1 2017-03-13 Jeff Law <law@redhat.com>
3 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
4 match_scratch operand is highest.
6 2017-03-13 Martin Liska <mliska@suse.cz>
9 * ipa-pure-const.c (warn_function_noreturn): If the declarations
10 is a CHKP clone, use original declaration.
12 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
14 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
15 (arc_conditional_register_usage): Use a different allocation order
16 when optimizing for size.
17 * common/config/arc/arc-common.c (arc_option_optimization_table):
18 Section anchors default on when optimizing for size.
20 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
24 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
27 * config/arc/arc.md (cpu_facility): Add cd variant.
28 (*movqi_insn): Add code density variant.
29 (*movhi_insn): Likewise.
30 (*movqi_insn): Likewise.
31 (*addsi3_mixed): Likewise.
32 (subsi3_insn): Likewise.
34 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36 * config/arc/arc.md (movsi_cond_exec): Update constraint.
38 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
40 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
41 expressions with MINUS and UNARY ops.
43 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
46 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
48 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
49 between vec_select and vector argument.
50 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
51 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
52 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
53 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
54 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
55 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
57 2017-03-13 Richard Biener <rguenther@suse.de>
60 * params.def (vect-max-peeling-for-alignment): Fix typo.
62 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
64 * doc/install.texi (Specific) <mips-*-*>: Remove description of
65 issue that only occurred with binutils below 2.18.
67 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
69 * doc/install.texi (Specific) <cris-axis-elf>: No longer
70 refer to binutils 2.11/2.12 minimum.
72 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
74 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
75 ftp.kernel.org and simplify binutils requirement.
77 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
79 * doc/invoke.texi (Warning Options): Fix spelling of link-time
81 (Optimize Options): Ditto. Also remove redundancy.
83 2017-03-10 David Malcolm <dmalcolm@redhat.com>
86 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
88 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
89 to G_ to avoid double translation.
91 2017-03-10 David Malcolm <dmalcolm@redhat.com>
94 * auto-profile.c (get_combined_location): Convert leading
95 character of diagnostics to lower case and remove trailing period.
96 (read_profile): Likewise for various diagnostics.
97 * config/arm/arm.c (arm_option_override): Remove trailing period
98 from various diagnostics.
99 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
100 (msp430_expand_delay_cycles): Likewise.
102 2017-03-10 David Malcolm <dmalcolm@redhat.com>
105 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
106 full command-line argument, rather than just "str".
107 (aarch64_validate_march): Likewise.
108 (aarch64_validate_mtune): Likewise.
110 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
112 PR rtl-optimization/78911
113 * lra-assigns.c (must_not_spill_p): New function.
116 2017-03-10 Jakub Jelinek <jakub@redhat.com>
118 PR tree-optimization/79981
119 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
120 ATOMIC_COMPARE_EXCHANGE ifn result.
121 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
122 IFN_ATOMIC_COMPARE_EXCHANGE.
124 2017-03-10 David Malcolm <dmalcolm@redhat.com>
127 * opts.c (parse_sanitizer_options): Add missing question mark to
128 "did you mean" message.
130 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
132 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
134 (VMULEUH_UNS): Likewise.
135 (VMULOUB_UNS): Likewise.
136 (VMULOUH_UNS): Likewise.
137 * config/rs6000/rs6000.c (builtin_function_type): Remove
138 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
140 2017-03-10 David Malcolm <dmalcolm@redhat.com>
143 * read-rtl-function.c (function_reader::read_rtx_operand): Update
144 x with result of extra_parsing_for_operand_code_0.
145 (function_reader::extra_parsing_for_operand_code_0): Convert
146 return type from void to rtx, returning x. When reading
147 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
148 larger size containing struct block_symbol.
150 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
152 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
153 -mfloat128-hardware without -m64.
155 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
158 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
159 entries to the case statement that marks unsigned arguments to
160 overloaded functions.
162 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
164 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
165 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
167 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
170 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
171 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
173 2017-03-10 Martin Liska <mliska@suse.cz>
177 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
178 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
181 2017-03-10 Olivier Hainque <hainque@adacore.com>
183 * tree-switch-conversion (array_value_type): Start by resetting
184 candidate type to it's main variant.
186 2017-03-10 Jakub Jelinek <jakub@redhat.com>
188 PR rtl-optimization/79909
189 * combine.c (try_combine): Use simplify_replace_rtx on individual
190 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
191 of the whole CALL_INSN_FUNCTION_USAGE.
193 PR tree-optimization/79972
194 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
195 get_range_info on SSA_NAMEs. Formatting fixes.
197 2017-03-10 Richard Biener <rguenther@suse.de>
198 Jakub Jelinek <jakub@redhat.com>
200 PR tree-optimization/77975
201 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
203 (get_val_for): For constant x return it. Formatting fix.
204 (loop_niter_by_eval): Avoid pointless looping if the next iteration
205 would use the same bases as the current one.
207 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
209 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
210 instead of vec_select for V1TImode.
211 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
213 (VSX_LE_128): Add V1TI to this mode iterator.
214 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
215 (*vsx_le_perm_store_<mode>): Likewise.
216 (pre-reload splitter for VSX stores): Likewise.
217 (post-reload splitter for VSX stores): Likewise.
218 (*vsx_xxpermdi2_le_<mode>): Likewise.
219 (*vsx_lxvd2x2_le_<mode>): Likewise.
220 (*vsx_stxvd2x2_le_<mode>): Likewise.
222 2017-03-09 Michael Eager <eager@eagercon.com>
224 Correct failures with --enable-checking=yes,rtl.
226 * config/microblaze/microblaze.c (microblaze_expand_shift):
227 Replace GET_CODE test with CONST_INT_P and INTVAL test with
229 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
230 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
232 2017-03-09 Richard Biener <rguenther@suse.de>
234 PR tree-optimization/79977
235 * graphite-scop-detection.c (scop_detection::merge_sese):
236 Handle the case of extra exits to blocks dominating the entry.
238 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
240 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
243 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
245 PR rtl-optimization/79949
246 * lra-constraints.c (process_alt_operands): Check memory when
247 trying to predict a cycle. Print about the overall increase.
249 2017-03-09 Richard Biener <rguenther@suse.de>
252 * gimple-expr.c (useless_type_conversion_p): Preserve
253 TYPE_SATURATING for fixed-point types.
255 2017-03-09 Richard Biener <rguenther@suse.de>
258 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
259 alignment of BLKmode params.
261 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
264 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
265 (VALL_NO_V2Q): Likewise.
267 * config/aarch64/aarch64-simd.md
268 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
270 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
271 VALL_NO_V2Q mode iterator.
272 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
274 2017-03-09 Martin Liska <mliska@suse.cz>
276 PR tree-optimization/79631
277 * tree-chkp-opt.c (chkp_is_constant_addr): Call
278 tree_int_cst_sign_bit just for INTEGER constants.
280 2017-03-09 Martin Liska <mliska@suse.cz>
284 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
287 2017-03-09 Marek Polacek <polacek@redhat.com>
290 * tree.c (inchash::add_expr): Handle TREE_VEC.
292 2017-03-09 Martin Liska <mliska@suse.cz>
295 (chkp_narrow_size_and_offset): New function.
296 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
297 (void chkp_parse_bit_field_ref): New function.
298 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
299 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
301 2017-03-09 Martin Liska <mliska@suse.cz>
304 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
305 (chkp_find_bounds_1): Remove gcc_unreachable.
307 2017-03-09 Jakub Jelinek <jakub@redhat.com>
310 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
311 BUILT_IN_SYNC*, determine the access type from the size suffix and
312 always build a MEM_REF with that type. Handle forgotten
313 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
316 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
317 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
318 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
319 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
320 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
321 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
322 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
323 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
324 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
325 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
326 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
327 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
328 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
329 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
330 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
331 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
332 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
333 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
334 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
335 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
336 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
337 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
338 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
339 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
340 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
341 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
342 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
343 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
344 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
345 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
346 definitions outside of __OPTIMIZE__ guarded section.
349 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
350 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
351 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
352 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
355 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
357 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
358 ("vfenez<mode>"): Add missing constraints.
360 2017-03-08 Martin Sebor <msebor@redhat.com>
363 * config/nds32/nds32.c (nds32_option_override):
365 2017-03-08 Jakub Jelinek <jakub@redhat.com>
368 * gimplify.c (gimplify_omp_for): Replace index var in outer
369 taskloop statement with an artificial variable and add
370 OMP_CLAUSE_PRIVATE clause for it.
372 2017-03-08 Richard Biener <rguenther@suse.de>
374 PR tree-optimization/79955
375 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
376 for accesses that are completely outside of the variable.
378 2017-03-08 Andrew Haley <aph@redhat.com>
380 PR tree-optimization/79943
381 * tree-ssa-loop-split.c (compute_new_first_bound): When
382 calculating the new upper bound, (END-BEG) should be added, not
385 2017-03-08 Jakub Jelinek <jakub@redhat.com>
387 * config/avr/avr.md (setmemhi): Make sure match_dup
388 operand number comes before match_scratch.
390 2017-03-08 Richard Biener <rguenther@suse.de>
392 PR tree-optimization/79920
393 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
394 with ncopies == 1 to ...
395 (vect_transform_slp_perm_load): ... here. Properly compute
396 all element loads by iterating VF times over the group. Do
397 not handle ncopies (computed in a broken way) in
398 vect_create_mask_and_perm.
400 2017-03-08 Jakub Jelinek <jakub@redhat.com>
403 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
404 is a uniform vector, use uniform_vector_p return value instead of
405 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
407 2017-03-07 Marek Polacek <polacek@redhat.com>
410 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
411 (alloca_call_type): Likewise.
413 2017-03-07 Martin Liska <mliska@suse.cz>
415 * gcov.c (process_args): Put comment to correct location.
417 2017-03-07 Martin Liska <mliska@suse.cz>
420 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
421 Use array_at_struct_end_p instead of DECL_CHAIN (field).
422 (chkp_narrow_bounds_for_field): Likewise.
423 (chkp_parse_array_and_component_ref): Pass one more argument to
426 2017-03-07 Richard Biener <rguenther@suse.de>
428 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
431 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
433 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
434 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
436 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
439 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
440 to end of description.
441 (PARAM_MAX_STORES_TO_MERGE): Likewise.
443 2017-03-07 Jakub Jelinek <jakub@redhat.com>
445 PR rtl-optimization/79901
446 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
448 (*avx512f_<code><mode>3<mask_name>): ... this.
449 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
450 iterator instead of VI8_AVX2_AVX512BW.
452 PR rtl-optimization/79901
453 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
454 min/max expander, expand it using expand_vec_cond_expr.
457 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
460 2017-03-06 Jakub Jelinek <jakub@redhat.com>
463 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
464 to void * for PCH reasons.
465 * dwarf2out.c (output_loc_operands, output_die): Cast
466 v.val_vec.array to unsigned char *.
468 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
471 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
474 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
476 PR rtl-optimization/79571
477 * lra-constraints.c (process_alt_operands): Calculate static
478 reject and subtract it from overall when only addresses will be
481 2017-03-06 Julia Koval <julia.koval@intel.com>
484 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
485 incoming stack boundary to 128 for 64-bit targets.
487 2017-03-06 Richard Biener <rguenther@suse.de>
489 PR tree-optimization/79894
490 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
491 to NULL after folding it.
493 2017-03-06 Richard Biener <rguenther@suse.de>
495 PR tree-optimization/79824
496 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
497 check disabling peeling for gaps.
499 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
501 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
502 attributes): Document gettimeofday.
504 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
506 * config/s390/s390.c (s390_option_override_internal): Set
507 PARAM_MIN_VECT_LOOP_BOUND
509 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
511 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
512 * config/s390/s390.md: Likewise.
514 2017-03-06 Jakub Jelinek <jakub@redhat.com>
517 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
518 (<avx2_avx512>_perm<mode>): Rename to ...
519 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
521 (<avx512>_perm<mode>_mask): Rename to ...
522 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
524 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
525 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
526 instead of VI8F_256_512.
527 (avx512f_perm<mode>): New define_expand.
528 (avx512f_perm<mode>_mask): Likewise.
529 (avx512f_perm<mode>_1<mask_name>): New define_insn.
530 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
532 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
534 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
535 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
537 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
539 2017-03-06 Martin Liska <mliska@suse.cz>
542 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
543 when having a SSA NAME w/o VAR_DECL assigned to it.
545 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
547 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
548 msa_dpsub_<su>_d): Fix MODE for vec_select.
550 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
552 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
554 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
556 2017-03-06 Richard Biener <rguenther@suse.de>
558 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
559 * plugin.c (register_plugin_info): Likewise.
560 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
562 2017-03-05 Jakub Jelinek <jakub@redhat.com>
564 * config/i386/sse.md (sse_storehps, sse_storelps,
565 avx_<castmode><avxsizesuffix>_<castmode>,
566 avx512f_<castmode><avxsizesuffix>_<castmode>,
567 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
568 in condition that at least one operand is not a MEM.
570 2017-03-03 Jakub Jelinek <jakub@redhat.com>
573 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
574 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
576 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
577 gimple_call_nothrow_p flag based on whether original builtin can throw.
578 If it can, emit following stmts on the fallthrough edge.
579 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
580 don't create new bb if inserting just debug stmts on the edge, try to
581 insert them on the fallthru bb or just reset debug stmts.
583 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
586 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
587 restore recog_data (including the operand rtxes inside it) around
588 the call to get_insn_template.
590 2017-03-03 Martin Sebor <msebor@redhat.com>
592 PR tree-optimization/79699
593 * context.c (context::~context): Free MPFR caches to avoid
594 a memory leak on program exit.
596 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
598 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
599 Use wide_int::ulow () instead of .elt (0).
601 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
603 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
604 (*pushxf): Limit oF constraint to 32bit targets and add oC
605 constraint for 64bit targets.
606 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
607 (*pushdf): Change rmF constraint to rmC.
609 2017-03-03 Martin Liska <mliska@suse.cz>
611 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
612 Remove unused variable.
614 2017-03-03 Jakub Jelinek <jakub@redhat.com>
617 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
618 is a memory operand, increase num_memory.
619 (ix86_expand_args_builtin): Likewise.
621 2017-03-03 Jan Hubicka <jh@suse.cz>
624 * ipa-devirt.c (maybe_record_node): Properly handle
625 __cxa_pure_virtual visibility.
627 2017-03-03 Martin Liska <mliska@suse.cz>
629 PR tree-optimization/79803
630 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
632 (pass_loop_prefetch::execute): Disabled optimization if an
633 assumption about L1 cache size is not met.
635 2017-03-03 Martin Liska <mliska@suse.cz>
637 PR rtl-optimization/79574
638 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
639 (hash_scan_set): Likewise.
640 (dump_hash_table): Likewise.
641 (hoist_code): Likewise.
643 2017-03-03 Richard Biener <rguenther@suse.de>
645 * fixed-value.c (fixed_from_string): Restore use of elt (1)
646 in place of uhigh ().
647 (fixed_convert_from_real): Likewise.
649 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
652 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
654 2017-03-03 Richard Biener <rguenther@suse.de>
657 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
658 TYPE_OVERFLOW_UNDEFINED check.
660 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
662 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
664 (vector_ae_<mode>_p): Likewise.
665 (vector_nez_<mode>_p): Likewise.
666 (vector_ne_v2di_p): Likewise.
667 (vector_ae_v2di_p): Likewise.
668 (vector_ne_<mode>_p): Likewise.
669 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
671 (vsx_tsqrt<mode>2_fe): Likewise.
673 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
676 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
678 2017-03-02 Jakub Jelinek <jakub@redhat.com>
680 PR rtl-optimization/79780
681 * cprop.c (one_cprop_pass): When second and further conditional trap
682 in a single basic block is turned into an unconditional trap, turn it
683 into a deleted note to avoid RTL verification failures.
685 2017-03-02 Richard Biener <rguenther@suse.de>
687 * fold-const.c (const_binop): Use ulow () instead of elt (0).
689 2017-03-02 Richard Biener <rguenther@suse.de>
691 PR tree-optimization/79345
693 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
694 param and abort the walk, returning -1 if it is hit.
695 (walk_aliased_vdefs): Take a limit param and pass it on.
696 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
697 defaulting to 0 and return a signed int.
698 * tree-ssa-uninit.c (struct check_defs_data): New struct.
699 (check_defs): New helper.
700 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
701 about uninitialized memory.
702 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
703 bogus uninitialized warning.
704 (fixed_convert_from_real): Likewise.
706 2017-03-02 Bin Cheng <bin.cheng@arm.com>
708 PR tree-optimization/66768
709 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
710 iv_use if base object can't be determined.
712 2017-03-02 Jakub Jelinek <jakub@redhat.com>
714 PR tree-optimization/79345
715 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
716 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
717 (get_pattern_stats): Initialize it.
718 * genemit.c (gen_expand): Verify match_scratch numbers come after
719 match_operand/match_dup numbers.
720 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
721 match_scratch numbers.
722 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
724 * config/s390/s390.md (trunctdsd2): Likewise.
726 2017-03-02 Richard Biener <rguenther@suse.de>
728 * wide-int.h (wide_int_storage::operator=): Implement in terms
731 2017-03-02 Richard Biener <rguenther@suse.de>
733 PR tree-optimization/79777
734 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
735 the to insert expression to sth existing.
737 2017-03-01 Martin Sebor <msebor@redhat.com>
740 * gimple-ssa-sprintf.c
741 (directive::known_width_and_precision): New function.
742 (format_integer): Use it.
743 (get_mpfr_format_length): Consider the full range of precision
744 when computing %g output with the # flag. Set the likely byte
745 count to 3 rather than 1 when precision is indeterminate.
746 (format_floating): Correct the lower bound of precision.
748 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
750 * doc/invoke.texi: Document default code model for 64-bit Linux.
752 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
755 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
756 udiv rather than div since input pattern is unsigned.
758 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
760 * config/i386/i386.c (print_reg): Warn for values of
761 unsupported size in integer register.
763 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
766 * config/rs6000/predicates.md (current_file_function_operand): Do
767 not allow self calls to be local if the function is replaceable.
769 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
772 * config/rs6000/altivec.h (vec_ctz and others): Change the
773 preprocessor macro that controls conditional compilation from
774 _ARCH_PWR9 to __POWER9_VECTOR__.
775 (vec_all_ne): Change parameterization of __altivec_scalar_pred
776 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
777 control (instead of _ARCH_PWR9 control) so that template
778 definition uses power9-specific function.
779 (vec_any_eq): Likewise.
780 (vec_all_ne): Change macro definition to use a power9-specific
781 expansion under #ifdef __POWER9_VECTOR__ control (instead of
783 (vec_any_eq) Likewise.
784 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
785 expansion for CMPNEF to remove support for xvcmpnesp instruction.
786 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
787 support for xvcmpnedp instruction.
788 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
789 macro expansion so that Power9 implementation of vec_all_ne does
790 not use the AltiVec predicate framework.
791 (VCMPNEH_P): Likewise.
792 (VCMPNEW_P): Likewise.
793 (VCMPNED_P): Likewise.
794 (VCMPNEFP_P): Likewise.
795 (VCMPNEDP_P): Likewise.
796 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
797 implementation of vec_any_eq to not use AltiVec predicate
799 (VCMPAEH_P): Likewise.
800 (VCMPAEW_P): Likewise.
801 (VCMPAED_P): Likewise.
802 (VCMPAEFP_P): Likewise.
803 (VCMPAEDP_P): Likewise.
804 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
805 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
806 not use the AltiVec predicate framework.
807 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
808 of vec_any_eq to not use AltiVec predicate framework.
809 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
810 support for predefined __POWER9_VECTOR__ macro to indicate that
811 Power9 instruction selection is enabled.
812 (altivec_overloaded_builtins): Remove extraneous
813 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
814 function argument types RS6000_BTI_bool_V16QI and
815 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
816 entry for overloaded function argument types RS6000_BTI_bool_V4SI
817 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
818 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
819 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
820 Power9 for implementations of vec_cmpne. Change the signature for
821 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
822 (representing vec_all_ne) to remove the previously described first
823 argument of type RS6000_BTI_INTSI, as this was an artifact of
824 reliance on the AltiVec predicate framework, which is no longer
825 used in the implementation of these functions. Add
826 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
827 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
828 since, unlike the AltiVec predicate framework implementation, we
829 do not share function descriptors between vec_alle and vec_anyeq.
830 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
831 set of modes that receive special treatment even when
832 TARGET_P9_VECTOR is true. The special treatment emits code that
833 does not depend on Power9 instructions.
834 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
835 define_expand to not rely on AltiVec predicate framework.
836 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
838 (vector_ne_v2di_p): Change this define_expand to not rely on
839 AltiVec predicate framework.
840 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
842 (vector_ne_<mode>_p): Change this define_expand to not rely on
843 AltiVec predicate framework.
844 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
846 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
847 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
849 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
850 define_insn pattern because the xvcmpne<VSs>. instruction is not
852 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
853 instruction is not supported.
855 2017-03-01 Jakub Jelinek <jakub@redhat.com>
857 * config/nvptx/nvptx.c: Include intl.h.
859 2017-03-01 Martin Jambor <mjambor@suse.cz>
862 * ipa-prop.h (ipa_bits): Removed field known.
863 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
864 to pointers. Adjusted their comments to warn about their sharing.
865 (ipcp_transformation_summary): Change bits to a vector of pointers.
866 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
867 (ipa_get_ipa_bits_for_value): Declare.
868 * tree-vrp.h (value_range): Mark as GTY((for_user)).
869 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
870 (ipa_bits_hash_table): Likewise.
871 (ipa_vr_ggc_hash_traits): Likewise.
872 (ipa_vr_hash_table): Likewise.
873 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
874 being pointers and vr_known being removed.
875 (ipa_set_jf_unknown): Likewise.
876 (ipa_get_ipa_bits_for_value): New function.
877 (ipa_set_jfunc_bits): Likewise.
878 (ipa_get_value_range): New overloaded functions.
879 (ipa_set_jfunc_vr): Likewise.
880 (ipa_compute_jump_functions_for_edge): Use the above functions to
881 construct bits and vr parts of jump functions.
882 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
883 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
885 (ipcp_grow_transformations_if_necessary): Also allocate
886 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
888 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
889 them. Fix too long lines.
890 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
891 vr_known being removed.
892 (ipa_read_jump_function): Use new setter functions to construct bits
893 and vr parts of jump functions or set them to NULL.
894 (write_ipcp_transformation_info): Adjust for bits being pointers.
895 (read_ipcp_transformation_info): Likewise.
896 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
898 Include gt-ipa-prop.h.
899 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
901 (ipcp_store_bits_results): Likewise.
902 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
903 Do not write to existing jump functions but use a temporary instead.
905 2017-03-01 Jakub Jelinek <jakub@redhat.com>
908 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
909 attempt to use its first operand as BIT_FIELD_REF base.
911 2017-03-01 Richard Biener <rguenther@suse.de>
914 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
915 interpolating formula in wrapping arithmetic.
916 (chrec_apply): Convert chrec_evaluate return value to wanted type.
918 2017-03-01 Jakub Jelinek <jakub@redhat.com>
920 PR tree-optimization/79734
921 * tree-vect-generic.c (expand_vector_condition): Optimize
922 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
923 Handle VEC_COND_EXPR where comparison has different inner width from
926 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
928 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
929 markup, and similar issues. Remove @opindex entries for things
930 that aren't options. Add missing -mmpy-option entries.
932 2017-02-28 Jakub Jelinek <jakub@redhat.com>
934 PR tree-optimization/79737
935 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
936 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
937 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
938 instead of byte_size. Formatting fix.
939 (shift_bytes_in_array_right): Formatting fix.
941 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
944 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
945 condition on optimize for the leaf function test.
947 2017-02-28 Martin Liska <mliska@suse.cz>
950 * read-rtl-function.c (function_reader::handle_unknown_directive):
951 Bail out when one uses -flto.
953 2017-02-28 Martin Liska <mliska@suse.cz>
955 * common.opt: Replace space with tabular for options of <number>
957 * config/i386/i386.opt: Show <number> value for
958 -mlarge-data-threshold.
959 * opts.c (print_filtered_help): Do not display number in hexadecimal
962 2017-02-28 Martin Liska <mliska@suse.cz>
964 * common.opt: Fix --help=option -Q for options which are of
967 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
969 * config/i386/i386.c (print_reg): Error out for values
970 of 8-bit size in invalid integer register.
972 2017-02-28 Martin Sebor <msebor@redhat.com>
974 PR tree-optimization/79691
975 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
977 2017-02-28 Jakub Jelinek <jakub@redhat.com>
980 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
981 gcc_unreachable with output_operand_lossage.
983 2017-02-28 Richard Biener <rguenther@suse.de>
985 PR tree-optimization/79740
986 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
988 (visit_nary_op): Insert the nary into the hashtable if we
990 * tree-ssa-pre.c (eliminate_insert): Robustify.
992 2017-02-28 Richard Biener <rguenther@suse.de>
995 * fold-const.c (decode_field_reference): Reject out-of-bound
998 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1000 * config/i386/i386.c: Include intl.h.
1001 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
1002 instead of just cond ? "..." : "...".
1003 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
1004 * coverage.c (read_counts_file): Likewise.
1005 * omp-offload.c: Include intl.h.
1006 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
1007 of just cond ? "..." : "...".
1008 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
1009 of just cond ? "..." : "...".
1011 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
1014 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
1016 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
1017 'tune for' CPU name.
1018 * config/arm/arm-cpu-data.h: Regenerated.
1020 2017-02-28 Richard Biener <rguenther@suse.de>
1022 PR tree-optimization/79732
1023 * tree-inline.c (expand_call_inline): Do not shadow var.
1025 2017-02-28 Richard Biener <rguenther@suse.de>
1027 PR tree-optimization/79723
1028 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
1029 address-space properly.
1031 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
1033 * doc/optinfo.texi (Optimization groups): Fix option used for
1035 * doc/invoke.texi (-fopt-info): Document "omp".
1036 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
1037 (OPTGROUP_ALL): Add OPTGROUP_OMP.
1038 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
1039 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
1040 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
1042 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
1044 * dumpfile.c (optgroup_options): Instead of "openmp", associate
1045 OPTGROUP_OMP with "omp".
1047 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
1050 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
1051 for arithmetic shift of unsigned V2DI.
1053 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
1055 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
1056 arc/linux.h headers.
1057 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
1058 (LINK_SPEC): Likewise.
1059 (ARC_TLS_EXTRA_START_SPEC): Likewise.
1060 (EXTRA_SPECS): Likewise.
1061 (STARTFILE_SPEC): Likewise.
1062 (ENDFILE_SPEC): Likewise.
1063 (LIB_SPEC): Likewise.
1064 (TARGET_SDATA_DEFAULT): Likewise.
1065 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
1066 (MULTILIB_DEFAULTS): Likewise.
1067 (DWARF2_UNWIND_INFO): Likewise.
1068 * config/arc/big.h: New file.
1069 * config/arc/elf.h: Likewise.
1070 * config/arc/linux.h: Likewise.
1071 * config/arc/t-uClibc: Remove.
1073 2017-02-27 Bin Cheng <bin.cheng@arm.com>
1075 PR tree-optimization/77536
1076 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
1077 (tree_transform_and_unroll_loop): Use above function to compute the
1078 estimated niter of unrolled loop and use it when scaling profile.
1079 Also use count info rather than frequency if it's non-zero.
1080 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
1081 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
1082 (vect_transform_loop): Call above function.
1084 2017-02-27 Richard Biener <rguenther@suse.de>
1086 PR tree-optimization/45397
1087 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
1088 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
1089 (visit_nary_op): Add pattern matching for CSEing sign-changed
1090 or truncated operations with wider ones.
1092 2017-02-27 Richard Biener <rguenther@suse.de>
1094 PR tree-optimization/79690
1095 * tree-vect-stmts.c (vectorizable_store): Use vector type
1096 built from the DR with address-space.
1098 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
1100 * doc/invoke.texi (Optimize Options): Refine the description
1101 of asan-use-after-return.
1103 2017-02-25 Alan Modra <amodra@gmail.com>
1105 PR rtl-optimization/79584
1106 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
1107 base, not ad->base_term, the reg within base. Remove assertion
1108 that ad->base == ad->base_term. Replace gen_int_mode using
1109 bogus mode with const0_rtx.
1111 2017-02-25 Jakub Jelinek <jakub@redhat.com>
1114 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
1115 FMA_EXPR like tcc_binary or tcc_unary.
1117 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
1120 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
1122 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
1123 (output_loc_operands): Handle DW_OP_call_ref and
1124 DW_OP_GNU_variable_value.
1125 (struct variable_value_struct): New type.
1126 (struct variable_value_hasher): Likewise.
1127 (variable_value_hash): New variable.
1128 (string_types): Remove.
1129 (copy_loc_descr): New function.
1130 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
1131 (prepend_loc_descr_to_each): New function.
1132 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
1133 instead of add_loc_descr_to_each if the first argument is single
1134 location list and the second has multiple.
1135 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
1136 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
1137 when looking for variable value which doesn't have other location info.
1138 (loc_list_from_tree): Formatting fix.
1139 (gen_array_type_die): Simplify DW_AT_string_length handling.
1140 (adjust_string_types): Remove.
1141 (gen_subprogram_die): Don't call adjust_string_types nor test/set
1142 string_types. Call resolve_variable_values.
1143 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
1144 (resolve_addr_in_expr): Likewise. Add A argument.
1145 (copy_deref_exprloc): Remove deref argument. Adjust for the
1146 original expression being DW_OP_GNU_variable_value with optionally
1147 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
1148 optionally after it.
1149 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
1150 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
1151 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
1152 (variable_value_hasher::hash, variable_value_hasher::equal): New
1154 (resolve_variable_value_in_expr, resolve_variable_value,
1155 resolve_variable_values, note_variable_value_in_expr,
1156 note_variable_value): New functions.
1157 (dwarf2out_early_finish): Call note_variable_value on all toplevel
1160 2017-02-24 Jakub Jelinek <jakub@redhat.com>
1163 * opts.h (handle_generated_option): Add GENERATED_P argument.
1164 * opts-common.c (handle_option): Adjust function comment.
1165 (handle_generated_option): Add GENERATED_P argument, pass it to
1167 (control_warning_option): Pass false to handle_generated_option
1169 * opts.c (maybe_default_option): Pass true to handle_generated_option
1171 * optc-gen.awk: Likewise.
1173 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1175 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
1176 a REG, look at the REG it is a SUBREG of.
1177 (splitter for cmpeqsi_t): Ditto.
1179 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1181 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
1182 the special USEs with the pattern of the insn, not the insn itself.
1184 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
1187 * doc/invoke.texi: Document -mload-store-pairs.
1189 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1190 Sandra Loosemore <sandra@codesourcery.com>
1192 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
1193 argument isn't a CONST_INT.
1194 (nios2_alternate_compare_const): Assert op is a CONST_INT.
1195 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
1196 (nios2_validate_compare): Bypass alternate compare logic if *op2
1198 (ldstwm_operation_p): Return false if first_base is not a REG or
1199 if first_offset is not a CONST_INT.
1201 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1203 * config/cris/cris.md: Use correct operand in a define_peephole2.
1205 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1207 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
1209 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1211 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
1212 this_insn if it is an INSN or JUMP_INSN.
1213 (force_offsettable): Look at base, not at addr.
1214 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
1215 on things that aren't necessarily CONST_INTs.
1217 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
1219 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
1220 -mfpmath=sse is the default also for x86-32 targets with SSE2
1221 instruction set when @option{-ffast-math} is enabled
1223 2017-02-24 Jeff Law <law@redhat.com>
1225 PR rtl-optimizatoin/79286
1226 * ira.c (update_equiv_regs): Drop may_trap_p exception to
1229 2017-02-24 Richard Biener <rguenther@suse.de>
1231 PR tree-optimization/79389
1232 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
1235 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
1237 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
1238 function comment to reflect reality.
1239 (loop_exits_before_overflow): Fix typo in function description.
1241 2017-02-24 Richard Biener <rguenther@suse.de>
1243 PR tree-optimization/79389
1244 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
1245 properly that a threading opportunity exists. Detect conditional
1246 copy/constant propagation opportunities.
1248 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
1250 * config/visium/visium.md (type): Add trap.
1251 (b): New mode attribute.
1252 (*btst): Rename into...
1253 (*btst<mode>): ...this and adjust.
1254 (*cbranchsi4_btst_insn): Rename into...
1255 (*cbranch<mode>4_btst_insn): ...this and adjust.
1256 (trap): New define_insn.
1258 2017-02-23 Jakub Jelinek <jakub@redhat.com>
1260 PR tree-optimization/79389
1261 * ifcvt.c (struct noce_if_info): Add rev_cond field.
1262 (noce_reversed_cond_code): New function.
1263 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
1264 reversed_comparison_code. Formatting fix.
1265 (noce_try_store_flag): Test rev_cond != NULL in addition to
1266 reversed_comparison_code.
1267 (noce_try_store_flag_constants): Likewise.
1268 (noce_try_store_flag_mask): Likewise.
1269 (noce_try_addcc): Use rev_cond if non-NULL instead of
1270 reversed_comparison_code.
1271 (noce_try_cmove_arith): Likewise. Formatting fixes.
1272 (noce_try_minmax, noce_try_abs): Clear rev_cond.
1273 (noce_find_if_block): Initialize rev_cond.
1274 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
1275 instead of false as last argument never attempt to reverse it
1278 2017-02-23 Bin Cheng <bin.cheng@arm.com>
1280 PR tree-optimization/79663
1281 * tree-predcom.c (combine_chains): Process refs in reverse order
1282 only for ZERO length chains, and add explaining comment.
1284 2017-02-23 Jeff Law <law@redhat.com>
1286 PR tree-optimization/79578
1287 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
1288 in call to operand_equal_p.
1290 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
1293 * config/i386/cpuid.h: Fix another undefined behavior.
1295 2017-02-23 Richard Biener <rguenther@suse.de>
1297 PR tree-optimization/79683
1298 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
1299 vector types for data-refs.
1301 2017-02-23 Martin Liska <mliska@suse.cz>
1303 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
1305 2017-02-23 Jakub Jelinek <jakub@redhat.com>
1308 * internal-fn.c (get_range_pos_neg): Moved to ...
1309 * tree.c (get_range_pos_neg): ... here. No longer static.
1310 * tree.h (get_range_pos_neg): New prototype.
1311 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
1312 are known to be in between 0 and signed maximum inclusive, try to
1313 expand both unsigned and signed divmod and use the cheaper one from
1316 2017-02-22 Jeff Law <law@redhat.com>
1318 PR tree-optimization/79578
1319 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
1320 to compare base operands.
1322 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
1325 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
1326 gpc_reg_operand instead of fpr_reg_operand.
1328 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
1330 * config/mips/mips.c (mips_return_in_memory): Force FP
1331 vector types to be returned in memory for o32 ABI.
1333 2017-02-22 Jakub Jelinek <jakub@redhat.com>
1335 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
1336 instead of DW_TAG_member for static data member declarations and don't
1337 set no_linkage_name for static inline data members.
1338 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
1341 2017-02-22 Martin Liska <mliska@suse.cz>
1343 * doc/invoke.texi: Replace inequality signs with square brackets
1346 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1348 PR tree-optimization/68644
1349 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
1351 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
1354 * lra-constraints.c (simplify_operand_subreg): Handle
1355 WORD_REGISTER_OPERATIONS targets.
1357 2017-02-22 Jakub Jelinek <jakub@redhat.com>
1360 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
1361 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
1362 elimination by swapping fld*.
1364 2017-02-22 Richard Biener <rguenther@suse.de>
1366 PR tree-optimization/79673
1367 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
1368 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
1369 irrelevant address-space qualifiers and avoiding a
1370 ADDR_SPACE_CONVERT_EXPR from fold_convert.
1372 2017-02-22 Richard Biener <rguenther@suse.de>
1374 PR tree-optimization/79666
1375 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
1376 to not symbolically negate if that may introduce undefined
1379 2017-02-22 Martin Liska <mliska@suse.cz>
1382 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
1383 * data-streamer-out.c (streamer_write_gcov_count_stream):
1385 * value-prof.c (stream_out_histogram_value): Make assert more
1386 precise based on type of counter.
1388 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
1391 * config/i386/i386.md (standard_x87sse_constant_load splitter):
1392 Use nonimmediate_operand instead of memory_operand for operand 1.
1393 (float-extend standard_x87sse_constant_load splitter): Ditto.
1395 2017-02-21 Jeff Law <law@redhat.com>
1397 PR tree-optimization/79621
1398 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
1399 blocks with edges to themselves.
1401 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1404 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
1405 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
1406 Use gimple_call_builtin_p.
1409 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
1410 on temporarily removed DEBUG_INSNs.
1412 PR tree-optimization/79649
1413 * tree-loop-distribution.c (classify_partition): Give up on
1414 non-generic address space loads/stores.
1416 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
1418 * doc/loop.texi (Loop manipulation): Remove nonexistent
1419 tree_ssa_loop_version from the documentation.
1420 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
1422 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1425 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
1426 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
1427 * config/rs6000/rs6000.c: Include except.h.
1428 (rs6000_expand_split_stack_prologue): Call
1429 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
1431 2017-02-21 Martin Jambor <mjambor@suse.cz>
1434 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
1437 2017-02-21 Martin Jambor <mjambor@suse.cz>
1439 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
1440 for backward compatibility only.
1441 * doc/invoke.texi (Option Summary): Remove all references to
1444 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
1448 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1450 * lra-constraints.c (curr_insn_transform): Handle
1451 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
1453 2017-02-21 Martin Liska <mliska@suse.cz>
1455 * config/i386/i386.opt: Replace -masm-dialect with -masm.
1457 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
1459 PR translation/79638
1460 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
1462 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
1465 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
1466 (arm_function_ok_for_sibcall): Return false for an indirect call by
1467 descriptor if all the argument registers are used.
1468 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
1469 alignment of the function.
1471 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1473 PR tree-optimization/61441
1474 * simplify-rtx.c (simplify_const_unary_operation): For
1475 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
1476 the sNaN unmodified.
1478 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
1480 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
1481 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
1482 instead of SYSTEM_HEADER_DIR.
1484 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
1485 Martin Liška <mliska@suse.cz>
1487 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
1488 Fix typos and grammar, use active voice, and clarify.
1490 2017-02-20 Marek Polacek <polacek@redhat.com>
1493 * gimplify.c (gimplify_expr): Handle unused *&&L;.
1496 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
1498 2017-02-20 Jakub Jelinek <jakub@redhat.com>
1501 * config/i386/i386.c (ix86_expand_builtin): Handle
1502 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
1503 ix86_builtins_isa[fcode].isa as a requirement of those
1504 flags and any other flag in the bitmask.
1505 (ix86_init_mmx_sse_builtins): Use 0 instead of
1506 ~OPTION_MASK_ISA_64BIT as mask.
1507 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
1508 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
1509 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
1510 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
1512 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1515 * lra-constraints.c (split_reg): Check requested split mode
1516 is supported by the register.
1518 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1520 * lra-constraints.c (simplify_operand_subreg): Remove early
1523 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1526 * lra-constraints.c (curr_insn_transform): Tighten condition
1527 for converting SUBREG reloads from OP_OUT to OP_INOUT.
1529 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1532 * lra-constraints.c (curr_insn_transform): Handle
1533 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
1535 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
1538 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
1540 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
1542 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
1545 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
1548 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1550 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
1553 2017-02-18 Jakub Jelinek <jakub@redhat.com>
1555 * final.c (last_columnnum, override_columnnum): New variables.
1556 (final_start_function): Set last_columnnum, pass it to begin_prologue
1557 hook and pass 0 to dwarf2out_begin_prologue.
1558 (final_scan_insn): Update override_columnnum. Pass last_columnnum
1559 to source_line debug hook.
1560 (notice_source_line): Compute last_columnnum and for debug_column_info
1561 return true on column changes.
1562 * debug.h (struct gcc_debug_hooks): Add column argument to
1563 source_line and begin_prologue hooks.
1564 (debug_nothing_int_charstar_int_bool): Remove prototype.
1565 (debug_nothing_int_int_charstar,
1566 debug_nothing_int_int_charstar_int_bool): New prototypes.
1567 (dwarf2out_begin_prologue): Add column argument.
1568 * debug.c (do_nothing_debug_hooks): Adjust source_line and
1569 begin_prologue hooks.
1570 (debug_nothing_int_charstar_int_bool): Remove.
1571 (debug_nothing_int_int_charstar,
1572 debug_nothing_int_int_charstar_int_bool): New functions.
1573 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
1574 through to dwarf2out_source_line.
1575 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
1576 (dwarf2out_source_line): Add column argument, emit it if requested.
1577 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
1579 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
1580 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
1581 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
1582 through to dwarf2out_begin_prologue.
1583 (vmsdbgout_source_line): Add column argument, pass it through to
1584 dwarf2out_source_line.
1585 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
1586 dbxout_source_line caller.
1587 (dbxout_source_line): Add column argument.
1589 * common.opt (gno-column-info, gcolumn-info): New options.
1590 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
1591 (check_die): Also test for multiple DW_AT_decl_column attributes.
1592 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
1593 DW_AT_decl_column if requested.
1594 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
1596 (gen_variable_die): Likewise.
1597 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
1598 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
1601 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
1602 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
1603 (ix86_handle_option): Handle OPT_m3dnowa.
1604 * doc/invoke.texi (-m3dnowa): Document.
1605 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
1606 -m3dnowa instead of -m3dnow -march=athlon.
1609 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
1610 instead of gcc_assert for K, r and R code checks. Formatting fixes.
1612 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1615 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
1616 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
1617 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
1618 generator for vsx_xxpermdi_<mode>_be.
1619 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
1620 force big-endian semantics.
1621 (vsx_xxpermdi_<mode>_be): New define_expand with same
1622 implementation as previous version of vsx_xxpermdi_<mode>.
1624 2017-02-17 Jakub Jelinek <jakub@redhat.com>
1626 PR tree-optimization/79327
1627 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
1628 variable, its initialization and use.
1630 2017-02-17 Julia Koval <julia.koval@intel.com>
1632 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
1633 (OPTION_MASK_ISA_PKU_UNSET): New.
1634 (ix86_handle_option): Handle -mrdpid.
1635 * config/i386/cpuid.h (bit_RDPID): New.
1636 * config/i386/driver-i386.c (host_detect_local_cpu):
1637 Detect RDPID feature.
1638 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
1639 * config/i386/i386-c.c (ix86_target_macros_internal):
1641 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
1642 (ix86_valid_target_attribute_inner_p): Add "rdpid".
1643 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
1644 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
1645 * config/i386/i386.md (define_insn "rdpid"): New.
1646 * config/i386/i386.opt Add -mrdpid.
1647 * config/i386/immintrin.h (_rdpid_u32): New.
1649 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
1651 PR rtl-optimization/79541
1652 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
1653 instead of transforming it into USE.
1655 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
1657 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
1658 If HONOR_SNANS (SFmode) force the input to a register.
1659 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
1660 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
1661 an frsp or similar insn.
1663 2017-02-17 Martin Liska <mliska@suse.cz>
1665 PR rtl-optimization/79577
1666 * params.def (selsched-max-sched-times): Increase minimum to 1.
1668 2017-02-17 Martin Liska <mliska@suse.cz>
1670 PR rtl-optimization/79574
1671 * gcse.c (want_to_gcse_p): Prevent integer overflow.
1673 2017-02-17 Martin Liska <mliska@suse.cz>
1675 PR tree-optimization/79529
1676 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
1677 ssa_defined_default_def_p to handle cases which are implicitly
1679 * tree-ssa.c (ssa_defined_default_def_p): New function.
1680 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
1681 which are implicitly defined.
1682 * tree-ssa.h (ssa_defined_default_def_p): Declare.
1684 2017-02-17 Richard Biener <rguenther@suse.de>
1687 * params.def (max-ssa-name-query-depth): Limit to 10.
1689 2017-02-17 Richard Biener <rguenther@suse.de>
1691 PR tree-optimization/79552
1692 * tree-ssa-structalias.c (visit_loadstore): Properly verify
1695 2017-02-17 Richard Biener <rguenther@suse.de>
1698 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
1700 2017-02-17 Marek Polacek <polacek@redhat.com>
1703 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
1704 (fold_negate_expr): New wrapper.
1706 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
1708 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
1709 Correct terminology and de-emphasize pre-standard behavior.
1711 2017-02-16 Alan Modra <amodra@gmail.com>
1713 PR rtl-optimization/79286
1714 * ira.c (def_dominates_uses): New function.
1715 (update_equiv_regs): Don't create an equivalence for insns that
1716 may trap where the register def does not dominate the use.
1718 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
1720 PR rtl-optimization/78127
1721 * lra.c (lra): Call lra_eliminate before finish the loop after
1724 2017-02-16 Richard Biener <rguenther@suse.de>
1726 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
1728 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
1729 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
1730 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
1731 (isl_val_int_from_wi): New function.
1732 (extract_affine_gmp): Rename to ...
1733 (extract_affine_wi): ... this, take a widest_int.
1734 (extract_affine_int): Just wrap extract_affine_wi.
1735 (add_param_constraints): Use isl_val_int_from_wi.
1736 (add_loop_constraints): Likewise, and extract_affine_wi.
1738 2017-02-15 Jeff Law <law@redhat.com>
1741 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
1742 ira_init_register_move_cost_if_necessary.
1744 2017-02-15 Martin Sebor <msebor@redhat.com>
1747 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
1748 removed in a prior commit.
1750 2017-02-15 Bin Cheng <bin.cheng@arm.com>
1752 PR tree-optimization/79347
1753 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
1754 counters during peeling.
1756 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
1758 * Makefile.in (site.exp): Remove "set ISLVER".
1760 2017-02-15 Jakub Jelinek <jakub@redhat.com>
1763 * real.c (real_from_integer): Call real_convert even for decimal.
1765 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
1768 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
1770 2017-02-14 Andrew Pinski <apinski@cavium.com>
1772 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
1773 cores and change the partno/implementer to be correct.
1774 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
1775 the 'B" as the implementer.
1776 * config/aarch64/aarch64-tune.md: Regenerate.
1778 2017-02-14 Carl Love <cel@us.ibm.com>
1780 * config/rs6000/rs6000.c: Add case statement entry to make the
1781 xvcvuxdsp built-in argument unsigned.
1782 * config/rs6000/vsx.md: Fix the source and return operand types so they
1783 match the instruction definitions from the ISA document. Fix typo
1784 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
1787 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
1790 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
1791 member early_clobber_alts.
1792 * lra-lives.c (reg_early_clobber_p): New.
1793 (process_bb_lives): Use it.
1794 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
1795 (debug_operand_data): Initialize early_clobber_alts.
1796 (setup_operand_alternative): Set up early_clobber_alts.
1797 (collect_non_operand_hard_regs): Ditto. Pass early clobber
1798 alternatives to new_insn_reg.
1799 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
1801 (lra_update_insn_regno_info): Pass the new arg.
1803 2017-02-14 Jakub Jelinek <jakub@redhat.com>
1806 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
1807 (new_oacc_loop_raw): Don't clear already cleared fields.
1810 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
1811 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
1812 _mm512_prefetch_i64gather_ps): New inline functions and macros.
1814 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
1817 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
1819 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
1822 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
1823 the extra instruction to the right place to store 128-bit constant
1826 2017-02-14 Martin Sebor <msebor@redhat.com>
1829 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
1830 warning for strings of unknown length.
1832 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
1834 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
1836 2017-02-14 Jeff Law <law@redhat.com>
1839 * ira-costs.c (scan_one_insn): Initialize register move costs
1840 for pseudos seen in USE/CLOBBER insns.
1842 PR tree-optimization/79095
1843 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
1844 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
1845 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
1846 if the operands are known to be not equal, then the resulting range
1848 (intersect_ranges): If the new range is ~[0,0] and the old range is
1849 wide, then prefer ~[0,0].
1850 * tree-vrp.c (overflow_comparison_p_1): New function.
1851 (overflow_comparison_p): New function.
1852 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
1853 if NAME is used in an overflow test.
1854 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
1855 overflow check that can be expressed as an equality test, then adjust
1856 ops to be that equality test.
1858 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1860 * config/s390/s390-builtin-types.def: Remove flags argument.
1861 * config/s390/s390.c (s390_init_builtins): Likewise.
1863 2017-02-14 Martin Liska <mliska@suse.cz>
1865 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
1866 vector. Fix trailing white spaces.
1868 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
1870 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
1873 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1875 PR rtl-optimization/68664
1876 * config/arm/arm.c (arm_sched_can_speculate_insn):
1877 New function. Declare prototype.
1878 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
1880 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1882 PR rtl-optimization/68664
1883 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
1885 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
1887 2017-02-14 Amit Pawar <amit.pawar@amd.com>
1889 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
1890 max skip bytes for function, loop and jump.
1892 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1894 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
1895 ABS_EXPR for gimple dump.
1897 2017-02-14 Jakub Jelinek <jakub@redhat.com>
1900 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
1902 PR tree-optimization/79408
1903 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
1904 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
1905 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
1906 also if rhs1 is INTEGER_CST.
1908 2017-02-14 Richard Biener <rguenther@suse.de>
1911 * tree-into-ssa.c (insert_phi_nodes): When the function can
1912 have abnormal edges rewrite SSA names with broken use-def
1913 dominance out of SSA and register them for PHI insertion.
1915 2017-02-13 Martin Sebor <msebor@redhat.com>
1918 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
1919 clearing info.nowrite flag when snprintf size argument is a range.
1921 2017-02-13 Jakub Jelinek <jakub@redhat.com>
1923 * cprop.c (cprop_jump): Add missing space in string literal.
1924 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
1925 (get_constraint_for_component_ref): Likewise.
1926 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
1927 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
1928 * lra-constraints.c (process_alt_operands): Likewise.
1929 * ipa-inline.c (inline_small_functions): Likewise.
1930 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
1931 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
1932 * trans-mem.c (diagnose_tm_1_op): Likewise.
1933 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
1934 (grid_parallel_clauses_gridifiable): Likewise.
1936 * config/nvptx/mkoffload.c (process): Add space in between
1939 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
1940 "MOD4_SSE_REGS" and "ALL_REGS".
1942 * spellcheck.c (test_data): Add , in between "foo" and "food".
1944 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1947 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
1948 boundary crossing check and subsequent code generation agree.
1950 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1952 * config/aarch64/aarch64.c (has_memory_op): Delete.
1953 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
1956 2017-02-13 Jakub Jelinek <jakub@redhat.com>
1958 PR rtl-optimization/79388
1959 PR rtl-optimization/79450
1960 * combine.c (distribute_notes): When removing TEM_INSN for which
1961 corresponding dest has last value recorded, invalidate that last
1964 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1966 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
1967 of explicit '@'. Add missing assembly comment marker on branch costs
1970 2017-02-13 Nathan Sidwell <nathan@acm.org>
1972 * gengtype-lex.l (<in_struct>): Add '/'.
1974 2017-02-13 Martin Liska <mliska@suse.cz>
1977 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
1979 2017-02-13 Richard Biener <rguenther@suse.de>
1981 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
1983 * configure: Re-generate.
1984 * config.in: Likewise.
1985 * graphite-dependences.c: Simplify as if
1986 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
1987 * graphite-isl-ast-to-gimple.c: Likewise.
1988 * graphite-optimize-isl.c: Likewise.
1989 * graphite-poly.c: Likewise.
1990 * graphite-sese-to-poly.c: Likewise.
1991 * graphite.h: Likewise.
1992 * toplev.c: Include isl/version.h and use isl_version () for
1993 printing the ISL version.
1994 * doc/install.texi: Update ISL requirement.
1996 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
1998 * doc/standards.texi (Standards): Update reference to
2001 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
2003 * doc/extend.texi (Named Address Spaces): sourceware.org now
2005 * doc/install.texi (Binaries): Ditto.
2008 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
2010 * doc/cpp.texi: Replace "stringify"/"stringification" with C
2011 standard terminology "stringize"/"stringizing" throughout.
2012 * doc/cppinternals.texi: Likewise.
2014 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
2016 * doc/extend.texi: Fix some spelling mistakes and typos.
2017 * doc/invoke.texi: Likewise.
2019 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
2022 * params.def (inline-min-speedup) Change from 10 to 8.
2024 2017-02-11 Jakub Jelinek <jakub@redhat.com>
2026 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
2029 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
2032 * ipa-inline-analysis.c (get_minimal_bb): New function.
2033 (record_modified): Use it.
2034 (remap_edge_change_prob): Handle also ancestor functions.
2036 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
2038 * doc/contrib.texi (Contributors): Remove broken link into
2039 the Mauve CVS repository.
2041 2017-02-11 Jakub Jelinek <jakub@redhat.com>
2044 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
2045 result computation whenever lhs doesn't have vector mode, not
2046 just when it has BLKmode.
2048 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
2050 * doc/makefile.texi (profiledbootstrap): Refer to the
2051 installation instructions only in textual form.
2053 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2056 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
2058 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
2060 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
2061 (Specific): Update mingw-w64 reference.
2063 (Specific): Remove broken link to Renesas RX processor.
2065 2017-02-10 Richard Biener <rguenther@suse.de>
2067 * toplev.c (process_options): Do not mention obsolete graphite
2068 options when printing sorry message about missing graphite support.
2069 Mention -floop-nest-optimize.
2071 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
2073 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
2074 (vtst_p16): Likewise.
2075 (vtstq_p8): Likewise.
2076 (vtstq_p16): Likewise.
2078 (vtstq_p64): Likewise.
2079 * config/arm/arm_neon.h (vgetq_lane_p64): New.
2080 (vset_lane_p64): New.
2081 (vsetq_lane_p64): New.
2083 2017-02-10 Jakub Jelinek <jakub@redhat.com>
2085 PR tree-optimization/79411
2086 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
2087 stmt operands are SSA_NAMEs used in abnormal phis.
2088 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
2091 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
2094 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
2097 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
2099 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
2101 2017-02-09 Jakub Jelinek <jakub@redhat.com>
2103 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
2107 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
2108 not arbitrary TREE_CONSTANT.
2111 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
2112 "omp declare target link" attribute unless is_global_var.
2113 * omp-offload.c (find_link_var_op): Likewise.
2115 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
2116 Chung-Lin Tang <cltang@codesourcery.com>
2118 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
2120 (gimplify_adjust_omp_clauses): Don't delete TILE.
2121 (gimplify_omp_for): Deal with TILE.
2122 * internal-fn.c (expand_GOACC_TILE): New function.
2123 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
2125 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
2126 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
2128 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
2129 avoid DIV for outermost collapse var.
2130 (expand_oacc_for): Insert tile element loop as needed. Adjust.
2131 Remove out of date comments, fix whitespace.
2132 * omp-general.c (omp_extract_for_data): Deal with tiling.
2133 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
2134 adjust OLF_DIM_BASE value.
2135 (struct omp_for_data): Add tiling field.
2136 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
2137 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
2138 for auto loops. Remove default auto determining, moved to
2139 oacc_loop_fixed_partitions.
2140 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
2141 stmts, add e_mask field.
2142 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
2143 (oacc_thread_numbers): Use oacc_dim_call.
2144 (oacc_xform_tile): New.
2145 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
2146 (finish_oacc_loop): Adjust for ifns vector.
2147 (oacc_loop_discover_walk): Append loop abstraction sites to list,
2148 add case for GOACC_TILE fns.
2149 (oacc_loop_xform_loop): Delete.
2150 (oacc_loop_process): Iterate over call list directly, and add
2151 handling for GOACC_TILE fns.
2152 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
2154 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
2155 vector partitioning to outer loops. Assign 2 partitions to loops
2156 when available. Add TILE handling.
2157 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
2158 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
2159 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
2160 * tree.c (omp_clause_num_ops): Adjust TILE ops.
2161 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
2163 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
2165 * configure.ac (ACX_BUGURL): Update.
2166 * configure: Regenerate.
2168 2017-02-09 Richard Biener <rguenther@suse.de>
2170 PR tree-optimization/69823
2171 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
2172 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
2174 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
2176 * config/arc/arc-c.def: Add __NPS400__ definition.
2177 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
2178 (TARGET_NPS400): Define.
2180 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
2182 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
2184 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
2186 (arc_cpu_types): Fill the arch_info field with a pointer into the
2187 arc_arch_types table.
2188 (arc_selected_cpu): Declare.
2189 * config/arc/arc.c (arc_selected_cpu): Make global.
2190 (arc_selected_arch): Delete.
2191 (arc_base_cpu): Delete.
2192 (arc_override_options): Remove references to deleted variables,
2193 update access to arch information.
2194 (ARC_OPT): Update access to arch information.
2195 (ARC_OPTX): Likewise.
2196 * config/arc/arc.h (arc_base_cpu): Remove declaration.
2197 (TARGET_ARC600): Update access to arch information.
2198 (TARGET_ARC601): Likewise.
2199 (TARGET_ARC700): Likewise.
2200 (TARGET_EM): Likewise.
2201 (TARGET_HS): Likewise.
2202 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
2205 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
2208 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
2209 condition/operands for integer GE/LE/GEU/LEU operations.
2211 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
2213 PR translation/79397
2214 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
2217 2017-02-08 Martin Jambor <mjambor@suse.cz>
2220 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
2221 whether allocation happened.
2222 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
2223 nothing was allocated.
2225 2017-02-08 Jakub Jelinek <jakub@redhat.com>
2227 PR tree-optimization/79408
2228 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
2229 constant, but SSA_NAME with a known integer range, use the minimum
2230 of that range instead of op1 to determine if modulo can be replaced
2231 with its first operand.
2233 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2235 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
2237 2017-02-08 Richard Biener <rguenther@suse.de>
2239 PR tree-optimization/71824
2240 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
2241 Check all loops contained in the merged region.
2243 2017-02-07 Andrew Pinski <apinski@cavium.com>
2245 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
2247 2017-02-07 Andrew Pinski <apinski@cavium.com>
2249 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
2250 (thunderxt88): Likewise.
2251 (thunderxt81): Disable LSE and change v8.1 to v8.
2252 (thunderxt83): Likewise.
2254 2017-02-07 Jakub Jelinek <jakub@redhat.com>
2255 Richard Biener <rguenther@suse.de>
2258 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
2259 type from int to size_t.
2260 * ira-costs.c (struct_costs_size): Change type from int to size_t.
2262 2017-02-07 Jakub Jelinek <jakub@redhat.com>
2264 PR rtl-optimization/79386
2265 * cprop.c (bypass_conditional_jumps): Initialize
2266 bypass_last_basic_block already before splitting bbs after
2267 unconditional traps...
2268 (bypass_conditional_jumps): ... rather than here.
2271 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
2272 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
2273 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
2274 fix -masm=intel patterns.
2276 2017-02-07 Richard Biener <rguenther@suse.de>
2278 PR tree-optimization/79256
2280 * builtins.c (get_object_alignment_2): Use min_align_of_type
2281 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
2282 and ADJUST_FIELD_ALIGN.
2284 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
2286 * doc/tm.texi: Regenerate.
2287 * stor-layout.c (layout_decl): Adjust.
2288 (update_alignment_for_field): Likewise.
2289 (place_field): Likewise.
2290 (min_align_of_type): Likewise.
2291 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
2292 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
2293 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
2294 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
2295 * config/frv/frv.c (frv_adjust_field_align): Likewise.
2296 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
2297 * config/i386/i386.c (x86_field_alignment): Likewise.
2298 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
2299 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
2300 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
2301 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
2302 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
2303 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
2307 2017-01-30 Richard Biener <rguenther@suse.de>
2309 PR tree-optimization/79256
2310 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
2311 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
2314 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
2316 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
2317 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
2318 builtins to SImode and emit a zero-extend, if necessary.
2320 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2322 * docs/invoke.texi (RISC-V Options): Alphabetize.
2324 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2326 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
2329 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2331 * config/riscv/riscv.c: New file.
2332 * gcc/common/config/riscv/riscv-common.c: Likewise.
2333 * config.gcc: Likewise.
2334 * config/riscv/constraints.md: Likewise.
2335 * config/riscv/elf.h: Likewise.
2336 * config/riscv/generic.md: Likewise.
2337 * config/riscv/linux.h: Likewise.
2338 * config/riscv/multilib-generator: Likewise.
2339 * config/riscv/peephole.md: Likewise.
2340 * config/riscv/pic.md: Likewise.
2341 * config/riscv/predicates.md: Likewise.
2342 * config/riscv/riscv-builtins.c: Likewise.
2343 * config/riscv/riscv-c.c: Likewise.
2344 * config/riscv/riscv-ftypes.def: Likewise.
2345 * config/riscv/riscv-modes.def: Likewise.
2346 * config/riscv/riscv-opts.h: Likewise.
2347 * config/riscv/riscv-protos.h: Likewise.
2348 * config/riscv/riscv.h: Likewise.
2349 * config/riscv/riscv.md: Likewise.
2350 * config/riscv/riscv.opt: Likewise.
2351 * config/riscv/sync.md: Likewise.
2352 * config/riscv/t-elf-multilib: Likewise.
2353 * config/riscv/t-linux: Likewise.
2354 * config/riscv/t-linux-multilib: Likewise.
2355 * config/riscv/t-riscv: Likewise.
2356 * configure.ac: Likewise.
2357 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
2358 Waterman as RISC-V maintainers.
2359 * doc/install.texi: Add RISC-V entries.
2360 * doc/invoke.texi: Add RISC-V options section.
2361 * doc/md.texi: Add RISC-V constraints section.
2362 * configure: Regenerated.
2364 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2367 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
2368 false values to be constant vectors with all 0 or all 1 bits set.
2369 (vcondu<mode><mode>): Likewise.
2370 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
2372 (fpmask_comparison_operator): Update comment.
2373 (vecint_comparison_operator): New predicate.
2374 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
2375 vector conditionals when the true and false values are constant
2376 vectors with all 0 bits or all 1 bits set.
2378 2017-02-06 Martin Sebor <msebor@redhat.com>
2380 PR tree-optimization/79376
2381 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
2383 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
2385 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
2386 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
2387 to simplify split condition.
2389 2017-02-06 Jakub Jelinek <jakub@redhat.com>
2391 * omp-expand.c (oxpand_omp_atomic_fetch_op,
2392 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
2395 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
2397 PR rtl-optimization/68664
2398 * target.def (can_speculate_insn): New hook.
2399 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
2400 * doc/tm.texi: Regenerate.
2401 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
2402 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
2403 (rs6000_sched_can_speculate_insn): New function.
2405 2017-02-06 Jakub Jelinek <jakub@redhat.com>
2407 PR tree-optimization/79284
2408 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
2409 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
2410 vectorizable_mask_load_store, vectorizable_operation,
2411 vect_is_simple_cond, get_same_sized_vectype): Use it instead
2412 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
2413 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
2414 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
2415 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
2416 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
2417 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
2418 is_gimple_assign (stmt). Replace another such test with
2419 is_gimple_assign (stmt).
2421 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
2424 * config/avr/avr.c (rtl-iter.h): Include it.
2425 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
2426 (avr_legitimate_combined_insn): ...and implementation.
2428 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
2430 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
2431 * config/s390/s390.c (s390_const_operand_ok)
2432 (s390_canonicalize_comparison, s390_extract_part)
2433 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
2434 (s390_contiguous_bitmask_p, s390_rtx_costs)
2435 (legitimize_pic_address): Likewise.
2436 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
2437 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
2438 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
2439 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
2440 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
2442 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
2444 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
2445 REGNO($0) == REGNO($1).
2447 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2449 * config/s390/linux.h(SIZE_TYPE): Add comment.
2451 2017-02-06 Julian Brown <julian@codesourcery.com>
2452 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
2453 Virendra Pathak <virendra.pathak@broadcom.com>
2455 * config/aarch64/aarch64-cores.def: Change the scheduler
2457 * config/aarch64/aarch64.md: Include thunderx2t99.md.
2458 * config/aarch64/thunderx2t99.md: New file.
2460 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2462 * doc/standards.texi (Go Language): Update link to language
2465 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
2467 * tree-eh.c (lower_resx): Sanitize profile.
2468 (cleanup_empty_eh_move_lp): Likewise.
2470 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
2473 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
2475 * cfgloopmanip.h (loop_version): Update prototype.
2476 * modulo-sched.c (sms_schedule): Update call of loop_version.
2477 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
2478 * tree-parloops.c (gen_parallel_loop): Likewise.
2479 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
2480 * tree-ssa-loop-split.c (split_loop): Likewise.
2481 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
2482 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
2484 2017-02-05 Martin Liska <mliska@suse.cz>
2487 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
2489 (print_operand_address): Initialize a struct to zero.
2491 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2493 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
2494 garbage collector only in textual form.
2496 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2498 * doc/extend.texi (x86 specific memory model extensions for
2499 transactional memory): Simplify a phrase.
2501 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
2504 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
2505 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
2506 (atomic_storedi_1): Likewise.
2508 2017-02-04 Jakub Jelinek <jakub@redhat.com>
2510 PR tree-optimization/79338
2511 * tree-parloops.c (gather_scalar_reductions): Don't call
2512 vect_analyze_loop_form for loop->inner before destroying loop's
2515 2017-02-03 Martin Sebor <msebor@redhat.com>
2517 PR tree-optimization/79327
2518 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
2519 when precision has resulted in leading zeros.
2520 (format_integer): Adjust the likely counter to assume an unknown
2521 argument that may be zero is non-zero.
2523 2017-02-03 Jason Merrill <jason@redhat.com>
2526 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
2527 avoid copying non-taken branch.
2529 2017-02-03 Jakub Jelinek <jakub@redhat.com>
2531 PR tree-optimization/79340
2532 * tree-vect-loop.c (vectorizable_reduction): Release
2533 vec_defs elements after safe_splicing them into other vectors.
2536 PR tree-optimization/79327
2537 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
2538 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
2540 (format_integer): Use wide_int_to_tree instead of build_int_cst
2541 + to_?hwi. If argmin is NULL, just set argmin and argmax to
2542 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
2543 of shortest and longest sequence.
2545 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
2547 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
2548 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
2550 2017-02-03 Walter Lee <walt@tilera.com>
2553 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
2554 after initial stackframe link reg save.
2555 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
2557 2017-02-03 Jakub Jelinek <jakub@redhat.com>
2560 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
2561 wu for stxssp alternative.
2563 2017-02-03 Martin Sebor <msebor@redhat.com>
2565 PR tree-optimization/79352
2566 * gimple-fold.c (get_range_strlen): Add argument.
2567 (get_range_strlen): Change return type to bool.
2568 (get_maxval_strlen): Pass in a dummy argument.
2569 * gimple-fold.h (get_range_strlen): Change return type to bool.
2570 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
2571 * tree.h (array_at_struct_end_p): Add argument.
2572 * tree.c (array_at_struct_end_p): Handle it.
2574 2017-02-03 Martin Liska <mliska@suse.cz>
2577 * multiple_target.c (create_dispatcher_calls): Redirect edge
2578 from a caller of a dispatcher.
2579 (expand_target_clones): Make the clones local.
2580 (ipa_target_clone): Do both target clones and resolvers.
2581 (ipa_dispatcher_calls): Remove the pass.
2582 (pass_dispatcher_calls::gate): Likewise.
2583 (make_pass_dispatcher_calls): Likewise.
2584 * passes.def (pass_target_clone): Put as very first IPA early
2587 2017-02-03 Martin Liska <mliska@suse.cz>
2589 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
2590 in case of a function with ifunc attribute.
2592 2017-02-03 Martin Liska <mliska@suse.cz>
2594 * cgraph.c (cgraph_node::dump): Dump function version info.
2595 * symtab.c (symtab_node::dump_base): Add missing new line.
2597 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
2599 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
2600 (ifcombine_ifandif): Use it.
2602 2017-02-03 Martin Liska <mliska@suse.cz>
2604 * doc/invoke.texi: Document default value for
2605 use-after-scope-direct-emission-threshold.
2607 2017-02-03 Martin Liska <mliska@suse.cz>
2609 PR tree-optimization/79339
2610 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
2611 (format_floating): Likewise.
2613 2017-02-03 Martin Liska <mliska@suse.cz>
2616 * ipa-prop.c (ipa_node_params_t::insert): Remove current
2618 (ipa_node_params_t::remove): Likewise.
2619 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
2620 initialization from removed ipa_node_params_t::insert.
2621 (ipa_node_params::~ipa_node_params): Move from removed
2622 ipa_node_params_t::release.
2623 * symbol-summary.h (symbol_summary::m_released): New member.
2624 Do not release a summary twice. Do not allow to call finalizer
2625 for types of a summary that live in GGC memory.
2627 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
2629 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
2632 2017-02-02 Martin Sebor <msebor@redhat.com>
2635 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
2636 (format_string): Tighten up the range of output for non-constant
2637 strings and correct the expected range for wide non-constant strings.
2639 2017-02-02 Martin Sebor <msebor@redhat.com>
2641 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
2644 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
2646 (-fdump-tree-@var): Add to index and document how to come up
2647 with pass-specific option and dump file names.
2648 (-fdump-passes): Clarify where to look for output.
2650 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
2653 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
2654 statistics of the analyzed path; allow threading for speed when
2655 any of BBs along the path are optimized for speed.
2657 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
2660 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
2661 settings of the virtual registers.
2664 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
2666 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2667 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
2670 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2672 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
2673 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
2675 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2677 * config/s390/s390.md: Add missing comments with the expanded
2679 * config/s390/vector.md: Likewise.
2680 * config/s390/vx-builtins.md: Likewise.
2682 2017-02-02 Jakub Jelinek <jakub@redhat.com>
2685 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
2686 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
2687 conditions on a single line.
2689 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2691 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
2692 __S390_VX__ to __VX__.
2694 2017-02-01 Andrew Pinski <apinski@cavium.com>
2696 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
2697 stmt_info to record_stmt_cost.
2698 (vect_get_known_peeling_cost): Pass stmt_info if known to
2700 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
2701 cpu_vector_cost field into
2702 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
2703 field into vec_int_stmt_cost and vec_fp_stmt_cost.
2704 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
2705 splitting of scalar_stmt_cost and vec_stmt_cost.
2706 (thunderx_vector_cost): Likewise.
2707 (cortexa57_vector_cost): LIkewise.
2708 (exynosm1_vector_cost): Likewise.
2709 (xgene1_vector_cost): Likewise.
2710 (thunderx2t99_vector_cost): Improve after the splitting of the two
2712 (aarch64_builtin_vectorization_cost): Update for the splitting of
2713 scalar_stmt_cost and vec_stmt_cost.
2715 2017-02-01 Torvald Riegel <triegel@redhat.com>
2716 Richard Henderson <rth@redhat.com>
2718 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
2719 conditional on existance of a fast atomic load.
2720 * optabs-query.c (can_atomic_load_p): New function.
2721 * optabs-query.h (can_atomic_load_p): Declare it.
2722 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
2723 no fast atomic load is available for the particular size of access.
2724 (expand_atomic_compare_and_swap): Likewise.
2725 (expand_atomic_load): Likewise.
2726 (expand_atomic_store): Likewise.
2727 (expand_atomic_fetch_op): Likewise.
2728 * testsuite/lib/target-supports.exp
2729 (check_effective_target_sync_int_128): Remove x86 because it provides
2730 no fast atomic load.
2731 (check_effective_target_sync_int_128_runtime): Likewise.
2733 2017-02-01 Richard Biener <rguenther@suse.de>
2735 * graphite.c: Include tree-vectorizer.h for find_loop_location.
2736 (graphite_transform_loops): Provide opt-info for optimized nests.
2737 * tree-parloop.c (parallelize_loops): Provide opt-info for
2740 2017-02-01 Richard Biener <rguenther@suse.de>
2743 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
2746 2017-02-01 Richard Biener <rguenther@suse.de>
2748 PR tree-optimization/71824
2749 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
2750 Verify the loops are valid in the merged SESE region.
2751 (scop_detection::can_represent_loop_1): Check analyzing the
2752 evolution of the number of iterations in the region succeeds.
2754 2017-01-31 Ian Lance Taylor <iant@golang.org>
2756 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
2757 REG_ARGS_SIZE note to 32-bit push insns and call insn.
2759 2017-01-31 David Malcolm <dmalcolm@redhat.com>
2761 PR preprocessor/79210
2762 * input.c (get_substring_ranges_for_loc): Replace line_width
2763 assertion with error-handling.
2765 2017-01-31 Richard Biener <rguenther@suse.de>
2767 PR tree-optimization/77318
2768 * graphite-sese-to-poly.c (extract_affine): Fix assert.
2769 (create_pw_aff_from_tree): Take loop parameter.
2770 (add_condition_to_pbb): Pass loop of the condition to
2771 create_pw_aff_from_tree.
2773 2017-01-31 Jakub Jelinek <jakub@redhat.com>
2775 * config/s390/s390.c (s390_asan_shadow_offset): New function.
2776 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
2778 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2782 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
2784 (convert_int_to_float128): Likewise.
2785 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
2786 (convert_int_to_float128): Likewise.
2787 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
2788 (UNSPEC_IEEE128_CONVERT): Likewise.
2789 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
2790 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
2791 Use local variables for IBM extended format.
2792 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
2793 (fix_trunc<mode>si2_fprs): Likewise.
2794 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
2795 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
2796 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
2797 to know that we can now have integers of all sizes in vector
2799 (fix<uns>_<mode>di2_hw): Likewise.
2800 (float<uns>_<mode>si2_hw): Likewise.
2801 (fix_<mode>si2_hw): Likewise.
2802 (fixuns_<mode>si2_hw): Likewise.
2803 (float<uns>_<mode>di2_hw): Likewise.
2804 (float_<mode>di2_hw): Likewise.
2805 (float_<mode>si2_hw): Likewise.
2806 (floatuns_<mode>di2_hw): Likewise.
2807 (floatuns_<mode>si2_hw): Likewise.
2808 (xscvqp<su>wz_<mode>): Delete, no longer used.
2809 (xscvqp<su>dz_<mode>): Likewise.
2810 (xscv<su>dqp_<mode>): Likewise.
2811 (ieee128_mfvsrd_64bit): Likewise.
2812 (ieee128_mfvsrd_32bit): Likewise.
2813 (ieee128_mfvsrwz): Likewise.
2814 (ieee128_mtvsrw): Likewise.
2815 (ieee128_mtvsrd_64bit): Likewise.
2816 (ieee128_mtvsrd_32bit): Likewise.
2818 2017-01-31 Martin Liska <mliska@suse.cz>
2821 * ipa-prop.c (ipa_free_all_node_params): Call release method
2822 instead of ~sumbol_summary to not to trigger double times
2825 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
2827 PR tree-optimization/71691
2828 * bitmap.h (class auto_bitmap): New.
2829 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
2830 is_maybe_undefined instead of ssa_undefined_value_p.
2832 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2834 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
2835 __S390_ARCH_LEVEL__ to __ARCH__.
2837 2017-01-31 Jakub Jelinek <jakub@redhat.com>
2839 PR tree-optimization/79267
2840 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
2841 if should_remove_lhs_p is true.
2843 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
2846 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
2847 (add_alignment_attribute): New.
2848 (base_type_die): Add alignment attribute.
2849 (subrange_type_die): Likewise.
2850 (modified_type_die): Likewise.
2851 (gen_array_type_die): Likewise.
2852 (gen_descr_array_type_die: Likewise.
2853 (gen_enumeration_type_die): Likewise.
2854 (gen_subprogram_die): Likewise.
2855 (gen_variable_die): Likewise.
2856 (gen_field_die): Likewise.
2857 (gen_ptr_to_mbr_type_die): Likewise.
2858 (gen_struct_or_union_type_die): Likewise.
2859 (gen_subroutine_type_die): Likewise.
2860 (gen_typedef_die): Likewise.
2861 (base_type_cmp): Compare alignment attribute.
2863 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2866 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
2867 (setb_unsigned) New pattern for setb with CCUNS.
2868 * config/rs6000/rs6000.c (expand_block_compare): Use a different
2869 subfc./subfe sequence to avoid overflow problems. Generate a
2870 shorter sequence with cmpld/setb for power9.
2871 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
2872 for generating subfc. instruction.
2873 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
2874 now uses this instruction.
2876 2017-01-30 Ian Lance Taylor <iant@google.com>
2879 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
2880 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
2882 2017-01-30 Martin Sebor <msebor@redhat.com>
2884 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
2885 Move constant to the right of a relational operator.
2886 (get_mpfr_format_length, format_character, format_string): Ditto.
2887 (should_warn_p, maybe_warn): Same.
2889 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
2891 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
2894 * asan.c (get_translation_unit_decl): Remove function.
2895 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
2897 2017-01-30 Martin Liska <mliska@suse.cz>
2899 PR gcov-profile/79259
2900 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
2903 2017-01-30 Martin Liska <mliska@suse.cz>
2906 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
2907 Initialize variables with NULL value.
2909 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
2912 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
2914 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
2916 2017-01-30 Richard Biener <rguenther@suse.de>
2918 PR tree-optimization/79276
2919 * tree-vrp.c (process_assert_insertions): Properly adjust common
2920 when removing a duplicate.
2922 * gcc.dg/torture/pr79276.c: New testcase.
2924 2017-01-30 Richard Biener <rguenther@suse.de>
2926 PR tree-optimization/79256
2927 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
2928 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
2930 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
2932 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
2935 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
2936 ("*r<noxa>sbg_<mode>_sll_bitmask")
2937 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
2938 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
2939 Use contiguous_bitmask_nowrap_operand.
2941 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2944 * config/rs6000/altivec.h (vec_xl): Revise #define.
2945 (vec_xst): Likewise.
2947 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
2949 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
2951 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
2953 PR rtl-optimization/79194
2954 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
2955 traps before call to bypass_conditional_jumps.
2957 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
2959 PR tree-optimization/71374
2960 * lra-constraints.c (check_conflict_input_operands): New.
2961 (match_reload): Use it.
2963 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
2966 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
2967 account to calculate conflict_set.
2969 2017-01-27 Bin Cheng <bin.cheng@arm.com>
2971 PR rtl-optimization/78559
2972 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
2973 other_insn in combine.
2975 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
2977 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
2978 uint16_type_node for BT_UINT16.
2980 2017-01-27 David Malcolm <dmalcolm@redhat.com>
2982 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
2983 "RTL Tests" to menu.
2984 (GIMPLE Tests): New node.
2985 (RTL Tests): New node.
2987 2017-01-27 Richard Biener <rguenther@suse.de>
2989 PR tree-optimization/79245
2990 * tree-loop-distribution.c (distribute_loop): Apply cost
2991 modeling also to detected patterns.
2993 2017-01-27 Richard Biener <rguenther@suse.de>
2995 PR tree-optimization/71433
2996 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
2997 (compare_assert_loc): New function.
2998 (process_assert_insertions): Sort and optimize assert locations
2999 to remove duplicates and push down identical assertions on
3000 edges to their destination block.
3002 2017-01-27 Richard Biener <rguenther@suse.de>
3004 PR tree-optimization/79244
3005 * tree-vrp.c (remove_range_assertions): Forcefully propagate
3006 out SSA names even if abnormal.
3008 2017-01-27 Jakub Jelinek <jakub@redhat.com>
3010 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
3011 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
3012 instead of MPFR_RNDN.
3014 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
3017 * arm.c (arm_option_override): Don't call build_target_option_node
3018 until after doing all option overrides.
3019 (arm_valid_target_attribute_tree): Likewise.
3021 2017-01-27 Martin Liska <mliska@suse.cz>
3023 * doc/invoke.texi (-fprofile-arcs): Document profiling support
3024 for {cd}tors and C++ {cd}tors.
3026 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
3028 * config/s390/s390.md ("*setmem_long_and")
3029 ("*setmem_long_and_31z"): Use zero_extend instead of and.
3031 2017-01-26 Martin Sebor <msebor@redhat.com>
3033 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
3036 2017-01-26 Martin Sebor <msebor@redhat.com>
3038 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
3039 HAVE_DFmode before using XFmode or DFmode.
3040 (parse_directive): Avoid using the z length modifier to avoid
3041 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
3044 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
3045 to accept adjustment as an array.
3046 (get_int_range): New function.
3047 (struct directive): Make width and prec arrays.
3048 (directive::set_width, directive::set_precision): Call get_int_range.
3049 (format_integer, format_floating): Handle width and precision ranges.
3050 (format_string, parse_directive): Same.
3052 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3055 * dwarf2out.c (generate_skeleton_bottom_up): For children with
3056 comdat_type_p set, just clone them, but keep the children in the
3060 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
3061 which have direct callers with -fvar-tracking-assignments enabled
3063 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
3064 inside of type units.
3066 2017-01-26 Martin Sebor <msebor@redhat.com>
3069 * gimple-ssa-sprintf.c (struct result_range): Add likely and
3071 (struct format_result): Replace number_chars, number_chars_min,
3072 and number_chars_max with a single member of struct result_range.
3074 (format_result::operator+=): Adjust.
3075 (struct fmtresult): Remove bounded. Handle likely and unlikely
3077 (fmtresult::adjust_for_width_or_precision): New function.
3078 (fmtresult:type_max_digits): New function.
3079 (bytes_remaining): Handle likely and unlikely counters.
3080 (min_bytes_remaining): Remove.
3081 (format_percent): Simplify.
3082 (format_integer, format_floating): Set likely and unlikely counters.
3083 (get_string_length, format_character, format_string): Same.
3084 (format_plain, should_warn_p): New function.
3085 (maybe_warn): Call should_warn_p. Update diagnostic messages
3086 and handle those for all directives, including plain strings.
3087 (format_directive): Handle likely and unlikely counters.
3088 Remove unnecessary quoting from diagnostics. Add an informational
3090 (add_bytes): Remove.
3091 (pass_sprintf_length::compute_format_length): Simplify.
3092 (try_substitute_return_value): Handle likely and unlikely counters.
3094 2017-01-26 Carl Love <cel@us.ibm.com>
3096 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
3097 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
3099 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
3102 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
3103 endianess for subregs into account.
3104 * lra-constraints.c (lra_constraints): Do risky transformations
3105 always on the first iteration.
3106 * lra-lives.c (check_pseudos_live_through_calls): Add arg
3107 last_call_used_reg_set.
3108 (process_bb_lives): Define and use last_call_used_reg_set.
3109 * lra.c (lra): Always continue after lra_constraints on the first
3112 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
3114 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
3116 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
3118 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3120 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
3121 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
3122 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
3123 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
3124 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
3125 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
3126 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
3127 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
3128 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
3130 2017-01-26 Marek Polacek <polacek@redhat.com>
3133 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
3134 for the third operand.
3136 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3139 * omp-low.c (struct omp_context): Add simt_stmt field.
3140 (scan_omp_for): Return omp_context *.
3141 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
3142 context to the _simt_ SIMD stmt.
3143 (lower_omp_for): For combined SIMD with sibling _simt_
3144 SIMD, make sure to use the same decls in _looptemp_
3145 clauses as in the sibling.
3147 2017-01-26 David Sherwood <david.sherwood@arm.com>
3150 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
3153 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3156 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
3157 emit fld b; fld a; if possible.
3159 * brig-builtins.def: Update copyright years.
3160 * config/arm/arm_acle_builtins.def: Update copyright years.
3162 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
3165 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
3166 constraint instead of o for the stxsd instruction.
3168 2017-01-25 Carl Love <cel@us.ibm.com>
3170 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
3171 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
3173 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
3175 * doc/invoke.texi (C++ Dialect Options): Fix typo.
3177 2017-01-25 Richard Biener <rguenther@suse.de>
3179 PR tree-optimization/69264
3180 * target.def (vector_alignment_reachable): Improve documentation.
3181 * doc/tm.texi: Regenerate.
3182 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
3184 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
3185 earlier changes with respect to TYPE_USER_ALIGN.
3186 (vector_alignment_reachable_p): Likewise. Improve dumping.
3188 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3191 * config/arm/arm.md (xordi3): Force constant operand into a register
3194 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3196 * doc/invoke.texi (-fstore-merging): Correct default optimization
3197 levels at which it is enabled.
3198 (-O): Move -fstore-merging from list to...
3201 2017-01-25 Richard Biener <rguenther@suse.de>
3204 * omp-expand.c: Include debug.h.
3205 (expand_omp_taskreg): Make sure to generate early debug before
3206 outlining anything from a function.
3207 (expand_omp_target): Likewise.
3208 (grid_expand_target_grid_body): Likewise.
3210 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
3213 * asan.c (get_translation_unit_decl): New function.
3214 (asan_add_global): Extract modules file name from globals
3215 TRANSLATION_UNIT_DECL name.
3217 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
3220 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
3221 for long calls with APCS frame and VFP.
3223 2017-01-24 David Malcolm <dmalcolm@redhat.com>
3225 * cfg.c (original_copy_tables_initialized_p): New function.
3226 * cfg.h (original_copy_tables_initialized_p): New decl.
3227 * cfgrtl.c (relink_block_chain): Guard the call to
3228 free_original_copy_tables with a call to
3229 original_copy_tables_initialized_p.
3230 * cgraph.h (symtab_node::native_rtl_p): New decl.
3231 * cgraphunit.c (symtab_node::native_rtl_p): New function.
3232 (symtab_node::needed_p): Don't assert for early assembly output
3233 for __RTL functions.
3234 (cgraph_node::finalize_function): Set "force_output" for __RTL
3236 (cgraph_node::analyze): Bail out early for __RTL functions.
3237 (analyze_functions): Update assertion to support __RTL functions.
3238 (cgraph_node::expand): Bail out early for __RTL functions.
3239 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
3241 * function.h (struct function): Update comment for field
3243 * gimple-expr.c: Include "tree-pass.h".
3244 (gimple_has_body_p): Return false for __RTL functions.
3245 * Makefile.in (OBJS): Add run-rtl-passes.o.
3246 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
3248 (gcc::pass_manager::get_clean_slate): New accessor.
3249 * passes.c: Include "insn-addr.h".
3250 (should_skip_pass_p): Add logging. Update logic for running
3251 "expand" to be compatible with both __GIMPLE and __RTL. Guard
3252 property-provider override so it is only done for gimple passes.
3254 (skip_pass): New function.
3255 (execute_one_pass): Call skip_pass when skipping passes.
3256 * read-md.c (md_reader::read_char): Support filtering
3257 the input to a subset of line numbers.
3258 (md_reader::md_reader): Initialize fields
3259 m_first_line and m_last_line.
3260 (md_reader::read_file_fragment): New function.
3261 * read-md.h (md_reader::read_file_fragment): New decl.
3262 (md_reader::m_first_line): New field.
3263 (md_reader::m_last_line): New field.
3264 * read-rtl-function.c (function_reader::create_function): Only
3265 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
3266 curr_properties. Set DECL_INITIAL to a dummy block.
3267 (read_rtl_function_body_from_file_range): New function.
3268 * read-rtl-function.h (read_rtl_function_body_from_file_range):
3270 * run-rtl-passes.c: New file.
3271 * run-rtl-passes.h: New file.
3273 2017-01-24 Jeff Law <law@redhat.com>
3275 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
3278 2017-01-24 Bin Cheng <bin.cheng@arm.com>
3280 PR tree-optimization/79159
3281 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
3282 (record_nonwrapping_iv): Improve boundary using above function if no
3283 value range information.
3285 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
3286 Martin Jambor <mjambor@suse.cz>
3288 * brig-builtins.def: New file.
3289 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
3290 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
3291 (DEF_HSAIL_SAT_BUILTIN): Likewise.
3292 (DEF_HSAIL_INTR_BUILTIN): Likewise.
3293 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
3294 * builtin-types.def (BT_INT8): New.
3295 (BT_INT16): Likewise.
3296 (BT_UINT8): Likewise.
3297 (BT_UINT16): Likewise.
3298 (BT_FN_ULONG): Likewise.
3299 (BT_FN_UINT_INT): Likewise.
3300 (BT_FN_UINT_ULONG): Likewise.
3301 (BT_FN_UINT_LONG): Likewise.
3302 (BT_FN_UINT_PTR): Likewise.
3303 (BT_FN_ULONG_PTR): Likewise.
3304 (BT_FN_INT8_FLOAT): Likewise.
3305 (BT_FN_INT16_FLOAT): Likewise.
3306 (BT_FN_UINT32_FLOAT): Likewise.
3307 (BT_FN_UINT16_FLOAT): Likewise.
3308 (BT_FN_UINT8_FLOAT): Likewise.
3309 (BT_FN_UINT64_FLOAT): Likewise.
3310 (BT_FN_UINT16_UINT32): Likewise.
3311 (BT_FN_UINT32_UINT16): Likewise.
3312 (BT_FN_UINT16_UINT16_UINT16): Likewise.
3313 (BT_FN_INT_PTR_INT): Likewise.
3314 (BT_FN_UINT_PTR_UINT): Likewise.
3315 (BT_FN_LONG_PTR_LONG): Likewise.
3316 (BT_FN_ULONG_PTR_ULONG): Likewise.
3317 (BT_FN_VOID_UINT64_UINT64): Likewise.
3318 (BT_FN_UINT8_UINT8_UINT8): Likewise.
3319 (BT_FN_INT8_INT8_INT8): Likewise.
3320 (BT_FN_INT16_INT16_INT16): Likewise.
3321 (BT_FN_INT_INT_INT): Likewise.
3322 (BT_FN_UINT_FLOAT_UINT): Likewise.
3323 (BT_FN_FLOAT_UINT_UINT): Likewise.
3324 (BT_FN_ULONG_UINT_UINT): Likewise.
3325 (BT_FN_ULONG_UINT_PTR): Likewise.
3326 (BT_FN_ULONG_ULONG_ULONG): Likewise.
3327 (BT_FN_UINT_UINT_UINT): Likewise.
3328 (BT_FN_VOID_UINT_PTR): Likewise.
3329 (BT_FN_UINT_UINT_PTR: Likewise.
3330 (BT_FN_UINT32_UINT64_PTR): Likewise.
3331 (BT_FN_INT_INT_UINT_UINT): Likewise.
3332 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
3333 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
3334 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
3335 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
3336 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
3337 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
3338 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
3339 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
3340 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
3341 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
3342 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
3343 * doc/frontends.texi: List BRIG FE.
3344 * doc/install.texi (Testing): Add BRIG tesring requirements.
3345 * doc/invoke.texi (Overall Options): Mention BRIG.
3346 * doc/standards.texi (Standards): Doucment BRIG HSA version.
3348 2017-01-24 Richard Biener <rguenther@suse.de>
3350 PR translation/79208
3351 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
3353 2017-01-24 Martin Jambor <mjambor@suse.cz>
3356 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
3357 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
3360 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
3363 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
3364 casts from signed to unsigned really don't have a range.
3366 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
3368 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
3369 GMP_RNDx for compatiblity.
3371 2017-01-24 Martin Liska <mliska@suse.cz>
3374 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
3375 that would prevent us to call alloca with -1 as argument.
3377 2017-01-24 Jakub Jelinek <jakub@redhat.com>
3379 * dwarf2out.c (output_compilation_unit_header, output_file_names):
3380 Avoid -Wformat-security warning.
3382 2017-01-23 Andrew Pinski <apinski@cavium.com>
3384 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
3387 2017-01-23 Martin Sebor <msebor@redhat.com>
3390 * gimple-ssa-sprintf.c (warn_level): New global.
3391 (format_integer): Use it here and throughout the rest of the file.
3392 Use the same switch to compute sign as base.
3393 (maybe_warn): New function.
3394 (format_directive): Factor out warnings into maybe_warn.
3395 Add debugging output. Use warn_level.
3396 (add_bytes): Use warn_level.
3397 (pass_sprintf_length::compute_format_length): Add debugging output.
3398 (try_substitute_return_value): Same.
3399 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
3402 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
3403 (struct fmtresult, format_integer, format_floating): Adjust.
3404 (fmtresult::fmtresult): Set max correctly in two argument ctor.
3405 (get_string_length, format_string,format_directive): Same.
3406 (pass_sprintf_length::compute_format_length): Same.
3407 (try_substitute_return_value): Simplify slightly.
3410 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
3411 (fmtresult::operator+=): Outlined.
3412 (struct fmtresult): Add ctors.
3413 (struct conversion_spec): Rename...
3414 (struct directive): ...to this. Add and remove data members.
3415 (directive::set_width, directive::set_precision): New functions.
3416 (format_percent): Use fmtresult ctor.
3417 (get_width_and_precision): Remove.
3418 (format_integer): Make naming changes. Avoid computing width and
3420 (format_floating): Same. Adjust indentation.
3421 (format_character, format_none): New functions.
3422 (format_string): Moved character handling to format_character.
3423 (format_directive): Remove arguments, change return type.
3424 (parse_directive): New function.
3425 (pass_sprintf_length::compute_format_length): Move directive
3426 parsing to parse_directive.
3428 2017-01-23 Jakub Jelinek <jakub@redhat.com>
3430 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
3431 (assign_assembler_name_if_needed): ... this.
3432 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
3433 (assign_assembler_name_if_needed): ... this.
3434 (free_lang_data_in_cgraph): Adjust callers.
3435 * cgraphunit.c (cgraph_node::analyze): Likewise.
3436 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
3439 2017-01-23 Richard Biener <rguenther@suse.de>
3441 PR tree-optimization/79088
3442 PR tree-optimization/79188
3443 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
3444 resetting loop bounds after last path deletion. Reset loop
3445 bounds of the target loop, make code match the comments.
3446 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
3447 Make sure loops need no fixups.
3449 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
3451 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
3452 exponent support with double type for first argument.
3453 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
3454 type returned by __builtin_vec_extract_sig,
3455 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
3456 functions from "vector int" to "vector unsigned int" or from
3457 "vector long long int" to "vector unsigned long long int".
3458 Changed type returned by __builtin_vec_extract_exp,
3459 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
3460 functions from "vector int" to "vector unsigned int" or from
3461 "vector long long int" to "vector unsigned long long int".
3462 Changed return type of __builtin_vec_test_data_class,
3463 __builtin_vec_test_data_class_sp, and
3464 __builtin_vec_test_data_class_dp from "vector int" to
3465 "vector bool int" or from "vector long long int" to "vector bool
3466 long long int" and changed second argument type from "unsigned
3467 int" to "int". Added new overloaded function forms "vector float
3468 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
3469 "vector float __builtin_vec_insert_exp_sp (vector float, vector
3470 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
3471 double, vector unsigned long long int)" and "vector double
3472 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
3473 long int)". Changed return type of
3474 __builtin_scalar_test_data_class and
3475 __builtin_scalar_test_data_class_sp and
3476 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
3477 int" and changed second argument from "unsigned int" to "int".
3478 Changed type returned by __builtin_scalar_test_neg,
3479 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
3480 from "int" to "bool int". Added new overloaded function form
3481 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
3482 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
3483 exponent double-precision with floating point first argument.
3484 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
3485 documentation of scalar_test_data_class, scalar_test_neg,
3486 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
3487 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
3488 vec_test_data_class built-in functions to reflect refinements in
3489 their type signatures.
3491 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
3493 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
3495 (aarch64_elf_asm_destructor): Likewise.
3497 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
3499 PR rtl-optimization/78634
3500 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
3501 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
3502 * ifcvt.c (noce_try_cmove): Add missing cost check.
3504 PR rtl-optimization/71724
3505 * combine.c (if_then_else_cond): Look for situations where it is
3506 beneficial to undo the work of one of the recursive calls.
3508 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3510 PR tree-optimization/70754
3511 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3512 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3513 combined stmt before it if not NULL.
3514 (combine_chains): Process refs reversely and compute dominance point
3517 2017-01-23 Martin Liska <mliska@suse.cz>
3519 PR tree-optimization/79196
3520 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
3521 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
3523 (strlen_optimize_stmt): Call the renamed function.
3525 2017-01-23 Michael Matz <matz@suse.de>
3527 PR tree-optimization/78384
3528 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
3530 2017-01-23 Richard Biener <rguenther@suse.de>
3532 PR tree-optimization/79186
3533 * tree-vrp.c (register_new_assert_for): Make sure we've seen
3534 both incoming edges before moving an assert.
3536 2017-01-23 Martin Jambor <mjambor@suse.cz>
3538 * ipa-prop.c (load_from_param_1): Removed.
3539 (load_from_unmodified_param): Bits from load_from_param_1 put back
3541 (load_from_param): Removed.
3542 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
3543 with stmt. Reverted back to use of load_from_unmodified_param.
3545 2017-01-23 Martin Jambor <mjambor@suse.cz>
3548 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
3549 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
3550 field a pointer to garbage collected vector, mark lattices and
3551 ipcp_orig_node with GTY((skip)).
3552 (ipa_get_param_count): Adjust to descriptors being a pointer.
3553 (ipa_get_param): Likewise.
3554 (ipa_get_type): Likewise.
3555 (ipa_get_param_move_cost): Likewise.
3556 (ipa_set_param_used): Likewise.
3557 (ipa_get_controlled_uses): Likewise.
3558 (ipa_set_controlled_uses): Likewise.
3559 (ipa_is_param_used): Likewise.
3560 (ipa_node_params_t): Move into garbage collector. New methods insert
3562 (ipa_node_params_sum): Annotate wth GTY(()).
3563 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
3565 (ipa_load_from_parm_agg): Adjust declaration.
3566 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
3567 * ipa-profile.c (ipa_profile): Likewise.
3568 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
3569 (ipa_populate_param_decls): Make descriptors parameter garbage
3571 (ipa_dump_param): Adjust to descriptors being a pointer.
3572 (ipa_alloc_node_params): Likewise.
3573 (ipa_initialize_node_params): Likewise.
3574 (load_from_param_1): Make descriptors parameter garbage collected.
3575 (load_from_unmodified_param): Likewise.
3576 (load_from_param): Likewise.
3577 (ipa_load_from_parm_agg): Likewise.
3578 (ipa_node_params::~ipa_node_params): Removed.
3579 (ipa_free_all_node_params): Remove call to delete operator.
3580 (ipa_node_params_t::insert): New.
3581 (ipa_node_params_t::remove): Likewise.
3582 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
3583 copy known_csts and known_contexts vectors.
3584 (ipa_read_node_info): Adjust to descriptors being a pointer.
3585 (ipcp_modif_dom_walker): Make m_descriptors field garbage
3587 (ipcp_transform_function): Make descriptors variable garbage
3590 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
3592 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
3593 * config/i386/avx512dqintrin.h: Ditto.
3594 * config/i386/avx512fintrin.h: Ditto.
3595 * gcc/config/i386/i386.c: Handle new builtins.
3596 * config/i386/i386-builtin.def: Add new builtins.
3597 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
3598 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
3600 2017-01-23 Jakub Jelinek <jakub@redhat.com>
3601 Martin Liska <mliska@suse.cz>
3603 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
3604 * asan.c (asan_expand_poison_ifn): Support stores and use
3605 appropriate ASAN report function.
3606 * internal-fn.c (expand_ASAN_POISON_USE): New function.
3607 * internal-fn.def (ASAN_POISON_USE): Declare.
3608 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
3609 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
3610 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
3611 ASAN_POISON calls w/o LHS.
3612 * tree-ssa.c (execute_update_addresses_taken): Create clobber
3613 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
3614 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
3615 * gimplify.c (asan_poison_variables): Add attribute
3616 use_after_scope_memory to variables that really needs to live
3618 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
3619 having the attribute.
3621 2017-01-23 Martin Liska <mliska@suse.cz>
3623 * asan.c (create_asan_shadow_var): New function.
3624 (asan_expand_poison_ifn): Likewise.
3625 * asan.h (asan_expand_poison_ifn): New declaration.
3626 * internal-fn.c (expand_ASAN_POISON): Likewise.
3627 * internal-fn.def (ASAN_POISON): New builtin.
3628 * sanopt.c (pass_sanopt::execute): Expand
3629 asan_expand_poison_ifn.
3630 * tree-inline.c (copy_decl_for_dup_finish): Make function
3632 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
3633 * tree-ssa.c (is_asan_mark_p): New function.
3634 (execute_update_addresses_taken): Rewrite local variables
3635 (identified just by use-after-scope as addressable) into SSA.
3637 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
3639 * doc/install.texi (Specific): opensource.apple.com uses https
3640 now. Remove trailing slash.
3642 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
3644 * README.Portability: Remove note on an Irix compatibility issue.
3646 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
3648 * gcov.c (INCLUDE_ALGORITHM): Define.
3649 (INCLUDE_VECTOR): Define.
3650 No longer include <vector> and <algorithm> directly.
3652 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
3654 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
3656 * doc/invoke.texi (Code Gen Options): Ditto.
3658 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
3661 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
3663 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
3665 rtl-optimization/79125
3666 * cprop.c (local_cprop_pass): Handle cases where we make an
3669 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
3673 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
3674 read from, for big endian.
3676 2017-01-20 Jiong Wang <jiong.wang@arm.com>
3678 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
3679 register pauth builtins for LP64 only.
3681 2017-01-20 Marek Polacek <polacek@redhat.com>
3684 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
3687 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
3689 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
3691 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
3692 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
3693 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
3695 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3698 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
3699 in big-endian mode when they are not a single duplicated value.
3701 2017-01-20 Richard Biener <rguenther@suse.de>
3703 * BASE-VER: Bump to 7.0.1.
3705 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
3707 * omp-low.c (omplow_simd_context): New struct. Use it...
3708 (lower_rec_simd_input_clauses): ...here and...
3709 (lower_rec_input_clauses): ...here to hold common data. Adjust all
3710 references to idx, lane, max_vf, is_simt.
3712 2017-01-20 Graham Markall <graham.markall@embecosm.com>
3714 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
3717 2017-01-20 Martin Jambor <mjambor@suse.cz>
3719 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
3720 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
3722 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
3723 (GTFILES): Rename hsa.c to hsa-common.c.
3724 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
3725 * hsa-dump.c: Likewise.
3726 * hsa-gen.c: Likewise.
3727 * hsa-regalloc.c: Likewise.
3728 * ipa-hsa.c: Likewise.
3729 * omp-expand.c: Likewise.
3730 * omp-low.c: Likewise.
3731 * toplev.c: Likewise.
3733 2017-01-20 Marek Polacek <polacek@redhat.com>
3736 * doc/invoke.texi: Document -Wduplicated-branches.
3737 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
3738 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
3739 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
3740 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
3741 return 0 only when not OEP_LEXICOGRAPHIC.
3742 (fold_build_cleanup_point_expr): Use the expression
3743 location when building CLEANUP_POINT_EXPR.
3744 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
3745 * tree.c (add_expr): Handle error_mark_node.
3747 2017-01-20 Martin Liska <mliska@suse.cz>
3750 * tree-profile.c (init_ic_make_global_vars): Do not call
3752 (gimple_init_gcov_profiler): Likewise.
3754 2017-01-20 Martin Liska <mliska@suse.cz>
3757 * cgraph.h (maybe_create_reference): Remove argument and
3759 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
3761 * ipa-cp.c (create_specialized_node): Likewise.
3762 * symtab.c (symtab_node::maybe_create_reference): Handle
3763 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
3765 2017-01-20 Martin Liska <mliska@suse.cz>
3767 * read-rtl-function.c (function_reader::create_function): Use
3768 build_decl instread of build_decl_stat.
3770 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
3772 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
3773 * config/i386/avx512dqintrin.h: Ditto.
3774 * config/i386/avx512fintrin.h: Ditto.
3775 * config/i386/i386-builtin-types.def: Add new types.
3776 * gcc/config/i386/i386.c: Handle new types.
3777 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
3778 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
3779 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
3780 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
3781 (__builtin_ia32_kshiftridi): New.
3782 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
3784 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
3788 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
3789 define to rs6000_init_stack_protect_guard.
3790 (rs6000_init_stack_protect_guard): New function.
3792 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
3793 Yunqiang Su <yunqiang.su@imgtec.com>
3795 * config.gcc (supported_defaults): Add madd4.
3796 (with_madd4): Add validation.
3797 (all_defaults): Add madd4.
3798 * config/mips/mips.opt (mmadd4): New option.
3799 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
3801 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
3803 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
3804 (ISA_HAS_FUSED_MADD4): Likewise.
3805 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
3806 * gcc/doc/install.texi (--with-madd4): Document the new option.
3808 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3810 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
3811 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
3812 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
3813 (aarch64_init_pauth_hint_builtins): New.
3814 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
3815 (aarch64_expand_builtin): Expand new builtins.
3817 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3819 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
3820 * combine-stack-adj.c (no_unhandled_cfa): Handle
3821 REG_CFA_TOGGLE_RA_MANGLE.
3822 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
3823 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
3824 info for return address signing.
3825 (aarch64_expand_epilogue): Likewise.
3827 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3829 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
3830 * config/aarch64/aarch64-protos.h
3831 (aarch64_return_address_signing_enabled): New declaration.
3832 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
3834 (aarch64_expand_prologue): Sign return address before it's pushed onto
3836 (aarch64_expand_epilogue): Authenticate return address fetched from
3838 (aarch64_override_options): Sanity check for ILP32 and ISA level.
3839 (aarch64_attributes): New function attributes for "sign-return-address".
3840 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
3841 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
3842 ("*do_return"): Generate combined instructions according to key index.
3843 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
3844 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
3846 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
3847 * config/aarch64/aarch64.opt (msign-return-address=): New.
3848 * doc/extend.texi (AArch64 Function Attributes): Documents
3849 "sign-return-address=".
3850 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
3852 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
3854 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
3855 overall option summary.
3857 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3859 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
3860 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
3861 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
3862 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
3864 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
3866 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
3867 -mpower9-minmax by default for -mcpu=power9.
3868 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
3869 128-bit floating point.
3871 2017-01-20 Alan Modra <amodra@gmail.com>
3873 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
3874 optimizing for size.
3876 2017-01-20 Alan Modra <amodra@gmail.com>
3879 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
3880 for strcmp and strncmp from corresponding builtin decl.
3882 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
3884 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
3885 instead of i386/rtems-64.h.
3886 * config/i386/rtems-64.h: Remove.
3888 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
3892 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
3894 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
3896 2017-01-19 Tamar Christina <tamar.christina@arm.com>
3898 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3899 Change int to HOST_WIDE_INT.
3900 * config/aarch64/aarch64-protos.h
3901 (aarch64_simd_gen_const_vector_dup): Likewise.
3902 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
3904 2017-01-19 David Malcolm <dmalcolm@redhat.com>
3906 * langhooks-def.h (lhd_type_for_size): New decl.
3907 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
3908 * langhooks.c (lhd_type_for_size): New function, taken from
3911 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
3913 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
3914 define_bypass for CR latency.
3915 (power9-cracked-alu): Update bypass latency and remove power9-branch.
3916 (power9-alu2): Add define_bypass for CR latency.
3918 (power9-mul): Update insn latency.
3919 (power9-mul-compare): Update insn latency, bypass latency and remove
3922 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3924 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
3926 * config/aarch64/aarch64.md
3927 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
3928 aarch64_nopcrelative_literal_loads.
3929 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
3931 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
3933 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
3935 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
3937 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
3940 * config.gcc (supported_defaults): Add lxc1-sxc1.
3941 (with_lxc1_sxc1): Add validation.
3942 (all_defaults): Add lxc1-sxc1.
3943 * config/mips/mips.opt (mlxc1-sxc1): New option.
3944 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
3946 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
3947 __mips_no_lxc1_sxc1.
3948 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
3949 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
3950 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
3952 2017-01-19 Richard Biener <rguenther@suse.de>
3954 PR tree-optimization/72488
3955 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
3956 sure to restore SSA info.
3957 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
3959 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
3961 PR rtl-optimization/79121
3962 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
3963 of the inner type when shifting an extended value.
3965 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
3968 * symtab.c (symtab_node::equal_address_to): Fix comparing of
3969 interposable aliases.
3971 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
3974 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
3975 Use the evmergelohi instruction.
3976 (mov_si<mode>_e500_subreg4_2_le): Likewise.
3977 (mov_sitf_e500_subreg8_2_be): Likewise.
3978 (mov_sitf_e500_subreg12_2_le): Likewise.
3979 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
3980 (mov_si<mode>_e500_subreg4_2_be): Likewise.
3981 (mov_sitf_e500_subreg8_2_le): Likewise.
3982 (mov_sitf_e500_subreg12_2_be): Likewise.
3984 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3986 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
3987 attribute from vecsimple to vecperm.
3988 (altivec_vbpermq2): Likewise.
3990 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3993 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
3995 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3996 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
3997 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
3998 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
3999 case where N arg is SIZE_MAX.
4000 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
4001 (cmpstrsi): Add pattern.
4003 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
4005 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4006 __builtin_vec_revb builtins.
4007 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
4008 built-in functions to support generation of the ISA 3.0 XXBR<x>
4009 vector byte reverse instructions.
4010 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
4011 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
4012 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
4013 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
4014 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
4015 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
4016 (P9V_BUILTIN_VEC_REVB): Likewise.
4017 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
4018 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
4019 (p9_xxbrq_v16qi): Likewise.
4020 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
4021 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
4022 (p9_xxbrh_v8hi): Likewise.
4023 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
4024 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
4025 vec_revb built-in functions.
4027 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
4029 PR rtl-optimization/78952
4030 * config/i386/i386.md (any_extract): New code iterator.
4031 (*insvqi_2): Use any_extract for source operand.
4032 (*insvqi_3): Use any_shiftrt for source operand.
4034 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
4036 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
4038 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
4040 2017-01-18 Matthias Klose <doko@ubuntu.com>
4042 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
4044 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4046 * config/rs6000/altivec.h (vec_bperm): Change #define.
4047 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
4048 (altivec_vbpermq2): New define_insn.
4049 (altivec_vbpermd): Likewise.
4050 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
4052 (VBPERMD): Likewise.
4053 (VBPERM): New polymorphic function interface.
4054 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
4055 Add entries for P9V_BUILTIN_VEC_VBPERM.
4056 * doc/extend.texi: Add interfaces for vec_bperm.
4058 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4060 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
4061 first letter of error messages.
4062 (s390_resolve_overloaded_builtin): Likewise.
4063 * config/s390/s390.c (s390_expand_builtin): Likewise.
4064 (s390_invalid_arg_for_unprototyped_fn): Likewise.
4065 (s390_valid_target_attribute_inner_p): Likewise.
4066 * config/s390/s390.md ("tabort"): Likewise.
4068 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
4070 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
4071 (ISA_AVOID_DIV_HILO): New macro.
4072 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
4073 (ISA_HAS_DDIV): Likewise.
4075 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4077 * doc/invoke.texi (fabi-version): Correct number of occurrences.
4079 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4081 * doc/invoke.texi (fabi-version): Spelling fix.
4083 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4086 * doc/invoke.texi (fabi-version): Mention mangling fix for
4089 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4092 * doc/invoke.texi (fabi-version): Document discriminator mangling.
4094 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
4097 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
4098 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
4100 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
4102 (stack_protect_test): Ditto.
4103 * config/rs6000/rs6000.opt (mstack-protector-guard=,
4104 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
4106 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
4107 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
4108 -mstack-protector-guard-offset=.
4109 (RS/6000 and PowerPC Options): Ditto.
4111 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
4113 * config/i386/i386.h (MASK_CLASS_P): New define.
4114 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
4115 there are no registers from different register sets also when
4116 mask registers are used. Update function comment.
4117 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
4118 to (*k/*r) and (*k/*km) alternatives.
4120 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
4122 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
4123 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
4124 (EH_RETURN_HANDLER_RTX): New define.
4125 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
4126 Force frame pointer in EH return functions.
4127 (aarch64_expand_epilogue): Add barrier for eh_return.
4128 (aarch64_final_eh_return_addr): Remove.
4129 (aarch64_eh_return_handler_rtx): New function.
4130 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
4132 (aarch64_eh_return_handler_rtx): New prototype.
4134 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4136 * config/rs6000/altivec.h (vec_rlmi): New #define.
4137 (vec_vrlnm): Likewise.
4138 (vec_rlnm): Likewise.
4139 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
4140 (UNSPEC_VRLNM): Likewise.
4141 (VIlong): New mode iterator.
4142 (altivec_vrl<VI_char>mi): New define_insn.
4143 (altivec_vrl<VI_char>nm): Likewise.
4144 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
4147 (RLNM): New polymorphic function entry.
4148 (VRLWMI): New monomorphic function entry.
4150 (RLMI): New polymorphic function entry.
4151 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
4152 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
4153 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
4156 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4159 * dwarf2out.c (field_byte_offset): Restore the
4160 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
4161 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
4162 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
4165 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
4168 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
4170 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4173 * dwarf2out.c (add_data_member_location_attribute): For constant
4174 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
4175 instead of DW_AT_data_member_location, DW_AT_bit_offset and
4176 DW_AT_byte_size attributes.
4178 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
4180 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
4181 after forcing to constant memory when the code model is medium.
4183 2017-01-17 Julia Koval <julia.koval@intel.com>
4186 * config/i386/avx512fintrin.h
4187 (_mm512_i32gather_ps): Change __addr type to void const*.
4188 (_mm512_mask_i32gather_ps): Ditto.
4189 (_mm512_i32gather_pd): Ditto.
4190 (_mm512_mask_i32gather_pd): Ditto.
4191 (_mm512_i64gather_ps): Ditto.
4192 (_mm512_mask_i64gather_ps): Ditto.
4193 (_mm512_i64gather_pd): Ditto.
4194 (_mm512_mask_i64gather_pd): Ditto.
4195 (_mm512_i32gather_epi32): Ditto.
4196 (_mm512_mask_i32gather_epi32): Ditto.
4197 (_mm512_i32gather_epi64): Ditto.
4198 (_mm512_mask_i32gather_epi64): Ditto.
4199 (_mm512_i64gather_epi32): Ditto.
4200 (_mm512_mask_i64gather_epi32): Ditto.
4201 (_mm512_i64gather_epi64): Ditto.
4202 (_mm512_mask_i64gather_epi64): Ditto.
4203 (_mm512_i32scatter_ps): Change __addr type to void*.
4204 (_mm512_mask_i32scatter_ps): Ditto.
4205 (_mm512_i32scatter_pd): Ditto.
4206 (_mm512_mask_i32scatter_pd): Ditto.
4207 (_mm512_i64scatter_ps): Ditto.
4208 (_mm512_mask_i64scatter_ps): Ditto.
4209 (_mm512_i64scatter_pd): Ditto.
4210 (_mm512_mask_i64scatter_pd): Ditto.
4211 (_mm512_i32scatter_epi32): Ditto.
4212 (_mm512_mask_i32scatter_epi32): Ditto.
4213 (_mm512_i32scatter_epi64): Ditto.
4214 (_mm512_mask_i32scatter_epi64): Ditto.
4215 (_mm512_i64scatter_epi32): Ditto.
4216 (_mm512_mask_i64scatter_epi32): Ditto.
4217 (_mm512_i64scatter_epi64): Ditto.
4218 (_mm512_mask_i64scatter_epi64): Ditto.
4219 * config/i386/avx512pfintrin.h
4220 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
4221 (_mm512_mask_prefetch_i32gather_ps): Ditto.
4222 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4223 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4224 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
4225 (_mm512_prefetch_i32scatter_ps): Ditto.
4226 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4227 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4228 (_mm512_prefetch_i64scatter_pd): Ditto.
4229 (_mm512_prefetch_i64scatter_ps): Ditto.
4230 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4231 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4232 * config/i386/avx512vlintrin.h
4233 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
4234 (_mm_mmask_i32gather_ps): Ditto.
4235 (_mm256_mmask_i32gather_pd): Ditto.
4236 (_mm_mmask_i32gather_pd): Ditto.
4237 (_mm256_mmask_i64gather_ps): Ditto.
4238 (_mm_mmask_i64gather_ps): Ditto.
4239 (_mm256_mmask_i64gather_pd): Ditto.
4240 (_mm_mmask_i64gather_pd): Ditto.
4241 (_mm256_mmask_i32gather_epi32): Ditto.
4242 (_mm_mmask_i32gather_epi32): Ditto.
4243 (_mm256_mmask_i32gather_epi64): Ditto.
4244 (_mm_mmask_i32gather_epi64): Ditto.
4245 (_mm256_mmask_i64gather_epi32): Ditto.
4246 (_mm_mmask_i64gather_epi32): Ditto.
4247 (_mm256_mmask_i64gather_epi64): Ditto.
4248 (_mm_mmask_i64gather_epi64): Ditto.
4249 (_mm256_i32scatter_ps): Change __addr type to void*.
4250 (_mm256_mask_i32scatter_ps): Ditto.
4251 (_mm_i32scatter_ps): Ditto.
4252 (_mm_mask_i32scatter_ps): Ditto.
4253 (_mm256_i32scatter_pd): Ditto.
4254 (_mm256_mask_i32scatter_pd): Ditto.
4255 (_mm_i32scatter_pd): Ditto.
4256 (_mm_mask_i32scatter_pd): Ditto.
4257 (_mm256_i64scatter_ps): Ditto.
4258 (_mm256_mask_i64scatter_ps): Ditto.
4259 (_mm_i64scatter_ps): Ditto.
4260 (_mm_mask_i64scatter_ps): Ditto.
4261 (_mm256_i64scatter_pd): Ditto.
4262 (_mm256_mask_i64scatter_pd): Ditto.
4263 (_mm_i64scatter_pd): Ditto.
4264 (_mm_mask_i64scatter_pd): Ditto.
4265 (_mm256_i32scatter_epi32): Ditto.
4266 (_mm256_mask_i32scatter_epi32): Ditto.
4267 (_mm_i32scatter_epi32): Ditto.
4268 (_mm_mask_i32scatter_epi32): Ditto.
4269 (_mm256_i32scatter_epi64): Ditto.
4270 (_mm256_mask_i32scatter_epi64): Ditto.
4271 (_mm_i32scatter_epi64): Ditto.
4272 (_mm_mask_i32scatter_epi64): Ditto.
4273 (_mm256_i64scatter_epi32): Ditto.
4274 (_mm256_mask_i64scatter_epi32): Ditto.
4275 (_mm_i64scatter_epi32): Ditto.
4276 (_mm_mask_i64scatter_epi32): Ditto.
4277 (_mm256_i64scatter_epi64): Ditto.
4278 (_mm256_mask_i64scatter_epi64): Ditto.
4279 (_mm_i64scatter_epi64): Ditto.
4280 (_mm_mask_i64scatter_epi64): Ditto.
4281 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
4282 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
4283 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
4284 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
4285 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
4286 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
4287 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
4288 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
4289 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
4290 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
4291 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
4292 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
4293 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
4294 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
4295 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
4296 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
4297 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
4298 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
4299 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
4300 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
4301 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
4302 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
4303 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
4304 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
4305 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
4306 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
4307 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
4308 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
4309 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
4310 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
4311 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
4312 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
4313 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
4314 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
4315 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
4316 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
4317 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
4318 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
4319 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
4320 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
4321 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
4322 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
4323 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
4324 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
4325 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
4326 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
4327 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
4328 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
4329 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
4330 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
4331 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
4332 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
4333 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
4334 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
4335 definitions accordingly.
4337 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
4338 Kuan-Lin Chen <kuanlinchentw@gmail.com>
4341 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
4344 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
4347 * ira-conflicts.c (ira_build_conflicts): Update total conflict
4348 hard regs for inner regno.
4350 2017-01-17 Martin Liska <mliska@suse.cz>
4353 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
4354 assumption and add comment.
4356 2017-01-17 Nathan Sidwell <nathan@acm.org>
4358 * ipa-visibility.c (localize_node): New function, broken out of ...
4359 (function_and_variable_visibility): ... here. Call it.
4361 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
4364 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
4365 correctly set frequency of oudgoing edge.
4366 (duplicate_thread_path): Fix profile updating.
4368 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4371 * configure.ac: Add GCC_BASE_VER.
4372 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
4373 version from BASE-VER file.
4374 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
4375 (gcc.o): Depend on $(BASEVER).
4376 * common.opt (dumpfullversion): New option.
4377 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
4378 * doc/invoke.texi: Document -dumpfullversion.
4379 * doc/install.texi: Document --with-gcc-major-version-only.
4380 * configure: Regenerated.
4382 2017-01-17 Richard Biener <rguenther@suse.de>
4384 PR tree-optimization/71433
4385 * tree-vrp.c (register_new_assert_for): Merge same asserts
4386 on all incoming edges.
4387 (process_assert_insertions_for): Handle insertions at the
4390 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
4392 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
4393 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
4395 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
4398 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
4401 2017-01-17 Alan Modra <amodra@gmail.com>
4404 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
4405 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
4406 symbolic stack limit when pic.
4408 2017-01-16 Martin Sebor <msebor@redhat.com>
4410 PR tree-optimization/78608
4411 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
4413 2017-01-16 Jeff Law <law@redhat.com>
4416 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
4417 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
4418 for several include directories that may be relative to sysroot.
4419 * config/i386/x-mingw32 (gplus_includedir): Define.
4420 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
4421 (native_system_includedir): Likewise.
4422 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
4423 override if TARGET_SYSTEM_ROOT is defined.
4424 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
4426 PR tree-optimization/79090
4427 PR tree-optimization/33562
4428 PR tree-optimization/61912
4429 PR tree-optimization/77485
4430 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
4431 and computed trims into the dump file.
4433 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
4435 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
4437 2017-01-16 Jakub Jelinek <jakub@redhat.com>
4440 * gimplify.c (gimplify_init_constructor): If want_value and
4441 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
4445 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
4446 sequence. Formatting fixes.
4447 (doloop_optimize): Formatting fixes.
4450 * gcc.c (debug_level_greater_than_spec_func): New function.
4451 (static_spec_functions): Add debug-level-gt spec function.
4452 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
4454 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
4455 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
4456 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
4457 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
4458 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
4459 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
4461 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
4463 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
4464 QImode fixups to general and mask registers only.
4466 2017-01-16 Carl Love <cel@us.ibm.com>
4468 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
4469 for built-in functions
4470 vector signed char vec_nabs (vector signed char)
4471 vector signed short vec_nabs (vector signed short)
4472 vector signed int vec_nabs (vector signed int)
4473 vector signed long long vec_nabs (vector signed long long)
4474 vector float vec_nabs (vector float)
4475 vector double vec_nabs (vector double)
4476 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
4478 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
4479 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
4480 * doc/extend.texi: Update the documentation file for the new built-in
4483 2017-01-16 Martin Sebor <msebor@redhat.com>
4485 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
4488 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4490 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
4491 UNSPEC_VSX__XXSPLTD to require special splat handling.
4493 2017-01-16 David Malcolm <dmalcolm@redhat.com>
4496 * system.h: Poison strndup.
4498 2017-01-16 Alan Modra <amodra@gmail.com>
4501 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
4504 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
4506 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
4508 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
4510 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
4511 call recog here. Assert that INSN_CODE (insn) is non-negative.
4513 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
4516 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
4518 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
4519 in the currently scheduled RTL fragment.
4521 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
4523 PR rtl-optimization/78751
4524 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
4527 2017-01-14 Jeff Law <law@redhat.com>
4529 PR tree-optimization/79090
4530 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
4531 variable length stores.
4532 (compute_trims): Delete dead assignment to *trim_tail.
4533 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
4536 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
4538 PR rtl-optimization/78626
4539 PR rtl-optimization/78727
4540 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
4541 of a block, and split such blocks after everything else is finished.
4543 2017-01-14 Alan Modra <amodra@gmail.com>
4546 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
4547 target legitimate_combined_insn.
4548 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
4549 (rs6000_legitimate_combined_insn): New function.
4550 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
4552 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
4553 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
4554 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
4556 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
4558 * doc/frontends.texi (G++ and GCC): Remove references to Java.
4560 2017-01-13 Jeff Law <law@redhat.com>
4562 PR tree-optimization/33562
4563 PR tree-optimization/61912
4564 PR tree-optimization/77485
4565 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
4567 (delete_dead_assignment): Likewise.
4568 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
4569 statement to delete_dead_call and delete_dead_assignment.
4571 2017-01-13 David Malcolm <dmalcolm@redhat.com>
4574 * substring-locations.c (format_warning_va): Strengthen case 1 so
4575 that both endpoints of the substring must be within the format
4576 range for just the substring to be printed.
4578 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
4580 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
4581 * config/i386/i386.c (ix86_target_string): Add missing options
4582 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
4583 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
4584 flags_other and ix86_target_other to flags2_other. Display unknown
4586 (ix86_valid_target_attribute_inner_p): Add missing options and
4587 reorder options by implied ISAs, as in ix86_target_string.
4589 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
4591 * hash-table.h (hash_table::too_empty_p): New function.
4592 (hash_table::expand): Use it.
4593 (hash_table::traverse): Likewise.
4594 (hash_table::empty_slot): Use sizeof (value_type) instead of
4595 sizeof (PTR) to convert bytes to elements. Shrink the table
4596 if the current size is excessive for the current number of
4599 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
4601 * ira-costs.c (record_reg_classes): Break from the inner loop
4602 early once alt_fail is known to be true. Update outer loop
4603 handling accordingly.
4605 2017-01-13 Jeff Law <law@redhat.com>
4607 * tree-ssa-dse.c (decrement_count): New function.
4608 (increment_start_addr, maybe_trim_memstar_call): Likewise.
4609 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
4610 when we know the partially dead statement is a mem* function.
4612 PR tree-optimization/61912
4613 PR tree-optimization/77485
4614 * tree-ssa-dse.c: Include expr.h.
4615 (maybe_trim_constructor_store): New function.
4616 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
4618 PR tree-optimization/33562
4619 PR tree-optimization/61912
4620 PR tree-optimization/77485
4621 * doc/invoke.texi: Document new dse-max-object-size param.
4622 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
4623 * tree-ssa-dse.c: Include params.h.
4624 (dse_store_status): New enum.
4625 (initialize_ao_ref_for_dse): New, partially extracted from
4627 (valid_ao_ref_for_dse, normalize_ref): New.
4628 (setup_live_bytes_from_ref, compute_trims): Likewise.
4629 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
4630 (maybe_trim_partially_dead_store): Likewise.
4631 (maybe_trim_complex_store): Likewise.
4632 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
4633 Track what bytes live from the original store. Return tri-state
4634 for dead, partially dead or live.
4635 (dse_dom_walker): Add constructor, destructor and new private members.
4636 (delete_dead_call, delete_dead_assignment): New extracted from
4638 (dse_optimize_stmt): Make a member of dse_dom_walker.
4639 Use initialize_ao_ref_for_dse.
4641 PR tree-optimization/33562
4642 PR tree-optimization/61912
4643 PR tree-optimization/77485
4644 * sbitmap.h (bitmap_count_bits): Prototype.
4645 (bitmap_clear_range, bitmap_set_range): Likewise.
4646 * sbitmap.c (bitmap_clear_range): New function.
4647 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
4649 2017-01-13 Martin Liska <mliska@suse.cz>
4652 * function.c (set_cfun): Add new argument force.
4653 * function.h (set_cfun): Likewise.
4654 * ipa-inline-transform.c (inline_call): Use the function when
4655 strict alising from is dropped for function we inline to.
4657 2017-01-13 Richard Biener <rguenther@suse.de>
4659 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
4660 for dumping GIMPLE INTEGER_CSTs.
4662 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4664 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
4665 to 201112L since C++17.
4667 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
4670 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
4671 if -fsanitize=kernel-address is present.
4673 2017-01-13 Richard Biener <rguenther@suse.de>
4675 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
4676 as _Literal ( type ) number in case usual suffixes do not
4677 preserve all information.
4679 2017-01-13 Richard Biener <rguenther@suse.de>
4681 PR tree-optimization/77283
4682 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
4683 and ssa-iterators.h.
4684 (is_feasible_trace): Implement a cost model based on joiner
4687 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4690 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
4691 char or short to __float128/_Float128 directly.
4693 2017-01-12 Martin Sebor <msebor@redhat.com>
4695 to -Wformat-overflow.
4696 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
4697 (min_bytes_remaining): Same.
4698 (get_string_length): Same.
4699 (format_string): Same.
4700 (format_directive): Same.
4702 (pass_sprintf_length::handle_gimple_call): Same.
4704 2017-01-12 Jakub Jelinek <jakub@redhat.com>
4706 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
4707 info.nowrite calls with no lhs that can't throw. Return bool
4708 whether gsi_remove has been called or not.
4709 (pass_sprintf_length::handle_gimple_call): Return bool whether
4710 try_substitute_return_value called gsi_remove. Formatting fix.
4711 (pass_sprintf_length::execute): Don't use gsi_remove if
4712 handle_gimple_call returned true.
4715 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
4716 be removed due to side-effects, don't remove following barrier nor
4717 turn the successor edge into fallthru edge.
4719 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4722 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
4723 element-reversing loads and stores as not swappable.
4725 2017-01-12 Nathan Sidwell <nathan@acm.org>
4726 Nicolai Stange <nicstange@gmail.com>
4728 * combine.c (try_combine): Don't ignore result of overlap checking
4729 loop. Combine overlap & asm check into single loop.
4731 2017-01-12 Richard Biener <rguenther@suse.de>
4733 * tree-pretty-print.c (dump_generic_node): Provide -gimple
4734 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
4736 2017-01-12 Richard Biener <rguenther@suse.de>
4738 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
4739 and TS_TARGET_OPTION directly derive from TS_BASE.
4740 * tree-core.h (tree_optimization_option): Derive from tree_base.
4741 (tree_target_option): Likewise.
4743 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
4745 * config/i386/i386.c (memory_address_length): Increase len
4746 only when rip_relative_addr_p returns false.
4748 2017-01-11 Julia Koval <julia.koval@intel.com>
4750 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
4751 (OPTION_MASK_ISA_SGX_SET): New.
4752 (ix86_handle_option): Handle OPT_msgx.
4753 * config.gcc: Added sgxintrin.h.
4754 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
4755 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
4756 * config/i386/i386.c (ix86_target_string): Add -msgx.
4758 (ix86_option_override_internal): Handle new options.
4759 (ix86_valid_target_attribute_inner_p): Add sgx.
4760 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
4761 * config/i386/i386.opt: Add msgx.
4762 * config/i386/sgxintrin.h: New file.
4763 * config/i386/x86intrin.h: Add sgxintrin.h.
4765 2017-01-11 Jakub Jelinek <jakub@redhat.com>
4768 * fold-const.c (maybe_nonzero_address): Return 1 for function
4770 (tree_single_nonzero_warnv_p): Don't handle function local objects
4774 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
4777 2017-01-11 David Malcolm <dmalcolm@redhat.com>
4780 * opts.c: Include "spellcheck.h"
4781 (struct string_fragment): New struct.
4782 (struct edit_distance_traits<const string_fragment &>): New
4784 (get_closest_sanitizer_option): New function.
4785 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
4787 2017-01-11 Jakub Jelinek <jakub@redhat.com>
4789 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
4791 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
4792 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
4793 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
4794 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
4795 for initial die_offset if dwarf_split_debug_info.
4796 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
4797 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
4799 (output_skeleton_debug_sections): Formatting fix. Use
4800 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
4801 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
4803 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
4805 * config/arm/cortex-a53.md: Add bypasses for
4807 (cortex_a53_r2f): Only use for transfers.
4808 (cortex_a53_f2r): Likewise.
4809 (cortex_a53_r2f_cvt): Add reservation for conversions.
4810 (cortex_a53_f2r_cvt): Likewise.
4812 2017-01-11 Tamar Christina <tamar.christina@arm.com>
4814 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
4815 to all inlined functions, change static to extern.
4817 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
4820 * config/arm/arm.c (legitimize_pic_address): Handle reference to
4822 (arm_assemble_integer): Likewise.
4824 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
4826 * config.gcc: Use new awk script to check CPU, FPU and architecture
4827 parameters for --with-... options.
4828 * config/arm/parsecpu.awk: New file
4829 * config/arm/arm-cpus.in: New file.
4830 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
4832 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
4834 * config/arm/t-arm: Update dependency rules.
4835 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
4836 of processing .def files.
4837 * config/arm/genopt.sh: Deleted.
4838 * config/arm/gentune.sh: Deleted.
4839 * config/arm/arm-cores.def: Deleted.
4840 * config/arm/arm-arches.def: Deleted.
4841 * config/arm/arm-fpus.def: Deleted.
4842 * config/arm/arm-tune.md: Regenerated.
4843 * config/arm/arm-tables.opt: Regenerated.
4844 * config/arm/arm-cpu.h: New generated file.
4845 * config/arm/arm-cpu-data.h: New generated file.
4846 * config/arm/arm-cpu-cdata.h: New generated file.
4848 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
4851 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
4853 (input_varpool_node): Unpack dynamically_initialized bit.
4855 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
4857 PR rtl-optimization/79032
4858 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
4859 the alignment of the adjusted memory reference against that of MODE,
4860 instead of the alignment of the original memory reference.
4862 2017-01-11 Martin Jambor <mjambor@suse.cz>
4864 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
4866 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
4867 decorated functions.
4869 2017-01-11 Richard Biener <rguenther@suse.de>
4871 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
4872 set range/nonnull info for PHI results. Do not set it on
4873 stmts marked for removal.
4875 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
4877 * expr.c (store_field): In the bitfield case, fetch the return value
4878 from the registers before applying a single big-endian adjustment.
4879 Always do a final load for a BLKmode value not larger than a word.
4881 2017-01-10 David Malcolm <dmalcolm@redhat.com>
4884 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
4885 that we correctly handle column numbers greater than
4886 LINE_MAP_MAX_COLUMN_NUMBER.
4888 2017-01-10 Martin Sebor <msebor@redhat.com>
4891 * gimple-ssa-sprintf.c (get_destination_size): Call
4892 {init,fini}object_sizes.
4893 * tree-object-size.c (addr_object_size): Adjust.
4894 (pass_through_call): Adjust.
4895 (pass_object_sizes::execute): Adjust.
4896 * tree-object-size.h (fini_object_sizes): Declare.
4898 2017-01-10 Martin Sebor <msebor@redhat.com>
4900 PR tree-optimization/78775
4901 * builtins.c (get_size_range): Move...
4902 * calls.c: ...to here.
4903 (alloc_max_size): Accept zero argument.
4904 (operand_signed_p): Remove.
4905 (maybe_warn_alloc_args_overflow): Call get_size_range.
4906 * calls.h (get_size_range): Declare.
4908 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
4910 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
4911 from TI's devices.csv file as of September 2016.
4912 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
4914 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
4916 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
4917 * doc/invoke.texi: Likewise.
4918 * doc/md.texi: Likewise.
4919 * doc/objc.texi: Likewise.
4921 2017-01-10 Joshua Conner <joshconner@google.com>
4923 * config/arm/fuchsia-elf.h: New file.
4924 * config/fuchsia.h: New file.
4925 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
4926 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
4928 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
4930 2016-01-10 Richard Biener <rguenther@suse.de>
4932 PR tree-optimization/79034
4933 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
4934 Propagate out degenerate PHIs in the joiner.
4936 2017-01-10 Martin Liska <mliska@suse.cz>
4938 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
4939 (sort_congruence_classes_by_decl_uid): Likewise.
4940 (sort_congruence_class_groups_by_decl_uid): Likewise.
4941 (sem_item_optimizer::merge_classes): Sort class, groups in these
4942 classes and members in the groups by DECL_UID of declarations.
4943 This would make merge operations stable.
4945 2017-01-10 Martin Liska <mliska@suse.cz>
4947 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
4948 usage of m_classes_vec.
4949 (sem_item_optimizer::~sem_item_optimizer): Likewise.
4950 (sem_item_optimizer::get_group_by_hash): Likewise.
4951 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
4952 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
4953 (sem_item_optimizer::verify_classes): Likewise.
4954 (sem_item_optimizer::process_cong_reduction): Likewise.
4955 (sem_item_optimizer::dump_cong_classes): Likewise.
4956 (sem_item_optimizer::merge_classes): Likewise.
4957 * ipa-icf.h (congruence_class_hash): Rename from
4958 congruence_class_group_hash. Remove declaration of m_classes_vec.
4960 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
4962 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
4963 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
4964 * config.gcc: Add avx512vpopcntdqintrin.h.
4965 * config/i386/avx512vpopcntdqintrin.h: New.
4966 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
4967 * config/i386/i386-builtin-types.def: Add new types.
4968 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
4969 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
4970 __builtin_ia32_vpopcountq_v8di_mask): New.
4971 * config/i386/i386-c.c (ix86_target_macros_internal): Define
4972 __AVX512VPOPCNTDQ__.
4973 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
4974 (PTA_AVX512VPOPCNTDQ): Define.
4975 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
4976 TARGET_AVX512VPOPCNTDQ_P): Define.
4977 * config/i386/i386.opt: Add mavx512vpopcntdq.
4978 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
4979 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
4981 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
4984 * predict.def (PRED_CALL): Set to 67.
4986 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
4988 * expr.c (store_field): In the bitfield case, if the value comes from
4989 a function call and is of an aggregate type returned in registers, do
4990 not modify the field mode; extract the value in all cases if the mode
4991 is BLKmode and the size is not larger than a word.
4993 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
4996 * config/i386/cpuid.h: Fix undefined behavior.
4998 2017-01-04 Jeff Law <law@redhat.com>
5000 PR tree-optimization/79007
5001 PR tree-optimization/67955
5002 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
5003 conservative for pt.null when flag_non_call_exceptions is on.
5005 2017-01-09 Jakub Jelinek <jakub@redhat.com>
5007 PR translation/79019
5008 PR translation/79020
5009 * params.def (PARAM_INLINE_MIN_SPEEDUP,
5010 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
5011 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
5013 * config/avr/avr.opt (maccumulate-args): Likewise.
5014 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
5015 * common.opt (freport-bug): Likewise.
5016 * cif-code.def (CIF_FINAL_ERROR): Likewise.
5017 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
5018 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
5019 translatable string.
5020 * config/i386/i386.c (function_value_32): Likewise.
5021 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
5022 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
5024 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
5025 * common/config/msp430/msp430-common.c (msp430_handle_option):
5027 * symtab.c (symtab_node::verify_base): Likewise.
5028 * opts.c (set_debug_level): Likewise.
5029 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
5030 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
5031 missing whitespace to translatable strings.
5032 * config/avr/avr.md (bswapsi2): Fix typo in comment.
5033 * config/sh/superh.h: Likewise.
5034 * config/i386/xopintrin.h: Likewise.
5035 * config/i386/znver1.md: Likewise.
5036 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
5037 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
5038 * double-int.h (struct double_int): Likewise.
5039 * double-int.c (div_and_round_double): Likewise.
5040 * wide-int.cc: Likewise.
5041 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
5042 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
5043 * cfgcleanup.c (crossjumps_occured): Renamed to ...
5044 (crossjumps_occurred): ... this.
5045 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
5048 PR tree-optimization/78899
5049 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
5050 returning bool return struct loop *, NULL for failure and the new
5052 (versionable_outer_loop_p): Don't version outer loop if it has
5053 dont_vectorized bit set.
5054 (tree_if_conversion): When versioning outer loop, ensure
5055 tree_if_conversion is performed also on the inner loop of the
5056 non-vectorizable outer loop copy.
5057 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
5058 LOOP_VECTORIZED in inner loop of the scalar outer loop and
5059 prevent vectorization of it.
5060 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
5061 the outer loop vectorization of the non-scalar version is attempted
5062 before vectorization of the inner loop in scalar version. If
5063 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
5064 vectorization of its inner loop.
5065 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
5066 has 2 inner loops, rename also on edges from bb whose single pred
5067 is outer_loop->header. Fix typo in function comment.
5069 2017-01-09 Martin Sebor <msebor@redhat.com>
5072 * asan.c (asan_emit_stack_protection): Increase local buffer size
5073 to avoid snprintf truncation warning.
5075 2017-01-09 Andrew Pinski <apinski@cavium.com>
5077 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
5078 to reference thunderx2t99 for the tuning structure
5079 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
5081 (thunderx2t99_extra_costs): This.
5082 * config/aarch64/aarch64-tune.md: Regenerate.
5083 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
5084 (vulcan_addrcost_table): This.
5085 (vulcan_regmove_cost): Rename to ...
5086 (thunderx2t99_regmove_cost): This.
5087 (vulcan_vector_cost): Rename to ...
5088 (thunderx2t99_vector_cost): this.
5089 (vulcan_branch_cost): Rename to ...
5090 (thunderx2t99_branch_cost): This.
5091 (vulcan_tunings): Rename to ...
5092 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
5093 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
5095 2017-01-09 Martin Jambor <mjambor@suse.cz>
5099 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
5100 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
5101 (propagate_vr_accross_jump_function): Use the above function for all
5102 value range computations for pass-through jump functions and type
5103 converasion from explicit value range values.
5104 (ipcp_propagate_stage): Do not attempt to deduce types of formal
5105 parameters from TYPE_ARG_TYPES.
5106 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
5107 (ipa_write_node_info): Stream type of the actual argument.
5108 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
5110 2017-01-09 Martin Liska <mliska@suse.cz>
5113 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
5114 (lookup_compiler): Do not show error message with have_E.
5116 2017-01-09 Jakub Jelinek <jakub@redhat.com>
5118 PR tree-optimization/78938
5119 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
5120 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
5121 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
5122 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
5125 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5127 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
5130 2017-01-09 Richard Biener <rguenther@suse.de>
5132 PR tree-optimization/78997
5133 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
5134 name condition properly.
5136 2017-01-09 Richard Biener <rguenther@suse.de>
5139 * dwarf2out.c (is_cxx): New overload with context.
5140 (is_naming_typedef_decl): Use it.
5142 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5144 * invoke.texi (Option Summary): Correct spacing in option lists
5145 and add line breaks to fix over-long lines.
5147 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5151 * extend.texi (Common Variable Attributes): Add xref to GCC
5152 Internals manual to explain mode attribute keywords.
5154 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5157 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
5158 and Preprocessor Options.
5159 (Options for Linking): Document -pthread here....
5160 (RS/6000 and PowerPC Options): ...not here.
5161 (Solaris 2 Options): ...or here.
5162 * doc/cppopts.texi: Document -pthread.
5164 2017-01-08 Martin Sebor <msebor@redhat.com>
5167 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
5168 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
5169 New member functions.
5170 (format_directive): Used them.
5172 (pass_sprintf_length::handle_gimple_call): Same.
5173 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
5174 to avoid truncation for any argument.
5175 (extract_affine_mul): Same.
5176 * tree.c (get_file_function_name): Same.
5178 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5181 * predict.def (PRED_INDIR_CALL): Set to 86.
5183 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5185 PR preprocessor/54124
5186 * doc/cppopts.texi: Reformat -d subtable to list the full name
5187 of the options. Add cross-reference to the docs for the general
5188 compiler -d options.
5189 * doc/invoke.texi (Developer Options): Add cross-reference to the
5190 preprocessor-specific -d option documentation.
5192 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5194 PR preprocessor/13498
5195 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
5196 redudant material, and reflect new command-line options.
5197 (System Headers): Likewise.
5199 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5201 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
5202 -isystem, and -idirafter. Copy-edit.
5203 * doc/cppopts.texi: Copy-edit. Remove contradiction about
5204 default for -ftrack-macro-expansion. Delete obsolete and
5205 badly-formatted implementation details about -fdebug-cpp output.
5206 * doc/cppwarnopts.texi: Copy-edit.
5208 2017-01-07 David Malcolm <dmalcolm@redhat.com>
5211 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
5212 that the transition from a max line width >= 1<<10 to narrower
5213 lines works correctly.
5215 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
5217 * doc/options.texi (PerFunction): New.
5218 * opt-functions.awk (switch_flags): Map both Optimization and
5219 PerFunction to CL_OPTIMIZATION.
5220 * opth-gen.awk: Test for PerFunction flag along with
5222 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
5223 it only when the latter is present. Skip those that don't in
5224 the hash function generator.
5225 * common.opt (fvar-tracking): Mark as PerFunction instead of
5227 (fvar-tracking-assignments): Likewise.
5228 (fvar-tracking-assignments-toggle): Likewise.
5229 (fvar-tracking-uninit): Likewise.
5231 2017-01-07 Jakub Jelinek <jakub@redhat.com>
5233 PR translation/79018
5234 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
5237 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
5240 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
5241 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
5242 TARGET_LONG_JUMP_TABLE_OFFSETS.
5243 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
5244 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
5245 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
5246 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
5247 * config/m68k/m68k.md (tablejump expander): Likewise.
5248 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
5249 TARGET_LONG_JUMP_TABLE_OFFSETS.
5250 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
5251 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
5253 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
5254 David Holsgrove <david.holsgrove@xilinx.com>
5256 * common/config/microblaze/microblaze-common.c
5257 (TARGET_EXCEPT_UNWIND_INFO): Remove.
5258 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
5260 * config/microblaze/microblaze.c (microblaze_must_save_register)
5261 (microblaze_expand_epilogue, microblaze_return_addr): Handle
5263 (microblaze_eh_return): New function.
5264 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
5265 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
5266 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
5267 * config/microblaze/microblaze.md (eh_return): New pattern.
5269 2017-01-06 Jakub Jelinek <jakub@redhat.com>
5271 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
5272 GCC_DIAGNOSTIC_STRINGIFY): Define.
5274 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
5276 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5278 * config/arm/arm.md (<mcrr>): New.
5280 * config/arm/arm.c (arm_arch5te): New.
5281 (arm_option_override): Set arm_arch5te.
5282 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
5284 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
5285 (arm_mcrr_qualifiers): ... this. New.
5286 (MRRC_QUALIFIERS): Define to...
5287 (arm_mrrc_qualifiers): ... this. New.
5288 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
5290 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
5291 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
5292 (MRRCI, mrrc, MRRC): New.
5293 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
5294 VUNSPEC_MRRC2): New.
5296 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5298 * config/arm/arm.md (<mcr>): New.
5300 * config/arm/arm.c (arm_coproc_builtin_available): Add
5301 support for mcr, mrc, mcr2 and mrc2.
5302 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
5303 (arm_mcr_qualifiers): ... this. New.
5304 (MRC_QUALIFIERS): Define to ...
5305 (arm_mrc_qualifiers): ... this. New.
5306 (MCR_QUALIFIERS): Define to ...
5307 (arm_mcr_qualifiers): ... this. New.
5308 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
5310 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
5311 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
5312 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
5315 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5317 * config/arm/arm.md (*ldc): New.
5321 * config/arm/arm.c (arm_coproc_builtin_available): Add
5322 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
5323 (arm_coproc_ldc_stc_legitimate_address): New.
5324 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
5325 'qualifier_const_pointer'.
5326 (LDC_QUALIFIERS): Define to...
5327 (arm_ldc_qualifiers): ... this. New.
5328 (STC_QUALIFIERS): Define to...
5329 (arm_stc_qualifiers): ... this. New.
5330 * config/arm/arm-protos.h
5331 (arm_coproc_ldc_stc_legitimate_address): New.
5332 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
5333 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
5334 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
5335 stc2, stcl, stc2l): New.
5336 * config/arm/constraints.md (Uz): New.
5337 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
5338 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
5339 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
5340 VUNSPEC_STC2L): New.
5342 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5344 * config/arm/arm.md (<cdp>): New.
5345 * config/arm/arm.c (neon_const_bounds): Rename this ...
5346 (arm_const_bounds): ... this.
5347 (arm_coproc_builtin_available): New.
5348 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
5349 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
5350 (CDP_QUALIFIERS): Define to...
5351 (arm_cdp_qualifiers): ... this. New.
5353 (arm_expand_builtin_args): Add case for 6 arguments.
5354 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
5355 (arm_const_bounds): ... this.
5356 (arm_coproc_builtin_available): New.
5357 * config/arm/arm_acle.h (__arm_cdp): New.
5359 * config/arm/arm_acle_builtins.def (cdp): New.
5361 * config/arm/iterators.md (CDPI,CDP,cdp): New.
5362 * config/arm/neon.md: Rename all 'neon_const_bounds' to
5364 * config/arm/types.md (coproc): New.
5365 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
5366 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
5367 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
5368 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
5370 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5372 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
5373 (UBINOP_QUALIFIERS): New.
5375 (acle_builtin_data): New. Change comment.
5376 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
5377 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
5378 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
5379 arm_acle_builtins.def.
5380 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
5381 (arm_init_acle_builtins): New.
5382 (CRC32_BUILTIN): Remove.
5383 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
5384 crc32cb, crc32ch and crc32cw.
5385 (arm_init_crc32_builtins): Remove.
5386 (arm_init_builtins): Use arm_init_acle_builtins rather
5387 than arm_init_crc32_builtins.
5388 (arm_expand_acle_builtin): New.
5389 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
5390 * config/arm/arm_acle_builtins.def: New.
5392 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5394 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
5395 (arm_builtin_datum): ... this.
5396 (arm_init_neon_builtin): Rename to ...
5397 (arm_init_builtin): ... this. Add a new parameters PREFIX
5398 and USE_SIG_IN_NAME.
5399 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
5400 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
5401 'arm_builtin_datum'.
5402 (arm_init_vfp_builtins): Likewise.
5403 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
5404 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
5405 (arm_expand_neon_args): Rename to ...
5406 (arm_expand_builtin_args): ... this. Rename builtin_arg
5407 enum values and differentiate between ARG_BUILTIN_MEMORY
5408 and ARG_BUILTIN_NEON_MEMORY.
5409 (arm_expand_neon_builtin_1): Rename to ...
5410 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
5411 values, arm_expand_builtin_args and add bool parameter NEON.
5412 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
5413 (arm_expand_vfp_builtin): Likewise.
5414 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
5416 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5419 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
5420 * predict.c (tree_estimate_probability_bb): Reverse direction of
5421 polymorphic call predictor.
5423 2017-01-06 David Malcolm <dmalcolm@redhat.com>
5425 * passes.c (execute_one_pass): Split out pass-skipping logic into...
5426 (determine_pass_name_match): ...this new function and...
5427 (should_skip_pass_p): ...this new function.
5429 2017-01-06 Nathan Sidwell <nathan@acm.org>
5431 * ipa-visibility.c (function_and_variable_visibility): Reformat
5432 comments and long lines. Remove extrneous if.
5433 * symtab.c (symtab_node::make_decl_local): Fix code format.
5434 (symtab_node::set_section_for_node): Fix comment typo.
5436 2017-01-06 Martin Liska <mliska@suse.cz>
5439 * lra-constraints.c: Rename invariant to lra_invariant.
5440 * predict.c (set_even_probabilities): Initialize e to NULL.
5442 2017-01-05 Martin Sebor <msebor@redhat.com>
5444 PR tree-optimization/78910
5445 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
5446 (format_integer): Correct off-by-one error in the handling
5447 of precision with negative numbers in signed conversions..
5449 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
5451 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
5453 2017-01-05 Jakub Jelinek <jakub@redhat.com>
5455 PR tree-optimization/71016
5456 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
5457 factor_out_conditional_conversion. Formatting fix.
5458 (factor_out_conditional_conversion): Add cond_stmt argument.
5459 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
5460 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
5463 2017-01-05 David Malcolm <dmalcolm@redhat.com>
5465 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
5466 read-rtl-function.o, and selftest-rtl.o.
5467 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
5468 (selftest::aarch64_test_loading_full_dump): New function.
5469 (selftest::aarch64_run_selftests): New function.
5470 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
5471 selftest::aarch64_run_selftests.
5472 * config/i386/i386.c
5473 (selftest::ix86_test_loading_dump_fragment_1): New function.
5474 (selftest::ix86_test_loading_call_insn): New function.
5475 (selftest::ix86_test_loading_full_dump): New function.
5476 (selftest::ix86_test_loading_unspec): New function.
5477 (selftest::ix86_run_selftests): Call the new functions.
5478 * emit-rtl.c (maybe_set_max_label_num): New function.
5479 * emit-rtl.h (maybe_set_max_label_num): New decl.
5480 * function.c (instantiate_decls): Guard call to
5481 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
5482 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
5484 * gensupport.c (gen_reader::gen_reader): Pass "false"
5485 for new "compact" param of rtx_reader.
5486 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
5487 rather than an empty string for NULL strings.
5488 * read-md.c: Potentially include config.h rather than bconfig.h.
5489 Wrap include of errors.h with #ifdef GENERATOR_FILE.
5490 (have_error): New global, copied from errors.c.
5491 (md_reader::read_name): Rename to...
5492 (md_reader::read_name_1): ...this, adding "out_loc" param,
5493 and converting "missing name or number" to returning false, rather
5495 (md_reader::read_name): Reimplement in terms of read_name_1.
5496 (md_reader::read_name_or_nil): New function.
5497 (md_reader::read_string): Handle "(nil)" by returning NULL.
5498 (md_reader::md_reader): Add new param "compact".
5499 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
5500 (md_reader::read_file): New method.
5501 * read-md.h (md_reader::md_reader): Add new param "compact".
5502 (md_reader::read_file): New method.
5503 (md_reader::is_compact): New accessor.
5504 (md_reader::read_name): Convert return type from void to file_location.
5505 (md_reader::read_name_or_nil): New decl.
5506 (md_reader::read_name_1): New decl.
5507 (md_reader::m_compact): New field.
5508 (noop_reader::noop_reader): Pass "false" for new "compact" param
5510 (rtx_reader::rtx_reader): Add new "compact" param.
5511 (rtx_reader::read_rtx_operand): Make virtual and convert return
5512 type from void to rtx.
5513 (rtx_reader::read_until): New decl.
5514 (rtx_reader::handle_any_trailing_information): New virtual function.
5515 (rtx_reader::postprocess): New virtual function.
5516 (rtx_reader::finalize_string): New virtual function.
5517 (rtx_reader::m_in_call_function_usage): New field.
5518 (rtx_reader::m_reuse_rtx_by_id): New field.
5519 * read-rtl-function.c: New file.
5520 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
5521 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
5522 (selftest::verify_three_block_rtl_cfg): New decl.
5523 * read-rtl-function.h: New file.
5524 * read-rtl.c: Potentially include config.h rather than bconfig.h.
5525 For host, include function.h, memmodel.h, and emit-rtl.h.
5526 (one_time_initialization): New function.
5527 (struct compact_insn_name): New struct.
5528 (compact_insn_names): New array.
5529 (find_code): Handle insn codes in compact dumps.
5530 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
5531 (bind_subst_iter_and_attr): Likewise.
5532 (add_condition_to_string): Likewise.
5533 (add_condition_to_rtx): Likewise.
5534 (apply_attribute_uses): Likewise.
5535 (add_current_iterators): Likewise.
5536 (apply_iterators): Likewise.
5537 (initialize_iterators): Guard usage of apply_subst_iterator with
5538 #ifdef GENERATOR_FILE.
5539 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
5540 (md_reader::read_mapping): Likewise.
5541 (add_define_attr_for_define_subst): Likewise.
5542 (add_define_subst_attr): Likewise.
5543 (read_subst_mapping): Likewise.
5544 (check_code_iterator): Likewise.
5545 (rtx_reader::read_rtx): Likewise. Move one-time initialization
5547 (one_time_initialization): New function.
5548 (rtx_reader::read_until): New method.
5549 (read_flags): New function.
5550 (parse_reg_note_name): New function.
5551 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
5552 Handle reuse_rtx ids.
5553 Wrap iterator lookup within #ifdef GENERATOR_FILE.
5554 Add parsing support for RTL dumps, mirroring the special-cases in
5555 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
5556 values, and calling handle_any_trailing_information.
5557 (rtx_reader::read_rtx_operand): Convert return type from void
5558 to rtx, returning return_rtx. Handle case 'e'. Call
5559 finalize_string on XSTR and XTMPL fields.
5560 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
5561 "(nil)" values were omitted. Call the postprocess vfunc on the
5563 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
5564 class ctor. Initialize m_in_call_function_usage. Call
5565 one_time_initialization.
5566 * rtl-tests.c (selftest::test_uncond_jump): Call
5567 set_new_first_and_last_insn.
5568 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
5569 * selftest-rtl.c: New file.
5570 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
5571 (selftest::get_insn_by_uid): New decl.
5572 * selftest-run-tests.c (selftest::run_tests): Call
5573 read_rtl_function_c_tests.
5574 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
5575 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
5578 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
5580 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
5581 operands in a special way. Assert that pos+len <= mode precision.
5583 2017-01-05 Jakub Jelinek <jakub@redhat.com>
5585 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
5586 3 argument Alias with unlimited for the negative form.
5587 (fno-vect-cost-model): Removed.
5589 2017-01-05 Martin Liska <mliska@suse.cz>
5591 * hsa-gen.c (gen_hsa_divmod): New function.
5592 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
5594 2017-01-05 Martin Liska <mliska@suse.cz>
5597 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
5600 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5602 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
5603 small constant length operands.
5605 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5607 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
5608 between loop iterations.
5610 2017-01-05 Martin Liska <mliska@suse.cz>
5613 * gimplify.c (gimplify_decl_expr): Compare to
5614 asan_poisoned_variables instread of checking flags.
5615 (gimplify_target_expr): Likewise.
5616 (gimplify_expr): Likewise.
5617 (gimplify_function_tree): Conditionally initialize
5618 asan_poisoned_variables.
5620 2017-01-04 Jeff Law <law@redhat.com>
5622 PR tree-optimizatin/78812
5623 * rtl.h (contains_mem_rtx_p): Prototype.
5624 * ifcvt.c (containts_mem_rtx_p): Move from here to...
5625 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
5626 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
5627 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
5628 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
5630 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5632 * input.c (assert_char_at_range): Default-initialize actual_range.
5634 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5636 * df-scan.c (df_ref_create_structure): Make regno unsigned,
5637 to match the caller.
5639 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5641 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
5642 insns after final jump in test to emit dummy move.
5644 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5646 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
5647 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
5649 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5651 * multiple_target.c (create_dispatcher_calls): Init e_next.
5652 * tree-ssa-loop-split.c (split_loop): Init border.
5653 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
5656 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
5661 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
5662 (altivec_register_operand): Do not return true if the operand
5663 contains a SUBREG mixing SImode and SFmode.
5664 (vsx_register_operand): Likewise.
5665 (vsx_reg_sfsubreg_ok): New predicate.
5666 (vfloat_operand): Do not return true if the operand contains a
5667 SUBREG mixing SImode and SFmode.
5668 (vint_operand): Likewise.
5669 (vlogical_operand): Likewise.
5670 (gpc_reg_operand): Likewise.
5671 (int_reg_operand): Likewise.
5672 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
5673 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
5674 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
5676 (rs6000_emit_move_si_sf_subreg): New helper function.
5677 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
5678 fixup SUBREGs involving SImode and SFmode.
5679 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
5680 numbers for the new peephole2 optimization.
5681 (peephole2 for SFmode unions): New peephole2 to optimize cases in
5682 the GLIBC math library that do AND/IOR/XOR operations on single
5683 precision floating point.
5684 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
5685 target macros to say whether we need to avoid SUBREGs mixing
5687 (TARGET_ALLOW_SF_SUBREG): Likewise.
5688 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
5689 (UNSPEC_SI_FROM_SF): Likewise.
5690 (iorxor): Change spacing.
5691 (and_ior_xor): New iterator for AND, IOR, and XOR.
5692 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
5693 (movdi_from_sf_zero_ext): Likewise.
5694 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
5695 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
5696 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
5697 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
5698 (fms<mode>4): Likewise.
5699 (fnma<mode>4): Likewise.
5700 (fnms<mode>4): Likewise.
5701 (nfma<mode>4): Likewise.
5702 (nfms<mode>4): Likewise.
5704 2017-01-04 Marek Polacek <polacek@redhat.com>
5707 * doc/invoke.texi: Document -Wpointer-compare.
5709 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5711 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
5714 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
5715 descriptions for -gdwarf-5 and emit them as uleb128 instead of
5718 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
5721 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
5722 documentation of the powerpc_popcntb_ok attribute.
5723 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
5724 code to issue warning messages if a requested CPU configuration is
5725 not supported by the binary (assembler and loader) toolchain.
5726 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
5727 made to define a built-in function that has been disabled.
5728 (paired_init_builtins): Add assertion to prevent ICE if attempt is
5729 made to define a built-in function that has been disabled.
5730 (altivec_init_builtins): Add comment explaining why definition
5731 of the DST built-in functions is not preceded by an assertion
5732 check. Add assertions to prevent ICE if attempts are made to
5733 define an altivec predicate or an abs* built-in function that has
5735 (htm_init_builtins): Add comment explaining why definition of the
5736 htm built-in functions is not preceded by an assertion check.
5738 2017-01-04 Jeff Law <law@redhat.com>
5740 PR tree-optimizatin/67955
5741 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
5742 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
5743 the points-to solution does not include pt_null. Use DECL_PT_UID
5746 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
5748 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
5749 Use gen_int_mode instead of gen_lopwart for const_int operands.
5751 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5753 PR tree-optimization/71563
5754 * match.pd: Simplify X << Y into X if Y is known to be 0 or
5755 out of range value - has low bits known to be zero.
5757 2017-01-04 Alan Modra <amodra@gmail.com>
5759 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
5760 * configure: Regenerate.
5761 * config.in: Regenerate.
5763 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5766 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
5767 a substring of the message, but strcmp with the whole message. Ifdef
5768 ENABLE_NLS, translate the message first using dgettext.
5770 2017-01-03 Jeff Law <law@redhat.com>
5772 PR tree-optimizatin/78856
5773 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
5774 (mark_threaded_blocks): Remove code to truncate thread paths that
5775 cross multiple loop headers. Instead invalidate the cached loop
5776 iteration information and handle case of a thread path walking
5777 into an irreducible region.
5779 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5782 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
5783 assertions. Add support for doing the signbit if the IEEE 128-bit
5784 floating point value is in a GPR.
5785 * config/rs6000/rs6000.md (Fsignbit): Delete.
5786 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
5787 Update the length attribute if the value is in a GPR.
5788 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
5789 the sign or zero extension instruction, since the value is always 0/1.
5790 (signbit<mode>2_dm2): Delete using <Fsignbit>.
5793 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
5794 extracting SImode to a GPR register so that we can generate a
5795 store, limit the vector to be in a traditional Altivec register
5796 for the vextuwrx instruction.
5798 2017-01-03 Ian Lance Taylor <iant@google.com>
5800 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
5802 2017-01-03 Martin Sebor <msebor@redhat.com>
5804 PR tree-optimization/78696
5805 * gimple-ssa-sprintf.c (format_floating): Correct handling of
5806 precision. Use MPFR for %f for greater fidelity. Correct handling
5808 (pass_sprintf_length::compute_format_length): Set width and precision
5809 specified by asrerisk to void_node for vararg functions.
5810 (try_substitute_return_value): Adjust dump output.
5812 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
5814 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
5816 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
5818 * doc/invoke.texi (SPARC options): Document -mlra as the default.
5819 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
5820 -mlra/-mno-lra was passed to the compiler.
5822 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
5824 PR rtl-optimization/65618
5825 * emit-rtl.c (try_split): Move initialization of "before" and
5826 "after" to just before the call to emit_insn_after_setloc.
5828 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
5830 * doc/md.texi (Standard Names): Remove reference to Java frontend.
5832 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
5834 * dwarf2out.c (gen_enumeration_type_die): When
5835 -gno-strict-dwarf, add a DW_AT_encoding attribute.
5837 2017-01-03 Jakub Jelinek <jakub@redhat.com>
5839 PR tree-optimization/78965
5840 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
5841 Change first argument from const call_info & to call_info &. For %n
5842 set info.nowrite to false.
5845 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
5846 possibly throwing calls.
5848 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
5849 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
5850 and fns handling, rather than in a separate case SSA_NAME.
5852 2017-01-02 Jeff Law <law@redhat.com>
5854 * config/darwin-driver.c (darwin_driver_init): Const-correctness
5855 fixes for first_period and second_period variables.
5857 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
5860 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
5861 (*insvqi_1): New insn pattern.
5862 (*insvqi_1_mem_rex64): Ditto.
5864 (*insvqi_3): Rename from *insvqi.
5866 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
5868 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
5870 * doc/cfg.texi (Edges): Remove reference to Java.
5871 (Maintaining the CFG): Ditto.
5873 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5876 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
5877 transparent aliases.
5879 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5882 * predict.def (PRED_CALL): Update hitrate.
5883 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
5884 * predict.c (tree_estimate_probability_bb): Split CALL predictor
5885 into direct/indirect/polymorphic variants.
5887 2017-01-01 Jakub Jelinek <jakub@redhat.com>
5889 Update copyright years.
5891 * gcc.c (process_command): Update copyright notice dates.
5892 * gcov-dump.c (print_version): Ditto.
5893 * gcov.c (print_version): Ditto.
5894 * gcov-tool.c (print_version): Ditto.
5895 * gengtype.c (create_file): Ditto.
5896 * doc/cpp.texi: Bump @copying's copyright year.
5897 * doc/cppinternals.texi: Ditto.
5898 * doc/gcc.texi: Ditto.
5899 * doc/gccint.texi: Ditto.
5900 * doc/gcov.texi: Ditto.
5901 * doc/install.texi: Ditto.
5902 * doc/invoke.texi: Ditto.
5904 Copyright (C) 2017 Free Software Foundation, Inc.
5906 Copying and distribution of this file, with or without modification,
5907 are permitted in any medium without royalty provided the copyright
5908 notice and this notice are preserved.