1 2017-03-14 Martin Liska <mliska@suse.cz>
4 * multiple_target.c (expand_target_clones): Drop local.local
5 flag for default implementation.
7 2017-03-14 Richard Biener <rguenther@suse.de>
9 PR tree-optimization/80030
10 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
12 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
14 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
15 gcc_fallthrough() instead of __attribute__((fallthrough));
17 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19 * doc/gcc.texi: Remove "up" link to (DIR).
20 * doc/gccint.texi: Ditto.
22 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
24 * doc/install.texi (Specific) <avr>: Remove reference to
27 2017-03-13 Jeff Law <law@redhat.com>
29 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
30 attribute rather than comments.
32 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
33 match_scratch operand is highest.
35 2017-03-13 Martin Liska <mliska@suse.cz>
38 * ipa-pure-const.c (warn_function_noreturn): If the declarations
39 is a CHKP clone, use original declaration.
41 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
43 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
44 (arc_conditional_register_usage): Use a different allocation order
45 when optimizing for size.
46 * common/config/arc/arc-common.c (arc_option_optimization_table):
47 Section anchors default on when optimizing for size.
49 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
51 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
53 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
55 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
56 * config/arc/arc.md (cpu_facility): Add cd variant.
57 (*movqi_insn): Add code density variant.
58 (*movhi_insn): Likewise.
59 (*movqi_insn): Likewise.
60 (*addsi3_mixed): Likewise.
61 (subsi3_insn): Likewise.
63 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
65 * config/arc/arc.md (movsi_cond_exec): Update constraint.
67 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
69 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
70 expressions with MINUS and UNARY ops.
72 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
75 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
77 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
78 between vec_select and vector argument.
79 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
80 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
81 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
82 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
83 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
84 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
86 2017-03-13 Richard Biener <rguenther@suse.de>
89 * params.def (vect-max-peeling-for-alignment): Fix typo.
91 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
93 * doc/install.texi (Specific) <mips-*-*>: Remove description of
94 issue that only occurred with binutils below 2.18.
96 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
98 * doc/install.texi (Specific) <cris-axis-elf>: No longer
99 refer to binutils 2.11/2.12 minimum.
101 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
103 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
104 ftp.kernel.org and simplify binutils requirement.
106 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
108 * doc/invoke.texi (Warning Options): Fix spelling of link-time
110 (Optimize Options): Ditto. Also remove redundancy.
112 2017-03-10 David Malcolm <dmalcolm@redhat.com>
115 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
117 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
118 to G_ to avoid double translation.
120 2017-03-10 David Malcolm <dmalcolm@redhat.com>
123 * auto-profile.c (get_combined_location): Convert leading
124 character of diagnostics to lower case and remove trailing period.
125 (read_profile): Likewise for various diagnostics.
126 * config/arm/arm.c (arm_option_override): Remove trailing period
127 from various diagnostics.
128 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
129 (msp430_expand_delay_cycles): Likewise.
131 2017-03-10 David Malcolm <dmalcolm@redhat.com>
134 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
135 full command-line argument, rather than just "str".
136 (aarch64_validate_march): Likewise.
137 (aarch64_validate_mtune): Likewise.
139 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
141 PR rtl-optimization/78911
142 * lra-assigns.c (must_not_spill_p): New function.
145 2017-03-10 Jakub Jelinek <jakub@redhat.com>
147 PR tree-optimization/79981
148 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
149 ATOMIC_COMPARE_EXCHANGE ifn result.
150 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
151 IFN_ATOMIC_COMPARE_EXCHANGE.
153 2017-03-10 David Malcolm <dmalcolm@redhat.com>
156 * opts.c (parse_sanitizer_options): Add missing question mark to
157 "did you mean" message.
159 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
161 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
163 (VMULEUH_UNS): Likewise.
164 (VMULOUB_UNS): Likewise.
165 (VMULOUH_UNS): Likewise.
166 * config/rs6000/rs6000.c (builtin_function_type): Remove
167 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
169 2017-03-10 David Malcolm <dmalcolm@redhat.com>
172 * read-rtl-function.c (function_reader::read_rtx_operand): Update
173 x with result of extra_parsing_for_operand_code_0.
174 (function_reader::extra_parsing_for_operand_code_0): Convert
175 return type from void to rtx, returning x. When reading
176 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
177 larger size containing struct block_symbol.
179 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
181 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
182 -mfloat128-hardware without -m64.
184 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
187 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
188 entries to the case statement that marks unsigned arguments to
189 overloaded functions.
191 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
193 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
194 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
196 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
199 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
200 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
202 2017-03-10 Martin Liska <mliska@suse.cz>
206 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
207 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
210 2017-03-10 Olivier Hainque <hainque@adacore.com>
212 * tree-switch-conversion (array_value_type): Start by resetting
213 candidate type to it's main variant.
215 2017-03-10 Jakub Jelinek <jakub@redhat.com>
217 PR rtl-optimization/79909
218 * combine.c (try_combine): Use simplify_replace_rtx on individual
219 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
220 of the whole CALL_INSN_FUNCTION_USAGE.
222 PR tree-optimization/79972
223 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
224 get_range_info on SSA_NAMEs. Formatting fixes.
226 2017-03-10 Richard Biener <rguenther@suse.de>
227 Jakub Jelinek <jakub@redhat.com>
229 PR tree-optimization/77975
230 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
232 (get_val_for): For constant x return it. Formatting fix.
233 (loop_niter_by_eval): Avoid pointless looping if the next iteration
234 would use the same bases as the current one.
236 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
238 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
239 instead of vec_select for V1TImode.
240 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
242 (VSX_LE_128): Add V1TI to this mode iterator.
243 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
244 (*vsx_le_perm_store_<mode>): Likewise.
245 (pre-reload splitter for VSX stores): Likewise.
246 (post-reload splitter for VSX stores): Likewise.
247 (*vsx_xxpermdi2_le_<mode>): Likewise.
248 (*vsx_lxvd2x2_le_<mode>): Likewise.
249 (*vsx_stxvd2x2_le_<mode>): Likewise.
251 2017-03-09 Michael Eager <eager@eagercon.com>
253 Correct failures with --enable-checking=yes,rtl.
255 * config/microblaze/microblaze.c (microblaze_expand_shift):
256 Replace GET_CODE test with CONST_INT_P and INTVAL test with
258 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
259 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
261 2017-03-09 Richard Biener <rguenther@suse.de>
263 PR tree-optimization/79977
264 * graphite-scop-detection.c (scop_detection::merge_sese):
265 Handle the case of extra exits to blocks dominating the entry.
267 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
269 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
272 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
274 PR rtl-optimization/79949
275 * lra-constraints.c (process_alt_operands): Check memory when
276 trying to predict a cycle. Print about the overall increase.
278 2017-03-09 Richard Biener <rguenther@suse.de>
281 * gimple-expr.c (useless_type_conversion_p): Preserve
282 TYPE_SATURATING for fixed-point types.
284 2017-03-09 Richard Biener <rguenther@suse.de>
287 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
288 alignment of BLKmode params.
290 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
293 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
294 (VALL_NO_V2Q): Likewise.
296 * config/aarch64/aarch64-simd.md
297 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
299 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
300 VALL_NO_V2Q mode iterator.
301 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
303 2017-03-09 Martin Liska <mliska@suse.cz>
305 PR tree-optimization/79631
306 * tree-chkp-opt.c (chkp_is_constant_addr): Call
307 tree_int_cst_sign_bit just for INTEGER constants.
309 2017-03-09 Martin Liska <mliska@suse.cz>
313 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
316 2017-03-09 Marek Polacek <polacek@redhat.com>
319 * tree.c (inchash::add_expr): Handle TREE_VEC.
321 2017-03-09 Martin Liska <mliska@suse.cz>
324 (chkp_narrow_size_and_offset): New function.
325 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
326 (void chkp_parse_bit_field_ref): New function.
327 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
328 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
330 2017-03-09 Martin Liska <mliska@suse.cz>
333 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
334 (chkp_find_bounds_1): Remove gcc_unreachable.
336 2017-03-09 Jakub Jelinek <jakub@redhat.com>
339 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
340 BUILT_IN_SYNC*, determine the access type from the size suffix and
341 always build a MEM_REF with that type. Handle forgotten
342 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
345 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
346 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
347 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
348 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
349 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
350 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
351 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
352 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
353 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
354 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
355 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
356 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
357 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
358 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
359 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
360 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
361 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
362 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
363 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
364 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
365 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
366 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
367 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
368 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
369 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
370 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
371 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
372 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
373 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
374 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
375 definitions outside of __OPTIMIZE__ guarded section.
378 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
379 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
380 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
381 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
384 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
386 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
387 ("vfenez<mode>"): Add missing constraints.
389 2017-03-08 Martin Sebor <msebor@redhat.com>
392 * config/nds32/nds32.c (nds32_option_override):
394 2017-03-08 Jakub Jelinek <jakub@redhat.com>
397 * gimplify.c (gimplify_omp_for): Replace index var in outer
398 taskloop statement with an artificial variable and add
399 OMP_CLAUSE_PRIVATE clause for it.
401 2017-03-08 Richard Biener <rguenther@suse.de>
403 PR tree-optimization/79955
404 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
405 for accesses that are completely outside of the variable.
407 2017-03-08 Andrew Haley <aph@redhat.com>
409 PR tree-optimization/79943
410 * tree-ssa-loop-split.c (compute_new_first_bound): When
411 calculating the new upper bound, (END-BEG) should be added, not
414 2017-03-08 Jakub Jelinek <jakub@redhat.com>
416 * config/avr/avr.md (setmemhi): Make sure match_dup
417 operand number comes before match_scratch.
419 2017-03-08 Richard Biener <rguenther@suse.de>
421 PR tree-optimization/79920
422 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
423 with ncopies == 1 to ...
424 (vect_transform_slp_perm_load): ... here. Properly compute
425 all element loads by iterating VF times over the group. Do
426 not handle ncopies (computed in a broken way) in
427 vect_create_mask_and_perm.
429 2017-03-08 Jakub Jelinek <jakub@redhat.com>
432 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
433 is a uniform vector, use uniform_vector_p return value instead of
434 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
436 2017-03-07 Marek Polacek <polacek@redhat.com>
439 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
440 (alloca_call_type): Likewise.
442 2017-03-07 Martin Liska <mliska@suse.cz>
444 * gcov.c (process_args): Put comment to correct location.
446 2017-03-07 Martin Liska <mliska@suse.cz>
449 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
450 Use array_at_struct_end_p instead of DECL_CHAIN (field).
451 (chkp_narrow_bounds_for_field): Likewise.
452 (chkp_parse_array_and_component_ref): Pass one more argument to
455 2017-03-07 Richard Biener <rguenther@suse.de>
457 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
460 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
462 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
463 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
465 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
468 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
469 to end of description.
470 (PARAM_MAX_STORES_TO_MERGE): Likewise.
472 2017-03-07 Jakub Jelinek <jakub@redhat.com>
474 PR rtl-optimization/79901
475 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
477 (*avx512f_<code><mode>3<mask_name>): ... this.
478 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
479 iterator instead of VI8_AVX2_AVX512BW.
481 PR rtl-optimization/79901
482 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
483 min/max expander, expand it using expand_vec_cond_expr.
486 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
489 2017-03-06 Jakub Jelinek <jakub@redhat.com>
492 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
493 to void * for PCH reasons.
494 * dwarf2out.c (output_loc_operands, output_die): Cast
495 v.val_vec.array to unsigned char *.
497 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
500 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
503 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
505 PR rtl-optimization/79571
506 * lra-constraints.c (process_alt_operands): Calculate static
507 reject and subtract it from overall when only addresses will be
510 2017-03-06 Julia Koval <julia.koval@intel.com>
513 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
514 incoming stack boundary to 128 for 64-bit targets.
516 2017-03-06 Richard Biener <rguenther@suse.de>
518 PR tree-optimization/79894
519 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
520 to NULL after folding it.
522 2017-03-06 Richard Biener <rguenther@suse.de>
524 PR tree-optimization/79824
525 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
526 check disabling peeling for gaps.
528 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
530 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
531 attributes): Document gettimeofday.
533 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
535 * config/s390/s390.c (s390_option_override_internal): Set
536 PARAM_MIN_VECT_LOOP_BOUND
538 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
540 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
541 * config/s390/s390.md: Likewise.
543 2017-03-06 Jakub Jelinek <jakub@redhat.com>
546 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
547 (<avx2_avx512>_perm<mode>): Rename to ...
548 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
550 (<avx512>_perm<mode>_mask): Rename to ...
551 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
553 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
554 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
555 instead of VI8F_256_512.
556 (avx512f_perm<mode>): New define_expand.
557 (avx512f_perm<mode>_mask): Likewise.
558 (avx512f_perm<mode>_1<mask_name>): New define_insn.
559 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
561 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
563 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
564 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
566 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
568 2017-03-06 Martin Liska <mliska@suse.cz>
571 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
572 when having a SSA NAME w/o VAR_DECL assigned to it.
574 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
576 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
577 msa_dpsub_<su>_d): Fix MODE for vec_select.
579 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
581 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
583 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
585 2017-03-06 Richard Biener <rguenther@suse.de>
587 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
588 * plugin.c (register_plugin_info): Likewise.
589 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
591 2017-03-05 Jakub Jelinek <jakub@redhat.com>
593 * config/i386/sse.md (sse_storehps, sse_storelps,
594 avx_<castmode><avxsizesuffix>_<castmode>,
595 avx512f_<castmode><avxsizesuffix>_<castmode>,
596 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
597 in condition that at least one operand is not a MEM.
599 2017-03-03 Jakub Jelinek <jakub@redhat.com>
602 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
603 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
605 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
606 gimple_call_nothrow_p flag based on whether original builtin can throw.
607 If it can, emit following stmts on the fallthrough edge.
608 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
609 don't create new bb if inserting just debug stmts on the edge, try to
610 insert them on the fallthru bb or just reset debug stmts.
612 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
615 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
616 restore recog_data (including the operand rtxes inside it) around
617 the call to get_insn_template.
619 2017-03-03 Martin Sebor <msebor@redhat.com>
621 PR tree-optimization/79699
622 * context.c (context::~context): Free MPFR caches to avoid
623 a memory leak on program exit.
625 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
627 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
628 Use wide_int::ulow () instead of .elt (0).
630 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
632 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
633 (*pushxf): Limit oF constraint to 32bit targets and add oC
634 constraint for 64bit targets.
635 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
636 (*pushdf): Change rmF constraint to rmC.
638 2017-03-03 Martin Liska <mliska@suse.cz>
640 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
641 Remove unused variable.
643 2017-03-03 Jakub Jelinek <jakub@redhat.com>
646 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
647 is a memory operand, increase num_memory.
648 (ix86_expand_args_builtin): Likewise.
650 2017-03-03 Jan Hubicka <jh@suse.cz>
653 * ipa-devirt.c (maybe_record_node): Properly handle
654 __cxa_pure_virtual visibility.
656 2017-03-03 Martin Liska <mliska@suse.cz>
658 PR tree-optimization/79803
659 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
661 (pass_loop_prefetch::execute): Disabled optimization if an
662 assumption about L1 cache size is not met.
664 2017-03-03 Martin Liska <mliska@suse.cz>
666 PR rtl-optimization/79574
667 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
668 (hash_scan_set): Likewise.
669 (dump_hash_table): Likewise.
670 (hoist_code): Likewise.
672 2017-03-03 Richard Biener <rguenther@suse.de>
674 * fixed-value.c (fixed_from_string): Restore use of elt (1)
675 in place of uhigh ().
676 (fixed_convert_from_real): Likewise.
678 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
681 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
683 2017-03-03 Richard Biener <rguenther@suse.de>
686 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
687 TYPE_OVERFLOW_UNDEFINED check.
689 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
691 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
693 (vector_ae_<mode>_p): Likewise.
694 (vector_nez_<mode>_p): Likewise.
695 (vector_ne_v2di_p): Likewise.
696 (vector_ae_v2di_p): Likewise.
697 (vector_ne_<mode>_p): Likewise.
698 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
700 (vsx_tsqrt<mode>2_fe): Likewise.
702 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
705 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
707 2017-03-02 Jakub Jelinek <jakub@redhat.com>
709 PR rtl-optimization/79780
710 * cprop.c (one_cprop_pass): When second and further conditional trap
711 in a single basic block is turned into an unconditional trap, turn it
712 into a deleted note to avoid RTL verification failures.
714 2017-03-02 Richard Biener <rguenther@suse.de>
716 * fold-const.c (const_binop): Use ulow () instead of elt (0).
718 2017-03-02 Richard Biener <rguenther@suse.de>
720 PR tree-optimization/79345
722 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
723 param and abort the walk, returning -1 if it is hit.
724 (walk_aliased_vdefs): Take a limit param and pass it on.
725 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
726 defaulting to 0 and return a signed int.
727 * tree-ssa-uninit.c (struct check_defs_data): New struct.
728 (check_defs): New helper.
729 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
730 about uninitialized memory.
731 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
732 bogus uninitialized warning.
733 (fixed_convert_from_real): Likewise.
735 2017-03-02 Bin Cheng <bin.cheng@arm.com>
737 PR tree-optimization/66768
738 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
739 iv_use if base object can't be determined.
741 2017-03-02 Jakub Jelinek <jakub@redhat.com>
743 PR tree-optimization/79345
744 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
745 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
746 (get_pattern_stats): Initialize it.
747 * genemit.c (gen_expand): Verify match_scratch numbers come after
748 match_operand/match_dup numbers.
749 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
750 match_scratch numbers.
751 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
753 * config/s390/s390.md (trunctdsd2): Likewise.
755 2017-03-02 Richard Biener <rguenther@suse.de>
757 * wide-int.h (wide_int_storage::operator=): Implement in terms
760 2017-03-02 Richard Biener <rguenther@suse.de>
762 PR tree-optimization/79777
763 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
764 the to insert expression to sth existing.
766 2017-03-01 Martin Sebor <msebor@redhat.com>
769 * gimple-ssa-sprintf.c
770 (directive::known_width_and_precision): New function.
771 (format_integer): Use it.
772 (get_mpfr_format_length): Consider the full range of precision
773 when computing %g output with the # flag. Set the likely byte
774 count to 3 rather than 1 when precision is indeterminate.
775 (format_floating): Correct the lower bound of precision.
777 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
779 * doc/invoke.texi: Document default code model for 64-bit Linux.
781 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
784 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
785 udiv rather than div since input pattern is unsigned.
787 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
789 * config/i386/i386.c (print_reg): Warn for values of
790 unsupported size in integer register.
792 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
795 * config/rs6000/predicates.md (current_file_function_operand): Do
796 not allow self calls to be local if the function is replaceable.
798 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
801 * config/rs6000/altivec.h (vec_ctz and others): Change the
802 preprocessor macro that controls conditional compilation from
803 _ARCH_PWR9 to __POWER9_VECTOR__.
804 (vec_all_ne): Change parameterization of __altivec_scalar_pred
805 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
806 control (instead of _ARCH_PWR9 control) so that template
807 definition uses power9-specific function.
808 (vec_any_eq): Likewise.
809 (vec_all_ne): Change macro definition to use a power9-specific
810 expansion under #ifdef __POWER9_VECTOR__ control (instead of
812 (vec_any_eq) Likewise.
813 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
814 expansion for CMPNEF to remove support for xvcmpnesp instruction.
815 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
816 support for xvcmpnedp instruction.
817 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
818 macro expansion so that Power9 implementation of vec_all_ne does
819 not use the AltiVec predicate framework.
820 (VCMPNEH_P): Likewise.
821 (VCMPNEW_P): Likewise.
822 (VCMPNED_P): Likewise.
823 (VCMPNEFP_P): Likewise.
824 (VCMPNEDP_P): Likewise.
825 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
826 implementation of vec_any_eq to not use AltiVec predicate
828 (VCMPAEH_P): Likewise.
829 (VCMPAEW_P): Likewise.
830 (VCMPAED_P): Likewise.
831 (VCMPAEFP_P): Likewise.
832 (VCMPAEDP_P): Likewise.
833 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
834 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
835 not use the AltiVec predicate framework.
836 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
837 of vec_any_eq to not use AltiVec predicate framework.
838 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
839 support for predefined __POWER9_VECTOR__ macro to indicate that
840 Power9 instruction selection is enabled.
841 (altivec_overloaded_builtins): Remove extraneous
842 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
843 function argument types RS6000_BTI_bool_V16QI and
844 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
845 entry for overloaded function argument types RS6000_BTI_bool_V4SI
846 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
847 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
848 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
849 Power9 for implementations of vec_cmpne. Change the signature for
850 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
851 (representing vec_all_ne) to remove the previously described first
852 argument of type RS6000_BTI_INTSI, as this was an artifact of
853 reliance on the AltiVec predicate framework, which is no longer
854 used in the implementation of these functions. Add
855 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
856 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
857 since, unlike the AltiVec predicate framework implementation, we
858 do not share function descriptors between vec_alle and vec_anyeq.
859 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
860 set of modes that receive special treatment even when
861 TARGET_P9_VECTOR is true. The special treatment emits code that
862 does not depend on Power9 instructions.
863 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
864 define_expand to not rely on AltiVec predicate framework.
865 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
867 (vector_ne_v2di_p): Change this define_expand to not rely on
868 AltiVec predicate framework.
869 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
871 (vector_ne_<mode>_p): Change this define_expand to not rely on
872 AltiVec predicate framework.
873 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
875 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
876 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
878 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
879 define_insn pattern because the xvcmpne<VSs>. instruction is not
881 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
882 instruction is not supported.
884 2017-03-01 Jakub Jelinek <jakub@redhat.com>
886 * config/nvptx/nvptx.c: Include intl.h.
888 2017-03-01 Martin Jambor <mjambor@suse.cz>
891 * ipa-prop.h (ipa_bits): Removed field known.
892 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
893 to pointers. Adjusted their comments to warn about their sharing.
894 (ipcp_transformation_summary): Change bits to a vector of pointers.
895 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
896 (ipa_get_ipa_bits_for_value): Declare.
897 * tree-vrp.h (value_range): Mark as GTY((for_user)).
898 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
899 (ipa_bits_hash_table): Likewise.
900 (ipa_vr_ggc_hash_traits): Likewise.
901 (ipa_vr_hash_table): Likewise.
902 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
903 being pointers and vr_known being removed.
904 (ipa_set_jf_unknown): Likewise.
905 (ipa_get_ipa_bits_for_value): New function.
906 (ipa_set_jfunc_bits): Likewise.
907 (ipa_get_value_range): New overloaded functions.
908 (ipa_set_jfunc_vr): Likewise.
909 (ipa_compute_jump_functions_for_edge): Use the above functions to
910 construct bits and vr parts of jump functions.
911 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
912 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
914 (ipcp_grow_transformations_if_necessary): Also allocate
915 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
917 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
918 them. Fix too long lines.
919 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
920 vr_known being removed.
921 (ipa_read_jump_function): Use new setter functions to construct bits
922 and vr parts of jump functions or set them to NULL.
923 (write_ipcp_transformation_info): Adjust for bits being pointers.
924 (read_ipcp_transformation_info): Likewise.
925 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
927 Include gt-ipa-prop.h.
928 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
930 (ipcp_store_bits_results): Likewise.
931 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
932 Do not write to existing jump functions but use a temporary instead.
934 2017-03-01 Jakub Jelinek <jakub@redhat.com>
937 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
938 attempt to use its first operand as BIT_FIELD_REF base.
940 2017-03-01 Richard Biener <rguenther@suse.de>
943 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
944 interpolating formula in wrapping arithmetic.
945 (chrec_apply): Convert chrec_evaluate return value to wanted type.
947 2017-03-01 Jakub Jelinek <jakub@redhat.com>
949 PR tree-optimization/79734
950 * tree-vect-generic.c (expand_vector_condition): Optimize
951 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
952 Handle VEC_COND_EXPR where comparison has different inner width from
955 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
957 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
958 markup, and similar issues. Remove @opindex entries for things
959 that aren't options. Add missing -mmpy-option entries.
961 2017-02-28 Jakub Jelinek <jakub@redhat.com>
963 PR tree-optimization/79737
964 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
965 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
966 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
967 instead of byte_size. Formatting fix.
968 (shift_bytes_in_array_right): Formatting fix.
970 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
973 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
974 condition on optimize for the leaf function test.
976 2017-02-28 Martin Liska <mliska@suse.cz>
979 * read-rtl-function.c (function_reader::handle_unknown_directive):
980 Bail out when one uses -flto.
982 2017-02-28 Martin Liska <mliska@suse.cz>
984 * common.opt: Replace space with tabular for options of <number>
986 * config/i386/i386.opt: Show <number> value for
987 -mlarge-data-threshold.
988 * opts.c (print_filtered_help): Do not display number in hexadecimal
991 2017-02-28 Martin Liska <mliska@suse.cz>
993 * common.opt: Fix --help=option -Q for options which are of
996 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
998 * config/i386/i386.c (print_reg): Error out for values
999 of 8-bit size in invalid integer register.
1001 2017-02-28 Martin Sebor <msebor@redhat.com>
1003 PR tree-optimization/79691
1004 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
1006 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1009 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
1010 gcc_unreachable with output_operand_lossage.
1012 2017-02-28 Richard Biener <rguenther@suse.de>
1014 PR tree-optimization/79740
1015 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
1017 (visit_nary_op): Insert the nary into the hashtable if we
1018 pattern-matched sth.
1019 * tree-ssa-pre.c (eliminate_insert): Robustify.
1021 2017-02-28 Richard Biener <rguenther@suse.de>
1024 * fold-const.c (decode_field_reference): Reject out-of-bound
1027 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1029 * config/i386/i386.c: Include intl.h.
1030 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
1031 instead of just cond ? "..." : "...".
1032 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
1033 * coverage.c (read_counts_file): Likewise.
1034 * omp-offload.c: Include intl.h.
1035 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
1036 of just cond ? "..." : "...".
1037 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
1038 of just cond ? "..." : "...".
1040 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
1043 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
1045 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
1046 'tune for' CPU name.
1047 * config/arm/arm-cpu-data.h: Regenerated.
1049 2017-02-28 Richard Biener <rguenther@suse.de>
1051 PR tree-optimization/79732
1052 * tree-inline.c (expand_call_inline): Do not shadow var.
1054 2017-02-28 Richard Biener <rguenther@suse.de>
1056 PR tree-optimization/79723
1057 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
1058 address-space properly.
1060 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
1062 * doc/optinfo.texi (Optimization groups): Fix option used for
1064 * doc/invoke.texi (-fopt-info): Document "omp".
1065 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
1066 (OPTGROUP_ALL): Add OPTGROUP_OMP.
1067 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
1068 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
1069 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
1071 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
1073 * dumpfile.c (optgroup_options): Instead of "openmp", associate
1074 OPTGROUP_OMP with "omp".
1076 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
1079 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
1080 for arithmetic shift of unsigned V2DI.
1082 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
1084 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
1085 arc/linux.h headers.
1086 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
1087 (LINK_SPEC): Likewise.
1088 (ARC_TLS_EXTRA_START_SPEC): Likewise.
1089 (EXTRA_SPECS): Likewise.
1090 (STARTFILE_SPEC): Likewise.
1091 (ENDFILE_SPEC): Likewise.
1092 (LIB_SPEC): Likewise.
1093 (TARGET_SDATA_DEFAULT): Likewise.
1094 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
1095 (MULTILIB_DEFAULTS): Likewise.
1096 (DWARF2_UNWIND_INFO): Likewise.
1097 * config/arc/big.h: New file.
1098 * config/arc/elf.h: Likewise.
1099 * config/arc/linux.h: Likewise.
1100 * config/arc/t-uClibc: Remove.
1102 2017-02-27 Bin Cheng <bin.cheng@arm.com>
1104 PR tree-optimization/77536
1105 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
1106 (tree_transform_and_unroll_loop): Use above function to compute the
1107 estimated niter of unrolled loop and use it when scaling profile.
1108 Also use count info rather than frequency if it's non-zero.
1109 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
1110 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
1111 (vect_transform_loop): Call above function.
1113 2017-02-27 Richard Biener <rguenther@suse.de>
1115 PR tree-optimization/45397
1116 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
1117 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
1118 (visit_nary_op): Add pattern matching for CSEing sign-changed
1119 or truncated operations with wider ones.
1121 2017-02-27 Richard Biener <rguenther@suse.de>
1123 PR tree-optimization/79690
1124 * tree-vect-stmts.c (vectorizable_store): Use vector type
1125 built from the DR with address-space.
1127 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
1129 * doc/invoke.texi (Optimize Options): Refine the description
1130 of asan-use-after-return.
1132 2017-02-25 Alan Modra <amodra@gmail.com>
1134 PR rtl-optimization/79584
1135 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
1136 base, not ad->base_term, the reg within base. Remove assertion
1137 that ad->base == ad->base_term. Replace gen_int_mode using
1138 bogus mode with const0_rtx.
1140 2017-02-25 Jakub Jelinek <jakub@redhat.com>
1143 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
1144 FMA_EXPR like tcc_binary or tcc_unary.
1146 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
1149 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
1151 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
1152 (output_loc_operands): Handle DW_OP_call_ref and
1153 DW_OP_GNU_variable_value.
1154 (struct variable_value_struct): New type.
1155 (struct variable_value_hasher): Likewise.
1156 (variable_value_hash): New variable.
1157 (string_types): Remove.
1158 (copy_loc_descr): New function.
1159 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
1160 (prepend_loc_descr_to_each): New function.
1161 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
1162 instead of add_loc_descr_to_each if the first argument is single
1163 location list and the second has multiple.
1164 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
1165 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
1166 when looking for variable value which doesn't have other location info.
1167 (loc_list_from_tree): Formatting fix.
1168 (gen_array_type_die): Simplify DW_AT_string_length handling.
1169 (adjust_string_types): Remove.
1170 (gen_subprogram_die): Don't call adjust_string_types nor test/set
1171 string_types. Call resolve_variable_values.
1172 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
1173 (resolve_addr_in_expr): Likewise. Add A argument.
1174 (copy_deref_exprloc): Remove deref argument. Adjust for the
1175 original expression being DW_OP_GNU_variable_value with optionally
1176 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
1177 optionally after it.
1178 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
1179 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
1180 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
1181 (variable_value_hasher::hash, variable_value_hasher::equal): New
1183 (resolve_variable_value_in_expr, resolve_variable_value,
1184 resolve_variable_values, note_variable_value_in_expr,
1185 note_variable_value): New functions.
1186 (dwarf2out_early_finish): Call note_variable_value on all toplevel
1189 2017-02-24 Jakub Jelinek <jakub@redhat.com>
1192 * opts.h (handle_generated_option): Add GENERATED_P argument.
1193 * opts-common.c (handle_option): Adjust function comment.
1194 (handle_generated_option): Add GENERATED_P argument, pass it to
1196 (control_warning_option): Pass false to handle_generated_option
1198 * opts.c (maybe_default_option): Pass true to handle_generated_option
1200 * optc-gen.awk: Likewise.
1202 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1204 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
1205 a REG, look at the REG it is a SUBREG of.
1206 (splitter for cmpeqsi_t): Ditto.
1208 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1210 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
1211 the special USEs with the pattern of the insn, not the insn itself.
1213 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
1216 * doc/invoke.texi: Document -mload-store-pairs.
1218 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1219 Sandra Loosemore <sandra@codesourcery.com>
1221 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
1222 argument isn't a CONST_INT.
1223 (nios2_alternate_compare_const): Assert op is a CONST_INT.
1224 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
1225 (nios2_validate_compare): Bypass alternate compare logic if *op2
1227 (ldstwm_operation_p): Return false if first_base is not a REG or
1228 if first_offset is not a CONST_INT.
1230 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1232 * config/cris/cris.md: Use correct operand in a define_peephole2.
1234 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1236 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
1238 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1240 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
1241 this_insn if it is an INSN or JUMP_INSN.
1242 (force_offsettable): Look at base, not at addr.
1243 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
1244 on things that aren't necessarily CONST_INTs.
1246 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
1248 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
1249 -mfpmath=sse is the default also for x86-32 targets with SSE2
1250 instruction set when @option{-ffast-math} is enabled
1252 2017-02-24 Jeff Law <law@redhat.com>
1254 PR rtl-optimizatoin/79286
1255 * ira.c (update_equiv_regs): Drop may_trap_p exception to
1258 2017-02-24 Richard Biener <rguenther@suse.de>
1260 PR tree-optimization/79389
1261 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
1264 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
1266 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
1267 function comment to reflect reality.
1268 (loop_exits_before_overflow): Fix typo in function description.
1270 2017-02-24 Richard Biener <rguenther@suse.de>
1272 PR tree-optimization/79389
1273 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
1274 properly that a threading opportunity exists. Detect conditional
1275 copy/constant propagation opportunities.
1277 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
1279 * config/visium/visium.md (type): Add trap.
1280 (b): New mode attribute.
1281 (*btst): Rename into...
1282 (*btst<mode>): ...this and adjust.
1283 (*cbranchsi4_btst_insn): Rename into...
1284 (*cbranch<mode>4_btst_insn): ...this and adjust.
1285 (trap): New define_insn.
1287 2017-02-23 Jakub Jelinek <jakub@redhat.com>
1289 PR tree-optimization/79389
1290 * ifcvt.c (struct noce_if_info): Add rev_cond field.
1291 (noce_reversed_cond_code): New function.
1292 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
1293 reversed_comparison_code. Formatting fix.
1294 (noce_try_store_flag): Test rev_cond != NULL in addition to
1295 reversed_comparison_code.
1296 (noce_try_store_flag_constants): Likewise.
1297 (noce_try_store_flag_mask): Likewise.
1298 (noce_try_addcc): Use rev_cond if non-NULL instead of
1299 reversed_comparison_code.
1300 (noce_try_cmove_arith): Likewise. Formatting fixes.
1301 (noce_try_minmax, noce_try_abs): Clear rev_cond.
1302 (noce_find_if_block): Initialize rev_cond.
1303 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
1304 instead of false as last argument never attempt to reverse it
1307 2017-02-23 Bin Cheng <bin.cheng@arm.com>
1309 PR tree-optimization/79663
1310 * tree-predcom.c (combine_chains): Process refs in reverse order
1311 only for ZERO length chains, and add explaining comment.
1313 2017-02-23 Jeff Law <law@redhat.com>
1315 PR tree-optimization/79578
1316 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
1317 in call to operand_equal_p.
1319 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
1322 * config/i386/cpuid.h: Fix another undefined behavior.
1324 2017-02-23 Richard Biener <rguenther@suse.de>
1326 PR tree-optimization/79683
1327 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
1328 vector types for data-refs.
1330 2017-02-23 Martin Liska <mliska@suse.cz>
1332 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
1334 2017-02-23 Jakub Jelinek <jakub@redhat.com>
1337 * internal-fn.c (get_range_pos_neg): Moved to ...
1338 * tree.c (get_range_pos_neg): ... here. No longer static.
1339 * tree.h (get_range_pos_neg): New prototype.
1340 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
1341 are known to be in between 0 and signed maximum inclusive, try to
1342 expand both unsigned and signed divmod and use the cheaper one from
1345 2017-02-22 Jeff Law <law@redhat.com>
1347 PR tree-optimization/79578
1348 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
1349 to compare base operands.
1351 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
1354 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
1355 gpc_reg_operand instead of fpr_reg_operand.
1357 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
1359 * config/mips/mips.c (mips_return_in_memory): Force FP
1360 vector types to be returned in memory for o32 ABI.
1362 2017-02-22 Jakub Jelinek <jakub@redhat.com>
1364 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
1365 instead of DW_TAG_member for static data member declarations and don't
1366 set no_linkage_name for static inline data members.
1367 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
1370 2017-02-22 Martin Liska <mliska@suse.cz>
1372 * doc/invoke.texi: Replace inequality signs with square brackets
1375 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1377 PR tree-optimization/68644
1378 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
1380 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
1383 * lra-constraints.c (simplify_operand_subreg): Handle
1384 WORD_REGISTER_OPERATIONS targets.
1386 2017-02-22 Jakub Jelinek <jakub@redhat.com>
1389 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
1390 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
1391 elimination by swapping fld*.
1393 2017-02-22 Richard Biener <rguenther@suse.de>
1395 PR tree-optimization/79673
1396 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
1397 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
1398 irrelevant address-space qualifiers and avoiding a
1399 ADDR_SPACE_CONVERT_EXPR from fold_convert.
1401 2017-02-22 Richard Biener <rguenther@suse.de>
1403 PR tree-optimization/79666
1404 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
1405 to not symbolically negate if that may introduce undefined
1408 2017-02-22 Martin Liska <mliska@suse.cz>
1411 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
1412 * data-streamer-out.c (streamer_write_gcov_count_stream):
1414 * value-prof.c (stream_out_histogram_value): Make assert more
1415 precise based on type of counter.
1417 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
1420 * config/i386/i386.md (standard_x87sse_constant_load splitter):
1421 Use nonimmediate_operand instead of memory_operand for operand 1.
1422 (float-extend standard_x87sse_constant_load splitter): Ditto.
1424 2017-02-21 Jeff Law <law@redhat.com>
1426 PR tree-optimization/79621
1427 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
1428 blocks with edges to themselves.
1430 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1433 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
1434 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
1435 Use gimple_call_builtin_p.
1438 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
1439 on temporarily removed DEBUG_INSNs.
1441 PR tree-optimization/79649
1442 * tree-loop-distribution.c (classify_partition): Give up on
1443 non-generic address space loads/stores.
1445 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
1447 * doc/loop.texi (Loop manipulation): Remove nonexistent
1448 tree_ssa_loop_version from the documentation.
1449 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
1451 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1454 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
1455 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
1456 * config/rs6000/rs6000.c: Include except.h.
1457 (rs6000_expand_split_stack_prologue): Call
1458 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
1460 2017-02-21 Martin Jambor <mjambor@suse.cz>
1463 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
1466 2017-02-21 Martin Jambor <mjambor@suse.cz>
1468 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
1469 for backward compatibility only.
1470 * doc/invoke.texi (Option Summary): Remove all references to
1473 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
1477 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1479 * lra-constraints.c (curr_insn_transform): Handle
1480 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
1482 2017-02-21 Martin Liska <mliska@suse.cz>
1484 * config/i386/i386.opt: Replace -masm-dialect with -masm.
1486 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
1488 PR translation/79638
1489 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
1491 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
1494 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
1495 (arm_function_ok_for_sibcall): Return false for an indirect call by
1496 descriptor if all the argument registers are used.
1497 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
1498 alignment of the function.
1500 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1502 PR tree-optimization/61441
1503 * simplify-rtx.c (simplify_const_unary_operation): For
1504 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
1505 the sNaN unmodified.
1507 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
1509 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
1510 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
1511 instead of SYSTEM_HEADER_DIR.
1513 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
1514 Martin Liška <mliska@suse.cz>
1516 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
1517 Fix typos and grammar, use active voice, and clarify.
1519 2017-02-20 Marek Polacek <polacek@redhat.com>
1522 * gimplify.c (gimplify_expr): Handle unused *&&L;.
1525 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
1527 2017-02-20 Jakub Jelinek <jakub@redhat.com>
1530 * config/i386/i386.c (ix86_expand_builtin): Handle
1531 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
1532 ix86_builtins_isa[fcode].isa as a requirement of those
1533 flags and any other flag in the bitmask.
1534 (ix86_init_mmx_sse_builtins): Use 0 instead of
1535 ~OPTION_MASK_ISA_64BIT as mask.
1536 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
1537 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
1538 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
1539 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
1541 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1544 * lra-constraints.c (split_reg): Check requested split mode
1545 is supported by the register.
1547 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1549 * lra-constraints.c (simplify_operand_subreg): Remove early
1552 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1555 * lra-constraints.c (curr_insn_transform): Tighten condition
1556 for converting SUBREG reloads from OP_OUT to OP_INOUT.
1558 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1561 * lra-constraints.c (curr_insn_transform): Handle
1562 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
1564 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
1567 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
1569 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
1571 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
1574 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
1577 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1579 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
1582 2017-02-18 Jakub Jelinek <jakub@redhat.com>
1584 * final.c (last_columnnum, override_columnnum): New variables.
1585 (final_start_function): Set last_columnnum, pass it to begin_prologue
1586 hook and pass 0 to dwarf2out_begin_prologue.
1587 (final_scan_insn): Update override_columnnum. Pass last_columnnum
1588 to source_line debug hook.
1589 (notice_source_line): Compute last_columnnum and for debug_column_info
1590 return true on column changes.
1591 * debug.h (struct gcc_debug_hooks): Add column argument to
1592 source_line and begin_prologue hooks.
1593 (debug_nothing_int_charstar_int_bool): Remove prototype.
1594 (debug_nothing_int_int_charstar,
1595 debug_nothing_int_int_charstar_int_bool): New prototypes.
1596 (dwarf2out_begin_prologue): Add column argument.
1597 * debug.c (do_nothing_debug_hooks): Adjust source_line and
1598 begin_prologue hooks.
1599 (debug_nothing_int_charstar_int_bool): Remove.
1600 (debug_nothing_int_int_charstar,
1601 debug_nothing_int_int_charstar_int_bool): New functions.
1602 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
1603 through to dwarf2out_source_line.
1604 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
1605 (dwarf2out_source_line): Add column argument, emit it if requested.
1606 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
1608 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
1609 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
1610 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
1611 through to dwarf2out_begin_prologue.
1612 (vmsdbgout_source_line): Add column argument, pass it through to
1613 dwarf2out_source_line.
1614 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
1615 dbxout_source_line caller.
1616 (dbxout_source_line): Add column argument.
1618 * common.opt (gno-column-info, gcolumn-info): New options.
1619 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
1620 (check_die): Also test for multiple DW_AT_decl_column attributes.
1621 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
1622 DW_AT_decl_column if requested.
1623 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
1625 (gen_variable_die): Likewise.
1626 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
1627 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
1630 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
1631 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
1632 (ix86_handle_option): Handle OPT_m3dnowa.
1633 * doc/invoke.texi (-m3dnowa): Document.
1634 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
1635 -m3dnowa instead of -m3dnow -march=athlon.
1638 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
1639 instead of gcc_assert for K, r and R code checks. Formatting fixes.
1641 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1644 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
1645 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
1646 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
1647 generator for vsx_xxpermdi_<mode>_be.
1648 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
1649 force big-endian semantics.
1650 (vsx_xxpermdi_<mode>_be): New define_expand with same
1651 implementation as previous version of vsx_xxpermdi_<mode>.
1653 2017-02-17 Jakub Jelinek <jakub@redhat.com>
1655 PR tree-optimization/79327
1656 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
1657 variable, its initialization and use.
1659 2017-02-17 Julia Koval <julia.koval@intel.com>
1661 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
1662 (OPTION_MASK_ISA_PKU_UNSET): New.
1663 (ix86_handle_option): Handle -mrdpid.
1664 * config/i386/cpuid.h (bit_RDPID): New.
1665 * config/i386/driver-i386.c (host_detect_local_cpu):
1666 Detect RDPID feature.
1667 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
1668 * config/i386/i386-c.c (ix86_target_macros_internal):
1670 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
1671 (ix86_valid_target_attribute_inner_p): Add "rdpid".
1672 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
1673 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
1674 * config/i386/i386.md (define_insn "rdpid"): New.
1675 * config/i386/i386.opt Add -mrdpid.
1676 * config/i386/immintrin.h (_rdpid_u32): New.
1678 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
1680 PR rtl-optimization/79541
1681 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
1682 instead of transforming it into USE.
1684 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
1686 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
1687 If HONOR_SNANS (SFmode) force the input to a register.
1688 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
1689 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
1690 an frsp or similar insn.
1692 2017-02-17 Martin Liska <mliska@suse.cz>
1694 PR rtl-optimization/79577
1695 * params.def (selsched-max-sched-times): Increase minimum to 1.
1697 2017-02-17 Martin Liska <mliska@suse.cz>
1699 PR rtl-optimization/79574
1700 * gcse.c (want_to_gcse_p): Prevent integer overflow.
1702 2017-02-17 Martin Liska <mliska@suse.cz>
1704 PR tree-optimization/79529
1705 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
1706 ssa_defined_default_def_p to handle cases which are implicitly
1708 * tree-ssa.c (ssa_defined_default_def_p): New function.
1709 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
1710 which are implicitly defined.
1711 * tree-ssa.h (ssa_defined_default_def_p): Declare.
1713 2017-02-17 Richard Biener <rguenther@suse.de>
1716 * params.def (max-ssa-name-query-depth): Limit to 10.
1718 2017-02-17 Richard Biener <rguenther@suse.de>
1720 PR tree-optimization/79552
1721 * tree-ssa-structalias.c (visit_loadstore): Properly verify
1724 2017-02-17 Richard Biener <rguenther@suse.de>
1727 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
1729 2017-02-17 Marek Polacek <polacek@redhat.com>
1732 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
1733 (fold_negate_expr): New wrapper.
1735 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
1737 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
1738 Correct terminology and de-emphasize pre-standard behavior.
1740 2017-02-16 Alan Modra <amodra@gmail.com>
1742 PR rtl-optimization/79286
1743 * ira.c (def_dominates_uses): New function.
1744 (update_equiv_regs): Don't create an equivalence for insns that
1745 may trap where the register def does not dominate the use.
1747 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
1749 PR rtl-optimization/78127
1750 * lra.c (lra): Call lra_eliminate before finish the loop after
1753 2017-02-16 Richard Biener <rguenther@suse.de>
1755 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
1757 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
1758 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
1759 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
1760 (isl_val_int_from_wi): New function.
1761 (extract_affine_gmp): Rename to ...
1762 (extract_affine_wi): ... this, take a widest_int.
1763 (extract_affine_int): Just wrap extract_affine_wi.
1764 (add_param_constraints): Use isl_val_int_from_wi.
1765 (add_loop_constraints): Likewise, and extract_affine_wi.
1767 2017-02-15 Jeff Law <law@redhat.com>
1770 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
1771 ira_init_register_move_cost_if_necessary.
1773 2017-02-15 Martin Sebor <msebor@redhat.com>
1776 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
1777 removed in a prior commit.
1779 2017-02-15 Bin Cheng <bin.cheng@arm.com>
1781 PR tree-optimization/79347
1782 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
1783 counters during peeling.
1785 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
1787 * Makefile.in (site.exp): Remove "set ISLVER".
1789 2017-02-15 Jakub Jelinek <jakub@redhat.com>
1792 * real.c (real_from_integer): Call real_convert even for decimal.
1794 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
1797 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
1799 2017-02-14 Andrew Pinski <apinski@cavium.com>
1801 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
1802 cores and change the partno/implementer to be correct.
1803 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
1804 the 'B" as the implementer.
1805 * config/aarch64/aarch64-tune.md: Regenerate.
1807 2017-02-14 Carl Love <cel@us.ibm.com>
1809 * config/rs6000/rs6000.c: Add case statement entry to make the
1810 xvcvuxdsp built-in argument unsigned.
1811 * config/rs6000/vsx.md: Fix the source and return operand types so they
1812 match the instruction definitions from the ISA document. Fix typo
1813 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
1816 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
1819 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
1820 member early_clobber_alts.
1821 * lra-lives.c (reg_early_clobber_p): New.
1822 (process_bb_lives): Use it.
1823 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
1824 (debug_operand_data): Initialize early_clobber_alts.
1825 (setup_operand_alternative): Set up early_clobber_alts.
1826 (collect_non_operand_hard_regs): Ditto. Pass early clobber
1827 alternatives to new_insn_reg.
1828 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
1830 (lra_update_insn_regno_info): Pass the new arg.
1832 2017-02-14 Jakub Jelinek <jakub@redhat.com>
1835 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
1836 (new_oacc_loop_raw): Don't clear already cleared fields.
1839 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
1840 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
1841 _mm512_prefetch_i64gather_ps): New inline functions and macros.
1843 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
1846 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
1848 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
1851 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
1852 the extra instruction to the right place to store 128-bit constant
1855 2017-02-14 Martin Sebor <msebor@redhat.com>
1858 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
1859 warning for strings of unknown length.
1861 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
1863 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
1865 2017-02-14 Jeff Law <law@redhat.com>
1868 * ira-costs.c (scan_one_insn): Initialize register move costs
1869 for pseudos seen in USE/CLOBBER insns.
1871 PR tree-optimization/79095
1872 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
1873 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
1874 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
1875 if the operands are known to be not equal, then the resulting range
1877 (intersect_ranges): If the new range is ~[0,0] and the old range is
1878 wide, then prefer ~[0,0].
1879 * tree-vrp.c (overflow_comparison_p_1): New function.
1880 (overflow_comparison_p): New function.
1881 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
1882 if NAME is used in an overflow test.
1883 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
1884 overflow check that can be expressed as an equality test, then adjust
1885 ops to be that equality test.
1887 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1889 * config/s390/s390-builtin-types.def: Remove flags argument.
1890 * config/s390/s390.c (s390_init_builtins): Likewise.
1892 2017-02-14 Martin Liska <mliska@suse.cz>
1894 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
1895 vector. Fix trailing white spaces.
1897 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
1899 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
1902 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1904 PR rtl-optimization/68664
1905 * config/arm/arm.c (arm_sched_can_speculate_insn):
1906 New function. Declare prototype.
1907 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
1909 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1911 PR rtl-optimization/68664
1912 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
1914 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
1916 2017-02-14 Amit Pawar <amit.pawar@amd.com>
1918 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
1919 max skip bytes for function, loop and jump.
1921 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1923 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
1924 ABS_EXPR for gimple dump.
1926 2017-02-14 Jakub Jelinek <jakub@redhat.com>
1929 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
1931 PR tree-optimization/79408
1932 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
1933 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
1934 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
1935 also if rhs1 is INTEGER_CST.
1937 2017-02-14 Richard Biener <rguenther@suse.de>
1940 * tree-into-ssa.c (insert_phi_nodes): When the function can
1941 have abnormal edges rewrite SSA names with broken use-def
1942 dominance out of SSA and register them for PHI insertion.
1944 2017-02-13 Martin Sebor <msebor@redhat.com>
1947 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
1948 clearing info.nowrite flag when snprintf size argument is a range.
1950 2017-02-13 Jakub Jelinek <jakub@redhat.com>
1952 * cprop.c (cprop_jump): Add missing space in string literal.
1953 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
1954 (get_constraint_for_component_ref): Likewise.
1955 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
1956 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
1957 * lra-constraints.c (process_alt_operands): Likewise.
1958 * ipa-inline.c (inline_small_functions): Likewise.
1959 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
1960 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
1961 * trans-mem.c (diagnose_tm_1_op): Likewise.
1962 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
1963 (grid_parallel_clauses_gridifiable): Likewise.
1965 * config/nvptx/mkoffload.c (process): Add space in between
1968 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
1969 "MOD4_SSE_REGS" and "ALL_REGS".
1971 * spellcheck.c (test_data): Add , in between "foo" and "food".
1973 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1976 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
1977 boundary crossing check and subsequent code generation agree.
1979 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1981 * config/aarch64/aarch64.c (has_memory_op): Delete.
1982 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
1985 2017-02-13 Jakub Jelinek <jakub@redhat.com>
1987 PR rtl-optimization/79388
1988 PR rtl-optimization/79450
1989 * combine.c (distribute_notes): When removing TEM_INSN for which
1990 corresponding dest has last value recorded, invalidate that last
1993 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1995 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
1996 of explicit '@'. Add missing assembly comment marker on branch costs
1999 2017-02-13 Nathan Sidwell <nathan@acm.org>
2001 * gengtype-lex.l (<in_struct>): Add '/'.
2003 2017-02-13 Martin Liska <mliska@suse.cz>
2006 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
2008 2017-02-13 Richard Biener <rguenther@suse.de>
2010 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
2012 * configure: Re-generate.
2013 * config.in: Likewise.
2014 * graphite-dependences.c: Simplify as if
2015 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
2016 * graphite-isl-ast-to-gimple.c: Likewise.
2017 * graphite-optimize-isl.c: Likewise.
2018 * graphite-poly.c: Likewise.
2019 * graphite-sese-to-poly.c: Likewise.
2020 * graphite.h: Likewise.
2021 * toplev.c: Include isl/version.h and use isl_version () for
2022 printing the ISL version.
2023 * doc/install.texi: Update ISL requirement.
2025 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
2027 * doc/standards.texi (Standards): Update reference to
2030 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
2032 * doc/extend.texi (Named Address Spaces): sourceware.org now
2034 * doc/install.texi (Binaries): Ditto.
2037 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
2039 * doc/cpp.texi: Replace "stringify"/"stringification" with C
2040 standard terminology "stringize"/"stringizing" throughout.
2041 * doc/cppinternals.texi: Likewise.
2043 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
2045 * doc/extend.texi: Fix some spelling mistakes and typos.
2046 * doc/invoke.texi: Likewise.
2048 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
2051 * params.def (inline-min-speedup) Change from 10 to 8.
2053 2017-02-11 Jakub Jelinek <jakub@redhat.com>
2055 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
2058 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
2061 * ipa-inline-analysis.c (get_minimal_bb): New function.
2062 (record_modified): Use it.
2063 (remap_edge_change_prob): Handle also ancestor functions.
2065 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
2067 * doc/contrib.texi (Contributors): Remove broken link into
2068 the Mauve CVS repository.
2070 2017-02-11 Jakub Jelinek <jakub@redhat.com>
2073 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
2074 result computation whenever lhs doesn't have vector mode, not
2075 just when it has BLKmode.
2077 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
2079 * doc/makefile.texi (profiledbootstrap): Refer to the
2080 installation instructions only in textual form.
2082 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2085 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
2087 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
2089 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
2090 (Specific): Update mingw-w64 reference.
2092 (Specific): Remove broken link to Renesas RX processor.
2094 2017-02-10 Richard Biener <rguenther@suse.de>
2096 * toplev.c (process_options): Do not mention obsolete graphite
2097 options when printing sorry message about missing graphite support.
2098 Mention -floop-nest-optimize.
2100 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
2102 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
2103 (vtst_p16): Likewise.
2104 (vtstq_p8): Likewise.
2105 (vtstq_p16): Likewise.
2107 (vtstq_p64): Likewise.
2108 * config/arm/arm_neon.h (vgetq_lane_p64): New.
2109 (vset_lane_p64): New.
2110 (vsetq_lane_p64): New.
2112 2017-02-10 Jakub Jelinek <jakub@redhat.com>
2114 PR tree-optimization/79411
2115 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
2116 stmt operands are SSA_NAMEs used in abnormal phis.
2117 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
2120 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
2123 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
2126 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
2128 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
2130 2017-02-09 Jakub Jelinek <jakub@redhat.com>
2132 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
2136 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
2137 not arbitrary TREE_CONSTANT.
2140 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
2141 "omp declare target link" attribute unless is_global_var.
2142 * omp-offload.c (find_link_var_op): Likewise.
2144 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
2145 Chung-Lin Tang <cltang@codesourcery.com>
2147 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
2149 (gimplify_adjust_omp_clauses): Don't delete TILE.
2150 (gimplify_omp_for): Deal with TILE.
2151 * internal-fn.c (expand_GOACC_TILE): New function.
2152 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
2154 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
2155 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
2157 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
2158 avoid DIV for outermost collapse var.
2159 (expand_oacc_for): Insert tile element loop as needed. Adjust.
2160 Remove out of date comments, fix whitespace.
2161 * omp-general.c (omp_extract_for_data): Deal with tiling.
2162 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
2163 adjust OLF_DIM_BASE value.
2164 (struct omp_for_data): Add tiling field.
2165 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
2166 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
2167 for auto loops. Remove default auto determining, moved to
2168 oacc_loop_fixed_partitions.
2169 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
2170 stmts, add e_mask field.
2171 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
2172 (oacc_thread_numbers): Use oacc_dim_call.
2173 (oacc_xform_tile): New.
2174 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
2175 (finish_oacc_loop): Adjust for ifns vector.
2176 (oacc_loop_discover_walk): Append loop abstraction sites to list,
2177 add case for GOACC_TILE fns.
2178 (oacc_loop_xform_loop): Delete.
2179 (oacc_loop_process): Iterate over call list directly, and add
2180 handling for GOACC_TILE fns.
2181 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
2183 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
2184 vector partitioning to outer loops. Assign 2 partitions to loops
2185 when available. Add TILE handling.
2186 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
2187 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
2188 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
2189 * tree.c (omp_clause_num_ops): Adjust TILE ops.
2190 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
2192 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
2194 * configure.ac (ACX_BUGURL): Update.
2195 * configure: Regenerate.
2197 2017-02-09 Richard Biener <rguenther@suse.de>
2199 PR tree-optimization/69823
2200 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
2201 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
2203 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
2205 * config/arc/arc-c.def: Add __NPS400__ definition.
2206 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
2207 (TARGET_NPS400): Define.
2209 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
2211 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
2213 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
2215 (arc_cpu_types): Fill the arch_info field with a pointer into the
2216 arc_arch_types table.
2217 (arc_selected_cpu): Declare.
2218 * config/arc/arc.c (arc_selected_cpu): Make global.
2219 (arc_selected_arch): Delete.
2220 (arc_base_cpu): Delete.
2221 (arc_override_options): Remove references to deleted variables,
2222 update access to arch information.
2223 (ARC_OPT): Update access to arch information.
2224 (ARC_OPTX): Likewise.
2225 * config/arc/arc.h (arc_base_cpu): Remove declaration.
2226 (TARGET_ARC600): Update access to arch information.
2227 (TARGET_ARC601): Likewise.
2228 (TARGET_ARC700): Likewise.
2229 (TARGET_EM): Likewise.
2230 (TARGET_HS): Likewise.
2231 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
2234 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
2237 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
2238 condition/operands for integer GE/LE/GEU/LEU operations.
2240 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
2242 PR translation/79397
2243 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
2246 2017-02-08 Martin Jambor <mjambor@suse.cz>
2249 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
2250 whether allocation happened.
2251 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
2252 nothing was allocated.
2254 2017-02-08 Jakub Jelinek <jakub@redhat.com>
2256 PR tree-optimization/79408
2257 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
2258 constant, but SSA_NAME with a known integer range, use the minimum
2259 of that range instead of op1 to determine if modulo can be replaced
2260 with its first operand.
2262 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2264 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
2266 2017-02-08 Richard Biener <rguenther@suse.de>
2268 PR tree-optimization/71824
2269 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
2270 Check all loops contained in the merged region.
2272 2017-02-07 Andrew Pinski <apinski@cavium.com>
2274 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
2276 2017-02-07 Andrew Pinski <apinski@cavium.com>
2278 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
2279 (thunderxt88): Likewise.
2280 (thunderxt81): Disable LSE and change v8.1 to v8.
2281 (thunderxt83): Likewise.
2283 2017-02-07 Jakub Jelinek <jakub@redhat.com>
2284 Richard Biener <rguenther@suse.de>
2287 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
2288 type from int to size_t.
2289 * ira-costs.c (struct_costs_size): Change type from int to size_t.
2291 2017-02-07 Jakub Jelinek <jakub@redhat.com>
2293 PR rtl-optimization/79386
2294 * cprop.c (bypass_conditional_jumps): Initialize
2295 bypass_last_basic_block already before splitting bbs after
2296 unconditional traps...
2297 (bypass_conditional_jumps): ... rather than here.
2300 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
2301 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
2302 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
2303 fix -masm=intel patterns.
2305 2017-02-07 Richard Biener <rguenther@suse.de>
2307 PR tree-optimization/79256
2309 * builtins.c (get_object_alignment_2): Use min_align_of_type
2310 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
2311 and ADJUST_FIELD_ALIGN.
2313 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
2315 * doc/tm.texi: Regenerate.
2316 * stor-layout.c (layout_decl): Adjust.
2317 (update_alignment_for_field): Likewise.
2318 (place_field): Likewise.
2319 (min_align_of_type): Likewise.
2320 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
2321 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
2322 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
2323 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
2324 * config/frv/frv.c (frv_adjust_field_align): Likewise.
2325 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
2326 * config/i386/i386.c (x86_field_alignment): Likewise.
2327 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
2328 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
2329 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
2330 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
2331 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
2332 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
2336 2017-01-30 Richard Biener <rguenther@suse.de>
2338 PR tree-optimization/79256
2339 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
2340 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
2343 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
2345 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
2346 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
2347 builtins to SImode and emit a zero-extend, if necessary.
2349 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2351 * docs/invoke.texi (RISC-V Options): Alphabetize.
2353 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2355 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
2358 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2360 * config/riscv/riscv.c: New file.
2361 * gcc/common/config/riscv/riscv-common.c: Likewise.
2362 * config.gcc: Likewise.
2363 * config/riscv/constraints.md: Likewise.
2364 * config/riscv/elf.h: Likewise.
2365 * config/riscv/generic.md: Likewise.
2366 * config/riscv/linux.h: Likewise.
2367 * config/riscv/multilib-generator: Likewise.
2368 * config/riscv/peephole.md: Likewise.
2369 * config/riscv/pic.md: Likewise.
2370 * config/riscv/predicates.md: Likewise.
2371 * config/riscv/riscv-builtins.c: Likewise.
2372 * config/riscv/riscv-c.c: Likewise.
2373 * config/riscv/riscv-ftypes.def: Likewise.
2374 * config/riscv/riscv-modes.def: Likewise.
2375 * config/riscv/riscv-opts.h: Likewise.
2376 * config/riscv/riscv-protos.h: Likewise.
2377 * config/riscv/riscv.h: Likewise.
2378 * config/riscv/riscv.md: Likewise.
2379 * config/riscv/riscv.opt: Likewise.
2380 * config/riscv/sync.md: Likewise.
2381 * config/riscv/t-elf-multilib: Likewise.
2382 * config/riscv/t-linux: Likewise.
2383 * config/riscv/t-linux-multilib: Likewise.
2384 * config/riscv/t-riscv: Likewise.
2385 * configure.ac: Likewise.
2386 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
2387 Waterman as RISC-V maintainers.
2388 * doc/install.texi: Add RISC-V entries.
2389 * doc/invoke.texi: Add RISC-V options section.
2390 * doc/md.texi: Add RISC-V constraints section.
2391 * configure: Regenerated.
2393 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2396 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
2397 false values to be constant vectors with all 0 or all 1 bits set.
2398 (vcondu<mode><mode>): Likewise.
2399 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
2401 (fpmask_comparison_operator): Update comment.
2402 (vecint_comparison_operator): New predicate.
2403 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
2404 vector conditionals when the true and false values are constant
2405 vectors with all 0 bits or all 1 bits set.
2407 2017-02-06 Martin Sebor <msebor@redhat.com>
2409 PR tree-optimization/79376
2410 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
2412 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
2414 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
2415 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
2416 to simplify split condition.
2418 2017-02-06 Jakub Jelinek <jakub@redhat.com>
2420 * omp-expand.c (oxpand_omp_atomic_fetch_op,
2421 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
2424 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
2426 PR rtl-optimization/68664
2427 * target.def (can_speculate_insn): New hook.
2428 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
2429 * doc/tm.texi: Regenerate.
2430 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
2431 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
2432 (rs6000_sched_can_speculate_insn): New function.
2434 2017-02-06 Jakub Jelinek <jakub@redhat.com>
2436 PR tree-optimization/79284
2437 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
2438 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
2439 vectorizable_mask_load_store, vectorizable_operation,
2440 vect_is_simple_cond, get_same_sized_vectype): Use it instead
2441 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
2442 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
2443 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
2444 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
2445 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
2446 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
2447 is_gimple_assign (stmt). Replace another such test with
2448 is_gimple_assign (stmt).
2450 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
2453 * config/avr/avr.c (rtl-iter.h): Include it.
2454 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
2455 (avr_legitimate_combined_insn): ...and implementation.
2457 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
2459 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
2460 * config/s390/s390.c (s390_const_operand_ok)
2461 (s390_canonicalize_comparison, s390_extract_part)
2462 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
2463 (s390_contiguous_bitmask_p, s390_rtx_costs)
2464 (legitimize_pic_address): Likewise.
2465 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
2466 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
2467 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
2468 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
2469 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
2471 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
2473 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
2474 REGNO($0) == REGNO($1).
2476 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2478 * config/s390/linux.h(SIZE_TYPE): Add comment.
2480 2017-02-06 Julian Brown <julian@codesourcery.com>
2481 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
2482 Virendra Pathak <virendra.pathak@broadcom.com>
2484 * config/aarch64/aarch64-cores.def: Change the scheduler
2486 * config/aarch64/aarch64.md: Include thunderx2t99.md.
2487 * config/aarch64/thunderx2t99.md: New file.
2489 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2491 * doc/standards.texi (Go Language): Update link to language
2494 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
2496 * tree-eh.c (lower_resx): Sanitize profile.
2497 (cleanup_empty_eh_move_lp): Likewise.
2499 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
2502 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
2504 * cfgloopmanip.h (loop_version): Update prototype.
2505 * modulo-sched.c (sms_schedule): Update call of loop_version.
2506 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
2507 * tree-parloops.c (gen_parallel_loop): Likewise.
2508 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
2509 * tree-ssa-loop-split.c (split_loop): Likewise.
2510 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
2511 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
2513 2017-02-05 Martin Liska <mliska@suse.cz>
2516 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
2518 (print_operand_address): Initialize a struct to zero.
2520 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2522 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
2523 garbage collector only in textual form.
2525 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2527 * doc/extend.texi (x86 specific memory model extensions for
2528 transactional memory): Simplify a phrase.
2530 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
2533 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
2534 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
2535 (atomic_storedi_1): Likewise.
2537 2017-02-04 Jakub Jelinek <jakub@redhat.com>
2539 PR tree-optimization/79338
2540 * tree-parloops.c (gather_scalar_reductions): Don't call
2541 vect_analyze_loop_form for loop->inner before destroying loop's
2544 2017-02-03 Martin Sebor <msebor@redhat.com>
2546 PR tree-optimization/79327
2547 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
2548 when precision has resulted in leading zeros.
2549 (format_integer): Adjust the likely counter to assume an unknown
2550 argument that may be zero is non-zero.
2552 2017-02-03 Jason Merrill <jason@redhat.com>
2555 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
2556 avoid copying non-taken branch.
2558 2017-02-03 Jakub Jelinek <jakub@redhat.com>
2560 PR tree-optimization/79340
2561 * tree-vect-loop.c (vectorizable_reduction): Release
2562 vec_defs elements after safe_splicing them into other vectors.
2565 PR tree-optimization/79327
2566 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
2567 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
2569 (format_integer): Use wide_int_to_tree instead of build_int_cst
2570 + to_?hwi. If argmin is NULL, just set argmin and argmax to
2571 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
2572 of shortest and longest sequence.
2574 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
2576 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
2577 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
2579 2017-02-03 Walter Lee <walt@tilera.com>
2582 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
2583 after initial stackframe link reg save.
2584 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
2586 2017-02-03 Jakub Jelinek <jakub@redhat.com>
2589 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
2590 wu for stxssp alternative.
2592 2017-02-03 Martin Sebor <msebor@redhat.com>
2594 PR tree-optimization/79352
2595 * gimple-fold.c (get_range_strlen): Add argument.
2596 (get_range_strlen): Change return type to bool.
2597 (get_maxval_strlen): Pass in a dummy argument.
2598 * gimple-fold.h (get_range_strlen): Change return type to bool.
2599 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
2600 * tree.h (array_at_struct_end_p): Add argument.
2601 * tree.c (array_at_struct_end_p): Handle it.
2603 2017-02-03 Martin Liska <mliska@suse.cz>
2606 * multiple_target.c (create_dispatcher_calls): Redirect edge
2607 from a caller of a dispatcher.
2608 (expand_target_clones): Make the clones local.
2609 (ipa_target_clone): Do both target clones and resolvers.
2610 (ipa_dispatcher_calls): Remove the pass.
2611 (pass_dispatcher_calls::gate): Likewise.
2612 (make_pass_dispatcher_calls): Likewise.
2613 * passes.def (pass_target_clone): Put as very first IPA early
2616 2017-02-03 Martin Liska <mliska@suse.cz>
2618 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
2619 in case of a function with ifunc attribute.
2621 2017-02-03 Martin Liska <mliska@suse.cz>
2623 * cgraph.c (cgraph_node::dump): Dump function version info.
2624 * symtab.c (symtab_node::dump_base): Add missing new line.
2626 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
2628 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
2629 (ifcombine_ifandif): Use it.
2631 2017-02-03 Martin Liska <mliska@suse.cz>
2633 * doc/invoke.texi: Document default value for
2634 use-after-scope-direct-emission-threshold.
2636 2017-02-03 Martin Liska <mliska@suse.cz>
2638 PR tree-optimization/79339
2639 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
2640 (format_floating): Likewise.
2642 2017-02-03 Martin Liska <mliska@suse.cz>
2645 * ipa-prop.c (ipa_node_params_t::insert): Remove current
2647 (ipa_node_params_t::remove): Likewise.
2648 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
2649 initialization from removed ipa_node_params_t::insert.
2650 (ipa_node_params::~ipa_node_params): Move from removed
2651 ipa_node_params_t::release.
2652 * symbol-summary.h (symbol_summary::m_released): New member.
2653 Do not release a summary twice. Do not allow to call finalizer
2654 for types of a summary that live in GGC memory.
2656 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
2658 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
2661 2017-02-02 Martin Sebor <msebor@redhat.com>
2664 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
2665 (format_string): Tighten up the range of output for non-constant
2666 strings and correct the expected range for wide non-constant strings.
2668 2017-02-02 Martin Sebor <msebor@redhat.com>
2670 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
2673 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
2675 (-fdump-tree-@var): Add to index and document how to come up
2676 with pass-specific option and dump file names.
2677 (-fdump-passes): Clarify where to look for output.
2679 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
2682 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
2683 statistics of the analyzed path; allow threading for speed when
2684 any of BBs along the path are optimized for speed.
2686 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
2689 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
2690 settings of the virtual registers.
2693 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
2695 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2696 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
2699 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2701 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
2702 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
2704 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2706 * config/s390/s390.md: Add missing comments with the expanded
2708 * config/s390/vector.md: Likewise.
2709 * config/s390/vx-builtins.md: Likewise.
2711 2017-02-02 Jakub Jelinek <jakub@redhat.com>
2714 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
2715 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
2716 conditions on a single line.
2718 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2720 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
2721 __S390_VX__ to __VX__.
2723 2017-02-01 Andrew Pinski <apinski@cavium.com>
2725 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
2726 stmt_info to record_stmt_cost.
2727 (vect_get_known_peeling_cost): Pass stmt_info if known to
2729 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
2730 cpu_vector_cost field into
2731 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
2732 field into vec_int_stmt_cost and vec_fp_stmt_cost.
2733 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
2734 splitting of scalar_stmt_cost and vec_stmt_cost.
2735 (thunderx_vector_cost): Likewise.
2736 (cortexa57_vector_cost): LIkewise.
2737 (exynosm1_vector_cost): Likewise.
2738 (xgene1_vector_cost): Likewise.
2739 (thunderx2t99_vector_cost): Improve after the splitting of the two
2741 (aarch64_builtin_vectorization_cost): Update for the splitting of
2742 scalar_stmt_cost and vec_stmt_cost.
2744 2017-02-01 Torvald Riegel <triegel@redhat.com>
2745 Richard Henderson <rth@redhat.com>
2747 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
2748 conditional on existance of a fast atomic load.
2749 * optabs-query.c (can_atomic_load_p): New function.
2750 * optabs-query.h (can_atomic_load_p): Declare it.
2751 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
2752 no fast atomic load is available for the particular size of access.
2753 (expand_atomic_compare_and_swap): Likewise.
2754 (expand_atomic_load): Likewise.
2755 (expand_atomic_store): Likewise.
2756 (expand_atomic_fetch_op): Likewise.
2757 * testsuite/lib/target-supports.exp
2758 (check_effective_target_sync_int_128): Remove x86 because it provides
2759 no fast atomic load.
2760 (check_effective_target_sync_int_128_runtime): Likewise.
2762 2017-02-01 Richard Biener <rguenther@suse.de>
2764 * graphite.c: Include tree-vectorizer.h for find_loop_location.
2765 (graphite_transform_loops): Provide opt-info for optimized nests.
2766 * tree-parloop.c (parallelize_loops): Provide opt-info for
2769 2017-02-01 Richard Biener <rguenther@suse.de>
2772 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
2775 2017-02-01 Richard Biener <rguenther@suse.de>
2777 PR tree-optimization/71824
2778 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
2779 Verify the loops are valid in the merged SESE region.
2780 (scop_detection::can_represent_loop_1): Check analyzing the
2781 evolution of the number of iterations in the region succeeds.
2783 2017-01-31 Ian Lance Taylor <iant@golang.org>
2785 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
2786 REG_ARGS_SIZE note to 32-bit push insns and call insn.
2788 2017-01-31 David Malcolm <dmalcolm@redhat.com>
2790 PR preprocessor/79210
2791 * input.c (get_substring_ranges_for_loc): Replace line_width
2792 assertion with error-handling.
2794 2017-01-31 Richard Biener <rguenther@suse.de>
2796 PR tree-optimization/77318
2797 * graphite-sese-to-poly.c (extract_affine): Fix assert.
2798 (create_pw_aff_from_tree): Take loop parameter.
2799 (add_condition_to_pbb): Pass loop of the condition to
2800 create_pw_aff_from_tree.
2802 2017-01-31 Jakub Jelinek <jakub@redhat.com>
2804 * config/s390/s390.c (s390_asan_shadow_offset): New function.
2805 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
2807 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2811 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
2813 (convert_int_to_float128): Likewise.
2814 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
2815 (convert_int_to_float128): Likewise.
2816 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
2817 (UNSPEC_IEEE128_CONVERT): Likewise.
2818 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
2819 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
2820 Use local variables for IBM extended format.
2821 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
2822 (fix_trunc<mode>si2_fprs): Likewise.
2823 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
2824 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
2825 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
2826 to know that we can now have integers of all sizes in vector
2828 (fix<uns>_<mode>di2_hw): Likewise.
2829 (float<uns>_<mode>si2_hw): Likewise.
2830 (fix_<mode>si2_hw): Likewise.
2831 (fixuns_<mode>si2_hw): Likewise.
2832 (float<uns>_<mode>di2_hw): Likewise.
2833 (float_<mode>di2_hw): Likewise.
2834 (float_<mode>si2_hw): Likewise.
2835 (floatuns_<mode>di2_hw): Likewise.
2836 (floatuns_<mode>si2_hw): Likewise.
2837 (xscvqp<su>wz_<mode>): Delete, no longer used.
2838 (xscvqp<su>dz_<mode>): Likewise.
2839 (xscv<su>dqp_<mode>): Likewise.
2840 (ieee128_mfvsrd_64bit): Likewise.
2841 (ieee128_mfvsrd_32bit): Likewise.
2842 (ieee128_mfvsrwz): Likewise.
2843 (ieee128_mtvsrw): Likewise.
2844 (ieee128_mtvsrd_64bit): Likewise.
2845 (ieee128_mtvsrd_32bit): Likewise.
2847 2017-01-31 Martin Liska <mliska@suse.cz>
2850 * ipa-prop.c (ipa_free_all_node_params): Call release method
2851 instead of ~sumbol_summary to not to trigger double times
2854 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
2856 PR tree-optimization/71691
2857 * bitmap.h (class auto_bitmap): New.
2858 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
2859 is_maybe_undefined instead of ssa_undefined_value_p.
2861 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2863 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
2864 __S390_ARCH_LEVEL__ to __ARCH__.
2866 2017-01-31 Jakub Jelinek <jakub@redhat.com>
2868 PR tree-optimization/79267
2869 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
2870 if should_remove_lhs_p is true.
2872 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
2875 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
2876 (add_alignment_attribute): New.
2877 (base_type_die): Add alignment attribute.
2878 (subrange_type_die): Likewise.
2879 (modified_type_die): Likewise.
2880 (gen_array_type_die): Likewise.
2881 (gen_descr_array_type_die: Likewise.
2882 (gen_enumeration_type_die): Likewise.
2883 (gen_subprogram_die): Likewise.
2884 (gen_variable_die): Likewise.
2885 (gen_field_die): Likewise.
2886 (gen_ptr_to_mbr_type_die): Likewise.
2887 (gen_struct_or_union_type_die): Likewise.
2888 (gen_subroutine_type_die): Likewise.
2889 (gen_typedef_die): Likewise.
2890 (base_type_cmp): Compare alignment attribute.
2892 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2895 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
2896 (setb_unsigned) New pattern for setb with CCUNS.
2897 * config/rs6000/rs6000.c (expand_block_compare): Use a different
2898 subfc./subfe sequence to avoid overflow problems. Generate a
2899 shorter sequence with cmpld/setb for power9.
2900 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
2901 for generating subfc. instruction.
2902 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
2903 now uses this instruction.
2905 2017-01-30 Ian Lance Taylor <iant@google.com>
2908 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
2909 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
2911 2017-01-30 Martin Sebor <msebor@redhat.com>
2913 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
2914 Move constant to the right of a relational operator.
2915 (get_mpfr_format_length, format_character, format_string): Ditto.
2916 (should_warn_p, maybe_warn): Same.
2918 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
2920 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
2923 * asan.c (get_translation_unit_decl): Remove function.
2924 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
2926 2017-01-30 Martin Liska <mliska@suse.cz>
2928 PR gcov-profile/79259
2929 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
2932 2017-01-30 Martin Liska <mliska@suse.cz>
2935 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
2936 Initialize variables with NULL value.
2938 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
2941 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
2943 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
2945 2017-01-30 Richard Biener <rguenther@suse.de>
2947 PR tree-optimization/79276
2948 * tree-vrp.c (process_assert_insertions): Properly adjust common
2949 when removing a duplicate.
2951 * gcc.dg/torture/pr79276.c: New testcase.
2953 2017-01-30 Richard Biener <rguenther@suse.de>
2955 PR tree-optimization/79256
2956 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
2957 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
2959 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
2961 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
2964 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
2965 ("*r<noxa>sbg_<mode>_sll_bitmask")
2966 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
2967 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
2968 Use contiguous_bitmask_nowrap_operand.
2970 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2973 * config/rs6000/altivec.h (vec_xl): Revise #define.
2974 (vec_xst): Likewise.
2976 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
2978 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
2980 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
2982 PR rtl-optimization/79194
2983 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
2984 traps before call to bypass_conditional_jumps.
2986 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
2988 PR tree-optimization/71374
2989 * lra-constraints.c (check_conflict_input_operands): New.
2990 (match_reload): Use it.
2992 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
2995 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
2996 account to calculate conflict_set.
2998 2017-01-27 Bin Cheng <bin.cheng@arm.com>
3000 PR rtl-optimization/78559
3001 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
3002 other_insn in combine.
3004 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
3006 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
3007 uint16_type_node for BT_UINT16.
3009 2017-01-27 David Malcolm <dmalcolm@redhat.com>
3011 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
3012 "RTL Tests" to menu.
3013 (GIMPLE Tests): New node.
3014 (RTL Tests): New node.
3016 2017-01-27 Richard Biener <rguenther@suse.de>
3018 PR tree-optimization/79245
3019 * tree-loop-distribution.c (distribute_loop): Apply cost
3020 modeling also to detected patterns.
3022 2017-01-27 Richard Biener <rguenther@suse.de>
3024 PR tree-optimization/71433
3025 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
3026 (compare_assert_loc): New function.
3027 (process_assert_insertions): Sort and optimize assert locations
3028 to remove duplicates and push down identical assertions on
3029 edges to their destination block.
3031 2017-01-27 Richard Biener <rguenther@suse.de>
3033 PR tree-optimization/79244
3034 * tree-vrp.c (remove_range_assertions): Forcefully propagate
3035 out SSA names even if abnormal.
3037 2017-01-27 Jakub Jelinek <jakub@redhat.com>
3039 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
3040 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
3041 instead of MPFR_RNDN.
3043 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
3046 * arm.c (arm_option_override): Don't call build_target_option_node
3047 until after doing all option overrides.
3048 (arm_valid_target_attribute_tree): Likewise.
3050 2017-01-27 Martin Liska <mliska@suse.cz>
3052 * doc/invoke.texi (-fprofile-arcs): Document profiling support
3053 for {cd}tors and C++ {cd}tors.
3055 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
3057 * config/s390/s390.md ("*setmem_long_and")
3058 ("*setmem_long_and_31z"): Use zero_extend instead of and.
3060 2017-01-26 Martin Sebor <msebor@redhat.com>
3062 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
3065 2017-01-26 Martin Sebor <msebor@redhat.com>
3067 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
3068 HAVE_DFmode before using XFmode or DFmode.
3069 (parse_directive): Avoid using the z length modifier to avoid
3070 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
3073 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
3074 to accept adjustment as an array.
3075 (get_int_range): New function.
3076 (struct directive): Make width and prec arrays.
3077 (directive::set_width, directive::set_precision): Call get_int_range.
3078 (format_integer, format_floating): Handle width and precision ranges.
3079 (format_string, parse_directive): Same.
3081 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3084 * dwarf2out.c (generate_skeleton_bottom_up): For children with
3085 comdat_type_p set, just clone them, but keep the children in the
3089 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
3090 which have direct callers with -fvar-tracking-assignments enabled
3092 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
3093 inside of type units.
3095 2017-01-26 Martin Sebor <msebor@redhat.com>
3098 * gimple-ssa-sprintf.c (struct result_range): Add likely and
3100 (struct format_result): Replace number_chars, number_chars_min,
3101 and number_chars_max with a single member of struct result_range.
3103 (format_result::operator+=): Adjust.
3104 (struct fmtresult): Remove bounded. Handle likely and unlikely
3106 (fmtresult::adjust_for_width_or_precision): New function.
3107 (fmtresult:type_max_digits): New function.
3108 (bytes_remaining): Handle likely and unlikely counters.
3109 (min_bytes_remaining): Remove.
3110 (format_percent): Simplify.
3111 (format_integer, format_floating): Set likely and unlikely counters.
3112 (get_string_length, format_character, format_string): Same.
3113 (format_plain, should_warn_p): New function.
3114 (maybe_warn): Call should_warn_p. Update diagnostic messages
3115 and handle those for all directives, including plain strings.
3116 (format_directive): Handle likely and unlikely counters.
3117 Remove unnecessary quoting from diagnostics. Add an informational
3119 (add_bytes): Remove.
3120 (pass_sprintf_length::compute_format_length): Simplify.
3121 (try_substitute_return_value): Handle likely and unlikely counters.
3123 2017-01-26 Carl Love <cel@us.ibm.com>
3125 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
3126 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
3128 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
3131 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
3132 endianess for subregs into account.
3133 * lra-constraints.c (lra_constraints): Do risky transformations
3134 always on the first iteration.
3135 * lra-lives.c (check_pseudos_live_through_calls): Add arg
3136 last_call_used_reg_set.
3137 (process_bb_lives): Define and use last_call_used_reg_set.
3138 * lra.c (lra): Always continue after lra_constraints on the first
3141 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
3143 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
3145 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
3147 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3149 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
3150 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
3151 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
3152 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
3153 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
3154 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
3155 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
3156 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
3157 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
3159 2017-01-26 Marek Polacek <polacek@redhat.com>
3162 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
3163 for the third operand.
3165 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3168 * omp-low.c (struct omp_context): Add simt_stmt field.
3169 (scan_omp_for): Return omp_context *.
3170 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
3171 context to the _simt_ SIMD stmt.
3172 (lower_omp_for): For combined SIMD with sibling _simt_
3173 SIMD, make sure to use the same decls in _looptemp_
3174 clauses as in the sibling.
3176 2017-01-26 David Sherwood <david.sherwood@arm.com>
3179 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
3182 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3185 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
3186 emit fld b; fld a; if possible.
3188 * brig-builtins.def: Update copyright years.
3189 * config/arm/arm_acle_builtins.def: Update copyright years.
3191 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
3194 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
3195 constraint instead of o for the stxsd instruction.
3197 2017-01-25 Carl Love <cel@us.ibm.com>
3199 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
3200 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
3202 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
3204 * doc/invoke.texi (C++ Dialect Options): Fix typo.
3206 2017-01-25 Richard Biener <rguenther@suse.de>
3208 PR tree-optimization/69264
3209 * target.def (vector_alignment_reachable): Improve documentation.
3210 * doc/tm.texi: Regenerate.
3211 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
3213 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
3214 earlier changes with respect to TYPE_USER_ALIGN.
3215 (vector_alignment_reachable_p): Likewise. Improve dumping.
3217 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3220 * config/arm/arm.md (xordi3): Force constant operand into a register
3223 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3225 * doc/invoke.texi (-fstore-merging): Correct default optimization
3226 levels at which it is enabled.
3227 (-O): Move -fstore-merging from list to...
3230 2017-01-25 Richard Biener <rguenther@suse.de>
3233 * omp-expand.c: Include debug.h.
3234 (expand_omp_taskreg): Make sure to generate early debug before
3235 outlining anything from a function.
3236 (expand_omp_target): Likewise.
3237 (grid_expand_target_grid_body): Likewise.
3239 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
3242 * asan.c (get_translation_unit_decl): New function.
3243 (asan_add_global): Extract modules file name from globals
3244 TRANSLATION_UNIT_DECL name.
3246 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
3249 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
3250 for long calls with APCS frame and VFP.
3252 2017-01-24 David Malcolm <dmalcolm@redhat.com>
3254 * cfg.c (original_copy_tables_initialized_p): New function.
3255 * cfg.h (original_copy_tables_initialized_p): New decl.
3256 * cfgrtl.c (relink_block_chain): Guard the call to
3257 free_original_copy_tables with a call to
3258 original_copy_tables_initialized_p.
3259 * cgraph.h (symtab_node::native_rtl_p): New decl.
3260 * cgraphunit.c (symtab_node::native_rtl_p): New function.
3261 (symtab_node::needed_p): Don't assert for early assembly output
3262 for __RTL functions.
3263 (cgraph_node::finalize_function): Set "force_output" for __RTL
3265 (cgraph_node::analyze): Bail out early for __RTL functions.
3266 (analyze_functions): Update assertion to support __RTL functions.
3267 (cgraph_node::expand): Bail out early for __RTL functions.
3268 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
3270 * function.h (struct function): Update comment for field
3272 * gimple-expr.c: Include "tree-pass.h".
3273 (gimple_has_body_p): Return false for __RTL functions.
3274 * Makefile.in (OBJS): Add run-rtl-passes.o.
3275 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
3277 (gcc::pass_manager::get_clean_slate): New accessor.
3278 * passes.c: Include "insn-addr.h".
3279 (should_skip_pass_p): Add logging. Update logic for running
3280 "expand" to be compatible with both __GIMPLE and __RTL. Guard
3281 property-provider override so it is only done for gimple passes.
3283 (skip_pass): New function.
3284 (execute_one_pass): Call skip_pass when skipping passes.
3285 * read-md.c (md_reader::read_char): Support filtering
3286 the input to a subset of line numbers.
3287 (md_reader::md_reader): Initialize fields
3288 m_first_line and m_last_line.
3289 (md_reader::read_file_fragment): New function.
3290 * read-md.h (md_reader::read_file_fragment): New decl.
3291 (md_reader::m_first_line): New field.
3292 (md_reader::m_last_line): New field.
3293 * read-rtl-function.c (function_reader::create_function): Only
3294 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
3295 curr_properties. Set DECL_INITIAL to a dummy block.
3296 (read_rtl_function_body_from_file_range): New function.
3297 * read-rtl-function.h (read_rtl_function_body_from_file_range):
3299 * run-rtl-passes.c: New file.
3300 * run-rtl-passes.h: New file.
3302 2017-01-24 Jeff Law <law@redhat.com>
3304 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
3307 2017-01-24 Bin Cheng <bin.cheng@arm.com>
3309 PR tree-optimization/79159
3310 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
3311 (record_nonwrapping_iv): Improve boundary using above function if no
3312 value range information.
3314 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
3315 Martin Jambor <mjambor@suse.cz>
3317 * brig-builtins.def: New file.
3318 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
3319 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
3320 (DEF_HSAIL_SAT_BUILTIN): Likewise.
3321 (DEF_HSAIL_INTR_BUILTIN): Likewise.
3322 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
3323 * builtin-types.def (BT_INT8): New.
3324 (BT_INT16): Likewise.
3325 (BT_UINT8): Likewise.
3326 (BT_UINT16): Likewise.
3327 (BT_FN_ULONG): Likewise.
3328 (BT_FN_UINT_INT): Likewise.
3329 (BT_FN_UINT_ULONG): Likewise.
3330 (BT_FN_UINT_LONG): Likewise.
3331 (BT_FN_UINT_PTR): Likewise.
3332 (BT_FN_ULONG_PTR): Likewise.
3333 (BT_FN_INT8_FLOAT): Likewise.
3334 (BT_FN_INT16_FLOAT): Likewise.
3335 (BT_FN_UINT32_FLOAT): Likewise.
3336 (BT_FN_UINT16_FLOAT): Likewise.
3337 (BT_FN_UINT8_FLOAT): Likewise.
3338 (BT_FN_UINT64_FLOAT): Likewise.
3339 (BT_FN_UINT16_UINT32): Likewise.
3340 (BT_FN_UINT32_UINT16): Likewise.
3341 (BT_FN_UINT16_UINT16_UINT16): Likewise.
3342 (BT_FN_INT_PTR_INT): Likewise.
3343 (BT_FN_UINT_PTR_UINT): Likewise.
3344 (BT_FN_LONG_PTR_LONG): Likewise.
3345 (BT_FN_ULONG_PTR_ULONG): Likewise.
3346 (BT_FN_VOID_UINT64_UINT64): Likewise.
3347 (BT_FN_UINT8_UINT8_UINT8): Likewise.
3348 (BT_FN_INT8_INT8_INT8): Likewise.
3349 (BT_FN_INT16_INT16_INT16): Likewise.
3350 (BT_FN_INT_INT_INT): Likewise.
3351 (BT_FN_UINT_FLOAT_UINT): Likewise.
3352 (BT_FN_FLOAT_UINT_UINT): Likewise.
3353 (BT_FN_ULONG_UINT_UINT): Likewise.
3354 (BT_FN_ULONG_UINT_PTR): Likewise.
3355 (BT_FN_ULONG_ULONG_ULONG): Likewise.
3356 (BT_FN_UINT_UINT_UINT): Likewise.
3357 (BT_FN_VOID_UINT_PTR): Likewise.
3358 (BT_FN_UINT_UINT_PTR: Likewise.
3359 (BT_FN_UINT32_UINT64_PTR): Likewise.
3360 (BT_FN_INT_INT_UINT_UINT): Likewise.
3361 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
3362 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
3363 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
3364 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
3365 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
3366 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
3367 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
3368 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
3369 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
3370 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
3371 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
3372 * doc/frontends.texi: List BRIG FE.
3373 * doc/install.texi (Testing): Add BRIG tesring requirements.
3374 * doc/invoke.texi (Overall Options): Mention BRIG.
3375 * doc/standards.texi (Standards): Doucment BRIG HSA version.
3377 2017-01-24 Richard Biener <rguenther@suse.de>
3379 PR translation/79208
3380 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
3382 2017-01-24 Martin Jambor <mjambor@suse.cz>
3385 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
3386 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
3389 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
3392 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
3393 casts from signed to unsigned really don't have a range.
3395 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
3397 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
3398 GMP_RNDx for compatiblity.
3400 2017-01-24 Martin Liska <mliska@suse.cz>
3403 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
3404 that would prevent us to call alloca with -1 as argument.
3406 2017-01-24 Jakub Jelinek <jakub@redhat.com>
3408 * dwarf2out.c (output_compilation_unit_header, output_file_names):
3409 Avoid -Wformat-security warning.
3411 2017-01-23 Andrew Pinski <apinski@cavium.com>
3413 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
3416 2017-01-23 Martin Sebor <msebor@redhat.com>
3419 * gimple-ssa-sprintf.c (warn_level): New global.
3420 (format_integer): Use it here and throughout the rest of the file.
3421 Use the same switch to compute sign as base.
3422 (maybe_warn): New function.
3423 (format_directive): Factor out warnings into maybe_warn.
3424 Add debugging output. Use warn_level.
3425 (add_bytes): Use warn_level.
3426 (pass_sprintf_length::compute_format_length): Add debugging output.
3427 (try_substitute_return_value): Same.
3428 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
3431 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
3432 (struct fmtresult, format_integer, format_floating): Adjust.
3433 (fmtresult::fmtresult): Set max correctly in two argument ctor.
3434 (get_string_length, format_string,format_directive): Same.
3435 (pass_sprintf_length::compute_format_length): Same.
3436 (try_substitute_return_value): Simplify slightly.
3439 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
3440 (fmtresult::operator+=): Outlined.
3441 (struct fmtresult): Add ctors.
3442 (struct conversion_spec): Rename...
3443 (struct directive): ...to this. Add and remove data members.
3444 (directive::set_width, directive::set_precision): New functions.
3445 (format_percent): Use fmtresult ctor.
3446 (get_width_and_precision): Remove.
3447 (format_integer): Make naming changes. Avoid computing width and
3449 (format_floating): Same. Adjust indentation.
3450 (format_character, format_none): New functions.
3451 (format_string): Moved character handling to format_character.
3452 (format_directive): Remove arguments, change return type.
3453 (parse_directive): New function.
3454 (pass_sprintf_length::compute_format_length): Move directive
3455 parsing to parse_directive.
3457 2017-01-23 Jakub Jelinek <jakub@redhat.com>
3459 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
3460 (assign_assembler_name_if_needed): ... this.
3461 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
3462 (assign_assembler_name_if_needed): ... this.
3463 (free_lang_data_in_cgraph): Adjust callers.
3464 * cgraphunit.c (cgraph_node::analyze): Likewise.
3465 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
3468 2017-01-23 Richard Biener <rguenther@suse.de>
3470 PR tree-optimization/79088
3471 PR tree-optimization/79188
3472 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
3473 resetting loop bounds after last path deletion. Reset loop
3474 bounds of the target loop, make code match the comments.
3475 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
3476 Make sure loops need no fixups.
3478 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
3480 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
3481 exponent support with double type for first argument.
3482 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
3483 type returned by __builtin_vec_extract_sig,
3484 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
3485 functions from "vector int" to "vector unsigned int" or from
3486 "vector long long int" to "vector unsigned long long int".
3487 Changed type returned by __builtin_vec_extract_exp,
3488 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
3489 functions from "vector int" to "vector unsigned int" or from
3490 "vector long long int" to "vector unsigned long long int".
3491 Changed return type of __builtin_vec_test_data_class,
3492 __builtin_vec_test_data_class_sp, and
3493 __builtin_vec_test_data_class_dp from "vector int" to
3494 "vector bool int" or from "vector long long int" to "vector bool
3495 long long int" and changed second argument type from "unsigned
3496 int" to "int". Added new overloaded function forms "vector float
3497 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
3498 "vector float __builtin_vec_insert_exp_sp (vector float, vector
3499 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
3500 double, vector unsigned long long int)" and "vector double
3501 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
3502 long int)". Changed return type of
3503 __builtin_scalar_test_data_class and
3504 __builtin_scalar_test_data_class_sp and
3505 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
3506 int" and changed second argument from "unsigned int" to "int".
3507 Changed type returned by __builtin_scalar_test_neg,
3508 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
3509 from "int" to "bool int". Added new overloaded function form
3510 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
3511 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
3512 exponent double-precision with floating point first argument.
3513 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
3514 documentation of scalar_test_data_class, scalar_test_neg,
3515 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
3516 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
3517 vec_test_data_class built-in functions to reflect refinements in
3518 their type signatures.
3520 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
3522 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
3524 (aarch64_elf_asm_destructor): Likewise.
3526 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
3528 PR rtl-optimization/78634
3529 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
3530 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
3531 * ifcvt.c (noce_try_cmove): Add missing cost check.
3533 PR rtl-optimization/71724
3534 * combine.c (if_then_else_cond): Look for situations where it is
3535 beneficial to undo the work of one of the recursive calls.
3537 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3539 PR tree-optimization/70754
3540 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3541 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3542 combined stmt before it if not NULL.
3543 (combine_chains): Process refs reversely and compute dominance point
3546 2017-01-23 Martin Liska <mliska@suse.cz>
3548 PR tree-optimization/79196
3549 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
3550 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
3552 (strlen_optimize_stmt): Call the renamed function.
3554 2017-01-23 Michael Matz <matz@suse.de>
3556 PR tree-optimization/78384
3557 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
3559 2017-01-23 Richard Biener <rguenther@suse.de>
3561 PR tree-optimization/79186
3562 * tree-vrp.c (register_new_assert_for): Make sure we've seen
3563 both incoming edges before moving an assert.
3565 2017-01-23 Martin Jambor <mjambor@suse.cz>
3567 * ipa-prop.c (load_from_param_1): Removed.
3568 (load_from_unmodified_param): Bits from load_from_param_1 put back
3570 (load_from_param): Removed.
3571 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
3572 with stmt. Reverted back to use of load_from_unmodified_param.
3574 2017-01-23 Martin Jambor <mjambor@suse.cz>
3577 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
3578 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
3579 field a pointer to garbage collected vector, mark lattices and
3580 ipcp_orig_node with GTY((skip)).
3581 (ipa_get_param_count): Adjust to descriptors being a pointer.
3582 (ipa_get_param): Likewise.
3583 (ipa_get_type): Likewise.
3584 (ipa_get_param_move_cost): Likewise.
3585 (ipa_set_param_used): Likewise.
3586 (ipa_get_controlled_uses): Likewise.
3587 (ipa_set_controlled_uses): Likewise.
3588 (ipa_is_param_used): Likewise.
3589 (ipa_node_params_t): Move into garbage collector. New methods insert
3591 (ipa_node_params_sum): Annotate wth GTY(()).
3592 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
3594 (ipa_load_from_parm_agg): Adjust declaration.
3595 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
3596 * ipa-profile.c (ipa_profile): Likewise.
3597 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
3598 (ipa_populate_param_decls): Make descriptors parameter garbage
3600 (ipa_dump_param): Adjust to descriptors being a pointer.
3601 (ipa_alloc_node_params): Likewise.
3602 (ipa_initialize_node_params): Likewise.
3603 (load_from_param_1): Make descriptors parameter garbage collected.
3604 (load_from_unmodified_param): Likewise.
3605 (load_from_param): Likewise.
3606 (ipa_load_from_parm_agg): Likewise.
3607 (ipa_node_params::~ipa_node_params): Removed.
3608 (ipa_free_all_node_params): Remove call to delete operator.
3609 (ipa_node_params_t::insert): New.
3610 (ipa_node_params_t::remove): Likewise.
3611 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
3612 copy known_csts and known_contexts vectors.
3613 (ipa_read_node_info): Adjust to descriptors being a pointer.
3614 (ipcp_modif_dom_walker): Make m_descriptors field garbage
3616 (ipcp_transform_function): Make descriptors variable garbage
3619 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
3621 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
3622 * config/i386/avx512dqintrin.h: Ditto.
3623 * config/i386/avx512fintrin.h: Ditto.
3624 * gcc/config/i386/i386.c: Handle new builtins.
3625 * config/i386/i386-builtin.def: Add new builtins.
3626 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
3627 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
3629 2017-01-23 Jakub Jelinek <jakub@redhat.com>
3630 Martin Liska <mliska@suse.cz>
3632 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
3633 * asan.c (asan_expand_poison_ifn): Support stores and use
3634 appropriate ASAN report function.
3635 * internal-fn.c (expand_ASAN_POISON_USE): New function.
3636 * internal-fn.def (ASAN_POISON_USE): Declare.
3637 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
3638 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
3639 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
3640 ASAN_POISON calls w/o LHS.
3641 * tree-ssa.c (execute_update_addresses_taken): Create clobber
3642 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
3643 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
3644 * gimplify.c (asan_poison_variables): Add attribute
3645 use_after_scope_memory to variables that really needs to live
3647 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
3648 having the attribute.
3650 2017-01-23 Martin Liska <mliska@suse.cz>
3652 * asan.c (create_asan_shadow_var): New function.
3653 (asan_expand_poison_ifn): Likewise.
3654 * asan.h (asan_expand_poison_ifn): New declaration.
3655 * internal-fn.c (expand_ASAN_POISON): Likewise.
3656 * internal-fn.def (ASAN_POISON): New builtin.
3657 * sanopt.c (pass_sanopt::execute): Expand
3658 asan_expand_poison_ifn.
3659 * tree-inline.c (copy_decl_for_dup_finish): Make function
3661 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
3662 * tree-ssa.c (is_asan_mark_p): New function.
3663 (execute_update_addresses_taken): Rewrite local variables
3664 (identified just by use-after-scope as addressable) into SSA.
3666 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
3668 * doc/install.texi (Specific): opensource.apple.com uses https
3669 now. Remove trailing slash.
3671 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
3673 * README.Portability: Remove note on an Irix compatibility issue.
3675 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
3677 * gcov.c (INCLUDE_ALGORITHM): Define.
3678 (INCLUDE_VECTOR): Define.
3679 No longer include <vector> and <algorithm> directly.
3681 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
3683 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
3685 * doc/invoke.texi (Code Gen Options): Ditto.
3687 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
3690 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
3692 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
3694 rtl-optimization/79125
3695 * cprop.c (local_cprop_pass): Handle cases where we make an
3698 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
3702 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
3703 read from, for big endian.
3705 2017-01-20 Jiong Wang <jiong.wang@arm.com>
3707 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
3708 register pauth builtins for LP64 only.
3710 2017-01-20 Marek Polacek <polacek@redhat.com>
3713 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
3716 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
3718 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
3720 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
3721 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
3722 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
3724 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3727 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
3728 in big-endian mode when they are not a single duplicated value.
3730 2017-01-20 Richard Biener <rguenther@suse.de>
3732 * BASE-VER: Bump to 7.0.1.
3734 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
3736 * omp-low.c (omplow_simd_context): New struct. Use it...
3737 (lower_rec_simd_input_clauses): ...here and...
3738 (lower_rec_input_clauses): ...here to hold common data. Adjust all
3739 references to idx, lane, max_vf, is_simt.
3741 2017-01-20 Graham Markall <graham.markall@embecosm.com>
3743 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
3746 2017-01-20 Martin Jambor <mjambor@suse.cz>
3748 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
3749 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
3751 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
3752 (GTFILES): Rename hsa.c to hsa-common.c.
3753 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
3754 * hsa-dump.c: Likewise.
3755 * hsa-gen.c: Likewise.
3756 * hsa-regalloc.c: Likewise.
3757 * ipa-hsa.c: Likewise.
3758 * omp-expand.c: Likewise.
3759 * omp-low.c: Likewise.
3760 * toplev.c: Likewise.
3762 2017-01-20 Marek Polacek <polacek@redhat.com>
3765 * doc/invoke.texi: Document -Wduplicated-branches.
3766 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
3767 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
3768 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
3769 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
3770 return 0 only when not OEP_LEXICOGRAPHIC.
3771 (fold_build_cleanup_point_expr): Use the expression
3772 location when building CLEANUP_POINT_EXPR.
3773 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
3774 * tree.c (add_expr): Handle error_mark_node.
3776 2017-01-20 Martin Liska <mliska@suse.cz>
3779 * tree-profile.c (init_ic_make_global_vars): Do not call
3781 (gimple_init_gcov_profiler): Likewise.
3783 2017-01-20 Martin Liska <mliska@suse.cz>
3786 * cgraph.h (maybe_create_reference): Remove argument and
3788 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
3790 * ipa-cp.c (create_specialized_node): Likewise.
3791 * symtab.c (symtab_node::maybe_create_reference): Handle
3792 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
3794 2017-01-20 Martin Liska <mliska@suse.cz>
3796 * read-rtl-function.c (function_reader::create_function): Use
3797 build_decl instread of build_decl_stat.
3799 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
3801 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
3802 * config/i386/avx512dqintrin.h: Ditto.
3803 * config/i386/avx512fintrin.h: Ditto.
3804 * config/i386/i386-builtin-types.def: Add new types.
3805 * gcc/config/i386/i386.c: Handle new types.
3806 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
3807 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
3808 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
3809 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
3810 (__builtin_ia32_kshiftridi): New.
3811 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
3813 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
3817 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
3818 define to rs6000_init_stack_protect_guard.
3819 (rs6000_init_stack_protect_guard): New function.
3821 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
3822 Yunqiang Su <yunqiang.su@imgtec.com>
3824 * config.gcc (supported_defaults): Add madd4.
3825 (with_madd4): Add validation.
3826 (all_defaults): Add madd4.
3827 * config/mips/mips.opt (mmadd4): New option.
3828 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
3830 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
3832 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
3833 (ISA_HAS_FUSED_MADD4): Likewise.
3834 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
3835 * gcc/doc/install.texi (--with-madd4): Document the new option.
3837 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3839 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
3840 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
3841 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
3842 (aarch64_init_pauth_hint_builtins): New.
3843 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
3844 (aarch64_expand_builtin): Expand new builtins.
3846 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3848 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
3849 * combine-stack-adj.c (no_unhandled_cfa): Handle
3850 REG_CFA_TOGGLE_RA_MANGLE.
3851 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
3852 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
3853 info for return address signing.
3854 (aarch64_expand_epilogue): Likewise.
3856 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3858 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
3859 * config/aarch64/aarch64-protos.h
3860 (aarch64_return_address_signing_enabled): New declaration.
3861 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
3863 (aarch64_expand_prologue): Sign return address before it's pushed onto
3865 (aarch64_expand_epilogue): Authenticate return address fetched from
3867 (aarch64_override_options): Sanity check for ILP32 and ISA level.
3868 (aarch64_attributes): New function attributes for "sign-return-address".
3869 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
3870 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
3871 ("*do_return"): Generate combined instructions according to key index.
3872 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
3873 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
3875 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
3876 * config/aarch64/aarch64.opt (msign-return-address=): New.
3877 * doc/extend.texi (AArch64 Function Attributes): Documents
3878 "sign-return-address=".
3879 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
3881 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
3883 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
3884 overall option summary.
3886 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3888 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
3889 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
3890 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
3891 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
3893 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
3895 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
3896 -mpower9-minmax by default for -mcpu=power9.
3897 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
3898 128-bit floating point.
3900 2017-01-20 Alan Modra <amodra@gmail.com>
3902 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
3903 optimizing for size.
3905 2017-01-20 Alan Modra <amodra@gmail.com>
3908 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
3909 for strcmp and strncmp from corresponding builtin decl.
3911 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
3913 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
3914 instead of i386/rtems-64.h.
3915 * config/i386/rtems-64.h: Remove.
3917 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
3921 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
3923 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
3925 2017-01-19 Tamar Christina <tamar.christina@arm.com>
3927 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3928 Change int to HOST_WIDE_INT.
3929 * config/aarch64/aarch64-protos.h
3930 (aarch64_simd_gen_const_vector_dup): Likewise.
3931 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
3933 2017-01-19 David Malcolm <dmalcolm@redhat.com>
3935 * langhooks-def.h (lhd_type_for_size): New decl.
3936 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
3937 * langhooks.c (lhd_type_for_size): New function, taken from
3940 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
3942 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
3943 define_bypass for CR latency.
3944 (power9-cracked-alu): Update bypass latency and remove power9-branch.
3945 (power9-alu2): Add define_bypass for CR latency.
3947 (power9-mul): Update insn latency.
3948 (power9-mul-compare): Update insn latency, bypass latency and remove
3951 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3953 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
3955 * config/aarch64/aarch64.md
3956 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
3957 aarch64_nopcrelative_literal_loads.
3958 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
3960 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
3962 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
3964 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
3966 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
3969 * config.gcc (supported_defaults): Add lxc1-sxc1.
3970 (with_lxc1_sxc1): Add validation.
3971 (all_defaults): Add lxc1-sxc1.
3972 * config/mips/mips.opt (mlxc1-sxc1): New option.
3973 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
3975 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
3976 __mips_no_lxc1_sxc1.
3977 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
3978 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
3979 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
3981 2017-01-19 Richard Biener <rguenther@suse.de>
3983 PR tree-optimization/72488
3984 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
3985 sure to restore SSA info.
3986 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
3988 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
3990 PR rtl-optimization/79121
3991 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
3992 of the inner type when shifting an extended value.
3994 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
3997 * symtab.c (symtab_node::equal_address_to): Fix comparing of
3998 interposable aliases.
4000 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
4003 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
4004 Use the evmergelohi instruction.
4005 (mov_si<mode>_e500_subreg4_2_le): Likewise.
4006 (mov_sitf_e500_subreg8_2_be): Likewise.
4007 (mov_sitf_e500_subreg12_2_le): Likewise.
4008 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
4009 (mov_si<mode>_e500_subreg4_2_be): Likewise.
4010 (mov_sitf_e500_subreg8_2_le): Likewise.
4011 (mov_sitf_e500_subreg12_2_be): Likewise.
4013 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4015 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
4016 attribute from vecsimple to vecperm.
4017 (altivec_vbpermq2): Likewise.
4019 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4022 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
4024 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4025 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
4026 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
4027 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
4028 case where N arg is SIZE_MAX.
4029 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
4030 (cmpstrsi): Add pattern.
4032 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
4034 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4035 __builtin_vec_revb builtins.
4036 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
4037 built-in functions to support generation of the ISA 3.0 XXBR<x>
4038 vector byte reverse instructions.
4039 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
4040 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
4041 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
4042 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
4043 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
4044 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
4045 (P9V_BUILTIN_VEC_REVB): Likewise.
4046 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
4047 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
4048 (p9_xxbrq_v16qi): Likewise.
4049 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
4050 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
4051 (p9_xxbrh_v8hi): Likewise.
4052 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
4053 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
4054 vec_revb built-in functions.
4056 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
4058 PR rtl-optimization/78952
4059 * config/i386/i386.md (any_extract): New code iterator.
4060 (*insvqi_2): Use any_extract for source operand.
4061 (*insvqi_3): Use any_shiftrt for source operand.
4063 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
4065 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
4067 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
4069 2017-01-18 Matthias Klose <doko@ubuntu.com>
4071 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
4073 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4075 * config/rs6000/altivec.h (vec_bperm): Change #define.
4076 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
4077 (altivec_vbpermq2): New define_insn.
4078 (altivec_vbpermd): Likewise.
4079 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
4081 (VBPERMD): Likewise.
4082 (VBPERM): New polymorphic function interface.
4083 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
4084 Add entries for P9V_BUILTIN_VEC_VBPERM.
4085 * doc/extend.texi: Add interfaces for vec_bperm.
4087 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4089 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
4090 first letter of error messages.
4091 (s390_resolve_overloaded_builtin): Likewise.
4092 * config/s390/s390.c (s390_expand_builtin): Likewise.
4093 (s390_invalid_arg_for_unprototyped_fn): Likewise.
4094 (s390_valid_target_attribute_inner_p): Likewise.
4095 * config/s390/s390.md ("tabort"): Likewise.
4097 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
4099 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
4100 (ISA_AVOID_DIV_HILO): New macro.
4101 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
4102 (ISA_HAS_DDIV): Likewise.
4104 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4106 * doc/invoke.texi (fabi-version): Correct number of occurrences.
4108 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4110 * doc/invoke.texi (fabi-version): Spelling fix.
4112 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4115 * doc/invoke.texi (fabi-version): Mention mangling fix for
4118 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4121 * doc/invoke.texi (fabi-version): Document discriminator mangling.
4123 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
4126 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
4127 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
4129 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
4131 (stack_protect_test): Ditto.
4132 * config/rs6000/rs6000.opt (mstack-protector-guard=,
4133 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
4135 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
4136 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
4137 -mstack-protector-guard-offset=.
4138 (RS/6000 and PowerPC Options): Ditto.
4140 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
4142 * config/i386/i386.h (MASK_CLASS_P): New define.
4143 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
4144 there are no registers from different register sets also when
4145 mask registers are used. Update function comment.
4146 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
4147 to (*k/*r) and (*k/*km) alternatives.
4149 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
4151 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
4152 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
4153 (EH_RETURN_HANDLER_RTX): New define.
4154 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
4155 Force frame pointer in EH return functions.
4156 (aarch64_expand_epilogue): Add barrier for eh_return.
4157 (aarch64_final_eh_return_addr): Remove.
4158 (aarch64_eh_return_handler_rtx): New function.
4159 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
4161 (aarch64_eh_return_handler_rtx): New prototype.
4163 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4165 * config/rs6000/altivec.h (vec_rlmi): New #define.
4166 (vec_vrlnm): Likewise.
4167 (vec_rlnm): Likewise.
4168 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
4169 (UNSPEC_VRLNM): Likewise.
4170 (VIlong): New mode iterator.
4171 (altivec_vrl<VI_char>mi): New define_insn.
4172 (altivec_vrl<VI_char>nm): Likewise.
4173 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
4176 (RLNM): New polymorphic function entry.
4177 (VRLWMI): New monomorphic function entry.
4179 (RLMI): New polymorphic function entry.
4180 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
4181 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
4182 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
4185 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4188 * dwarf2out.c (field_byte_offset): Restore the
4189 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
4190 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
4191 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
4194 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
4197 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
4199 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4202 * dwarf2out.c (add_data_member_location_attribute): For constant
4203 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
4204 instead of DW_AT_data_member_location, DW_AT_bit_offset and
4205 DW_AT_byte_size attributes.
4207 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
4209 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
4210 after forcing to constant memory when the code model is medium.
4212 2017-01-17 Julia Koval <julia.koval@intel.com>
4215 * config/i386/avx512fintrin.h
4216 (_mm512_i32gather_ps): Change __addr type to void const*.
4217 (_mm512_mask_i32gather_ps): Ditto.
4218 (_mm512_i32gather_pd): Ditto.
4219 (_mm512_mask_i32gather_pd): Ditto.
4220 (_mm512_i64gather_ps): Ditto.
4221 (_mm512_mask_i64gather_ps): Ditto.
4222 (_mm512_i64gather_pd): Ditto.
4223 (_mm512_mask_i64gather_pd): Ditto.
4224 (_mm512_i32gather_epi32): Ditto.
4225 (_mm512_mask_i32gather_epi32): Ditto.
4226 (_mm512_i32gather_epi64): Ditto.
4227 (_mm512_mask_i32gather_epi64): Ditto.
4228 (_mm512_i64gather_epi32): Ditto.
4229 (_mm512_mask_i64gather_epi32): Ditto.
4230 (_mm512_i64gather_epi64): Ditto.
4231 (_mm512_mask_i64gather_epi64): Ditto.
4232 (_mm512_i32scatter_ps): Change __addr type to void*.
4233 (_mm512_mask_i32scatter_ps): Ditto.
4234 (_mm512_i32scatter_pd): Ditto.
4235 (_mm512_mask_i32scatter_pd): Ditto.
4236 (_mm512_i64scatter_ps): Ditto.
4237 (_mm512_mask_i64scatter_ps): Ditto.
4238 (_mm512_i64scatter_pd): Ditto.
4239 (_mm512_mask_i64scatter_pd): Ditto.
4240 (_mm512_i32scatter_epi32): Ditto.
4241 (_mm512_mask_i32scatter_epi32): Ditto.
4242 (_mm512_i32scatter_epi64): Ditto.
4243 (_mm512_mask_i32scatter_epi64): Ditto.
4244 (_mm512_i64scatter_epi32): Ditto.
4245 (_mm512_mask_i64scatter_epi32): Ditto.
4246 (_mm512_i64scatter_epi64): Ditto.
4247 (_mm512_mask_i64scatter_epi64): Ditto.
4248 * config/i386/avx512pfintrin.h
4249 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
4250 (_mm512_mask_prefetch_i32gather_ps): Ditto.
4251 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4252 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4253 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
4254 (_mm512_prefetch_i32scatter_ps): Ditto.
4255 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4256 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4257 (_mm512_prefetch_i64scatter_pd): Ditto.
4258 (_mm512_prefetch_i64scatter_ps): Ditto.
4259 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4260 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4261 * config/i386/avx512vlintrin.h
4262 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
4263 (_mm_mmask_i32gather_ps): Ditto.
4264 (_mm256_mmask_i32gather_pd): Ditto.
4265 (_mm_mmask_i32gather_pd): Ditto.
4266 (_mm256_mmask_i64gather_ps): Ditto.
4267 (_mm_mmask_i64gather_ps): Ditto.
4268 (_mm256_mmask_i64gather_pd): Ditto.
4269 (_mm_mmask_i64gather_pd): Ditto.
4270 (_mm256_mmask_i32gather_epi32): Ditto.
4271 (_mm_mmask_i32gather_epi32): Ditto.
4272 (_mm256_mmask_i32gather_epi64): Ditto.
4273 (_mm_mmask_i32gather_epi64): Ditto.
4274 (_mm256_mmask_i64gather_epi32): Ditto.
4275 (_mm_mmask_i64gather_epi32): Ditto.
4276 (_mm256_mmask_i64gather_epi64): Ditto.
4277 (_mm_mmask_i64gather_epi64): Ditto.
4278 (_mm256_i32scatter_ps): Change __addr type to void*.
4279 (_mm256_mask_i32scatter_ps): Ditto.
4280 (_mm_i32scatter_ps): Ditto.
4281 (_mm_mask_i32scatter_ps): Ditto.
4282 (_mm256_i32scatter_pd): Ditto.
4283 (_mm256_mask_i32scatter_pd): Ditto.
4284 (_mm_i32scatter_pd): Ditto.
4285 (_mm_mask_i32scatter_pd): Ditto.
4286 (_mm256_i64scatter_ps): Ditto.
4287 (_mm256_mask_i64scatter_ps): Ditto.
4288 (_mm_i64scatter_ps): Ditto.
4289 (_mm_mask_i64scatter_ps): Ditto.
4290 (_mm256_i64scatter_pd): Ditto.
4291 (_mm256_mask_i64scatter_pd): Ditto.
4292 (_mm_i64scatter_pd): Ditto.
4293 (_mm_mask_i64scatter_pd): Ditto.
4294 (_mm256_i32scatter_epi32): Ditto.
4295 (_mm256_mask_i32scatter_epi32): Ditto.
4296 (_mm_i32scatter_epi32): Ditto.
4297 (_mm_mask_i32scatter_epi32): Ditto.
4298 (_mm256_i32scatter_epi64): Ditto.
4299 (_mm256_mask_i32scatter_epi64): Ditto.
4300 (_mm_i32scatter_epi64): Ditto.
4301 (_mm_mask_i32scatter_epi64): Ditto.
4302 (_mm256_i64scatter_epi32): Ditto.
4303 (_mm256_mask_i64scatter_epi32): Ditto.
4304 (_mm_i64scatter_epi32): Ditto.
4305 (_mm_mask_i64scatter_epi32): Ditto.
4306 (_mm256_i64scatter_epi64): Ditto.
4307 (_mm256_mask_i64scatter_epi64): Ditto.
4308 (_mm_i64scatter_epi64): Ditto.
4309 (_mm_mask_i64scatter_epi64): Ditto.
4310 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
4311 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
4312 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
4313 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
4314 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
4315 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
4316 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
4317 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
4318 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
4319 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
4320 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
4321 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
4322 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
4323 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
4324 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
4325 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
4326 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
4327 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
4328 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
4329 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
4330 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
4331 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
4332 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
4333 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
4334 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
4335 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
4336 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
4337 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
4338 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
4339 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
4340 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
4341 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
4342 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
4343 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
4344 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
4345 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
4346 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
4347 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
4348 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
4349 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
4350 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
4351 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
4352 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
4353 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
4354 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
4355 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
4356 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
4357 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
4358 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
4359 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
4360 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
4361 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
4362 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
4363 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
4364 definitions accordingly.
4366 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
4367 Kuan-Lin Chen <kuanlinchentw@gmail.com>
4370 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
4373 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
4376 * ira-conflicts.c (ira_build_conflicts): Update total conflict
4377 hard regs for inner regno.
4379 2017-01-17 Martin Liska <mliska@suse.cz>
4382 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
4383 assumption and add comment.
4385 2017-01-17 Nathan Sidwell <nathan@acm.org>
4387 * ipa-visibility.c (localize_node): New function, broken out of ...
4388 (function_and_variable_visibility): ... here. Call it.
4390 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
4393 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
4394 correctly set frequency of oudgoing edge.
4395 (duplicate_thread_path): Fix profile updating.
4397 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4400 * configure.ac: Add GCC_BASE_VER.
4401 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
4402 version from BASE-VER file.
4403 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
4404 (gcc.o): Depend on $(BASEVER).
4405 * common.opt (dumpfullversion): New option.
4406 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
4407 * doc/invoke.texi: Document -dumpfullversion.
4408 * doc/install.texi: Document --with-gcc-major-version-only.
4409 * configure: Regenerated.
4411 2017-01-17 Richard Biener <rguenther@suse.de>
4413 PR tree-optimization/71433
4414 * tree-vrp.c (register_new_assert_for): Merge same asserts
4415 on all incoming edges.
4416 (process_assert_insertions_for): Handle insertions at the
4419 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
4421 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
4422 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
4424 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
4427 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
4430 2017-01-17 Alan Modra <amodra@gmail.com>
4433 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
4434 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
4435 symbolic stack limit when pic.
4437 2017-01-16 Martin Sebor <msebor@redhat.com>
4439 PR tree-optimization/78608
4440 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
4442 2017-01-16 Jeff Law <law@redhat.com>
4445 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
4446 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
4447 for several include directories that may be relative to sysroot.
4448 * config/i386/x-mingw32 (gplus_includedir): Define.
4449 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
4450 (native_system_includedir): Likewise.
4451 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
4452 override if TARGET_SYSTEM_ROOT is defined.
4453 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
4455 PR tree-optimization/79090
4456 PR tree-optimization/33562
4457 PR tree-optimization/61912
4458 PR tree-optimization/77485
4459 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
4460 and computed trims into the dump file.
4462 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
4464 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
4466 2017-01-16 Jakub Jelinek <jakub@redhat.com>
4469 * gimplify.c (gimplify_init_constructor): If want_value and
4470 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
4474 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
4475 sequence. Formatting fixes.
4476 (doloop_optimize): Formatting fixes.
4479 * gcc.c (debug_level_greater_than_spec_func): New function.
4480 (static_spec_functions): Add debug-level-gt spec function.
4481 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
4483 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
4484 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
4485 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
4486 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
4487 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
4488 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
4490 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
4492 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
4493 QImode fixups to general and mask registers only.
4495 2017-01-16 Carl Love <cel@us.ibm.com>
4497 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
4498 for built-in functions
4499 vector signed char vec_nabs (vector signed char)
4500 vector signed short vec_nabs (vector signed short)
4501 vector signed int vec_nabs (vector signed int)
4502 vector signed long long vec_nabs (vector signed long long)
4503 vector float vec_nabs (vector float)
4504 vector double vec_nabs (vector double)
4505 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
4507 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
4508 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
4509 * doc/extend.texi: Update the documentation file for the new built-in
4512 2017-01-16 Martin Sebor <msebor@redhat.com>
4514 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
4517 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4519 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
4520 UNSPEC_VSX__XXSPLTD to require special splat handling.
4522 2017-01-16 David Malcolm <dmalcolm@redhat.com>
4525 * system.h: Poison strndup.
4527 2017-01-16 Alan Modra <amodra@gmail.com>
4530 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
4533 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
4535 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
4537 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
4539 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
4540 call recog here. Assert that INSN_CODE (insn) is non-negative.
4542 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
4545 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
4547 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
4548 in the currently scheduled RTL fragment.
4550 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
4552 PR rtl-optimization/78751
4553 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
4556 2017-01-14 Jeff Law <law@redhat.com>
4558 PR tree-optimization/79090
4559 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
4560 variable length stores.
4561 (compute_trims): Delete dead assignment to *trim_tail.
4562 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
4565 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
4567 PR rtl-optimization/78626
4568 PR rtl-optimization/78727
4569 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
4570 of a block, and split such blocks after everything else is finished.
4572 2017-01-14 Alan Modra <amodra@gmail.com>
4575 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
4576 target legitimate_combined_insn.
4577 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
4578 (rs6000_legitimate_combined_insn): New function.
4579 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
4581 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
4582 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
4583 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
4585 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
4587 * doc/frontends.texi (G++ and GCC): Remove references to Java.
4589 2017-01-13 Jeff Law <law@redhat.com>
4591 PR tree-optimization/33562
4592 PR tree-optimization/61912
4593 PR tree-optimization/77485
4594 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
4596 (delete_dead_assignment): Likewise.
4597 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
4598 statement to delete_dead_call and delete_dead_assignment.
4600 2017-01-13 David Malcolm <dmalcolm@redhat.com>
4603 * substring-locations.c (format_warning_va): Strengthen case 1 so
4604 that both endpoints of the substring must be within the format
4605 range for just the substring to be printed.
4607 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
4609 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
4610 * config/i386/i386.c (ix86_target_string): Add missing options
4611 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
4612 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
4613 flags_other and ix86_target_other to flags2_other. Display unknown
4615 (ix86_valid_target_attribute_inner_p): Add missing options and
4616 reorder options by implied ISAs, as in ix86_target_string.
4618 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
4620 * hash-table.h (hash_table::too_empty_p): New function.
4621 (hash_table::expand): Use it.
4622 (hash_table::traverse): Likewise.
4623 (hash_table::empty_slot): Use sizeof (value_type) instead of
4624 sizeof (PTR) to convert bytes to elements. Shrink the table
4625 if the current size is excessive for the current number of
4628 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
4630 * ira-costs.c (record_reg_classes): Break from the inner loop
4631 early once alt_fail is known to be true. Update outer loop
4632 handling accordingly.
4634 2017-01-13 Jeff Law <law@redhat.com>
4636 * tree-ssa-dse.c (decrement_count): New function.
4637 (increment_start_addr, maybe_trim_memstar_call): Likewise.
4638 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
4639 when we know the partially dead statement is a mem* function.
4641 PR tree-optimization/61912
4642 PR tree-optimization/77485
4643 * tree-ssa-dse.c: Include expr.h.
4644 (maybe_trim_constructor_store): New function.
4645 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
4647 PR tree-optimization/33562
4648 PR tree-optimization/61912
4649 PR tree-optimization/77485
4650 * doc/invoke.texi: Document new dse-max-object-size param.
4651 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
4652 * tree-ssa-dse.c: Include params.h.
4653 (dse_store_status): New enum.
4654 (initialize_ao_ref_for_dse): New, partially extracted from
4656 (valid_ao_ref_for_dse, normalize_ref): New.
4657 (setup_live_bytes_from_ref, compute_trims): Likewise.
4658 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
4659 (maybe_trim_partially_dead_store): Likewise.
4660 (maybe_trim_complex_store): Likewise.
4661 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
4662 Track what bytes live from the original store. Return tri-state
4663 for dead, partially dead or live.
4664 (dse_dom_walker): Add constructor, destructor and new private members.
4665 (delete_dead_call, delete_dead_assignment): New extracted from
4667 (dse_optimize_stmt): Make a member of dse_dom_walker.
4668 Use initialize_ao_ref_for_dse.
4670 PR tree-optimization/33562
4671 PR tree-optimization/61912
4672 PR tree-optimization/77485
4673 * sbitmap.h (bitmap_count_bits): Prototype.
4674 (bitmap_clear_range, bitmap_set_range): Likewise.
4675 * sbitmap.c (bitmap_clear_range): New function.
4676 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
4678 2017-01-13 Martin Liska <mliska@suse.cz>
4681 * function.c (set_cfun): Add new argument force.
4682 * function.h (set_cfun): Likewise.
4683 * ipa-inline-transform.c (inline_call): Use the function when
4684 strict alising from is dropped for function we inline to.
4686 2017-01-13 Richard Biener <rguenther@suse.de>
4688 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
4689 for dumping GIMPLE INTEGER_CSTs.
4691 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4693 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
4694 to 201112L since C++17.
4696 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
4699 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
4700 if -fsanitize=kernel-address is present.
4702 2017-01-13 Richard Biener <rguenther@suse.de>
4704 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
4705 as _Literal ( type ) number in case usual suffixes do not
4706 preserve all information.
4708 2017-01-13 Richard Biener <rguenther@suse.de>
4710 PR tree-optimization/77283
4711 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
4712 and ssa-iterators.h.
4713 (is_feasible_trace): Implement a cost model based on joiner
4716 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4719 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
4720 char or short to __float128/_Float128 directly.
4722 2017-01-12 Martin Sebor <msebor@redhat.com>
4724 to -Wformat-overflow.
4725 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
4726 (min_bytes_remaining): Same.
4727 (get_string_length): Same.
4728 (format_string): Same.
4729 (format_directive): Same.
4731 (pass_sprintf_length::handle_gimple_call): Same.
4733 2017-01-12 Jakub Jelinek <jakub@redhat.com>
4735 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
4736 info.nowrite calls with no lhs that can't throw. Return bool
4737 whether gsi_remove has been called or not.
4738 (pass_sprintf_length::handle_gimple_call): Return bool whether
4739 try_substitute_return_value called gsi_remove. Formatting fix.
4740 (pass_sprintf_length::execute): Don't use gsi_remove if
4741 handle_gimple_call returned true.
4744 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
4745 be removed due to side-effects, don't remove following barrier nor
4746 turn the successor edge into fallthru edge.
4748 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4751 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
4752 element-reversing loads and stores as not swappable.
4754 2017-01-12 Nathan Sidwell <nathan@acm.org>
4755 Nicolai Stange <nicstange@gmail.com>
4757 * combine.c (try_combine): Don't ignore result of overlap checking
4758 loop. Combine overlap & asm check into single loop.
4760 2017-01-12 Richard Biener <rguenther@suse.de>
4762 * tree-pretty-print.c (dump_generic_node): Provide -gimple
4763 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
4765 2017-01-12 Richard Biener <rguenther@suse.de>
4767 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
4768 and TS_TARGET_OPTION directly derive from TS_BASE.
4769 * tree-core.h (tree_optimization_option): Derive from tree_base.
4770 (tree_target_option): Likewise.
4772 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
4774 * config/i386/i386.c (memory_address_length): Increase len
4775 only when rip_relative_addr_p returns false.
4777 2017-01-11 Julia Koval <julia.koval@intel.com>
4779 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
4780 (OPTION_MASK_ISA_SGX_SET): New.
4781 (ix86_handle_option): Handle OPT_msgx.
4782 * config.gcc: Added sgxintrin.h.
4783 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
4784 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
4785 * config/i386/i386.c (ix86_target_string): Add -msgx.
4787 (ix86_option_override_internal): Handle new options.
4788 (ix86_valid_target_attribute_inner_p): Add sgx.
4789 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
4790 * config/i386/i386.opt: Add msgx.
4791 * config/i386/sgxintrin.h: New file.
4792 * config/i386/x86intrin.h: Add sgxintrin.h.
4794 2017-01-11 Jakub Jelinek <jakub@redhat.com>
4797 * fold-const.c (maybe_nonzero_address): Return 1 for function
4799 (tree_single_nonzero_warnv_p): Don't handle function local objects
4803 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
4806 2017-01-11 David Malcolm <dmalcolm@redhat.com>
4809 * opts.c: Include "spellcheck.h"
4810 (struct string_fragment): New struct.
4811 (struct edit_distance_traits<const string_fragment &>): New
4813 (get_closest_sanitizer_option): New function.
4814 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
4816 2017-01-11 Jakub Jelinek <jakub@redhat.com>
4818 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
4820 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
4821 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
4822 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
4823 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
4824 for initial die_offset if dwarf_split_debug_info.
4825 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
4826 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
4828 (output_skeleton_debug_sections): Formatting fix. Use
4829 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
4830 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
4832 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
4834 * config/arm/cortex-a53.md: Add bypasses for
4836 (cortex_a53_r2f): Only use for transfers.
4837 (cortex_a53_f2r): Likewise.
4838 (cortex_a53_r2f_cvt): Add reservation for conversions.
4839 (cortex_a53_f2r_cvt): Likewise.
4841 2017-01-11 Tamar Christina <tamar.christina@arm.com>
4843 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
4844 to all inlined functions, change static to extern.
4846 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
4849 * config/arm/arm.c (legitimize_pic_address): Handle reference to
4851 (arm_assemble_integer): Likewise.
4853 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
4855 * config.gcc: Use new awk script to check CPU, FPU and architecture
4856 parameters for --with-... options.
4857 * config/arm/parsecpu.awk: New file
4858 * config/arm/arm-cpus.in: New file.
4859 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
4861 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
4863 * config/arm/t-arm: Update dependency rules.
4864 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
4865 of processing .def files.
4866 * config/arm/genopt.sh: Deleted.
4867 * config/arm/gentune.sh: Deleted.
4868 * config/arm/arm-cores.def: Deleted.
4869 * config/arm/arm-arches.def: Deleted.
4870 * config/arm/arm-fpus.def: Deleted.
4871 * config/arm/arm-tune.md: Regenerated.
4872 * config/arm/arm-tables.opt: Regenerated.
4873 * config/arm/arm-cpu.h: New generated file.
4874 * config/arm/arm-cpu-data.h: New generated file.
4875 * config/arm/arm-cpu-cdata.h: New generated file.
4877 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
4880 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
4882 (input_varpool_node): Unpack dynamically_initialized bit.
4884 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
4886 PR rtl-optimization/79032
4887 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
4888 the alignment of the adjusted memory reference against that of MODE,
4889 instead of the alignment of the original memory reference.
4891 2017-01-11 Martin Jambor <mjambor@suse.cz>
4893 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
4895 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
4896 decorated functions.
4898 2017-01-11 Richard Biener <rguenther@suse.de>
4900 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
4901 set range/nonnull info for PHI results. Do not set it on
4902 stmts marked for removal.
4904 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
4906 * expr.c (store_field): In the bitfield case, fetch the return value
4907 from the registers before applying a single big-endian adjustment.
4908 Always do a final load for a BLKmode value not larger than a word.
4910 2017-01-10 David Malcolm <dmalcolm@redhat.com>
4913 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
4914 that we correctly handle column numbers greater than
4915 LINE_MAP_MAX_COLUMN_NUMBER.
4917 2017-01-10 Martin Sebor <msebor@redhat.com>
4920 * gimple-ssa-sprintf.c (get_destination_size): Call
4921 {init,fini}object_sizes.
4922 * tree-object-size.c (addr_object_size): Adjust.
4923 (pass_through_call): Adjust.
4924 (pass_object_sizes::execute): Adjust.
4925 * tree-object-size.h (fini_object_sizes): Declare.
4927 2017-01-10 Martin Sebor <msebor@redhat.com>
4929 PR tree-optimization/78775
4930 * builtins.c (get_size_range): Move...
4931 * calls.c: ...to here.
4932 (alloc_max_size): Accept zero argument.
4933 (operand_signed_p): Remove.
4934 (maybe_warn_alloc_args_overflow): Call get_size_range.
4935 * calls.h (get_size_range): Declare.
4937 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
4939 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
4940 from TI's devices.csv file as of September 2016.
4941 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
4943 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
4945 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
4946 * doc/invoke.texi: Likewise.
4947 * doc/md.texi: Likewise.
4948 * doc/objc.texi: Likewise.
4950 2017-01-10 Joshua Conner <joshconner@google.com>
4952 * config/arm/fuchsia-elf.h: New file.
4953 * config/fuchsia.h: New file.
4954 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
4955 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
4957 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
4959 2016-01-10 Richard Biener <rguenther@suse.de>
4961 PR tree-optimization/79034
4962 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
4963 Propagate out degenerate PHIs in the joiner.
4965 2017-01-10 Martin Liska <mliska@suse.cz>
4967 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
4968 (sort_congruence_classes_by_decl_uid): Likewise.
4969 (sort_congruence_class_groups_by_decl_uid): Likewise.
4970 (sem_item_optimizer::merge_classes): Sort class, groups in these
4971 classes and members in the groups by DECL_UID of declarations.
4972 This would make merge operations stable.
4974 2017-01-10 Martin Liska <mliska@suse.cz>
4976 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
4977 usage of m_classes_vec.
4978 (sem_item_optimizer::~sem_item_optimizer): Likewise.
4979 (sem_item_optimizer::get_group_by_hash): Likewise.
4980 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
4981 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
4982 (sem_item_optimizer::verify_classes): Likewise.
4983 (sem_item_optimizer::process_cong_reduction): Likewise.
4984 (sem_item_optimizer::dump_cong_classes): Likewise.
4985 (sem_item_optimizer::merge_classes): Likewise.
4986 * ipa-icf.h (congruence_class_hash): Rename from
4987 congruence_class_group_hash. Remove declaration of m_classes_vec.
4989 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
4991 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
4992 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
4993 * config.gcc: Add avx512vpopcntdqintrin.h.
4994 * config/i386/avx512vpopcntdqintrin.h: New.
4995 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
4996 * config/i386/i386-builtin-types.def: Add new types.
4997 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
4998 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
4999 __builtin_ia32_vpopcountq_v8di_mask): New.
5000 * config/i386/i386-c.c (ix86_target_macros_internal): Define
5001 __AVX512VPOPCNTDQ__.
5002 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
5003 (PTA_AVX512VPOPCNTDQ): Define.
5004 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
5005 TARGET_AVX512VPOPCNTDQ_P): Define.
5006 * config/i386/i386.opt: Add mavx512vpopcntdq.
5007 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
5008 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
5010 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5013 * predict.def (PRED_CALL): Set to 67.
5015 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
5017 * expr.c (store_field): In the bitfield case, if the value comes from
5018 a function call and is of an aggregate type returned in registers, do
5019 not modify the field mode; extract the value in all cases if the mode
5020 is BLKmode and the size is not larger than a word.
5022 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
5025 * config/i386/cpuid.h: Fix undefined behavior.
5027 2017-01-04 Jeff Law <law@redhat.com>
5029 PR tree-optimization/79007
5030 PR tree-optimization/67955
5031 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
5032 conservative for pt.null when flag_non_call_exceptions is on.
5034 2017-01-09 Jakub Jelinek <jakub@redhat.com>
5036 PR translation/79019
5037 PR translation/79020
5038 * params.def (PARAM_INLINE_MIN_SPEEDUP,
5039 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
5040 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
5042 * config/avr/avr.opt (maccumulate-args): Likewise.
5043 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
5044 * common.opt (freport-bug): Likewise.
5045 * cif-code.def (CIF_FINAL_ERROR): Likewise.
5046 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
5047 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
5048 translatable string.
5049 * config/i386/i386.c (function_value_32): Likewise.
5050 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
5051 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
5053 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
5054 * common/config/msp430/msp430-common.c (msp430_handle_option):
5056 * symtab.c (symtab_node::verify_base): Likewise.
5057 * opts.c (set_debug_level): Likewise.
5058 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
5059 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
5060 missing whitespace to translatable strings.
5061 * config/avr/avr.md (bswapsi2): Fix typo in comment.
5062 * config/sh/superh.h: Likewise.
5063 * config/i386/xopintrin.h: Likewise.
5064 * config/i386/znver1.md: Likewise.
5065 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
5066 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
5067 * double-int.h (struct double_int): Likewise.
5068 * double-int.c (div_and_round_double): Likewise.
5069 * wide-int.cc: Likewise.
5070 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
5071 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
5072 * cfgcleanup.c (crossjumps_occured): Renamed to ...
5073 (crossjumps_occurred): ... this.
5074 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
5077 PR tree-optimization/78899
5078 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
5079 returning bool return struct loop *, NULL for failure and the new
5081 (versionable_outer_loop_p): Don't version outer loop if it has
5082 dont_vectorized bit set.
5083 (tree_if_conversion): When versioning outer loop, ensure
5084 tree_if_conversion is performed also on the inner loop of the
5085 non-vectorizable outer loop copy.
5086 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
5087 LOOP_VECTORIZED in inner loop of the scalar outer loop and
5088 prevent vectorization of it.
5089 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
5090 the outer loop vectorization of the non-scalar version is attempted
5091 before vectorization of the inner loop in scalar version. If
5092 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
5093 vectorization of its inner loop.
5094 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
5095 has 2 inner loops, rename also on edges from bb whose single pred
5096 is outer_loop->header. Fix typo in function comment.
5098 2017-01-09 Martin Sebor <msebor@redhat.com>
5101 * asan.c (asan_emit_stack_protection): Increase local buffer size
5102 to avoid snprintf truncation warning.
5104 2017-01-09 Andrew Pinski <apinski@cavium.com>
5106 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
5107 to reference thunderx2t99 for the tuning structure
5108 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
5110 (thunderx2t99_extra_costs): This.
5111 * config/aarch64/aarch64-tune.md: Regenerate.
5112 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
5113 (vulcan_addrcost_table): This.
5114 (vulcan_regmove_cost): Rename to ...
5115 (thunderx2t99_regmove_cost): This.
5116 (vulcan_vector_cost): Rename to ...
5117 (thunderx2t99_vector_cost): this.
5118 (vulcan_branch_cost): Rename to ...
5119 (thunderx2t99_branch_cost): This.
5120 (vulcan_tunings): Rename to ...
5121 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
5122 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
5124 2017-01-09 Martin Jambor <mjambor@suse.cz>
5128 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
5129 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
5130 (propagate_vr_accross_jump_function): Use the above function for all
5131 value range computations for pass-through jump functions and type
5132 converasion from explicit value range values.
5133 (ipcp_propagate_stage): Do not attempt to deduce types of formal
5134 parameters from TYPE_ARG_TYPES.
5135 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
5136 (ipa_write_node_info): Stream type of the actual argument.
5137 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
5139 2017-01-09 Martin Liska <mliska@suse.cz>
5142 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
5143 (lookup_compiler): Do not show error message with have_E.
5145 2017-01-09 Jakub Jelinek <jakub@redhat.com>
5147 PR tree-optimization/78938
5148 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
5149 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
5150 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
5151 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
5154 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5156 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
5159 2017-01-09 Richard Biener <rguenther@suse.de>
5161 PR tree-optimization/78997
5162 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
5163 name condition properly.
5165 2017-01-09 Richard Biener <rguenther@suse.de>
5168 * dwarf2out.c (is_cxx): New overload with context.
5169 (is_naming_typedef_decl): Use it.
5171 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5173 * invoke.texi (Option Summary): Correct spacing in option lists
5174 and add line breaks to fix over-long lines.
5176 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5180 * extend.texi (Common Variable Attributes): Add xref to GCC
5181 Internals manual to explain mode attribute keywords.
5183 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5186 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
5187 and Preprocessor Options.
5188 (Options for Linking): Document -pthread here....
5189 (RS/6000 and PowerPC Options): ...not here.
5190 (Solaris 2 Options): ...or here.
5191 * doc/cppopts.texi: Document -pthread.
5193 2017-01-08 Martin Sebor <msebor@redhat.com>
5196 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
5197 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
5198 New member functions.
5199 (format_directive): Used them.
5201 (pass_sprintf_length::handle_gimple_call): Same.
5202 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
5203 to avoid truncation for any argument.
5204 (extract_affine_mul): Same.
5205 * tree.c (get_file_function_name): Same.
5207 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5210 * predict.def (PRED_INDIR_CALL): Set to 86.
5212 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5214 PR preprocessor/54124
5215 * doc/cppopts.texi: Reformat -d subtable to list the full name
5216 of the options. Add cross-reference to the docs for the general
5217 compiler -d options.
5218 * doc/invoke.texi (Developer Options): Add cross-reference to the
5219 preprocessor-specific -d option documentation.
5221 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5223 PR preprocessor/13498
5224 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
5225 redudant material, and reflect new command-line options.
5226 (System Headers): Likewise.
5228 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5230 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
5231 -isystem, and -idirafter. Copy-edit.
5232 * doc/cppopts.texi: Copy-edit. Remove contradiction about
5233 default for -ftrack-macro-expansion. Delete obsolete and
5234 badly-formatted implementation details about -fdebug-cpp output.
5235 * doc/cppwarnopts.texi: Copy-edit.
5237 2017-01-07 David Malcolm <dmalcolm@redhat.com>
5240 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
5241 that the transition from a max line width >= 1<<10 to narrower
5242 lines works correctly.
5244 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
5246 * doc/options.texi (PerFunction): New.
5247 * opt-functions.awk (switch_flags): Map both Optimization and
5248 PerFunction to CL_OPTIMIZATION.
5249 * opth-gen.awk: Test for PerFunction flag along with
5251 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
5252 it only when the latter is present. Skip those that don't in
5253 the hash function generator.
5254 * common.opt (fvar-tracking): Mark as PerFunction instead of
5256 (fvar-tracking-assignments): Likewise.
5257 (fvar-tracking-assignments-toggle): Likewise.
5258 (fvar-tracking-uninit): Likewise.
5260 2017-01-07 Jakub Jelinek <jakub@redhat.com>
5262 PR translation/79018
5263 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
5266 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
5269 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
5270 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
5271 TARGET_LONG_JUMP_TABLE_OFFSETS.
5272 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
5273 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
5274 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
5275 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
5276 * config/m68k/m68k.md (tablejump expander): Likewise.
5277 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
5278 TARGET_LONG_JUMP_TABLE_OFFSETS.
5279 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
5280 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
5282 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
5283 David Holsgrove <david.holsgrove@xilinx.com>
5285 * common/config/microblaze/microblaze-common.c
5286 (TARGET_EXCEPT_UNWIND_INFO): Remove.
5287 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
5289 * config/microblaze/microblaze.c (microblaze_must_save_register)
5290 (microblaze_expand_epilogue, microblaze_return_addr): Handle
5292 (microblaze_eh_return): New function.
5293 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
5294 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
5295 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
5296 * config/microblaze/microblaze.md (eh_return): New pattern.
5298 2017-01-06 Jakub Jelinek <jakub@redhat.com>
5300 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
5301 GCC_DIAGNOSTIC_STRINGIFY): Define.
5303 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
5305 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5307 * config/arm/arm.md (<mcrr>): New.
5309 * config/arm/arm.c (arm_arch5te): New.
5310 (arm_option_override): Set arm_arch5te.
5311 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
5313 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
5314 (arm_mcrr_qualifiers): ... this. New.
5315 (MRRC_QUALIFIERS): Define to...
5316 (arm_mrrc_qualifiers): ... this. New.
5317 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
5319 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
5320 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
5321 (MRRCI, mrrc, MRRC): New.
5322 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
5323 VUNSPEC_MRRC2): New.
5325 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5327 * config/arm/arm.md (<mcr>): New.
5329 * config/arm/arm.c (arm_coproc_builtin_available): Add
5330 support for mcr, mrc, mcr2 and mrc2.
5331 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
5332 (arm_mcr_qualifiers): ... this. New.
5333 (MRC_QUALIFIERS): Define to ...
5334 (arm_mrc_qualifiers): ... this. New.
5335 (MCR_QUALIFIERS): Define to ...
5336 (arm_mcr_qualifiers): ... this. New.
5337 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
5339 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
5340 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
5341 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
5344 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5346 * config/arm/arm.md (*ldc): New.
5350 * config/arm/arm.c (arm_coproc_builtin_available): Add
5351 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
5352 (arm_coproc_ldc_stc_legitimate_address): New.
5353 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
5354 'qualifier_const_pointer'.
5355 (LDC_QUALIFIERS): Define to...
5356 (arm_ldc_qualifiers): ... this. New.
5357 (STC_QUALIFIERS): Define to...
5358 (arm_stc_qualifiers): ... this. New.
5359 * config/arm/arm-protos.h
5360 (arm_coproc_ldc_stc_legitimate_address): New.
5361 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
5362 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
5363 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
5364 stc2, stcl, stc2l): New.
5365 * config/arm/constraints.md (Uz): New.
5366 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
5367 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
5368 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
5369 VUNSPEC_STC2L): New.
5371 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5373 * config/arm/arm.md (<cdp>): New.
5374 * config/arm/arm.c (neon_const_bounds): Rename this ...
5375 (arm_const_bounds): ... this.
5376 (arm_coproc_builtin_available): New.
5377 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
5378 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
5379 (CDP_QUALIFIERS): Define to...
5380 (arm_cdp_qualifiers): ... this. New.
5382 (arm_expand_builtin_args): Add case for 6 arguments.
5383 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
5384 (arm_const_bounds): ... this.
5385 (arm_coproc_builtin_available): New.
5386 * config/arm/arm_acle.h (__arm_cdp): New.
5388 * config/arm/arm_acle_builtins.def (cdp): New.
5390 * config/arm/iterators.md (CDPI,CDP,cdp): New.
5391 * config/arm/neon.md: Rename all 'neon_const_bounds' to
5393 * config/arm/types.md (coproc): New.
5394 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
5395 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
5396 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
5397 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
5399 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5401 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
5402 (UBINOP_QUALIFIERS): New.
5404 (acle_builtin_data): New. Change comment.
5405 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
5406 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
5407 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
5408 arm_acle_builtins.def.
5409 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
5410 (arm_init_acle_builtins): New.
5411 (CRC32_BUILTIN): Remove.
5412 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
5413 crc32cb, crc32ch and crc32cw.
5414 (arm_init_crc32_builtins): Remove.
5415 (arm_init_builtins): Use arm_init_acle_builtins rather
5416 than arm_init_crc32_builtins.
5417 (arm_expand_acle_builtin): New.
5418 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
5419 * config/arm/arm_acle_builtins.def: New.
5421 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5423 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
5424 (arm_builtin_datum): ... this.
5425 (arm_init_neon_builtin): Rename to ...
5426 (arm_init_builtin): ... this. Add a new parameters PREFIX
5427 and USE_SIG_IN_NAME.
5428 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
5429 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
5430 'arm_builtin_datum'.
5431 (arm_init_vfp_builtins): Likewise.
5432 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
5433 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
5434 (arm_expand_neon_args): Rename to ...
5435 (arm_expand_builtin_args): ... this. Rename builtin_arg
5436 enum values and differentiate between ARG_BUILTIN_MEMORY
5437 and ARG_BUILTIN_NEON_MEMORY.
5438 (arm_expand_neon_builtin_1): Rename to ...
5439 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
5440 values, arm_expand_builtin_args and add bool parameter NEON.
5441 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
5442 (arm_expand_vfp_builtin): Likewise.
5443 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
5445 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5448 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
5449 * predict.c (tree_estimate_probability_bb): Reverse direction of
5450 polymorphic call predictor.
5452 2017-01-06 David Malcolm <dmalcolm@redhat.com>
5454 * passes.c (execute_one_pass): Split out pass-skipping logic into...
5455 (determine_pass_name_match): ...this new function and...
5456 (should_skip_pass_p): ...this new function.
5458 2017-01-06 Nathan Sidwell <nathan@acm.org>
5460 * ipa-visibility.c (function_and_variable_visibility): Reformat
5461 comments and long lines. Remove extrneous if.
5462 * symtab.c (symtab_node::make_decl_local): Fix code format.
5463 (symtab_node::set_section_for_node): Fix comment typo.
5465 2017-01-06 Martin Liska <mliska@suse.cz>
5468 * lra-constraints.c: Rename invariant to lra_invariant.
5469 * predict.c (set_even_probabilities): Initialize e to NULL.
5471 2017-01-05 Martin Sebor <msebor@redhat.com>
5473 PR tree-optimization/78910
5474 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
5475 (format_integer): Correct off-by-one error in the handling
5476 of precision with negative numbers in signed conversions..
5478 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
5480 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
5482 2017-01-05 Jakub Jelinek <jakub@redhat.com>
5484 PR tree-optimization/71016
5485 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
5486 factor_out_conditional_conversion. Formatting fix.
5487 (factor_out_conditional_conversion): Add cond_stmt argument.
5488 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
5489 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
5492 2017-01-05 David Malcolm <dmalcolm@redhat.com>
5494 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
5495 read-rtl-function.o, and selftest-rtl.o.
5496 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
5497 (selftest::aarch64_test_loading_full_dump): New function.
5498 (selftest::aarch64_run_selftests): New function.
5499 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
5500 selftest::aarch64_run_selftests.
5501 * config/i386/i386.c
5502 (selftest::ix86_test_loading_dump_fragment_1): New function.
5503 (selftest::ix86_test_loading_call_insn): New function.
5504 (selftest::ix86_test_loading_full_dump): New function.
5505 (selftest::ix86_test_loading_unspec): New function.
5506 (selftest::ix86_run_selftests): Call the new functions.
5507 * emit-rtl.c (maybe_set_max_label_num): New function.
5508 * emit-rtl.h (maybe_set_max_label_num): New decl.
5509 * function.c (instantiate_decls): Guard call to
5510 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
5511 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
5513 * gensupport.c (gen_reader::gen_reader): Pass "false"
5514 for new "compact" param of rtx_reader.
5515 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
5516 rather than an empty string for NULL strings.
5517 * read-md.c: Potentially include config.h rather than bconfig.h.
5518 Wrap include of errors.h with #ifdef GENERATOR_FILE.
5519 (have_error): New global, copied from errors.c.
5520 (md_reader::read_name): Rename to...
5521 (md_reader::read_name_1): ...this, adding "out_loc" param,
5522 and converting "missing name or number" to returning false, rather
5524 (md_reader::read_name): Reimplement in terms of read_name_1.
5525 (md_reader::read_name_or_nil): New function.
5526 (md_reader::read_string): Handle "(nil)" by returning NULL.
5527 (md_reader::md_reader): Add new param "compact".
5528 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
5529 (md_reader::read_file): New method.
5530 * read-md.h (md_reader::md_reader): Add new param "compact".
5531 (md_reader::read_file): New method.
5532 (md_reader::is_compact): New accessor.
5533 (md_reader::read_name): Convert return type from void to file_location.
5534 (md_reader::read_name_or_nil): New decl.
5535 (md_reader::read_name_1): New decl.
5536 (md_reader::m_compact): New field.
5537 (noop_reader::noop_reader): Pass "false" for new "compact" param
5539 (rtx_reader::rtx_reader): Add new "compact" param.
5540 (rtx_reader::read_rtx_operand): Make virtual and convert return
5541 type from void to rtx.
5542 (rtx_reader::read_until): New decl.
5543 (rtx_reader::handle_any_trailing_information): New virtual function.
5544 (rtx_reader::postprocess): New virtual function.
5545 (rtx_reader::finalize_string): New virtual function.
5546 (rtx_reader::m_in_call_function_usage): New field.
5547 (rtx_reader::m_reuse_rtx_by_id): New field.
5548 * read-rtl-function.c: New file.
5549 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
5550 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
5551 (selftest::verify_three_block_rtl_cfg): New decl.
5552 * read-rtl-function.h: New file.
5553 * read-rtl.c: Potentially include config.h rather than bconfig.h.
5554 For host, include function.h, memmodel.h, and emit-rtl.h.
5555 (one_time_initialization): New function.
5556 (struct compact_insn_name): New struct.
5557 (compact_insn_names): New array.
5558 (find_code): Handle insn codes in compact dumps.
5559 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
5560 (bind_subst_iter_and_attr): Likewise.
5561 (add_condition_to_string): Likewise.
5562 (add_condition_to_rtx): Likewise.
5563 (apply_attribute_uses): Likewise.
5564 (add_current_iterators): Likewise.
5565 (apply_iterators): Likewise.
5566 (initialize_iterators): Guard usage of apply_subst_iterator with
5567 #ifdef GENERATOR_FILE.
5568 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
5569 (md_reader::read_mapping): Likewise.
5570 (add_define_attr_for_define_subst): Likewise.
5571 (add_define_subst_attr): Likewise.
5572 (read_subst_mapping): Likewise.
5573 (check_code_iterator): Likewise.
5574 (rtx_reader::read_rtx): Likewise. Move one-time initialization
5576 (one_time_initialization): New function.
5577 (rtx_reader::read_until): New method.
5578 (read_flags): New function.
5579 (parse_reg_note_name): New function.
5580 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
5581 Handle reuse_rtx ids.
5582 Wrap iterator lookup within #ifdef GENERATOR_FILE.
5583 Add parsing support for RTL dumps, mirroring the special-cases in
5584 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
5585 values, and calling handle_any_trailing_information.
5586 (rtx_reader::read_rtx_operand): Convert return type from void
5587 to rtx, returning return_rtx. Handle case 'e'. Call
5588 finalize_string on XSTR and XTMPL fields.
5589 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
5590 "(nil)" values were omitted. Call the postprocess vfunc on the
5592 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
5593 class ctor. Initialize m_in_call_function_usage. Call
5594 one_time_initialization.
5595 * rtl-tests.c (selftest::test_uncond_jump): Call
5596 set_new_first_and_last_insn.
5597 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
5598 * selftest-rtl.c: New file.
5599 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
5600 (selftest::get_insn_by_uid): New decl.
5601 * selftest-run-tests.c (selftest::run_tests): Call
5602 read_rtl_function_c_tests.
5603 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
5604 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
5607 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
5609 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
5610 operands in a special way. Assert that pos+len <= mode precision.
5612 2017-01-05 Jakub Jelinek <jakub@redhat.com>
5614 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
5615 3 argument Alias with unlimited for the negative form.
5616 (fno-vect-cost-model): Removed.
5618 2017-01-05 Martin Liska <mliska@suse.cz>
5620 * hsa-gen.c (gen_hsa_divmod): New function.
5621 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
5623 2017-01-05 Martin Liska <mliska@suse.cz>
5626 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
5629 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5631 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
5632 small constant length operands.
5634 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5636 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
5637 between loop iterations.
5639 2017-01-05 Martin Liska <mliska@suse.cz>
5642 * gimplify.c (gimplify_decl_expr): Compare to
5643 asan_poisoned_variables instread of checking flags.
5644 (gimplify_target_expr): Likewise.
5645 (gimplify_expr): Likewise.
5646 (gimplify_function_tree): Conditionally initialize
5647 asan_poisoned_variables.
5649 2017-01-04 Jeff Law <law@redhat.com>
5651 PR tree-optimizatin/78812
5652 * rtl.h (contains_mem_rtx_p): Prototype.
5653 * ifcvt.c (containts_mem_rtx_p): Move from here to...
5654 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
5655 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
5656 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
5657 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
5659 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5661 * input.c (assert_char_at_range): Default-initialize actual_range.
5663 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5665 * df-scan.c (df_ref_create_structure): Make regno unsigned,
5666 to match the caller.
5668 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5670 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
5671 insns after final jump in test to emit dummy move.
5673 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5675 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
5676 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
5678 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5680 * multiple_target.c (create_dispatcher_calls): Init e_next.
5681 * tree-ssa-loop-split.c (split_loop): Init border.
5682 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
5685 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
5690 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
5691 (altivec_register_operand): Do not return true if the operand
5692 contains a SUBREG mixing SImode and SFmode.
5693 (vsx_register_operand): Likewise.
5694 (vsx_reg_sfsubreg_ok): New predicate.
5695 (vfloat_operand): Do not return true if the operand contains a
5696 SUBREG mixing SImode and SFmode.
5697 (vint_operand): Likewise.
5698 (vlogical_operand): Likewise.
5699 (gpc_reg_operand): Likewise.
5700 (int_reg_operand): Likewise.
5701 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
5702 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
5703 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
5705 (rs6000_emit_move_si_sf_subreg): New helper function.
5706 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
5707 fixup SUBREGs involving SImode and SFmode.
5708 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
5709 numbers for the new peephole2 optimization.
5710 (peephole2 for SFmode unions): New peephole2 to optimize cases in
5711 the GLIBC math library that do AND/IOR/XOR operations on single
5712 precision floating point.
5713 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
5714 target macros to say whether we need to avoid SUBREGs mixing
5716 (TARGET_ALLOW_SF_SUBREG): Likewise.
5717 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
5718 (UNSPEC_SI_FROM_SF): Likewise.
5719 (iorxor): Change spacing.
5720 (and_ior_xor): New iterator for AND, IOR, and XOR.
5721 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
5722 (movdi_from_sf_zero_ext): Likewise.
5723 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
5724 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
5725 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
5726 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
5727 (fms<mode>4): Likewise.
5728 (fnma<mode>4): Likewise.
5729 (fnms<mode>4): Likewise.
5730 (nfma<mode>4): Likewise.
5731 (nfms<mode>4): Likewise.
5733 2017-01-04 Marek Polacek <polacek@redhat.com>
5736 * doc/invoke.texi: Document -Wpointer-compare.
5738 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5740 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
5743 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
5744 descriptions for -gdwarf-5 and emit them as uleb128 instead of
5747 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
5750 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
5751 documentation of the powerpc_popcntb_ok attribute.
5752 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
5753 code to issue warning messages if a requested CPU configuration is
5754 not supported by the binary (assembler and loader) toolchain.
5755 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
5756 made to define a built-in function that has been disabled.
5757 (paired_init_builtins): Add assertion to prevent ICE if attempt is
5758 made to define a built-in function that has been disabled.
5759 (altivec_init_builtins): Add comment explaining why definition
5760 of the DST built-in functions is not preceded by an assertion
5761 check. Add assertions to prevent ICE if attempts are made to
5762 define an altivec predicate or an abs* built-in function that has
5764 (htm_init_builtins): Add comment explaining why definition of the
5765 htm built-in functions is not preceded by an assertion check.
5767 2017-01-04 Jeff Law <law@redhat.com>
5769 PR tree-optimizatin/67955
5770 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
5771 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
5772 the points-to solution does not include pt_null. Use DECL_PT_UID
5775 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
5777 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
5778 Use gen_int_mode instead of gen_lopwart for const_int operands.
5780 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5782 PR tree-optimization/71563
5783 * match.pd: Simplify X << Y into X if Y is known to be 0 or
5784 out of range value - has low bits known to be zero.
5786 2017-01-04 Alan Modra <amodra@gmail.com>
5788 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
5789 * configure: Regenerate.
5790 * config.in: Regenerate.
5792 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5795 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
5796 a substring of the message, but strcmp with the whole message. Ifdef
5797 ENABLE_NLS, translate the message first using dgettext.
5799 2017-01-03 Jeff Law <law@redhat.com>
5801 PR tree-optimizatin/78856
5802 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
5803 (mark_threaded_blocks): Remove code to truncate thread paths that
5804 cross multiple loop headers. Instead invalidate the cached loop
5805 iteration information and handle case of a thread path walking
5806 into an irreducible region.
5808 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5811 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
5812 assertions. Add support for doing the signbit if the IEEE 128-bit
5813 floating point value is in a GPR.
5814 * config/rs6000/rs6000.md (Fsignbit): Delete.
5815 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
5816 Update the length attribute if the value is in a GPR.
5817 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
5818 the sign or zero extension instruction, since the value is always 0/1.
5819 (signbit<mode>2_dm2): Delete using <Fsignbit>.
5822 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
5823 extracting SImode to a GPR register so that we can generate a
5824 store, limit the vector to be in a traditional Altivec register
5825 for the vextuwrx instruction.
5827 2017-01-03 Ian Lance Taylor <iant@google.com>
5829 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
5831 2017-01-03 Martin Sebor <msebor@redhat.com>
5833 PR tree-optimization/78696
5834 * gimple-ssa-sprintf.c (format_floating): Correct handling of
5835 precision. Use MPFR for %f for greater fidelity. Correct handling
5837 (pass_sprintf_length::compute_format_length): Set width and precision
5838 specified by asrerisk to void_node for vararg functions.
5839 (try_substitute_return_value): Adjust dump output.
5841 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
5843 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
5845 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
5847 * doc/invoke.texi (SPARC options): Document -mlra as the default.
5848 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
5849 -mlra/-mno-lra was passed to the compiler.
5851 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
5853 PR rtl-optimization/65618
5854 * emit-rtl.c (try_split): Move initialization of "before" and
5855 "after" to just before the call to emit_insn_after_setloc.
5857 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
5859 * doc/md.texi (Standard Names): Remove reference to Java frontend.
5861 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
5863 * dwarf2out.c (gen_enumeration_type_die): When
5864 -gno-strict-dwarf, add a DW_AT_encoding attribute.
5866 2017-01-03 Jakub Jelinek <jakub@redhat.com>
5868 PR tree-optimization/78965
5869 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
5870 Change first argument from const call_info & to call_info &. For %n
5871 set info.nowrite to false.
5874 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
5875 possibly throwing calls.
5877 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
5878 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
5879 and fns handling, rather than in a separate case SSA_NAME.
5881 2017-01-02 Jeff Law <law@redhat.com>
5883 * config/darwin-driver.c (darwin_driver_init): Const-correctness
5884 fixes for first_period and second_period variables.
5886 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
5889 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
5890 (*insvqi_1): New insn pattern.
5891 (*insvqi_1_mem_rex64): Ditto.
5893 (*insvqi_3): Rename from *insvqi.
5895 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
5897 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
5899 * doc/cfg.texi (Edges): Remove reference to Java.
5900 (Maintaining the CFG): Ditto.
5902 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5905 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
5906 transparent aliases.
5908 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5911 * predict.def (PRED_CALL): Update hitrate.
5912 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
5913 * predict.c (tree_estimate_probability_bb): Split CALL predictor
5914 into direct/indirect/polymorphic variants.
5916 2017-01-01 Jakub Jelinek <jakub@redhat.com>
5918 Update copyright years.
5920 * gcc.c (process_command): Update copyright notice dates.
5921 * gcov-dump.c (print_version): Ditto.
5922 * gcov.c (print_version): Ditto.
5923 * gcov-tool.c (print_version): Ditto.
5924 * gengtype.c (create_file): Ditto.
5925 * doc/cpp.texi: Bump @copying's copyright year.
5926 * doc/cppinternals.texi: Ditto.
5927 * doc/gcc.texi: Ditto.
5928 * doc/gccint.texi: Ditto.
5929 * doc/gcov.texi: Ditto.
5930 * doc/install.texi: Ditto.
5931 * doc/invoke.texi: Ditto.
5933 Copyright (C) 2017 Free Software Foundation, Inc.
5935 Copying and distribution of this file, with or without modification,
5936 are permitted in any medium without royalty provided the copyright
5937 notice and this notice are preserved.