/arc-c.def: Add __NPS400__ definition.
37918 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
37919 (TARGET_NPS400): Define.
37921 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37923 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
37925 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
37926 pointer, arch_info.
37927 (arc_cpu_types): Fill the arch_info field with a pointer into the
37928 arc_arch_types table.
37929 (arc_selected_cpu): Declare.
37930 * config/arc/arc.c (arc_selected_cpu): Make global.
37931 (arc_selected_arch): Delete.
37932 (arc_base_cpu): Delete.
37933 (arc_override_options): Remove references to deleted variables,
37934 update access to arch information.
37935 (ARC_OPT): Update access to arch information.
37936 (ARC_OPTX): Likewise.
37937 * config/arc/arc.h (arc_base_cpu): Remove declaration.
37938 (TARGET_ARC600): Update access to arch information.
37939 (TARGET_ARC601): Likewise.
37940 (TARGET_ARC700): Likewise.
37941 (TARGET_EM): Likewise.
37942 (TARGET_HS): Likewise.
37943 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
37946 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
37949 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
37950 condition/operands for integer GE/LE/GEU/LEU operations.
37952 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
37954 PR translation/79397
37955 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
37958 2017-02-08 Martin Jambor <mjambor@suse.cz>
37961 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
37962 whether allocation happened.
37963 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
37964 nothing was allocated.
37966 2017-02-08 Jakub Jelinek <jakub@redhat.com>
37968 PR tree-optimization/79408
37969 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
37970 constant, but SSA_NAME with a known integer range, use the minimum
37971 of that range instead of op1 to determine if modulo can be replaced
37972 with its first operand.
37974 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37976 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
37978 2017-02-08 Richard Biener <rguenther@suse.de>
37980 PR tree-optimization/71824
37981 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37982 Check all loops contained in the merged region.
37984 2017-02-07 Andrew Pinski <apinski@cavium.com>
37986 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
37988 2017-02-07 Andrew Pinski <apinski@cavium.com>
37990 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
37991 (thunderxt88): Likewise.
37992 (thunderxt81): Disable LSE and change v8.1 to v8.
37993 (thunderxt83): Likewise.
37995 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37996 Richard Biener <rguenther@suse.de>
37998 PR middle-end/79399
37999 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38000 type from int to size_t.
38001 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38003 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38005 PR rtl-optimization/79386
38006 * cprop.c (bypass_conditional_jumps): Initialize
38007 bypass_last_basic_block already before splitting bbs after
38008 unconditional traps...
38009 (bypass_conditional_jumps): ... rather than here.
38012 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38013 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38014 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38015 fix -masm=intel patterns.
38017 2017-02-07 Richard Biener <rguenther@suse.de>
38019 PR tree-optimization/79256
38020 PR middle-end/79278
38021 * builtins.c (get_object_alignment_2): Use min_align_of_type
38022 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38023 and ADJUST_FIELD_ALIGN.
38025 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38027 * doc/tm.texi: Regenerate.
38028 * stor-layout.c (layout_decl): Adjust.
38029 (update_alignment_for_field): Likewise.
38030 (place_field): Likewise.
38031 (min_align_of_type): Likewise.
38032 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38033 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38034 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38035 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38036 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38037 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38038 * config/i386/i386.c (x86_field_alignment): Likewise.
38039 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38040 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38041 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38042 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38043 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38044 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38048 2017-01-30 Richard Biener <rguenther@suse.de>
38050 PR tree-optimization/79256
38051 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38052 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38055 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38057 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38058 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38059 builtins to SImode and emit a zero-extend, if necessary.
38061 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38063 * docs/invoke.texi (RISC-V Options): Alphabetize.
38065 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38067 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
38070 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38072 * config/riscv/riscv.c: New file.
38073 * common/config/riscv/riscv-common.c: Likewise.
38074 * config.gcc: Likewise.
38075 * config/riscv/constraints.md: Likewise.
38076 * config/riscv/elf.h: Likewise.
38077 * config/riscv/generic.md: Likewise.
38078 * config/riscv/linux.h: Likewise.
38079 * config/riscv/multilib-generator: Likewise.
38080 * config/riscv/peephole.md: Likewise.
38081 * config/riscv/pic.md: Likewise.
38082 * config/riscv/predicates.md: Likewise.
38083 * config/riscv/riscv-builtins.c: Likewise.
38084 * config/riscv/riscv-c.c: Likewise.
38085 * config/riscv/riscv-ftypes.def: Likewise.
38086 * config/riscv/riscv-modes.def: Likewise.
38087 * config/riscv/riscv-opts.h: Likewise.
38088 * config/riscv/riscv-protos.h: Likewise.
38089 * config/riscv/riscv.h: Likewise.
38090 * config/riscv/riscv.md: Likewise.
38091 * config/riscv/riscv.opt: Likewise.
38092 * config/riscv/sync.md: Likewise.
38093 * config/riscv/t-elf-multilib: Likewise.
38094 * config/riscv/t-linux: Likewise.
38095 * config/riscv/t-linux-multilib: Likewise.
38096 * config/riscv/t-riscv: Likewise.
38097 * configure.ac: Likewise.
38098 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
38099 Waterman as RISC-V maintainers.
38100 * doc/install.texi: Add RISC-V entries.
38101 * doc/invoke.texi: Add RISC-V options section.
38102 * doc/md.texi: Add RISC-V constraints section.
38103 * configure: Regenerated.
38105 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
38108 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
38109 false values to be constant vectors with all 0 or all 1 bits set.
38110 (vcondu<mode><mode>): Likewise.
38111 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
38113 (fpmask_comparison_operator): Update comment.
38114 (vecint_comparison_operator): New predicate.
38115 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
38116 vector conditionals when the true and false values are constant
38117 vectors with all 0 bits or all 1 bits set.
38119 2017-02-06 Martin Sebor <msebor@redhat.com>
38121 PR tree-optimization/79376
38122 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
38124 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
38126 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
38127 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
38128 to simplify split condition.
38130 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38132 * omp-expand.c (oxpand_omp_atomic_fetch_op,
38133 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
38136 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
38138 PR rtl-optimization/68664
38139 * target.def (can_speculate_insn): New hook.
38140 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
38141 * doc/tm.texi: Regenerate.
38142 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
38143 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
38144 (rs6000_sched_can_speculate_insn): New function.
38146 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38148 PR tree-optimization/79284
38149 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
38150 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
38151 vectorizable_mask_load_store, vectorizable_operation,
38152 vect_is_simple_cond, get_same_sized_vectype): Use it instead
38153 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
38154 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
38155 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
38156 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
38157 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
38158 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
38159 is_gimple_assign (stmt). Replace another such test with
38160 is_gimple_assign (stmt).
38162 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38165 * config/avr/avr.c (rtl-iter.h): Include it.
38166 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
38167 (avr_legitimate_combined_insn): ...and implementation.
38169 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
38171 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
38172 * config/s390/s390.c (s390_const_operand_ok)
38173 (s390_canonicalize_comparison, s390_extract_part)
38174 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
38175 (s390_contiguous_bitmask_p, s390_rtx_costs)
38176 (legitimize_pic_address): Likewise.
38177 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
38178 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
38179 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
38180 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
38181 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
38183 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38185 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
38186 REGNO($0) == REGNO($1).
38188 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38190 * config/s390/linux.h(SIZE_TYPE): Add comment.
38192 2017-02-06 Julian Brown <julian@codesourcery.com>
38193 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38194 Virendra Pathak <virendra.pathak@broadcom.com>
38196 * config/aarch64/aarch64-cores.def: Change the scheduler
38198 * config/aarch64/aarch64.md: Include thunderx2t99.md.
38199 * config/aarch64/thunderx2t99.md: New file.
38201 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38203 * doc/standards.texi (Go Language): Update link to language
38206 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38208 * tree-eh.c (lower_resx): Sanitize profile.
38209 (cleanup_empty_eh_move_lp): Likewise.
38211 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38214 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
38216 * cfgloopmanip.h (loop_version): Update prototype.
38217 * modulo-sched.c (sms_schedule): Update call of loop_version.
38218 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
38219 * tree-parloops.c (gen_parallel_loop): Likewise.
38220 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
38221 * tree-ssa-loop-split.c (split_loop): Likewise.
38222 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
38223 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
38225 2017-02-05 Martin Liska <mliska@suse.cz>
38228 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
38230 (print_operand_address): Initialize a struct to zero.
38232 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38234 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
38235 garbage collector only in textual form.
38237 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38239 * doc/extend.texi (x86 specific memory model extensions for
38240 transactional memory): Simplify a phrase.
38242 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
38245 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
38246 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
38247 (atomic_storedi_1): Likewise.
38249 2017-02-04 Jakub Jelinek <jakub@redhat.com>
38251 PR tree-optimization/79338
38252 * tree-parloops.c (gather_scalar_reductions): Don't call
38253 vect_analyze_loop_form for loop->inner before destroying loop's
38256 2017-02-03 Martin Sebor <msebor@redhat.com>
38258 PR tree-optimization/79327
38259 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
38260 when precision has resulted in leading zeros.
38261 (format_integer): Adjust the likely counter to assume an unknown
38262 argument that may be zero is non-zero.
38264 2017-02-03 Jason Merrill <jason@redhat.com>
38267 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
38268 avoid copying non-taken branch.
38270 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38272 PR tree-optimization/79340
38273 * tree-vect-loop.c (vectorizable_reduction): Release
38274 vec_defs elements after safe_splicing them into other vectors.
38277 PR tree-optimization/79327
38278 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
38279 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
38281 (format_integer): Use wide_int_to_tree instead of build_int_cst
38282 + to_?hwi. If argmin is NULL, just set argmin and argmax to
38283 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
38284 of shortest and longest sequence.
38286 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
38288 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
38289 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
38291 2017-02-03 Walter Lee <walt@tilera.com>
38294 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
38295 after initial stackframe link reg save.
38296 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
38298 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38301 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
38302 wu for stxssp alternative.
38304 2017-02-03 Martin Sebor <msebor@redhat.com>
38306 PR tree-optimization/79352
38307 * gimple-fold.c (get_range_strlen): Add argument.
38308 (get_range_strlen): Change return type to bool.
38309 (get_maxval_strlen): Pass in a dummy argument.
38310 * gimple-fold.h (get_range_strlen): Change return type to bool.
38311 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
38312 * tree.h (array_at_struct_end_p): Add argument.
38313 * tree.c (array_at_struct_end_p): Handle it.
38315 2017-02-03 Martin Liska <mliska@suse.cz>
38318 * multiple_target.c (create_dispatcher_calls): Redirect edge
38319 from a caller of a dispatcher.
38320 (expand_target_clones): Make the clones local.
38321 (ipa_target_clone): Do both target clones and resolvers.
38322 (ipa_dispatcher_calls): Remove the pass.
38323 (pass_dispatcher_calls::gate): Likewise.
38324 (make_pass_dispatcher_calls): Likewise.
38325 * passes.def (pass_target_clone): Put as very first IPA early
38328 2017-02-03 Martin Liska <mliska@suse.cz>
38330 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
38331 in case of a function with ifunc attribute.
38333 2017-02-03 Martin Liska <mliska@suse.cz>
38335 * cgraph.c (cgraph_node::dump): Dump function version info.
38336 * symtab.c (symtab_node::dump_base): Add missing new line.
38338 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38340 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
38341 (ifcombine_ifandif): Use it.
38343 2017-02-03 Martin Liska <mliska@suse.cz>
38345 * doc/invoke.texi: Document default value for
38346 use-after-scope-direct-emission-threshold.
38348 2017-02-03 Martin Liska <mliska@suse.cz>
38350 PR tree-optimization/79339
38351 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
38352 (format_floating): Likewise.
38354 2017-02-03 Martin Liska <mliska@suse.cz>
38357 * ipa-prop.c (ipa_node_params_t::insert): Remove current
38359 (ipa_node_params_t::remove): Likewise.
38360 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
38361 initialization from removed ipa_node_params_t::insert.
38362 (ipa_node_params::~ipa_node_params): Move from removed
38363 ipa_node_params_t::release.
38364 * symbol-summary.h (symbol_summary::m_released): New member.
38365 Do not release a summary twice. Do not allow to call finalizer
38366 for types of a summary that live in GGC memory.
38368 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38370 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
38373 2017-02-02 Martin Sebor <msebor@redhat.com>
38375 PR middle-end/79275
38376 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
38377 (format_string): Tighten up the range of output for non-constant
38378 strings and correct the expected range for wide non-constant strings.
38380 2017-02-02 Martin Sebor <msebor@redhat.com>
38382 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
38384 PR middle-end/32003
38385 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
38387 (-fdump-tree-@var): Add to index and document how to come up
38388 with pass-specific option and dump file names.
38389 (-fdump-passes): Clarify where to look for output.
38391 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38393 PR middle-end/77445
38394 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
38395 statistics of the analyzed path; allow threading for speed when
38396 any of BBs along the path are optimized for speed.
38398 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
38400 PR middle-end/78468
38401 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
38402 settings of the virtual registers.
38405 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
38407 * explow.c (get_dynamic_stack_size): Take known alignment of stack
38408 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
38411 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38413 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
38414 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
38416 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38418 * config/s390/s390.md: Add missing comments with the expanded
38420 * config/s390/vector.md: Likewise.
38421 * config/s390/vx-builtins.md: Likewise.
38423 2017-02-02 Jakub Jelinek <jakub@redhat.com>
38426 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
38427 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
38428 conditions on a single line.
38430 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38432 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38433 __S390_VX__ to __VX__.
38435 2017-02-01 Andrew Pinski <apinski@cavium.com>
38437 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
38438 stmt_info to record_stmt_cost.
38439 (vect_get_known_peeling_cost): Pass stmt_info if known to
38441 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
38442 cpu_vector_cost field into
38443 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
38444 field into vec_int_stmt_cost and vec_fp_stmt_cost.
38445 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
38446 splitting of scalar_stmt_cost and vec_stmt_cost.
38447 (thunderx_vector_cost): Likewise.
38448 (cortexa57_vector_cost): LIkewise.
38449 (exynosm1_vector_cost): Likewise.
38450 (xgene1_vector_cost): Likewise.
38451 (thunderx2t99_vector_cost): Improve after the splitting of the two
38453 (aarch64_builtin_vectorization_cost): Update for the splitting of
38454 scalar_stmt_cost and vec_stmt_cost.
38456 2017-02-01 Torvald Riegel <triegel@redhat.com>
38457 Richard Henderson <rth@redhat.com>
38459 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
38460 conditional on existance of a fast atomic load.
38461 * optabs-query.c (can_atomic_load_p): New function.
38462 * optabs-query.h (can_atomic_load_p): Declare it.
38463 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
38464 no fast atomic load is available for the particular size of access.
38465 (expand_atomic_compare_and_swap): Likewise.
38466 (expand_atomic_load): Likewise.
38467 (expand_atomic_store): Likewise.
38468 (expand_atomic_fetch_op): Likewise.
38469 * testsuite/lib/target-supports.exp
38470 (check_effective_target_sync_int_128): Remove x86 because it provides
38471 no fast atomic load.
38472 (check_effective_target_sync_int_128_runtime): Likewise.
38474 2017-02-01 Richard Biener <rguenther@suse.de>
38476 * graphite.c: Include tree-vectorizer.h for find_loop_location.
38477 (graphite_transform_loops): Provide opt-info for optimized nests.
38478 * tree-parloop.c (parallelize_loops): Provide opt-info for
38479 parallelized loops.
38481 2017-02-01 Richard Biener <rguenther@suse.de>
38483 PR middle-end/79315
38484 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
38485 was not set before.
38487 2017-02-01 Richard Biener <rguenther@suse.de>
38489 PR tree-optimization/71824
38490 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38491 Verify the loops are valid in the merged SESE region.
38492 (scop_detection::can_represent_loop_1): Check analyzing the
38493 evolution of the number of iterations in the region succeeds.
38495 2017-01-31 Ian Lance Taylor <iant@golang.org>
38497 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
38498 REG_ARGS_SIZE note to 32-bit push insns and call insn.
38500 2017-01-31 David Malcolm <dmalcolm@redhat.com>
38502 PR preprocessor/79210
38503 * input.c (get_substring_ranges_for_loc): Replace line_width
38504 assertion with error-handling.
38506 2017-01-31 Richard Biener <rguenther@suse.de>
38508 PR tree-optimization/77318
38509 * graphite-sese-to-poly.c (extract_affine): Fix assert.
38510 (create_pw_aff_from_tree): Take loop parameter.
38511 (add_condition_to_pbb): Pass loop of the condition to
38512 create_pw_aff_from_tree.
38514 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38516 * config/s390/s390.c (s390_asan_shadow_offset): New function.
38517 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
38519 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
38523 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
38525 (convert_int_to_float128): Likewise.
38526 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
38527 (convert_int_to_float128): Likewise.
38528 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
38529 (UNSPEC_IEEE128_CONVERT): Likewise.
38530 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
38531 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
38532 Use local variables for IBM extended format.
38533 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
38534 (fix_trunc<mode>si2_fprs): Likewise.
38535 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
38536 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
38537 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
38538 to know that we can now have integers of all sizes in vector
38540 (fix<uns>_<mode>di2_hw): Likewise.
38541 (float<uns>_<mode>si2_hw): Likewise.
38542 (fix_<mode>si2_hw): Likewise.
38543 (fixuns_<mode>si2_hw): Likewise.
38544 (float<uns>_<mode>di2_hw): Likewise.
38545 (float_<mode>di2_hw): Likewise.
38546 (float_<mode>si2_hw): Likewise.
38547 (floatuns_<mode>di2_hw): Likewise.
38548 (floatuns_<mode>si2_hw): Likewise.
38549 (xscvqp<su>wz_<mode>): Delete, no longer used.
38550 (xscvqp<su>dz_<mode>): Likewise.
38551 (xscv<su>dqp_<mode>): Likewise.
38552 (ieee128_mfvsrd_64bit): Likewise.
38553 (ieee128_mfvsrd_32bit): Likewise.
38554 (ieee128_mfvsrwz): Likewise.
38555 (ieee128_mtvsrw): Likewise.
38556 (ieee128_mtvsrd_64bit): Likewise.
38557 (ieee128_mtvsrd_32bit): Likewise.
38559 2017-01-31 Martin Liska <mliska@suse.cz>
38562 * ipa-prop.c (ipa_free_all_node_params): Call release method
38563 instead of ~sumbol_summary to not to trigger double times
38566 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
38568 PR tree-optimization/71691
38569 * bitmap.h (class auto_bitmap): New.
38570 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
38571 is_maybe_undefined instead of ssa_undefined_value_p.
38573 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38575 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38576 __S390_ARCH_LEVEL__ to __ARCH__.
38578 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38580 PR tree-optimization/79267
38581 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38582 if should_remove_lhs_p is true.
38584 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38587 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38588 (add_alignment_attribute): New.
38589 (base_type_die): Add alignment attribute.
38590 (subrange_type_die): Likewise.
38591 (modified_type_die): Likewise.
38592 (gen_array_type_die): Likewise.
38593 (gen_descr_array_type_die: Likewise.
38594 (gen_enumeration_type_die): Likewise.
38595 (gen_subprogram_die): Likewise.
38596 (gen_variable_die): Likewise.
38597 (gen_field_die): Likewise.
38598 (gen_ptr_to_mbr_type_die): Likewise.
38599 (gen_struct_or_union_type_die): Likewise.
38600 (gen_subroutine_type_die): Likewise.
38601 (gen_typedef_die): Likewise.
38602 (base_type_cmp): Compare alignment attribute.
38604 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38607 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38608 (setb_unsigned) New pattern for setb with CCUNS.
38609 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38610 subfc./subfe sequence to avoid overflow problems. Generate a
38611 shorter sequence with cmpld/setb for power9.
38612 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38613 for generating subfc. instruction.
38614 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38615 now uses this instruction.
38617 2017-01-30 Ian Lance Taylor <iant@google.com>
38620 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38621 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38623 2017-01-30 Martin Sebor <msebor@redhat.com>
38625 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38626 Move constant to the right of a relational operator.
38627 (get_mpfr_format_length, format_character, format_string): Ditto.
38628 (should_warn_p, maybe_warn): Same.
38630 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38632 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38635 * asan.c (get_translation_unit_decl): Remove function.
38636 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38638 2017-01-30 Martin Liska <mliska@suse.cz>
38640 PR gcov-profile/79259
38641 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38642 -fprofile-generate.
38644 2017-01-30 Martin Liska <mliska@suse.cz>
38647 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38648 Initialize variables with NULL value.
38650 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38653 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38655 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38657 2017-01-30 Richard Biener <rguenther@suse.de>
38659 PR tree-optimization/79276
38660 * tree-vrp.c (process_assert_insertions): Properly adjust common
38661 when removing a duplicate.
38663 2017-01-30 Richard Biener <rguenther@suse.de>
38665 PR tree-optimization/79256
38666 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38667 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38669 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38671 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38674 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38675 ("*r<noxa>sbg_<mode>_sll_bitmask")
38676 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38677 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38678 Use contiguous_bitmask_nowrap_operand.
38680 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38683 * config/rs6000/altivec.h (vec_xl): Revise #define.
38684 (vec_xst): Likewise.
38686 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38688 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38690 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38692 PR rtl-optimization/79194
38693 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38694 traps before call to bypass_conditional_jumps.
38696 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38698 PR tree-optimization/71374
38699 * lra-constraints.c (check_conflict_input_operands): New.
38700 (match_reload): Use it.
38702 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38705 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38706 account to calculate conflict_set.
38708 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38710 PR rtl-optimization/78559
38711 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38712 other_insn in combine.
38714 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38716 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38717 uint16_type_node for BT_UINT16.
38719 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38721 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38722 "RTL Tests" to menu.
38723 (GIMPLE Tests): New node.
38724 (RTL Tests): New node.
38726 2017-01-27 Richard Biener <rguenther@suse.de>
38728 PR tree-optimization/79245
38729 * tree-loop-distribution.c (distribute_loop): Apply cost
38730 modeling also to detected patterns.
38732 2017-01-27 Richard Biener <rguenther@suse.de>
38734 PR tree-optimization/71433
38735 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38736 (compare_assert_loc): New function.
38737 (process_assert_insertions): Sort and optimize assert locations
38738 to remove duplicates and push down identical assertions on
38739 edges to their destination block.
38741 2017-01-27 Richard Biener <rguenther@suse.de>
38743 PR tree-optimization/79244
38744 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38745 out SSA names even if abnormal.
38747 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38749 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38750 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38751 instead of MPFR_RNDN.
38753 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38756 * arm.c (arm_option_override): Don't call build_target_option_node
38757 until after doing all option overrides.
38758 (arm_valid_target_attribute_tree): Likewise.
38760 2017-01-27 Martin Liska <mliska@suse.cz>
38762 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38763 for {cd}tors and C++ {cd}tors.
38765 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38767 * config/s390/s390.md ("*setmem_long_and")
38768 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38770 2017-01-26 Martin Sebor <msebor@redhat.com>
38772 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38775 2017-01-26 Martin Sebor <msebor@redhat.com>
38777 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38778 HAVE_DFmode before using XFmode or DFmode.
38779 (parse_directive): Avoid using the z length modifier to avoid
38780 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38782 PR middle-end/78703
38783 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38784 to accept adjustment as an array.
38785 (get_int_range): New function.
38786 (struct directive): Make width and prec arrays.
38787 (directive::set_width, directive::set_precision): Call get_int_range.
38788 (format_integer, format_floating): Handle width and precision ranges.
38789 (format_string, parse_directive): Same.
38791 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38794 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38795 comdat_type_p set, just clone them, but keep the children in the
38799 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38800 which have direct callers with -fvar-tracking-assignments enabled
38802 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38803 inside of type units.
38805 2017-01-26 Martin Sebor <msebor@redhat.com>
38807 PR middle-end/78703
38808 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38810 (struct format_result): Replace number_chars, number_chars_min,
38811 and number_chars_max with a single member of struct result_range.
38813 (format_result::operator+=): Adjust.
38814 (struct fmtresult): Remove bounded. Handle likely and unlikely
38816 (fmtresult::adjust_for_width_or_precision): New function.
38817 (fmtresult:type_max_digits): New function.
38818 (bytes_remaining): Handle likely and unlikely counters.
38819 (min_bytes_remaining): Remove.
38820 (format_percent): Simplify.
38821 (format_integer, format_floating): Set likely and unlikely counters.
38822 (get_string_length, format_character, format_string): Same.
38823 (format_plain, should_warn_p): New function.
38824 (maybe_warn): Call should_warn_p. Update diagnostic messages
38825 and handle those for all directives, including plain strings.
38826 (format_directive): Handle likely and unlikely counters.
38827 Remove unnecessary quoting from diagnostics. Add an informational
38829 (add_bytes): Remove.
38830 (pass_sprintf_length::compute_format_length): Simplify.
38831 (try_substitute_return_value): Handle likely and unlikely counters.
38833 2017-01-26 Carl Love <cel@us.ibm.com>
38835 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
38836 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
38838 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
38841 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
38842 endianess for subregs into account.
38843 * lra-constraints.c (lra_constraints): Do risky transformations
38844 always on the first iteration.
38845 * lra-lives.c (check_pseudos_live_through_calls): Add arg
38846 last_call_used_reg_set.
38847 (process_bb_lives): Define and use last_call_used_reg_set.
38848 * lra.c (lra): Always continue after lra_constraints on the first
38851 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
38853 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
38855 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
38857 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38859 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
38860 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
38861 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
38862 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
38863 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
38864 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
38865 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
38866 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
38867 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
38869 2017-01-26 Marek Polacek <polacek@redhat.com>
38872 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
38873 for the third operand.
38875 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38877 PR middle-end/79236
38878 * omp-low.c (struct omp_context): Add simt_stmt field.
38879 (scan_omp_for): Return omp_context *.
38880 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
38881 context to the _simt_ SIMD stmt.
38882 (lower_omp_for): For combined SIMD with sibling _simt_
38883 SIMD, make sure to use the same decls in _looptemp_
38884 clauses as in the sibling.
38886 2017-01-26 David Sherwood <david.sherwood@arm.com>
38888 PR middle-end/79212
38889 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
38892 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38895 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
38896 emit fld b; fld a; if possible.
38898 * brig-builtins.def: Update copyright years.
38899 * config/arm/arm_acle_builtins.def: Update copyright years.
38901 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
38904 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
38905 constraint instead of o for the stxsd instruction.
38907 2017-01-25 Carl Love <cel@us.ibm.com>
38909 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
38910 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
38912 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
38914 * doc/invoke.texi (C++ Dialect Options): Fix typo.
38916 2017-01-25 Richard Biener <rguenther@suse.de>
38918 PR tree-optimization/69264
38919 * target.def (vector_alignment_reachable): Improve documentation.
38920 * doc/tm.texi: Regenerate.
38921 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
38923 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
38924 earlier changes with respect to TYPE_USER_ALIGN.
38925 (vector_alignment_reachable_p): Likewise. Improve dumping.
38927 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38930 * config/arm/arm.md (xordi3): Force constant operand into a register
38933 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38935 * doc/invoke.texi (-fstore-merging): Correct default optimization
38936 levels at which it is enabled.
38937 (-O): Move -fstore-merging from list to...
38940 2017-01-25 Richard Biener <rguenther@suse.de>
38943 * omp-expand.c: Include debug.h.
38944 (expand_omp_taskreg): Make sure to generate early debug before
38945 outlining anything from a function.
38946 (expand_omp_target): Likewise.
38947 (grid_expand_target_grid_body): Likewise.
38949 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
38952 * asan.c (get_translation_unit_decl): New function.
38953 (asan_add_global): Extract modules file name from globals
38954 TRANSLATION_UNIT_DECL name.
38956 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
38959 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
38960 for long calls with APCS frame and VFP.
38962 2017-01-24 David Malcolm <dmalcolm@redhat.com>
38964 * cfg.c (original_copy_tables_initialized_p): New function.
38965 * cfg.h (original_copy_tables_initialized_p): New decl.
38966 * cfgrtl.c (relink_block_chain): Guard the call to
38967 free_original_copy_tables with a call to
38968 original_copy_tables_initialized_p.
38969 * cgraph.h (symtab_node::native_rtl_p): New decl.
38970 * cgraphunit.c (symtab_node::native_rtl_p): New function.
38971 (symtab_node::needed_p): Don't assert for early assembly output
38972 for __RTL functions.
38973 (cgraph_node::finalize_function): Set "force_output" for __RTL
38975 (cgraph_node::analyze): Bail out early for __RTL functions.
38976 (analyze_functions): Update assertion to support __RTL functions.
38977 (cgraph_node::expand): Bail out early for __RTL functions.
38978 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
38980 * function.h (struct function): Update comment for field
38982 * gimple-expr.c: Include "tree-pass.h".
38983 (gimple_has_body_p): Return false for __RTL functions.
38984 * Makefile.in (OBJS): Add run-rtl-passes.o.
38985 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
38987 (gcc::pass_manager::get_clean_slate): New accessor.
38988 * passes.c: Include "insn-addr.h".
38989 (should_skip_pass_p): Add logging. Update logic for running
38990 "expand" to be compatible with both __GIMPLE and __RTL. Guard
38991 property-provider override so it is only done for gimple passes.
38993 (skip_pass): New function.
38994 (execute_one_pass): Call skip_pass when skipping passes.
38995 * read-md.c (md_reader::read_char): Support filtering
38996 the input to a subset of line numbers.
38997 (md_reader::md_reader): Initialize fields
38998 m_first_line and m_last_line.
38999 (md_reader::read_file_fragment): New function.
39000 * read-md.h (md_reader::read_file_fragment): New decl.
39001 (md_reader::m_first_line): New field.
39002 (md_reader::m_last_line): New field.
39003 * read-rtl-function.c (function_reader::create_function): Only
39004 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39005 curr_properties. Set DECL_INITIAL to a dummy block.
39006 (read_rtl_function_body_from_file_range): New function.
39007 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39009 * run-rtl-passes.c: New file.
39010 * run-rtl-passes.h: New file.
39012 2017-01-24 Jeff Law <law@redhat.com>
39014 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39017 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39019 PR tree-optimization/79159
39020 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39021 (record_nonwrapping_iv): Improve boundary using above function if no
39022 value range information.
39024 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39025 Martin Jambor <mjambor@suse.cz>
39027 * brig-builtins.def: New file.
39028 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39029 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39030 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39031 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39032 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39033 * builtin-types.def (BT_INT8): New.
39034 (BT_INT16): Likewise.
39035 (BT_UINT8): Likewise.
39036 (BT_UINT16): Likewise.
39037 (BT_FN_ULONG): Likewise.
39038 (BT_FN_UINT_INT): Likewise.
39039 (BT_FN_UINT_ULONG): Likewise.
39040 (BT_FN_UINT_LONG): Likewise.
39041 (BT_FN_UINT_PTR): Likewise.
39042 (BT_FN_ULONG_PTR): Likewise.
39043 (BT_FN_INT8_FLOAT): Likewise.
39044 (BT_FN_INT16_FLOAT): Likewise.
39045 (BT_FN_UINT32_FLOAT): Likewise.
39046 (BT_FN_UINT16_FLOAT): Likewise.
39047 (BT_FN_UINT8_FLOAT): Likewise.
39048 (BT_FN_UINT64_FLOAT): Likewise.
39049 (BT_FN_UINT16_UINT32): Likewise.
39050 (BT_FN_UINT32_UINT16): Likewise.
39051 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39052 (BT_FN_INT_PTR_INT): Likewise.
39053 (BT_FN_UINT_PTR_UINT): Likewise.
39054 (BT_FN_LONG_PTR_LONG): Likewise.
39055 (BT_FN_ULONG_PTR_ULONG): Likewise.
39056 (BT_FN_VOID_UINT64_UINT64): Likewise.
39057 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39058 (BT_FN_INT8_INT8_INT8): Likewise.
39059 (BT_FN_INT16_INT16_INT16): Likewise.
39060 (BT_FN_INT_INT_INT): Likewise.
39061 (BT_FN_UINT_FLOAT_UINT): Likewise.
39062 (BT_FN_FLOAT_UINT_UINT): Likewise.
39063 (BT_FN_ULONG_UINT_UINT): Likewise.
39064 (BT_FN_ULONG_UINT_PTR): Likewise.
39065 (BT_FN_ULONG_ULONG_ULONG): Likewise.
39066 (BT_FN_UINT_UINT_UINT): Likewise.
39067 (BT_FN_VOID_UINT_PTR): Likewise.
39068 (BT_FN_UINT_UINT_PTR: Likewise.
39069 (BT_FN_UINT32_UINT64_PTR): Likewise.
39070 (BT_FN_INT_INT_UINT_UINT): Likewise.
39071 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
39072 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
39073 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
39074 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
39075 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
39076 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
39077 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
39078 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
39079 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
39080 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
39081 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
39082 * doc/frontends.texi: List BRIG FE.
39083 * doc/install.texi (Testing): Add BRIG tesring requirements.
39084 * doc/invoke.texi (Overall Options): Mention BRIG.
39085 * doc/standards.texi (Standards): Doucment BRIG HSA version.
39087 2017-01-24 Richard Biener <rguenther@suse.de>
39089 PR translation/79208
39090 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
39092 2017-01-24 Martin Jambor <mjambor@suse.cz>
39095 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
39096 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
39097 and known_contexts.
39099 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
39101 PR middle-end/79123
39102 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
39103 casts from signed to unsigned really don't have a range.
39105 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
39107 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
39108 GMP_RNDx for compatiblity.
39110 2017-01-24 Martin Liska <mliska@suse.cz>
39113 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
39114 that would prevent us to call alloca with -1 as argument.
39116 2017-01-24 Jakub Jelinek <jakub@redhat.com>
39118 * dwarf2out.c (output_compilation_unit_header, output_file_names):
39119 Avoid -Wformat-security warning.
39121 2017-01-23 Andrew Pinski <apinski@cavium.com>
39123 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
39126 2017-01-23 Martin Sebor <msebor@redhat.com>
39128 PR middle-end/78703
39129 * gimple-ssa-sprintf.c (warn_level): New global.
39130 (format_integer): Use it here and throughout the rest of the file.
39131 Use the same switch to compute sign as base.
39132 (maybe_warn): New function.
39133 (format_directive): Factor out warnings into maybe_warn.
39134 Add debugging output. Use warn_level.
39135 (add_bytes): Use warn_level.
39136 (pass_sprintf_length::compute_format_length): Add debugging output.
39137 (try_substitute_return_value): Same.
39138 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
39140 PR middle-end/78703
39141 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
39142 (struct fmtresult, format_integer, format_floating): Adjust.
39143 (fmtresult::fmtresult): Set max correctly in two argument ctor.
39144 (get_string_length, format_string,format_directive): Same.
39145 (pass_sprintf_length::compute_format_length): Same.
39146 (try_substitute_return_value): Simplify slightly.
39148 PR middle-end/78703
39149 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
39150 (fmtresult::operator+=): Outlined.
39151 (struct fmtresult): Add ctors.
39152 (struct conversion_spec): Rename...
39153 (struct directive): ...to this. Add and remove data members.
39154 (directive::set_width, directive::set_precision): New functions.
39155 (format_percent): Use fmtresult ctor.
39156 (get_width_and_precision): Remove.
39157 (format_integer): Make naming changes. Avoid computing width and
39159 (format_floating): Same. Adjust indentation.
39160 (format_character, format_none): New functions.
39161 (format_string): Moved character handling to format_character.
39162 (format_directive): Remove arguments, change return type.
39163 (parse_directive): New function.
39164 (pass_sprintf_length::compute_format_length): Move directive
39165 parsing to parse_directive.
39167 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39169 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
39170 (assign_assembler_name_if_needed): ... this.
39171 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
39172 (assign_assembler_name_if_needed): ... this.
39173 (free_lang_data_in_cgraph): Adjust callers.
39174 * cgraphunit.c (cgraph_node::analyze): Likewise.
39175 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
39178 2017-01-23 Richard Biener <rguenther@suse.de>
39180 PR tree-optimization/79088
39181 PR tree-optimization/79188
39182 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
39183 resetting loop bounds after last path deletion. Reset loop
39184 bounds of the target loop, make code match the comments.
39185 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
39186 Make sure loops need no fixups.
39188 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
39190 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
39191 exponent support with double type for first argument.
39192 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
39193 type returned by __builtin_vec_extract_sig,
39194 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
39195 functions from "vector int" to "vector unsigned int" or from
39196 "vector long long int" to "vector unsigned long long int".
39197 Changed type returned by __builtin_vec_extract_exp,
39198 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
39199 functions from "vector int" to "vector unsigned int" or from
39200 "vector long long int" to "vector unsigned long long int".
39201 Changed return type of __builtin_vec_test_data_class,
39202 __builtin_vec_test_data_class_sp, and
39203 __builtin_vec_test_data_class_dp from "vector int" to
39204 "vector bool int" or from "vector long long int" to "vector bool
39205 long long int" and changed second argument type from "unsigned
39206 int" to "int". Added new overloaded function forms "vector float
39207 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
39208 "vector float __builtin_vec_insert_exp_sp (vector float, vector
39209 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
39210 double, vector unsigned long long int)" and "vector double
39211 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
39212 long int)". Changed return type of
39213 __builtin_scalar_test_data_class and
39214 __builtin_scalar_test_data_class_sp and
39215 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
39216 int" and changed second argument from "unsigned int" to "int".
39217 Changed type returned by __builtin_scalar_test_neg,
39218 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
39219 from "int" to "bool int". Added new overloaded function form
39220 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
39221 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
39222 exponent double-precision with floating point first argument.
39223 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
39224 documentation of scalar_test_data_class, scalar_test_neg,
39225 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
39226 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
39227 vec_test_data_class built-in functions to reflect refinements in
39228 their type signatures.
39230 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
39232 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
39234 (aarch64_elf_asm_destructor): Likewise.
39236 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
39238 PR rtl-optimization/78634
39239 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
39240 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
39241 * ifcvt.c (noce_try_cmove): Add missing cost check.
39243 PR rtl-optimization/71724
39244 * combine.c (if_then_else_cond): Look for situations where it is
39245 beneficial to undo the work of one of the recursive calls.
39247 2017-01-23 Bin Cheng <bin.cheng@arm.com>
39249 PR tree-optimization/70754
39250 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
39251 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
39252 combined stmt before it if not NULL.
39253 (combine_chains): Process refs reversely and compute dominance point
39256 2017-01-23 Martin Liska <mliska@suse.cz>
39258 PR tree-optimization/79196
39259 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
39260 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
39262 (strlen_optimize_stmt): Call the renamed function.
39264 2017-01-23 Michael Matz <matz@suse.de>
39266 PR tree-optimization/78384
39267 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
39269 2017-01-23 Richard Biener <rguenther@suse.de>
39271 PR tree-optimization/79186
39272 * tree-vrp.c (register_new_assert_for): Make sure we've seen
39273 both incoming edges before moving an assert.
39275 2017-01-23 Martin Jambor <mjambor@suse.cz>
39277 * ipa-prop.c (load_from_param_1): Removed.
39278 (load_from_unmodified_param): Bits from load_from_param_1 put back
39280 (load_from_param): Removed.
39281 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
39282 with stmt. Reverted back to use of load_from_unmodified_param.
39284 2017-01-23 Martin Jambor <mjambor@suse.cz>
39287 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
39288 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
39289 field a pointer to garbage collected vector, mark lattices and
39290 ipcp_orig_node with GTY((skip)).
39291 (ipa_get_param_count): Adjust to descriptors being a pointer.
39292 (ipa_get_param): Likewise.
39293 (ipa_get_type): Likewise.
39294 (ipa_get_param_move_cost): Likewise.
39295 (ipa_set_param_used): Likewise.
39296 (ipa_get_controlled_uses): Likewise.
39297 (ipa_set_controlled_uses): Likewise.
39298 (ipa_is_param_used): Likewise.
39299 (ipa_node_params_t): Move into garbage collector. New methods insert
39301 (ipa_node_params_sum): Annotate wth GTY(()).
39302 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
39304 (ipa_load_from_parm_agg): Adjust declaration.
39305 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
39306 * ipa-profile.c (ipa_profile): Likewise.
39307 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
39308 (ipa_populate_param_decls): Make descriptors parameter garbage
39310 (ipa_dump_param): Adjust to descriptors being a pointer.
39311 (ipa_alloc_node_params): Likewise.
39312 (ipa_initialize_node_params): Likewise.
39313 (load_from_param_1): Make descriptors parameter garbage collected.
39314 (load_from_unmodified_param): Likewise.
39315 (load_from_param): Likewise.
39316 (ipa_load_from_parm_agg): Likewise.
39317 (ipa_node_params::~ipa_node_params): Removed.
39318 (ipa_free_all_node_params): Remove call to delete operator.
39319 (ipa_node_params_t::insert): New.
39320 (ipa_node_params_t::remove): Likewise.
39321 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
39322 copy known_csts and known_contexts vectors.
39323 (ipa_read_node_info): Adjust to descriptors being a pointer.
39324 (ipcp_modif_dom_walker): Make m_descriptors field garbage
39326 (ipcp_transform_function): Make descriptors variable garbage
39329 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
39331 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
39332 * config/i386/avx512dqintrin.h: Ditto.
39333 * config/i386/avx512fintrin.h: Ditto.
39334 * config/i386/i386.c: Handle new builtins.
39335 * config/i386/i386-builtin.def: Add new builtins.
39336 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
39337 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
39339 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39340 Martin Liska <mliska@suse.cz>
39342 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
39343 * asan.c (asan_expand_poison_ifn): Support stores and use
39344 appropriate ASAN report function.
39345 * internal-fn.c (expand_ASAN_POISON_USE): New function.
39346 * internal-fn.def (ASAN_POISON_USE): Declare.
39347 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
39348 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
39349 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
39350 ASAN_POISON calls w/o LHS.
39351 * tree-ssa.c (execute_update_addresses_taken): Create clobber
39352 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
39353 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
39354 * gimplify.c (asan_poison_variables): Add attribute
39355 use_after_scope_memory to variables that really needs to live
39357 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
39358 having the attribute.
39360 2017-01-23 Martin Liska <mliska@suse.cz>
39362 * asan.c (create_asan_shadow_var): New function.
39363 (asan_expand_poison_ifn): Likewise.
39364 * asan.h (asan_expand_poison_ifn): New declaration.
39365 * internal-fn.c (expand_ASAN_POISON): Likewise.
39366 * internal-fn.def (ASAN_POISON): New builtin.
39367 * sanopt.c (pass_sanopt::execute): Expand
39368 asan_expand_poison_ifn.
39369 * tree-inline.c (copy_decl_for_dup_finish): Make function
39371 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
39372 * tree-ssa.c (is_asan_mark_p): New function.
39373 (execute_update_addresses_taken): Rewrite local variables
39374 (identified just by use-after-scope as addressable) into SSA.
39376 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39378 * doc/install.texi (Specific): opensource.apple.com uses https
39379 now. Remove trailing slash.
39381 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39383 * README.Portability: Remove note on an Irix compatibility issue.
39385 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
39387 * gcov.c (INCLUDE_ALGORITHM): Define.
39388 (INCLUDE_VECTOR): Define.
39389 No longer include <vector> and <algorithm> directly.
39391 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
39393 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
39395 * doc/invoke.texi (Code Gen Options): Ditto.
39397 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
39400 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
39402 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
39404 rtl-optimization/79125
39405 * cprop.c (local_cprop_pass): Handle cases where we make an
39406 unconditional trap.
39408 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
39412 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
39413 read from, for big endian.
39415 2017-01-20 Jiong Wang <jiong.wang@arm.com>
39417 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
39418 register pauth builtins for LP64 only.
39420 2017-01-20 Marek Polacek <polacek@redhat.com>
39423 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
39426 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39428 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
39430 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
39431 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
39432 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
39434 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39437 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
39438 in big-endian mode when they are not a single duplicated value.
39440 2017-01-20 Richard Biener <rguenther@suse.de>
39442 * BASE-VER: Bump to 7.0.1.
39444 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39446 * omp-low.c (omplow_simd_context): New struct. Use it...
39447 (lower_rec_simd_input_clauses): ...here and...
39448 (lower_rec_input_clauses): ...here to hold common data. Adjust all
39449 references to idx, lane, max_vf, is_simt.
39451 2017-01-20 Graham Markall <graham.markall@embecosm.com>
39453 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
39456 2017-01-20 Martin Jambor <mjambor@suse.cz>
39458 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
39459 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
39461 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
39462 (GTFILES): Rename hsa.c to hsa-common.c.
39463 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
39464 * hsa-dump.c: Likewise.
39465 * hsa-gen.c: Likewise.
39466 * hsa-regalloc.c: Likewise.
39467 * ipa-hsa.c: Likewise.
39468 * omp-expand.c: Likewise.
39469 * omp-low.c: Likewise.
39470 * toplev.c: Likewise.
39472 2017-01-20 Marek Polacek <polacek@redhat.com>
39475 * doc/invoke.texi: Document -Wduplicated-branches.
39476 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
39477 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
39478 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
39479 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
39480 return 0 only when not OEP_LEXICOGRAPHIC.
39481 (fold_build_cleanup_point_expr): Use the expression
39482 location when building CLEANUP_POINT_EXPR.
39483 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
39484 * tree.c (add_expr): Handle error_mark_node.
39486 2017-01-20 Martin Liska <mliska@suse.cz>
39489 * tree-profile.c (init_ic_make_global_vars): Do not call
39491 (gimple_init_gcov_profiler): Likewise.
39493 2017-01-20 Martin Liska <mliska@suse.cz>
39496 * cgraph.h (maybe_create_reference): Remove argument and
39498 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
39500 * ipa-cp.c (create_specialized_node): Likewise.
39501 * symtab.c (symtab_node::maybe_create_reference): Handle
39502 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
39504 2017-01-20 Martin Liska <mliska@suse.cz>
39506 * read-rtl-function.c (function_reader::create_function): Use
39507 build_decl instread of build_decl_stat.
39509 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
39511 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
39512 * config/i386/avx512dqintrin.h: Ditto.
39513 * config/i386/avx512fintrin.h: Ditto.
39514 * config/i386/i386-builtin-types.def: Add new types.
39515 * config/i386/i386.c: Handle new types.
39516 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
39517 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
39518 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
39519 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
39520 (__builtin_ia32_kshiftridi): New.
39521 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
39523 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
39527 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
39528 define to rs6000_init_stack_protect_guard.
39529 (rs6000_init_stack_protect_guard): New function.
39531 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39532 Yunqiang Su <yunqiang.su@imgtec.com>
39534 * config.gcc (supported_defaults): Add madd4.
39535 (with_madd4): Add validation.
39536 (all_defaults): Add madd4.
39537 * config/mips/mips.opt (mmadd4): New option.
39538 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39540 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39542 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
39543 (ISA_HAS_FUSED_MADD4): Likewise.
39544 * doc/invoke.texi (-mmadd4): Document the new option.
39545 * doc/install.texi (--with-madd4): Document the new option.
39547 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39549 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
39550 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
39551 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
39552 (aarch64_init_pauth_hint_builtins): New.
39553 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
39554 (aarch64_expand_builtin): Expand new builtins.
39556 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39558 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
39559 * combine-stack-adj.c (no_unhandled_cfa): Handle
39560 REG_CFA_TOGGLE_RA_MANGLE.
39561 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
39562 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
39563 info for return address signing.
39564 (aarch64_expand_epilogue): Likewise.
39566 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39568 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
39569 * config/aarch64/aarch64-protos.h
39570 (aarch64_return_address_signing_enabled): New declaration.
39571 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
39573 (aarch64_expand_prologue): Sign return address before it's pushed onto
39575 (aarch64_expand_epilogue): Authenticate return address fetched from
39577 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39578 (aarch64_attributes): New function attributes for "sign-return-address".
39579 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39580 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39581 ("*do_return"): Generate combined instructions according to key index.
39582 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39583 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39585 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39586 * config/aarch64/aarch64.opt (msign-return-address=): New.
39587 * doc/extend.texi (AArch64 Function Attributes): Documents
39588 "sign-return-address=".
39589 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39591 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39593 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39594 overall option summary.
39596 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39598 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39599 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39600 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39601 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39603 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39605 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39606 -mpower9-minmax by default for -mcpu=power9.
39607 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39608 128-bit floating point.
39610 2017-01-20 Alan Modra <amodra@gmail.com>
39612 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39613 optimizing for size.
39615 2017-01-20 Alan Modra <amodra@gmail.com>
39618 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39619 for strcmp and strncmp from corresponding builtin decl.
39621 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39623 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39624 instead of i386/rtems-64.h.
39625 * config/i386/rtems-64.h: Remove.
39627 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39631 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39633 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39635 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39637 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39638 Change int to HOST_WIDE_INT.
39639 * config/aarch64/aarch64-protos.h
39640 (aarch64_simd_gen_const_vector_dup): Likewise.
39641 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39643 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39645 * langhooks-def.h (lhd_type_for_size): New decl.
39646 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39647 * langhooks.c (lhd_type_for_size): New function, taken from
39650 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39652 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39653 define_bypass for CR latency.
39654 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39655 (power9-alu2): Add define_bypass for CR latency.
39657 (power9-mul): Update insn latency.
39658 (power9-mul-compare): Update insn latency, bypass latency and remove
39661 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39663 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39665 * config/aarch64/aarch64.md
39666 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39667 aarch64_nopcrelative_literal_loads.
39668 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39670 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39672 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39673 TARGET_LOONGSON_3A.
39674 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39676 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39679 * config.gcc (supported_defaults): Add lxc1-sxc1.
39680 (with_lxc1_sxc1): Add validation.
39681 (all_defaults): Add lxc1-sxc1.
39682 * config/mips/mips.opt (mlxc1-sxc1): New option.
39683 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39685 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39686 __mips_no_lxc1_sxc1.
39687 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39688 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39689 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39691 2017-01-19 Richard Biener <rguenther@suse.de>
39693 PR tree-optimization/72488
39694 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39695 sure to restore SSA info.
39696 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39698 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39700 PR rtl-optimization/79121
39701 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39702 of the inner type when shifting an extended value.
39704 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39707 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39708 interposable aliases.
39710 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39713 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39714 Use the evmergelohi instruction.
39715 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39716 (mov_sitf_e500_subreg8_2_be): Likewise.
39717 (mov_sitf_e500_subreg12_2_le): Likewise.
39718 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39719 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39720 (mov_sitf_e500_subreg8_2_le): Likewise.
39721 (mov_sitf_e500_subreg12_2_be): Likewise.
39723 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39725 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39726 attribute from vecsimple to vecperm.
39727 (altivec_vbpermq2): Likewise.
39729 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39732 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39734 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39735 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39736 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39737 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39738 case where N arg is SIZE_MAX.
39739 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39740 (cmpstrsi): Add pattern.
39742 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39744 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39745 __builtin_vec_revb builtins.
39746 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39747 built-in functions to support generation of the ISA 3.0 XXBR<x>
39748 vector byte reverse instructions.
39749 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39750 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39751 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39752 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39753 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39754 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39755 (P9V_BUILTIN_VEC_REVB): Likewise.
39756 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39757 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39758 (p9_xxbrq_v16qi): Likewise.
39759 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39760 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39761 (p9_xxbrh_v8hi): Likewise.
39762 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39763 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39764 vec_revb built-in functions.
39766 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39768 PR rtl-optimization/78952
39769 * config/i386/i386.md (any_extract): New code iterator.
39770 (*insvqi_2): Use any_extract for source operand.
39771 (*insvqi_3): Use any_shiftrt for source operand.
39773 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39775 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39777 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39779 2017-01-18 Matthias Klose <doko@ubuntu.com>
39781 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39783 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39785 * config/rs6000/altivec.h (vec_bperm): Change #define.
39786 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39787 (altivec_vbpermq2): New define_insn.
39788 (altivec_vbpermd): Likewise.
39789 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39790 function interface.
39791 (VBPERMD): Likewise.
39792 (VBPERM): New polymorphic function interface.
39793 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39794 Add entries for P9V_BUILTIN_VEC_VBPERM.
39795 * doc/extend.texi: Add interfaces for vec_bperm.
39797 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39799 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39800 first letter of error messages.
39801 (s390_resolve_overloaded_builtin): Likewise.
39802 * config/s390/s390.c (s390_expand_builtin): Likewise.
39803 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39804 (s390_valid_target_attribute_inner_p): Likewise.
39805 * config/s390/s390.md ("tabort"): Likewise.
39807 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39809 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39810 (ISA_AVOID_DIV_HILO): New macro.
39811 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39812 (ISA_HAS_DDIV): Likewise.
39814 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39816 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39818 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39820 * doc/invoke.texi (fabi-version): Spelling fix.
39822 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39825 * doc/invoke.texi (fabi-version): Mention mangling fix for
39828 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39831 * doc/invoke.texi (fabi-version): Document discriminator mangling.
39833 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
39836 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
39837 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
39839 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
39841 (stack_protect_test): Ditto.
39842 * config/rs6000/rs6000.opt (mstack-protector-guard=,
39843 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
39845 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
39846 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
39847 -mstack-protector-guard-offset=.
39848 (RS/6000 and PowerPC Options): Ditto.
39850 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39852 * config/i386/i386.h (MASK_CLASS_P): New define.
39853 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
39854 there are no registers from different register sets also when
39855 mask registers are used. Update function comment.
39856 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
39857 to (*k/*r) and (*k/*km) alternatives.
39859 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
39861 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
39862 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
39863 (EH_RETURN_HANDLER_RTX): New define.
39864 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
39865 Force frame pointer in EH return functions.
39866 (aarch64_expand_epilogue): Add barrier for eh_return.
39867 (aarch64_final_eh_return_addr): Remove.
39868 (aarch64_eh_return_handler_rtx): New function.
39869 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
39871 (aarch64_eh_return_handler_rtx): New prototype.
39873 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39875 * config/rs6000/altivec.h (vec_rlmi): New #define.
39876 (vec_vrlnm): Likewise.
39877 (vec_rlnm): Likewise.
39878 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
39879 (UNSPEC_VRLNM): Likewise.
39880 (VIlong): New mode iterator.
39881 (altivec_vrl<VI_char>mi): New define_insn.
39882 (altivec_vrl<VI_char>nm): Likewise.
39883 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
39885 (VRLDNM): Likewise.
39886 (RLNM): New polymorphic function entry.
39887 (VRLWMI): New monomorphic function entry.
39888 (VRLDMI): Likewise.
39889 (RLMI): New polymorphic function entry.
39890 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
39891 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
39892 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
39895 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39898 * dwarf2out.c (field_byte_offset): Restore the
39899 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
39900 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
39901 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
39904 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39907 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
39909 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39912 * dwarf2out.c (add_data_member_location_attribute): For constant
39913 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
39914 instead of DW_AT_data_member_location, DW_AT_bit_offset and
39915 DW_AT_byte_size attributes.
39917 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39919 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
39920 after forcing to constant memory when the code model is medium.
39922 2017-01-17 Julia Koval <julia.koval@intel.com>
39925 * config/i386/avx512fintrin.h
39926 (_mm512_i32gather_ps): Change __addr type to void const*.
39927 (_mm512_mask_i32gather_ps): Ditto.
39928 (_mm512_i32gather_pd): Ditto.
39929 (_mm512_mask_i32gather_pd): Ditto.
39930 (_mm512_i64gather_ps): Ditto.
39931 (_mm512_mask_i64gather_ps): Ditto.
39932 (_mm512_i64gather_pd): Ditto.
39933 (_mm512_mask_i64gather_pd): Ditto.
39934 (_mm512_i32gather_epi32): Ditto.
39935 (_mm512_mask_i32gather_epi32): Ditto.
39936 (_mm512_i32gather_epi64): Ditto.
39937 (_mm512_mask_i32gather_epi64): Ditto.
39938 (_mm512_i64gather_epi32): Ditto.
39939 (_mm512_mask_i64gather_epi32): Ditto.
39940 (_mm512_i64gather_epi64): Ditto.
39941 (_mm512_mask_i64gather_epi64): Ditto.
39942 (_mm512_i32scatter_ps): Change __addr type to void*.
39943 (_mm512_mask_i32scatter_ps): Ditto.
39944 (_mm512_i32scatter_pd): Ditto.
39945 (_mm512_mask_i32scatter_pd): Ditto.
39946 (_mm512_i64scatter_ps): Ditto.
39947 (_mm512_mask_i64scatter_ps): Ditto.
39948 (_mm512_i64scatter_pd): Ditto.
39949 (_mm512_mask_i64scatter_pd): Ditto.
39950 (_mm512_i32scatter_epi32): Ditto.
39951 (_mm512_mask_i32scatter_epi32): Ditto.
39952 (_mm512_i32scatter_epi64): Ditto.
39953 (_mm512_mask_i32scatter_epi64): Ditto.
39954 (_mm512_i64scatter_epi32): Ditto.
39955 (_mm512_mask_i64scatter_epi32): Ditto.
39956 (_mm512_i64scatter_epi64): Ditto.
39957 (_mm512_mask_i64scatter_epi64): Ditto.
39958 * config/i386/avx512pfintrin.h
39959 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
39960 (_mm512_mask_prefetch_i32gather_ps): Ditto.
39961 (_mm512_mask_prefetch_i64gather_pd): Ditto.
39962 (_mm512_mask_prefetch_i64gather_ps): Ditto.
39963 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
39964 (_mm512_prefetch_i32scatter_ps): Ditto.
39965 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
39966 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
39967 (_mm512_prefetch_i64scatter_pd): Ditto.
39968 (_mm512_prefetch_i64scatter_ps): Ditto.
39969 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
39970 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
39971 * config/i386/avx512vlintrin.h
39972 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
39973 (_mm_mmask_i32gather_ps): Ditto.
39974 (_mm256_mmask_i32gather_pd): Ditto.
39975 (_mm_mmask_i32gather_pd): Ditto.
39976 (_mm256_mmask_i64gather_ps): Ditto.
39977 (_mm_mmask_i64gather_ps): Ditto.
39978 (_mm256_mmask_i64gather_pd): Ditto.
39979 (_mm_mmask_i64gather_pd): Ditto.
39980 (_mm256_mmask_i32gather_epi32): Ditto.
39981 (_mm_mmask_i32gather_epi32): Ditto.
39982 (_mm256_mmask_i32gather_epi64): Ditto.
39983 (_mm_mmask_i32gather_epi64): Ditto.
39984 (_mm256_mmask_i64gather_epi32): Ditto.
39985 (_mm_mmask_i64gather_epi32): Ditto.
39986 (_mm256_mmask_i64gather_epi64): Ditto.
39987 (_mm_mmask_i64gather_epi64): Ditto.
39988 (_mm256_i32scatter_ps): Change __addr type to void*.
39989 (_mm256_mask_i32scatter_ps): Ditto.
39990 (_mm_i32scatter_ps): Ditto.
39991 (_mm_mask_i32scatter_ps): Ditto.
39992 (_mm256_i32scatter_pd): Ditto.
39993 (_mm256_mask_i32scatter_pd): Ditto.
39994 (_mm_i32scatter_pd): Ditto.
39995 (_mm_mask_i32scatter_pd): Ditto.
39996 (_mm256_i64scatter_ps): Ditto.
39997 (_mm256_mask_i64scatter_ps): Ditto.
39998 (_mm_i64scatter_ps): Ditto.
39999 (_mm_mask_i64scatter_ps): Ditto.
40000 (_mm256_i64scatter_pd): Ditto.
40001 (_mm256_mask_i64scatter_pd): Ditto.
40002 (_mm_i64scatter_pd): Ditto.
40003 (_mm_mask_i64scatter_pd): Ditto.
40004 (_mm256_i32scatter_epi32): Ditto.
40005 (_mm256_mask_i32scatter_epi32): Ditto.
40006 (_mm_i32scatter_epi32): Ditto.
40007 (_mm_mask_i32scatter_epi32): Ditto.
40008 (_mm256_i32scatter_epi64): Ditto.
40009 (_mm256_mask_i32scatter_epi64): Ditto.
40010 (_mm_i32scatter_epi64): Ditto.
40011 (_mm_mask_i32scatter_epi64): Ditto.
40012 (_mm256_i64scatter_epi32): Ditto.
40013 (_mm256_mask_i64scatter_epi32): Ditto.
40014 (_mm_i64scatter_epi32): Ditto.
40015 (_mm_mask_i64scatter_epi32): Ditto.
40016 (_mm256_i64scatter_epi64): Ditto.
40017 (_mm256_mask_i64scatter_epi64): Ditto.
40018 (_mm_i64scatter_epi64): Ditto.
40019 (_mm_mask_i64scatter_epi64): Ditto.
40020 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40021 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40022 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40023 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40024 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40025 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40026 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40027 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40028 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40029 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40030 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40031 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40032 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40033 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40034 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40035 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40036 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40037 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40038 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40039 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40040 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40041 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40042 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40043 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40044 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40045 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40046 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40047 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40048 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40049 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40050 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40051 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40052 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40053 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40054 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40055 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40056 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40057 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40058 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40059 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40060 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40061 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40062 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40063 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40064 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
40065 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
40066 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
40067 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
40068 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
40069 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
40070 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
40071 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
40072 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
40073 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
40074 definitions accordingly.
40076 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
40077 Kuan-Lin Chen <kuanlinchentw@gmail.com>
40080 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
40083 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
40086 * ira-conflicts.c (ira_build_conflicts): Update total conflict
40087 hard regs for inner regno.
40089 2017-01-17 Martin Liska <mliska@suse.cz>
40092 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
40093 assumption and add comment.
40095 2017-01-17 Nathan Sidwell <nathan@acm.org>
40097 * ipa-visibility.c (localize_node): New function, broken out of ...
40098 (function_and_variable_visibility): ... here. Call it.
40100 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40102 PR middle-end/77445
40103 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
40104 correctly set frequency of oudgoing edge.
40105 (duplicate_thread_path): Fix profile updating.
40107 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40110 * configure.ac: Add GCC_BASE_VER.
40111 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
40112 version from BASE-VER file.
40113 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
40114 (gcc.o): Depend on $(BASEVER).
40115 * common.opt (dumpfullversion): New option.
40116 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
40117 * doc/invoke.texi: Document -dumpfullversion.
40118 * doc/install.texi: Document --with-gcc-major-version-only.
40119 * configure: Regenerated.
40121 2017-01-17 Richard Biener <rguenther@suse.de>
40123 PR tree-optimization/71433
40124 * tree-vrp.c (register_new_assert_for): Merge same asserts
40125 on all incoming edges.
40126 (process_assert_insertions_for): Handle insertions at the
40129 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
40131 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
40132 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
40134 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
40137 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
40140 2017-01-17 Alan Modra <amodra@gmail.com>
40143 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
40144 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
40145 symbolic stack limit when pic.
40147 2017-01-16 Martin Sebor <msebor@redhat.com>
40149 PR tree-optimization/78608
40150 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
40152 2017-01-16 Jeff Law <law@redhat.com>
40155 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
40156 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
40157 for several include directories that may be relative to sysroot.
40158 * config/i386/x-mingw32 (gplus_includedir): Define.
40159 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
40160 (native_system_includedir): Likewise.
40161 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
40162 override if TARGET_SYSTEM_ROOT is defined.
40163 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
40165 PR tree-optimization/79090
40166 PR tree-optimization/33562
40167 PR tree-optimization/61912
40168 PR tree-optimization/77485
40169 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
40170 and computed trims into the dump file.
40172 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40174 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
40176 2017-01-16 Jakub Jelinek <jakub@redhat.com>
40179 * gimplify.c (gimplify_init_constructor): If want_value and
40180 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
40184 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
40185 sequence. Formatting fixes.
40186 (doloop_optimize): Formatting fixes.
40189 * gcc.c (debug_level_greater_than_spec_func): New function.
40190 (static_spec_functions): Add debug-level-gt spec function.
40191 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
40193 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40194 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40195 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
40196 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
40197 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
40198 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
40200 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
40202 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
40203 QImode fixups to general and mask registers only.
40205 2017-01-16 Carl Love <cel@us.ibm.com>
40207 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
40208 for built-in functions
40209 vector signed char vec_nabs (vector signed char)
40210 vector signed short vec_nabs (vector signed short)
40211 vector signed int vec_nabs (vector signed int)
40212 vector signed long long vec_nabs (vector signed long long)
40213 vector float vec_nabs (vector float)
40214 vector double vec_nabs (vector double)
40215 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
40217 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
40218 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
40219 * doc/extend.texi: Update the documentation file for the new built-in
40222 2017-01-16 Martin Sebor <msebor@redhat.com>
40224 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
40227 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40229 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
40230 UNSPEC_VSX__XXSPLTD to require special splat handling.
40232 2017-01-16 David Malcolm <dmalcolm@redhat.com>
40235 * system.h: Poison strndup.
40237 2017-01-16 Alan Modra <amodra@gmail.com>
40240 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
40243 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
40245 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
40247 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
40249 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
40250 call recog here. Assert that INSN_CODE (insn) is non-negative.
40252 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40255 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
40257 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
40258 in the currently scheduled RTL fragment.
40260 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40262 PR rtl-optimization/78751
40263 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
40266 2017-01-14 Jeff Law <law@redhat.com>
40268 PR tree-optimization/79090
40269 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
40270 variable length stores.
40271 (compute_trims): Delete dead assignment to *trim_tail.
40272 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
40275 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
40277 PR rtl-optimization/78626
40278 PR rtl-optimization/78727
40279 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
40280 of a block, and split such blocks after everything else is finished.
40282 2017-01-14 Alan Modra <amodra@gmail.com>
40285 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
40286 target legitimate_combined_insn.
40287 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
40288 (rs6000_legitimate_combined_insn): New function.
40289 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
40291 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
40292 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
40293 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
40295 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
40297 * doc/frontends.texi (G++ and GCC): Remove references to Java.
40299 2017-01-13 Jeff Law <law@redhat.com>
40301 PR tree-optimization/33562
40302 PR tree-optimization/61912
40303 PR tree-optimization/77485
40304 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
40306 (delete_dead_assignment): Likewise.
40307 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
40308 statement to delete_dead_call and delete_dead_assignment.
40310 2017-01-13 David Malcolm <dmalcolm@redhat.com>
40313 * substring-locations.c (format_warning_va): Strengthen case 1 so
40314 that both endpoints of the substring must be within the format
40315 range for just the substring to be printed.
40317 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
40319 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
40320 * config/i386/i386.c (ix86_target_string): Add missing options
40321 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
40322 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
40323 flags_other and ix86_target_other to flags2_other. Display unknown
40325 (ix86_valid_target_attribute_inner_p): Add missing options and
40326 reorder options by implied ISAs, as in ix86_target_string.
40328 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40330 * hash-table.h (hash_table::too_empty_p): New function.
40331 (hash_table::expand): Use it.
40332 (hash_table::traverse): Likewise.
40333 (hash_table::empty_slot): Use sizeof (value_type) instead of
40334 sizeof (PTR) to convert bytes to elements. Shrink the table
40335 if the current size is excessive for the current number of
40338 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40340 * ira-costs.c (record_reg_classes): Break from the inner loop
40341 early once alt_fail is known to be true. Update outer loop
40342 handling accordingly.
40344 2017-01-13 Jeff Law <law@redhat.com>
40346 * tree-ssa-dse.c (decrement_count): New function.
40347 (increment_start_addr, maybe_trim_memstar_call): Likewise.
40348 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
40349 when we know the partially dead statement is a mem* function.
40351 PR tree-optimization/61912
40352 PR tree-optimization/77485
40353 * tree-ssa-dse.c: Include expr.h.
40354 (maybe_trim_constructor_store): New function.
40355 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
40357 PR tree-optimization/33562
40358 PR tree-optimization/61912
40359 PR tree-optimization/77485
40360 * doc/invoke.texi: Document new dse-max-object-size param.
40361 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
40362 * tree-ssa-dse.c: Include params.h.
40363 (dse_store_status): New enum.
40364 (initialize_ao_ref_for_dse): New, partially extracted from
40366 (valid_ao_ref_for_dse, normalize_ref): New.
40367 (setup_live_bytes_from_ref, compute_trims): Likewise.
40368 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
40369 (maybe_trim_partially_dead_store): Likewise.
40370 (maybe_trim_complex_store): Likewise.
40371 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
40372 Track what bytes live from the original store. Return tri-state
40373 for dead, partially dead or live.
40374 (dse_dom_walker): Add constructor, destructor and new private members.
40375 (delete_dead_call, delete_dead_assignment): New extracted from
40377 (dse_optimize_stmt): Make a member of dse_dom_walker.
40378 Use initialize_ao_ref_for_dse.
40380 PR tree-optimization/33562
40381 PR tree-optimization/61912
40382 PR tree-optimization/77485
40383 * sbitmap.h (bitmap_count_bits): Prototype.
40384 (bitmap_clear_range, bitmap_set_range): Likewise.
40385 * sbitmap.c (bitmap_clear_range): New function.
40386 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
40388 2017-01-13 Martin Liska <mliska@suse.cz>
40391 * function.c (set_cfun): Add new argument force.
40392 * function.h (set_cfun): Likewise.
40393 * ipa-inline-transform.c (inline_call): Use the function when
40394 strict alising from is dropped for function we inline to.
40396 2017-01-13 Richard Biener <rguenther@suse.de>
40398 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
40399 for dumping GIMPLE INTEGER_CSTs.
40401 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
40403 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
40404 to 201112L since C++17.
40406 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
40409 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
40410 if -fsanitize=kernel-address is present.
40412 2017-01-13 Richard Biener <rguenther@suse.de>
40414 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
40415 as _Literal ( type ) number in case usual suffixes do not
40416 preserve all information.
40418 2017-01-13 Richard Biener <rguenther@suse.de>
40420 PR tree-optimization/77283
40421 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
40422 and ssa-iterators.h.
40423 (is_feasible_trace): Implement a cost model based on joiner
40426 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
40429 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
40430 char or short to __float128/_Float128 directly.
40432 2017-01-12 Martin Sebor <msebor@redhat.com>
40434 to -Wformat-overflow.
40435 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
40436 (min_bytes_remaining): Same.
40437 (get_string_length): Same.
40438 (format_string): Same.
40439 (format_directive): Same.
40441 (pass_sprintf_length::handle_gimple_call): Same.
40443 2017-01-12 Jakub Jelinek <jakub@redhat.com>
40445 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
40446 info.nowrite calls with no lhs that can't throw. Return bool
40447 whether gsi_remove has been called or not.
40448 (pass_sprintf_length::handle_gimple_call): Return bool whether
40449 try_substitute_return_value called gsi_remove. Formatting fix.
40450 (pass_sprintf_length::execute): Don't use gsi_remove if
40451 handle_gimple_call returned true.
40454 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
40455 be removed due to side-effects, don't remove following barrier nor
40456 turn the successor edge into fallthru edge.
40458 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40461 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
40462 element-reversing loads and stores as not swappable.
40464 2017-01-12 Nathan Sidwell <nathan@acm.org>
40465 Nicolai Stange <nicstange@gmail.com>
40467 * combine.c (try_combine): Don't ignore result of overlap checking
40468 loop. Combine overlap & asm check into single loop.
40470 2017-01-12 Richard Biener <rguenther@suse.de>
40472 * tree-pretty-print.c (dump_generic_node): Provide -gimple
40473 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
40475 2017-01-12 Richard Biener <rguenther@suse.de>
40477 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
40478 and TS_TARGET_OPTION directly derive from TS_BASE.
40479 * tree-core.h (tree_optimization_option): Derive from tree_base.
40480 (tree_target_option): Likewise.
40482 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
40484 * config/i386/i386.c (memory_address_length): Increase len
40485 only when rip_relative_addr_p returns false.
40487 2017-01-11 Julia Koval <julia.koval@intel.com>
40489 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
40490 (OPTION_MASK_ISA_SGX_SET): New.
40491 (ix86_handle_option): Handle OPT_msgx.
40492 * config.gcc: Added sgxintrin.h.
40493 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
40494 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
40495 * config/i386/i386.c (ix86_target_string): Add -msgx.
40497 (ix86_option_override_internal): Handle new options.
40498 (ix86_valid_target_attribute_inner_p): Add sgx.
40499 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
40500 * config/i386/i386.opt: Add msgx.
40501 * config/i386/sgxintrin.h: New file.
40502 * config/i386/x86intrin.h: Add sgxintrin.h.
40504 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40507 * fold-const.c (maybe_nonzero_address): Return 1 for function
40509 (tree_single_nonzero_warnv_p): Don't handle function local objects
40513 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
40516 2017-01-11 David Malcolm <dmalcolm@redhat.com>
40519 * opts.c: Include "spellcheck.h"
40520 (struct string_fragment): New struct.
40521 (struct edit_distance_traits<const string_fragment &>): New
40523 (get_closest_sanitizer_option): New function.
40524 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
40526 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40528 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
40530 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
40531 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
40532 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
40533 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
40534 for initial die_offset if dwarf_split_debug_info.
40535 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
40536 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
40538 (output_skeleton_debug_sections): Formatting fix. Use
40539 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
40540 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
40542 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
40544 * config/arm/cortex-a53.md: Add bypasses for
40545 cortex_a53_r2f_cvt.
40546 (cortex_a53_r2f): Only use for transfers.
40547 (cortex_a53_f2r): Likewise.
40548 (cortex_a53_r2f_cvt): Add reservation for conversions.
40549 (cortex_a53_f2r_cvt): Likewise.
40551 2017-01-11 Tamar Christina <tamar.christina@arm.com>
40553 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
40554 to all inlined functions, change static to extern.
40556 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
40559 * config/arm/arm.c (legitimize_pic_address): Handle reference to
40561 (arm_assemble_integer): Likewise.
40563 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
40565 * config.gcc: Use new awk script to check CPU, FPU and architecture
40566 parameters for --with-... options.
40567 * config/arm/parsecpu.awk: New file
40568 * config/arm/arm-cpus.in: New file.
40569 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
40571 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
40573 * config/arm/t-arm: Update dependency rules.
40574 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
40575 of processing .def files.
40576 * config/arm/genopt.sh: Deleted.
40577 * config/arm/gentune.sh: Deleted.
40578 * config/arm/arm-cores.def: Deleted.
40579 * config/arm/arm-arches.def: Deleted.
40580 * config/arm/arm-fpus.def: Deleted.
40581 * config/arm/arm-tune.md: Regenerated.
40582 * config/arm/arm-tables.opt: Regenerated.
40583 * config/arm/arm-cpu.h: New generated file.
40584 * config/arm/arm-cpu-data.h: New generated file.
40585 * config/arm/arm-cpu-cdata.h: New generated file.
40587 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40590 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40592 (input_varpool_node): Unpack dynamically_initialized bit.
40594 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40596 PR rtl-optimization/79032
40597 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40598 the alignment of the adjusted memory reference against that of MODE,
40599 instead of the alignment of the original memory reference.
40601 2017-01-11 Martin Jambor <mjambor@suse.cz>
40603 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40605 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40606 decorated functions.
40608 2017-01-11 Richard Biener <rguenther@suse.de>
40610 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40611 set range/nonnull info for PHI results. Do not set it on
40612 stmts marked for removal.
40614 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40616 * expr.c (store_field): In the bitfield case, fetch the return value
40617 from the registers before applying a single big-endian adjustment.
40618 Always do a final load for a BLKmode value not larger than a word.
40620 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40623 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40624 that we correctly handle column numbers greater than
40625 LINE_MAP_MAX_COLUMN_NUMBER.
40627 2017-01-10 Martin Sebor <msebor@redhat.com>
40629 PR middle-end/78245
40630 * gimple-ssa-sprintf.c (get_destination_size): Call
40631 {init,fini}object_sizes.
40632 * tree-object-size.c (addr_object_size): Adjust.
40633 (pass_through_call): Adjust.
40634 (pass_object_sizes::execute): Adjust.
40635 * tree-object-size.h (fini_object_sizes): Declare.
40637 2017-01-10 Martin Sebor <msebor@redhat.com>
40639 PR tree-optimization/78775
40640 * builtins.c (get_size_range): Move...
40641 * calls.c: ...to here.
40642 (alloc_max_size): Accept zero argument.
40643 (operand_signed_p): Remove.
40644 (maybe_warn_alloc_args_overflow): Call get_size_range.
40645 * calls.h (get_size_range): Declare.
40647 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40649 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40650 from TI's devices.csv file as of September 2016.
40651 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40653 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40655 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40656 * doc/invoke.texi: Likewise.
40657 * doc/md.texi: Likewise.
40658 * doc/objc.texi: Likewise.
40660 2017-01-10 Joshua Conner <joshconner@google.com>
40662 * config/arm/fuchsia-elf.h: New file.
40663 * config/fuchsia.h: New file.
40664 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40665 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40667 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40669 2016-01-10 Richard Biener <rguenther@suse.de>
40671 PR tree-optimization/79034
40672 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40673 Propagate out degenerate PHIs in the joiner.
40675 2017-01-10 Martin Liska <mliska@suse.cz>
40677 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40678 (sort_congruence_classes_by_decl_uid): Likewise.
40679 (sort_congruence_class_groups_by_decl_uid): Likewise.
40680 (sem_item_optimizer::merge_classes): Sort class, groups in these
40681 classes and members in the groups by DECL_UID of declarations.
40682 This would make merge operations stable.
40684 2017-01-10 Martin Liska <mliska@suse.cz>
40686 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40687 usage of m_classes_vec.
40688 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40689 (sem_item_optimizer::get_group_by_hash): Likewise.
40690 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40691 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40692 (sem_item_optimizer::verify_classes): Likewise.
40693 (sem_item_optimizer::process_cong_reduction): Likewise.
40694 (sem_item_optimizer::dump_cong_classes): Likewise.
40695 (sem_item_optimizer::merge_classes): Likewise.
40696 * ipa-icf.h (congruence_class_hash): Rename from
40697 congruence_class_group_hash. Remove declaration of m_classes_vec.
40699 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40701 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40702 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40703 * config.gcc: Add avx512vpopcntdqintrin.h.
40704 * config/i386/avx512vpopcntdqintrin.h: New.
40705 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40706 * config/i386/i386-builtin-types.def: Add new types.
40707 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40708 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40709 __builtin_ia32_vpopcountq_v8di_mask): New.
40710 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40711 __AVX512VPOPCNTDQ__.
40712 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40713 (PTA_AVX512VPOPCNTDQ): Define.
40714 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40715 TARGET_AVX512VPOPCNTDQ_P): Define.
40716 * config/i386/i386.opt: Add mavx512vpopcntdq.
40717 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40718 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40720 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40722 PR middle-end/77484
40723 * predict.def (PRED_CALL): Set to 67.
40725 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40727 * expr.c (store_field): In the bitfield case, if the value comes from
40728 a function call and is of an aggregate type returned in registers, do
40729 not modify the field mode; extract the value in all cases if the mode
40730 is BLKmode and the size is not larger than a word.
40732 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40735 * config/i386/cpuid.h: Fix undefined behavior.
40737 2017-01-04 Jeff Law <law@redhat.com>
40739 PR tree-optimization/79007
40740 PR tree-optimization/67955
40741 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40742 conservative for pt.null when flag_non_call_exceptions is on.
40744 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40746 PR translation/79019
40747 PR translation/79020
40748 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40749 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40750 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40752 * config/avr/avr.opt (maccumulate-args): Likewise.
40753 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40754 * common.opt (freport-bug): Likewise.
40755 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40756 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40757 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40758 translatable string.
40759 * config/i386/i386.c (function_value_32): Likewise.
40760 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40761 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40763 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40764 * common/config/msp430/msp430-common.c (msp430_handle_option):
40766 * symtab.c (symtab_node::verify_base): Likewise.
40767 * opts.c (set_debug_level): Likewise.
40768 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40769 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40770 missing whitespace to translatable strings.
40771 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40772 * config/sh/superh.h: Likewise.
40773 * config/i386/xopintrin.h: Likewise.
40774 * config/i386/znver1.md: Likewise.
40775 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40776 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40777 * double-int.h (struct double_int): Likewise.
40778 * double-int.c (div_and_round_double): Likewise.
40779 * wide-int.cc: Likewise.
40780 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40781 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40782 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40783 (crossjumps_occurred): ... this.
40784 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40787 PR tree-optimization/78899
40788 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40789 returning bool return struct loop *, NULL for failure and the new
40791 (versionable_outer_loop_p): Don't version outer loop if it has
40792 dont_vectorized bit set.
40793 (tree_if_conversion): When versioning outer loop, ensure
40794 tree_if_conversion is performed also on the inner loop of the
40795 non-vectorizable outer loop copy.
40796 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40797 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40798 prevent vectorization of it.
40799 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40800 the outer loop vectorization of the non-scalar version is attempted
40801 before vectorization of the inner loop in scalar version. If
40802 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40803 vectorization of its inner loop.
40804 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40805 has 2 inner loops, rename also on edges from bb whose single pred
40806 is outer_loop->header. Fix typo in function comment.
40808 2017-01-09 Martin Sebor <msebor@redhat.com>
40811 * asan.c (asan_emit_stack_protection): Increase local buffer size
40812 to avoid snprintf truncation warning.
40814 2017-01-09 Andrew Pinski <apinski@cavium.com>
40816 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40817 to reference thunderx2t99 for the tuning structure
40818 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40820 (thunderx2t99_extra_costs): This.
40821 * config/aarch64/aarch64-tune.md: Regenerate.
40822 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
40823 (vulcan_addrcost_table): This.
40824 (vulcan_regmove_cost): Rename to ...
40825 (thunderx2t99_regmove_cost): This.
40826 (vulcan_vector_cost): Rename to ...
40827 (thunderx2t99_vector_cost): this.
40828 (vulcan_branch_cost): Rename to ...
40829 (thunderx2t99_branch_cost): This.
40830 (vulcan_tunings): Rename to ...
40831 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
40832 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
40834 2017-01-09 Martin Jambor <mjambor@suse.cz>
40838 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
40839 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
40840 (propagate_vr_accross_jump_function): Use the above function for all
40841 value range computations for pass-through jump functions and type
40842 converasion from explicit value range values.
40843 (ipcp_propagate_stage): Do not attempt to deduce types of formal
40844 parameters from TYPE_ARG_TYPES.
40845 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
40846 (ipa_write_node_info): Stream type of the actual argument.
40847 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
40849 2017-01-09 Martin Liska <mliska@suse.cz>
40852 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
40853 (lookup_compiler): Do not show error message with have_E.
40855 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40857 PR tree-optimization/78938
40858 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
40859 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
40860 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
40861 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
40864 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40866 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
40869 2017-01-09 Richard Biener <rguenther@suse.de>
40871 PR tree-optimization/78997
40872 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
40873 name condition properly.
40875 2017-01-09 Richard Biener <rguenther@suse.de>
40878 * dwarf2out.c (is_cxx): New overload with context.
40879 (is_naming_typedef_decl): Use it.
40881 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40883 * invoke.texi (Option Summary): Correct spacing in option lists
40884 and add line breaks to fix over-long lines.
40886 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40888 PR middle-end/17660
40890 * extend.texi (Common Variable Attributes): Add xref to GCC
40891 Internals manual to explain mode attribute keywords.
40893 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40896 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
40897 and Preprocessor Options.
40898 (Options for Linking): Document -pthread here....
40899 (RS/6000 and PowerPC Options): ...not here.
40900 (Solaris 2 Options): ...or here.
40901 * doc/cppopts.texi: Document -pthread.
40903 2017-01-08 Martin Sebor <msebor@redhat.com>
40905 PR middle-end/77708
40906 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
40907 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
40908 New member functions.
40909 (format_directive): Used them.
40911 (pass_sprintf_length::handle_gimple_call): Same.
40912 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
40913 to avoid truncation for any argument.
40914 (extract_affine_mul): Same.
40915 * tree.c (get_file_function_name): Same.
40917 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40919 PR middle-end/77484
40920 * predict.def (PRED_INDIR_CALL): Set to 86.
40922 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40924 PR preprocessor/54124
40925 * doc/cppopts.texi: Reformat -d subtable to list the full name
40926 of the options. Add cross-reference to the docs for the general
40927 compiler -d options.
40928 * doc/invoke.texi (Developer Options): Add cross-reference to the
40929 preprocessor-specific -d option documentation.
40931 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40933 PR preprocessor/13498
40934 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
40935 redudant material, and reflect new command-line options.
40936 (System Headers): Likewise.
40938 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40940 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
40941 -isystem, and -idirafter. Copy-edit.
40942 * doc/cppopts.texi: Copy-edit. Remove contradiction about
40943 default for -ftrack-macro-expansion. Delete obsolete and
40944 badly-formatted implementation details about -fdebug-cpp output.
40945 * doc/cppwarnopts.texi: Copy-edit.
40947 2017-01-07 David Malcolm <dmalcolm@redhat.com>
40950 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40951 that the transition from a max line width >= 1<<10 to narrower
40952 lines works correctly.
40954 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
40956 * doc/options.texi (PerFunction): New.
40957 * opt-functions.awk (switch_flags): Map both Optimization and
40958 PerFunction to CL_OPTIMIZATION.
40959 * opth-gen.awk: Test for PerFunction flag along with
40961 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
40962 it only when the latter is present. Skip those that don't in
40963 the hash function generator.
40964 * common.opt (fvar-tracking): Mark as PerFunction instead of
40966 (fvar-tracking-assignments): Likewise.
40967 (fvar-tracking-assignments-toggle): Likewise.
40968 (fvar-tracking-uninit): Likewise.
40970 2017-01-07 Jakub Jelinek <jakub@redhat.com>
40972 PR translation/79018
40973 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
40976 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
40979 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
40980 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
40981 TARGET_LONG_JUMP_TABLE_OFFSETS.
40982 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
40983 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
40984 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
40985 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
40986 * config/m68k/m68k.md (tablejump expander): Likewise.
40987 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
40988 TARGET_LONG_JUMP_TABLE_OFFSETS.
40989 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
40990 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
40992 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
40993 David Holsgrove <david.holsgrove@xilinx.com>
40995 * common/config/microblaze/microblaze-common.c
40996 (TARGET_EXCEPT_UNWIND_INFO): Remove.
40997 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
40999 * config/microblaze/microblaze.c (microblaze_must_save_register)
41000 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41002 (microblaze_eh_return): New function.
41003 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41004 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41005 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41006 * config/microblaze/microblaze.md (eh_return): New pattern.
41008 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41010 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41011 GCC_DIAGNOSTIC_STRINGIFY): Define.
41013 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41015 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41017 * config/arm/arm.md (<mcrr>): New.
41019 * config/arm/arm.c (arm_arch5te): New.
41020 (arm_option_override): Set arm_arch5te.
41021 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41023 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41024 (arm_mcrr_qualifiers): ... this. New.
41025 (MRRC_QUALIFIERS): Define to...
41026 (arm_mrrc_qualifiers): ... this. New.
41027 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41029 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41030 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41031 (MRRCI, mrrc, MRRC): New.
41032 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41033 VUNSPEC_MRRC2): New.
41035 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41037 * config/arm/arm.md (<mcr>): New.
41039 * config/arm/arm.c (arm_coproc_builtin_available): Add
41040 support for mcr, mrc, mcr2 and mrc2.
41041 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41042 (arm_mcr_qualifiers): ... this. New.
41043 (MRC_QUALIFIERS): Define to ...
41044 (arm_mrc_qualifiers): ... this. New.
41045 (MCR_QUALIFIERS): Define to ...
41046 (arm_mcr_qualifiers): ... this. New.
41047 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41049 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41050 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41051 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41052 VUNSPEC_MRC2): New.
41054 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41056 * config/arm/arm.md (*ldc): New.
41060 * config/arm/arm.c (arm_coproc_builtin_available): Add
41061 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41062 (arm_coproc_ldc_stc_legitimate_address): New.
41063 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41064 'qualifier_const_pointer'.
41065 (LDC_QUALIFIERS): Define to...
41066 (arm_ldc_qualifiers): ... this. New.
41067 (STC_QUALIFIERS): Define to...
41068 (arm_stc_qualifiers): ... this. New.
41069 * config/arm/arm-protos.h
41070 (arm_coproc_ldc_stc_legitimate_address): New.
41071 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
41072 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
41073 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
41074 stc2, stcl, stc2l): New.
41075 * config/arm/constraints.md (Uz): New.
41076 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
41077 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
41078 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
41079 VUNSPEC_STC2L): New.
41081 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41083 * config/arm/arm.md (<cdp>): New.
41084 * config/arm/arm.c (neon_const_bounds): Rename this ...
41085 (arm_const_bounds): ... this.
41086 (arm_coproc_builtin_available): New.
41087 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
41088 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
41089 (CDP_QUALIFIERS): Define to...
41090 (arm_cdp_qualifiers): ... this. New.
41092 (arm_expand_builtin_args): Add case for 6 arguments.
41093 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
41094 (arm_const_bounds): ... this.
41095 (arm_coproc_builtin_available): New.
41096 * config/arm/arm_acle.h (__arm_cdp): New.
41098 * config/arm/arm_acle_builtins.def (cdp): New.
41100 * config/arm/iterators.md (CDPI,CDP,cdp): New.
41101 * config/arm/neon.md: Rename all 'neon_const_bounds' to
41102 'arm_const_bounds'.
41103 * config/arm/types.md (coproc): New.
41104 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
41105 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
41106 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
41107 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
41109 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41111 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
41112 (UBINOP_QUALIFIERS): New.
41114 (acle_builtin_data): New. Change comment.
41115 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
41116 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
41117 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
41118 arm_acle_builtins.def.
41119 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
41120 (arm_init_acle_builtins): New.
41121 (CRC32_BUILTIN): Remove.
41122 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
41123 crc32cb, crc32ch and crc32cw.
41124 (arm_init_crc32_builtins): Remove.
41125 (arm_init_builtins): Use arm_init_acle_builtins rather
41126 than arm_init_crc32_builtins.
41127 (arm_expand_acle_builtin): New.
41128 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
41129 * config/arm/arm_acle_builtins.def: New.
41131 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41133 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
41134 (arm_builtin_datum): ... this.
41135 (arm_init_neon_builtin): Rename to ...
41136 (arm_init_builtin): ... this. Add a new parameters PREFIX
41137 and USE_SIG_IN_NAME.
41138 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
41139 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
41140 'arm_builtin_datum'.
41141 (arm_init_vfp_builtins): Likewise.
41142 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
41143 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
41144 (arm_expand_neon_args): Rename to ...
41145 (arm_expand_builtin_args): ... this. Rename builtin_arg
41146 enum values and differentiate between ARG_BUILTIN_MEMORY
41147 and ARG_BUILTIN_NEON_MEMORY.
41148 (arm_expand_neon_builtin_1): Rename to ...
41149 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
41150 values, arm_expand_builtin_args and add bool parameter NEON.
41151 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
41152 (arm_expand_vfp_builtin): Likewise.
41153 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
41155 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41157 PR middle-end/77484
41158 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
41159 * predict.c (tree_estimate_probability_bb): Reverse direction of
41160 polymorphic call predictor.
41162 2017-01-06 David Malcolm <dmalcolm@redhat.com>
41164 * passes.c (execute_one_pass): Split out pass-skipping logic into...
41165 (determine_pass_name_match): ...this new function and...
41166 (should_skip_pass_p): ...this new function.
41168 2017-01-06 Nathan Sidwell <nathan@acm.org>
41170 * ipa-visibility.c (function_and_variable_visibility): Reformat
41171 comments and long lines. Remove extrneous if.
41172 * symtab.c (symtab_node::make_decl_local): Fix code format.
41173 (symtab_node::set_section_for_node): Fix comment typo.
41175 2017-01-06 Martin Liska <mliska@suse.cz>
41178 * lra-constraints.c: Rename invariant to lra_invariant.
41179 * predict.c (set_even_probabilities): Initialize e to NULL.
41181 2017-01-05 Martin Sebor <msebor@redhat.com>
41183 PR tree-optimization/78910
41184 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
41185 (format_integer): Correct off-by-one error in the handling
41186 of precision with negative numbers in signed conversions..
41188 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
41190 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
41192 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41194 PR tree-optimization/71016
41195 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
41196 factor_out_conditional_conversion. Formatting fix.
41197 (factor_out_conditional_conversion): Add cond_stmt argument.
41198 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
41199 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
41202 2017-01-05 David Malcolm <dmalcolm@redhat.com>
41204 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
41205 read-rtl-function.o, and selftest-rtl.o.
41206 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
41207 (selftest::aarch64_test_loading_full_dump): New function.
41208 (selftest::aarch64_run_selftests): New function.
41209 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
41210 selftest::aarch64_run_selftests.
41211 * config/i386/i386.c
41212 (selftest::ix86_test_loading_dump_fragment_1): New function.
41213 (selftest::ix86_test_loading_call_insn): New function.
41214 (selftest::ix86_test_loading_full_dump): New function.
41215 (selftest::ix86_test_loading_unspec): New function.
41216 (selftest::ix86_run_selftests): Call the new functions.
41217 * emit-rtl.c (maybe_set_max_label_num): New function.
41218 * emit-rtl.h (maybe_set_max_label_num): New decl.
41219 * function.c (instantiate_decls): Guard call to
41220 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
41221 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
41223 * gensupport.c (gen_reader::gen_reader): Pass "false"
41224 for new "compact" param of rtx_reader.
41225 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
41226 rather than an empty string for NULL strings.
41227 * read-md.c: Potentially include config.h rather than bconfig.h.
41228 Wrap include of errors.h with #ifdef GENERATOR_FILE.
41229 (have_error): New global, copied from errors.c.
41230 (md_reader::read_name): Rename to...
41231 (md_reader::read_name_1): ...this, adding "out_loc" param,
41232 and converting "missing name or number" to returning false, rather
41234 (md_reader::read_name): Reimplement in terms of read_name_1.
41235 (md_reader::read_name_or_nil): New function.
41236 (md_reader::read_string): Handle "(nil)" by returning NULL.
41237 (md_reader::md_reader): Add new param "compact".
41238 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
41239 (md_reader::read_file): New method.
41240 * read-md.h (md_reader::md_reader): Add new param "compact".
41241 (md_reader::read_file): New method.
41242 (md_reader::is_compact): New accessor.
41243 (md_reader::read_name): Convert return type from void to file_location.
41244 (md_reader::read_name_or_nil): New decl.
41245 (md_reader::read_name_1): New decl.
41246 (md_reader::m_compact): New field.
41247 (noop_reader::noop_reader): Pass "false" for new "compact" param
41249 (rtx_reader::rtx_reader): Add new "compact" param.
41250 (rtx_reader::read_rtx_operand): Make virtual and convert return
41251 type from void to rtx.
41252 (rtx_reader::read_until): New decl.
41253 (rtx_reader::handle_any_trailing_information): New virtual function.
41254 (rtx_reader::postprocess): New virtual function.
41255 (rtx_reader::finalize_string): New virtual function.
41256 (rtx_reader::m_in_call_function_usage): New field.
41257 (rtx_reader::m_reuse_rtx_by_id): New field.
41258 * read-rtl-function.c: New file.
41259 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
41260 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
41261 (selftest::verify_three_block_rtl_cfg): New decl.
41262 * read-rtl-function.h: New file.
41263 * read-rtl.c: Potentially include config.h rather than bconfig.h.
41264 For host, include function.h, memmodel.h, and emit-rtl.h.
41265 (one_time_initialization): New function.
41266 (struct compact_insn_name): New struct.
41267 (compact_insn_names): New array.
41268 (find_code): Handle insn codes in compact dumps.
41269 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
41270 (bind_subst_iter_and_attr): Likewise.
41271 (add_condition_to_string): Likewise.
41272 (add_condition_to_rtx): Likewise.
41273 (apply_attribute_uses): Likewise.
41274 (add_current_iterators): Likewise.
41275 (apply_iterators): Likewise.
41276 (initialize_iterators): Guard usage of apply_subst_iterator with
41277 #ifdef GENERATOR_FILE.
41278 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
41279 (md_reader::read_mapping): Likewise.
41280 (add_define_attr_for_define_subst): Likewise.
41281 (add_define_subst_attr): Likewise.
41282 (read_subst_mapping): Likewise.
41283 (check_code_iterator): Likewise.
41284 (rtx_reader::read_rtx): Likewise. Move one-time initialization
41286 (one_time_initialization): New function.
41287 (rtx_reader::read_until): New method.
41288 (read_flags): New function.
41289 (parse_reg_note_name): New function.
41290 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
41291 Handle reuse_rtx ids.
41292 Wrap iterator lookup within #ifdef GENERATOR_FILE.
41293 Add parsing support for RTL dumps, mirroring the special-cases in
41294 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
41295 values, and calling handle_any_trailing_information.
41296 (rtx_reader::read_rtx_operand): Convert return type from void
41297 to rtx, returning return_rtx. Handle case 'e'. Call
41298 finalize_string on XSTR and XTMPL fields.
41299 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
41300 "(nil)" values were omitted. Call the postprocess vfunc on the
41302 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
41303 class ctor. Initialize m_in_call_function_usage. Call
41304 one_time_initialization.
41305 * rtl-tests.c (selftest::test_uncond_jump): Call
41306 set_new_first_and_last_insn.
41307 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
41308 * selftest-rtl.c: New file.
41309 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
41310 (selftest::get_insn_by_uid): New decl.
41311 * selftest-run-tests.c (selftest::run_tests): Call
41312 read_rtl_function_c_tests.
41313 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
41314 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
41317 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
41319 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
41320 operands in a special way. Assert that pos+len <= mode precision.
41322 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41324 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
41325 3 argument Alias with unlimited for the negative form.
41326 (fno-vect-cost-model): Removed.
41328 2017-01-05 Martin Liska <mliska@suse.cz>
41330 * hsa-gen.c (gen_hsa_divmod): New function.
41331 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
41333 2017-01-05 Martin Liska <mliska@suse.cz>
41336 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
41339 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41341 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
41342 small constant length operands.
41344 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41346 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
41347 between loop iterations.
41349 2017-01-05 Martin Liska <mliska@suse.cz>
41352 * gimplify.c (gimplify_decl_expr): Compare to
41353 asan_poisoned_variables instread of checking flags.
41354 (gimplify_target_expr): Likewise.
41355 (gimplify_expr): Likewise.
41356 (gimplify_function_tree): Conditionally initialize
41357 asan_poisoned_variables.
41359 2017-01-04 Jeff Law <law@redhat.com>
41361 PR tree-optimizatin/78812
41362 * rtl.h (contains_mem_rtx_p): Prototype.
41363 * ifcvt.c (containts_mem_rtx_p): Move from here to...
41364 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
41365 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
41366 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
41367 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
41369 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41371 * input.c (assert_char_at_range): Default-initialize actual_range.
41373 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41375 * df-scan.c (df_ref_create_structure): Make regno unsigned,
41376 to match the caller.
41378 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41380 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
41381 insns after final jump in test to emit dummy move.
41383 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41385 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
41386 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
41388 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41390 * multiple_target.c (create_dispatcher_calls): Init e_next.
41391 * tree-ssa-loop-split.c (split_loop): Init border.
41392 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
41395 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
41400 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
41401 (altivec_register_operand): Do not return true if the operand
41402 contains a SUBREG mixing SImode and SFmode.
41403 (vsx_register_operand): Likewise.
41404 (vsx_reg_sfsubreg_ok): New predicate.
41405 (vfloat_operand): Do not return true if the operand contains a
41406 SUBREG mixing SImode and SFmode.
41407 (vint_operand): Likewise.
41408 (vlogical_operand): Likewise.
41409 (gpc_reg_operand): Likewise.
41410 (int_reg_operand): Likewise.
41411 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
41412 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
41413 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
41415 (rs6000_emit_move_si_sf_subreg): New helper function.
41416 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
41417 fixup SUBREGs involving SImode and SFmode.
41418 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
41419 numbers for the new peephole2 optimization.
41420 (peephole2 for SFmode unions): New peephole2 to optimize cases in
41421 the GLIBC math library that do AND/IOR/XOR operations on single
41422 precision floating point.
41423 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
41424 target macros to say whether we need to avoid SUBREGs mixing
41426 (TARGET_ALLOW_SF_SUBREG): Likewise.
41427 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
41428 (UNSPEC_SI_FROM_SF): Likewise.
41429 (iorxor): Change spacing.
41430 (and_ior_xor): New iterator for AND, IOR, and XOR.
41431 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
41432 (movdi_from_sf_zero_ext): Likewise.
41433 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
41434 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
41435 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
41436 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
41437 (fms<mode>4): Likewise.
41438 (fnma<mode>4): Likewise.
41439 (fnms<mode>4): Likewise.
41440 (nfma<mode>4): Likewise.
41441 (nfms<mode>4): Likewise.
41443 2017-01-04 Marek Polacek <polacek@redhat.com>
41446 * doc/invoke.texi: Document -Wpointer-compare.
41448 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41450 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
41453 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
41454 descriptions for -gdwarf-5 and emit them as uleb128 instead of
41457 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
41460 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
41461 documentation of the powerpc_popcntb_ok attribute.
41462 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
41463 code to issue warning messages if a requested CPU configuration is
41464 not supported by the binary (assembler and loader) toolchain.
41465 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
41466 made to define a built-in function that has been disabled.
41467 (paired_init_builtins): Add assertion to prevent ICE if attempt is
41468 made to define a built-in function that has been disabled.
41469 (altivec_init_builtins): Add comment explaining why definition
41470 of the DST built-in functions is not preceded by an assertion
41471 check. Add assertions to prevent ICE if attempts are made to
41472 define an altivec predicate or an abs* built-in function that has
41474 (htm_init_builtins): Add comment explaining why definition of the
41475 htm built-in functions is not preceded by an assertion check.
41477 2017-01-04 Jeff Law <law@redhat.com>
41479 PR tree-optimizatin/67955
41480 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
41481 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
41482 the points-to solution does not include pt_null. Use DECL_PT_UID
41485 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
41487 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
41488 Use gen_int_mode instead of gen_lopwart for const_int operands.
41490 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41492 PR tree-optimization/71563
41493 * match.pd: Simplify X << Y into X if Y is known to be 0 or
41494 out of range value - has low bits known to be zero.
41496 2017-01-04 Alan Modra <amodra@gmail.com>
41498 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
41499 * configure: Regenerate.
41500 * config.in: Regenerate.
41502 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41505 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
41506 a substring of the message, but strcmp with the whole message. Ifdef
41507 ENABLE_NLS, translate the message first using dgettext.
41509 2017-01-03 Jeff Law <law@redhat.com>
41511 PR tree-optimizatin/78856
41512 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
41513 (mark_threaded_blocks): Remove code to truncate thread paths that
41514 cross multiple loop headers. Instead invalidate the cached loop
41515 iteration information and handle case of a thread path walking
41516 into an irreducible region.
41518 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
41521 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
41522 assertions. Add support for doing the signbit if the IEEE 128-bit
41523 floating point value is in a GPR.
41524 * config/rs6000/rs6000.md (Fsignbit): Delete.
41525 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
41526 Update the length attribute if the value is in a GPR.
41527 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
41528 the sign or zero extension instruction, since the value is always 0/1.
41529 (signbit<mode>2_dm2): Delete using <Fsignbit>.
41532 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
41533 extracting SImode to a GPR register so that we can generate a
41534 store, limit the vector to be in a traditional Altivec register
41535 for the vextuwrx instruction.
41537 2017-01-03 Ian Lance Taylor <iant@google.com>
41539 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
41541 2017-01-03 Martin Sebor <msebor@redhat.com>
41543 PR tree-optimization/78696
41544 * gimple-ssa-sprintf.c (format_floating): Correct handling of
41545 precision. Use MPFR for %f for greater fidelity. Correct handling
41547 (pass_sprintf_length::compute_format_length): Set width and precision
41548 specified by asrerisk to void_node for vararg functions.
41549 (try_substitute_return_value): Adjust dump output.
41551 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
41553 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
41555 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
41557 * doc/invoke.texi (SPARC options): Document -mlra as the default.
41558 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
41559 -mlra/-mno-lra was passed to the compiler.
41561 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
41563 PR rtl-optimization/65618
41564 * emit-rtl.c (try_split): Move initialization of "before" and
41565 "after" to just before the call to emit_insn_after_setloc.
41567 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
41569 * doc/md.texi (Standard Names): Remove reference to Java frontend.
41571 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
41573 * dwarf2out.c (gen_enumeration_type_die): When
41574 -gno-strict-dwarf, add a DW_AT_encoding attribute.
41576 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41578 PR tree-optimization/78965
41579 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41580 Change first argument from const call_info & to call_info &. For %n
41581 set info.nowrite to false.
41583 PR middle-end/78901
41584 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41585 possibly throwing calls.
41587 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41588 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41589 and fns handling, rather than in a separate case SSA_NAME.
41591 2017-01-02 Jeff Law <law@redhat.com>
41593 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41594 fixes for first_period and second_period variables.
41596 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41599 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41600 (*insvqi_1): New insn pattern.
41601 (*insvqi_1_mem_rex64): Ditto.
41602 (*insvqi_2): Ditto.
41603 (*insvqi_3): Rename from *insvqi.
41605 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41607 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41609 * doc/cfg.texi (Edges): Remove reference to Java.
41610 (Maintaining the CFG): Ditto.
41612 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41614 PR middle-end/77674
41615 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41616 transparent aliases.
41618 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41620 PR middle-end/77484
41621 * predict.def (PRED_CALL): Update hitrate.
41622 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41623 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41624 into direct/indirect/polymorphic variants.
41626 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41628 Update copyright years.
41630 * gcc.c (process_command): Update copyright notice dates.
41631 * gcov-dump.c (print_version): Ditto.
41632 * gcov.c (print_version): Ditto.
41633 * gcov-tool.c (print_version): Ditto.
41634 * gengtype.c (create_file): Ditto.
41635 * doc/cpp.texi: Bump @copying's copyright year.
41636 * doc/cppinternals.texi: Ditto.
41637 * doc/gcc.texi: Ditto.
41638 * doc/gccint.texi: Ditto.
41639 * doc/gcov.texi: Ditto.
41640 * doc/install.texi: Ditto.
41641 * doc/invoke.texi: Ditto.
41643 Copyright (C) 2017 Free Software Foundation, Inc.
41645 Copying and distribution of this file, with or without modification,
41646 are permitted in any medium without royalty provided the copyright
41647 notice and this notice are preserved.
projects / platform / upstream / gcc.git / blob