1 2017-03-16 Jakub Jelinek <jakub@redhat.com>
4 * tree-diagnostic.c (default_tree_printer): No longer static.
5 * tree-diagnostic.h (default_tree_printer): New prototype.
7 2017-03-16 Tamar Christina <tamar.christina@arm.com>
9 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
12 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
15 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
16 Use h_con constraint for operand 1.
17 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
18 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
20 2017-03-15 Jeff Law <law@redhat.com>
22 PR tree-optimization/71437
23 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
24 (record_temporary_equivalences): Use it.
26 PR tree-optimization/71437
27 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
28 tree-ssa-scopedtables.
29 (lookup_avail_expr, build_and_record_new_cond): Likewise.
30 (record_conditions, record_cond, vuse_eq): Likewise.
31 (record_edge_info): Adjust to API tweak of record_conditions.
32 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
33 (record_temporary_equivalences, optimize_stmt): Likewise.
34 (eliminate_redundant_computations): Likewise.
35 (record_equivalences_from_stmt): Likewise.
36 * tree-ssa-scopedtables.c: Include options.h and params.h.
37 (vuse_eq): New function, moved from tree-ssa-dom.c
38 (build_and_record_new_cond): Likewise.
39 (record_conditions): Likewise. Accept vector of conditions rather
40 than edge_equivalence structure for first argument.
41 for the first argument.
42 (avail_exprs_stack::lookup_avail_expr): New member function, moved
44 (avail_exprs_stack::record_cond): Likewise.
45 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
47 (avail_exprs_stack): Add new member functions lookup_avail_expr
49 (record_conditions): Declare.
51 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
54 * lra-constraints.c (process_alt_operands): Increase reject for
55 reloading an input/output operand.
57 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
60 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
61 insns to convert from signed/unsigned char/short to IEEE 128-bit
63 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
65 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
68 * config/i386/i386.c (ix86_vector_duplicate_value): Create
69 subreg of inner mode for values already in registers.
71 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
73 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
74 iteration reg is used after the loop.
76 2017-03-14 Martin Sebor <msebor@redhat.com>
78 PR tree-optimization/79800
79 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
80 precision in negative-positive range.
81 (format_floating): Call non-const overload with adjusted precision.
83 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
86 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
89 2017-03-14 Martin Sebor <msebor@redhat.com>
92 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
93 * builtins.def (aligned_alloc): Use it.
96 * Makefile.in (GTFILES): Add calls.c.
97 * calls.c: Include "gt-calls.h".
99 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
101 PR rtl-optimization/79728
102 * regs.h (struct target_regs): New field
103 x_contains_allocatable_regs_of_mode.
104 (contains_allocatable_regs_of_mode): New macro.
105 * reginfo.c (init_reg_sets_1): Initialize it, and change
106 contains_reg_of_mode so it includes global regs as well.
107 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
108 rather than contains_regs_of_mode.
110 2017-03-14 Martin Liska <mliska@suse.cz>
112 * doc/invoke.texi: Document options that can't be combined with
113 -fcheck-pointer-bounds.
115 2017-03-14 Martin Liska <mliska@suse.cz>
118 * doc/invoke.texi (-Wchkp): Document the option.
120 2017-03-14 Martin Liska <mliska@suse.cz>
122 * Makefile.in: Install gcov-dump.
124 2017-03-14 Martin Liska <mliska@suse.cz>
126 * multiple_target.c (expand_target_clones): Bail out for
127 an invalid attribute.
129 2017-03-14 Richard Biener <rguenther@suse.de>
131 * alias.c (struct alias_set_entry): Pack properly.
132 * cfgloop.h (struct loop): Likewise.
133 * cse.c (struct set): Likewise.
134 * ipa-utils.c (struct searchc_env): Likewise.
135 * loop-invariant.c (struct invariant): Likewise.
136 * lra-remat.c (struct cand): Likewise.
137 * recog.c (struct change_t): Likewise.
138 * rtl.h (struct address_info): Likewise.
139 * symbol-summary.h (function_summary): Likewise.
140 * tree-loop-distribution.c (struct partition): Likewise.
141 * tree-object-size.c (struct object_size_info): Likewise.
142 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
143 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
144 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
145 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
146 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
147 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
148 (struct _stmt_vec_info): Likewise.
150 2017-03-14 Martin Liska <mliska@suse.cz>
153 * multiple_target.c (create_dispatcher_calls): Check that
154 a target can create a function dispatcher.
156 2017-03-14 Martin Liska <mliska@suse.cz>
159 * multiple_target.c (expand_target_clones): Drop local.local
160 flag for default implementation.
162 2017-03-14 Richard Biener <rguenther@suse.de>
164 PR tree-optimization/80030
165 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
167 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
169 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
170 gcc_fallthrough() instead of __attribute__((fallthrough));
172 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
174 * doc/gcc.texi: Remove "up" link to (DIR).
175 * doc/gccint.texi: Ditto.
177 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
179 * doc/install.texi (Specific) <avr>: Remove reference to
182 2017-03-13 Jeff Law <law@redhat.com>
184 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
185 attribute rather than comments.
187 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
188 match_scratch operand is highest.
190 2017-03-13 Martin Liska <mliska@suse.cz>
193 * ipa-pure-const.c (warn_function_noreturn): If the declarations
194 is a CHKP clone, use original declaration.
196 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
198 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
199 (arc_conditional_register_usage): Use a different allocation order
200 when optimizing for size.
201 * common/config/arc/arc-common.c (arc_option_optimization_table):
202 Section anchors default on when optimizing for size.
204 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
206 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
208 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
210 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
211 * config/arc/arc.md (cpu_facility): Add cd variant.
212 (*movqi_insn): Add code density variant.
213 (*movhi_insn): Likewise.
214 (*movqi_insn): Likewise.
215 (*addsi3_mixed): Likewise.
216 (subsi3_insn): Likewise.
218 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
220 * config/arc/arc.md (movsi_cond_exec): Update constraint.
222 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
224 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
225 expressions with MINUS and UNARY ops.
227 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
230 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
232 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
233 between vec_select and vector argument.
234 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
235 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
236 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
237 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
238 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
239 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
241 2017-03-13 Richard Biener <rguenther@suse.de>
244 * params.def (vect-max-peeling-for-alignment): Fix typo.
246 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
248 * doc/install.texi (Specific) <mips-*-*>: Remove description of
249 issue that only occurred with binutils below 2.18.
251 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
253 * doc/install.texi (Specific) <cris-axis-elf>: No longer
254 refer to binutils 2.11/2.12 minimum.
256 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
258 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
259 ftp.kernel.org and simplify binutils requirement.
261 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
263 * doc/invoke.texi (Warning Options): Fix spelling of link-time
265 (Optimize Options): Ditto. Also remove redundancy.
267 2017-03-10 David Malcolm <dmalcolm@redhat.com>
270 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
272 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
273 to G_ to avoid double translation.
275 2017-03-10 David Malcolm <dmalcolm@redhat.com>
278 * auto-profile.c (get_combined_location): Convert leading
279 character of diagnostics to lower case and remove trailing period.
280 (read_profile): Likewise for various diagnostics.
281 * config/arm/arm.c (arm_option_override): Remove trailing period
282 from various diagnostics.
283 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
284 (msp430_expand_delay_cycles): Likewise.
286 2017-03-10 David Malcolm <dmalcolm@redhat.com>
289 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
290 full command-line argument, rather than just "str".
291 (aarch64_validate_march): Likewise.
292 (aarch64_validate_mtune): Likewise.
294 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
296 PR rtl-optimization/78911
297 * lra-assigns.c (must_not_spill_p): New function.
300 2017-03-10 Jakub Jelinek <jakub@redhat.com>
302 PR tree-optimization/79981
303 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
304 ATOMIC_COMPARE_EXCHANGE ifn result.
305 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
306 IFN_ATOMIC_COMPARE_EXCHANGE.
308 2017-03-10 David Malcolm <dmalcolm@redhat.com>
311 * opts.c (parse_sanitizer_options): Add missing question mark to
312 "did you mean" message.
314 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
316 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
318 (VMULEUH_UNS): Likewise.
319 (VMULOUB_UNS): Likewise.
320 (VMULOUH_UNS): Likewise.
321 * config/rs6000/rs6000.c (builtin_function_type): Remove
322 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
324 2017-03-10 David Malcolm <dmalcolm@redhat.com>
327 * read-rtl-function.c (function_reader::read_rtx_operand): Update
328 x with result of extra_parsing_for_operand_code_0.
329 (function_reader::extra_parsing_for_operand_code_0): Convert
330 return type from void to rtx, returning x. When reading
331 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
332 larger size containing struct block_symbol.
334 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
336 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
337 -mfloat128-hardware without -m64.
339 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
342 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
343 entries to the case statement that marks unsigned arguments to
344 overloaded functions.
346 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
348 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
349 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
351 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
354 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
355 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
357 2017-03-10 Martin Liska <mliska@suse.cz>
361 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
362 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
365 2017-03-10 Olivier Hainque <hainque@adacore.com>
367 * tree-switch-conversion (array_value_type): Start by resetting
368 candidate type to it's main variant.
370 2017-03-10 Jakub Jelinek <jakub@redhat.com>
372 PR rtl-optimization/79909
373 * combine.c (try_combine): Use simplify_replace_rtx on individual
374 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
375 of the whole CALL_INSN_FUNCTION_USAGE.
377 PR tree-optimization/79972
378 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
379 get_range_info on SSA_NAMEs. Formatting fixes.
381 2017-03-10 Richard Biener <rguenther@suse.de>
382 Jakub Jelinek <jakub@redhat.com>
384 PR tree-optimization/77975
385 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
387 (get_val_for): For constant x return it. Formatting fix.
388 (loop_niter_by_eval): Avoid pointless looping if the next iteration
389 would use the same bases as the current one.
391 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
393 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
394 instead of vec_select for V1TImode.
395 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
397 (VSX_LE_128): Add V1TI to this mode iterator.
398 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
399 (*vsx_le_perm_store_<mode>): Likewise.
400 (pre-reload splitter for VSX stores): Likewise.
401 (post-reload splitter for VSX stores): Likewise.
402 (*vsx_xxpermdi2_le_<mode>): Likewise.
403 (*vsx_lxvd2x2_le_<mode>): Likewise.
404 (*vsx_stxvd2x2_le_<mode>): Likewise.
406 2017-03-09 Michael Eager <eager@eagercon.com>
408 Correct failures with --enable-checking=yes,rtl.
410 * config/microblaze/microblaze.c (microblaze_expand_shift):
411 Replace GET_CODE test with CONST_INT_P and INTVAL test with
413 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
414 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
416 2017-03-09 Richard Biener <rguenther@suse.de>
418 PR tree-optimization/79977
419 * graphite-scop-detection.c (scop_detection::merge_sese):
420 Handle the case of extra exits to blocks dominating the entry.
422 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
424 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
427 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
429 PR rtl-optimization/79949
430 * lra-constraints.c (process_alt_operands): Check memory when
431 trying to predict a cycle. Print about the overall increase.
433 2017-03-09 Richard Biener <rguenther@suse.de>
436 * gimple-expr.c (useless_type_conversion_p): Preserve
437 TYPE_SATURATING for fixed-point types.
439 2017-03-09 Richard Biener <rguenther@suse.de>
442 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
443 alignment of BLKmode params.
445 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
448 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
449 (VALL_NO_V2Q): Likewise.
451 * config/aarch64/aarch64-simd.md
452 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
454 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
455 VALL_NO_V2Q mode iterator.
456 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
458 2017-03-09 Martin Liska <mliska@suse.cz>
460 PR tree-optimization/79631
461 * tree-chkp-opt.c (chkp_is_constant_addr): Call
462 tree_int_cst_sign_bit just for INTEGER constants.
464 2017-03-09 Martin Liska <mliska@suse.cz>
468 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
471 2017-03-09 Marek Polacek <polacek@redhat.com>
474 * tree.c (inchash::add_expr): Handle TREE_VEC.
476 2017-03-09 Martin Liska <mliska@suse.cz>
479 (chkp_narrow_size_and_offset): New function.
480 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
481 (void chkp_parse_bit_field_ref): New function.
482 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
483 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
485 2017-03-09 Martin Liska <mliska@suse.cz>
488 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
489 (chkp_find_bounds_1): Remove gcc_unreachable.
491 2017-03-09 Jakub Jelinek <jakub@redhat.com>
494 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
495 BUILT_IN_SYNC*, determine the access type from the size suffix and
496 always build a MEM_REF with that type. Handle forgotten
497 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
500 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
501 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
502 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
503 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
504 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
505 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
506 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
507 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
508 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
509 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
510 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
511 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
512 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
513 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
514 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
515 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
516 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
517 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
518 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
519 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
520 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
521 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
522 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
523 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
524 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
525 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
526 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
527 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
528 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
529 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
530 definitions outside of __OPTIMIZE__ guarded section.
533 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
534 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
535 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
536 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
539 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
541 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
542 ("vfenez<mode>"): Add missing constraints.
544 2017-03-08 Martin Sebor <msebor@redhat.com>
547 * config/nds32/nds32.c (nds32_option_override):
549 2017-03-08 Jakub Jelinek <jakub@redhat.com>
552 * gimplify.c (gimplify_omp_for): Replace index var in outer
553 taskloop statement with an artificial variable and add
554 OMP_CLAUSE_PRIVATE clause for it.
556 2017-03-08 Richard Biener <rguenther@suse.de>
558 PR tree-optimization/79955
559 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
560 for accesses that are completely outside of the variable.
562 2017-03-08 Andrew Haley <aph@redhat.com>
564 PR tree-optimization/79943
565 * tree-ssa-loop-split.c (compute_new_first_bound): When
566 calculating the new upper bound, (END-BEG) should be added, not
569 2017-03-08 Jakub Jelinek <jakub@redhat.com>
571 * config/avr/avr.md (setmemhi): Make sure match_dup
572 operand number comes before match_scratch.
574 2017-03-08 Richard Biener <rguenther@suse.de>
576 PR tree-optimization/79920
577 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
578 with ncopies == 1 to ...
579 (vect_transform_slp_perm_load): ... here. Properly compute
580 all element loads by iterating VF times over the group. Do
581 not handle ncopies (computed in a broken way) in
582 vect_create_mask_and_perm.
584 2017-03-08 Jakub Jelinek <jakub@redhat.com>
587 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
588 is a uniform vector, use uniform_vector_p return value instead of
589 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
591 2017-03-07 Marek Polacek <polacek@redhat.com>
594 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
595 (alloca_call_type): Likewise.
597 2017-03-07 Martin Liska <mliska@suse.cz>
599 * gcov.c (process_args): Put comment to correct location.
601 2017-03-07 Martin Liska <mliska@suse.cz>
604 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
605 Use array_at_struct_end_p instead of DECL_CHAIN (field).
606 (chkp_narrow_bounds_for_field): Likewise.
607 (chkp_parse_array_and_component_ref): Pass one more argument to
610 2017-03-07 Richard Biener <rguenther@suse.de>
612 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
615 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
617 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
618 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
620 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
623 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
624 to end of description.
625 (PARAM_MAX_STORES_TO_MERGE): Likewise.
627 2017-03-07 Jakub Jelinek <jakub@redhat.com>
629 PR rtl-optimization/79901
630 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
632 (*avx512f_<code><mode>3<mask_name>): ... this.
633 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
634 iterator instead of VI8_AVX2_AVX512BW.
636 PR rtl-optimization/79901
637 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
638 min/max expander, expand it using expand_vec_cond_expr.
641 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
644 2017-03-06 Jakub Jelinek <jakub@redhat.com>
647 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
648 to void * for PCH reasons.
649 * dwarf2out.c (output_loc_operands, output_die): Cast
650 v.val_vec.array to unsigned char *.
652 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
655 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
658 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
660 PR rtl-optimization/79571
661 * lra-constraints.c (process_alt_operands): Calculate static
662 reject and subtract it from overall when only addresses will be
665 2017-03-06 Julia Koval <julia.koval@intel.com>
668 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
669 incoming stack boundary to 128 for 64-bit targets.
671 2017-03-06 Richard Biener <rguenther@suse.de>
673 PR tree-optimization/79894
674 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
675 to NULL after folding it.
677 2017-03-06 Richard Biener <rguenther@suse.de>
679 PR tree-optimization/79824
680 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
681 check disabling peeling for gaps.
683 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
685 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
686 attributes): Document gettimeofday.
688 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
690 * config/s390/s390.c (s390_option_override_internal): Set
691 PARAM_MIN_VECT_LOOP_BOUND
693 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
695 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
696 * config/s390/s390.md: Likewise.
698 2017-03-06 Jakub Jelinek <jakub@redhat.com>
701 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
702 (<avx2_avx512>_perm<mode>): Rename to ...
703 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
705 (<avx512>_perm<mode>_mask): Rename to ...
706 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
708 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
709 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
710 instead of VI8F_256_512.
711 (avx512f_perm<mode>): New define_expand.
712 (avx512f_perm<mode>_mask): Likewise.
713 (avx512f_perm<mode>_1<mask_name>): New define_insn.
714 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
716 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
718 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
719 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
721 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
723 2017-03-06 Martin Liska <mliska@suse.cz>
726 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
727 when having a SSA NAME w/o VAR_DECL assigned to it.
729 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
731 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
732 msa_dpsub_<su>_d): Fix MODE for vec_select.
734 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
736 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
738 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
740 2017-03-06 Richard Biener <rguenther@suse.de>
742 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
743 * plugin.c (register_plugin_info): Likewise.
744 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
746 2017-03-05 Jakub Jelinek <jakub@redhat.com>
748 * config/i386/sse.md (sse_storehps, sse_storelps,
749 avx_<castmode><avxsizesuffix>_<castmode>,
750 avx512f_<castmode><avxsizesuffix>_<castmode>,
751 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
752 in condition that at least one operand is not a MEM.
754 2017-03-03 Jakub Jelinek <jakub@redhat.com>
757 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
758 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
760 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
761 gimple_call_nothrow_p flag based on whether original builtin can throw.
762 If it can, emit following stmts on the fallthrough edge.
763 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
764 don't create new bb if inserting just debug stmts on the edge, try to
765 insert them on the fallthru bb or just reset debug stmts.
767 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
770 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
771 restore recog_data (including the operand rtxes inside it) around
772 the call to get_insn_template.
774 2017-03-03 Martin Sebor <msebor@redhat.com>
776 PR tree-optimization/79699
777 * context.c (context::~context): Free MPFR caches to avoid
778 a memory leak on program exit.
780 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
782 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
783 Use wide_int::ulow () instead of .elt (0).
785 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
787 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
788 (*pushxf): Limit oF constraint to 32bit targets and add oC
789 constraint for 64bit targets.
790 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
791 (*pushdf): Change rmF constraint to rmC.
793 2017-03-03 Martin Liska <mliska@suse.cz>
795 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
796 Remove unused variable.
798 2017-03-03 Jakub Jelinek <jakub@redhat.com>
801 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
802 is a memory operand, increase num_memory.
803 (ix86_expand_args_builtin): Likewise.
805 2017-03-03 Jan Hubicka <jh@suse.cz>
808 * ipa-devirt.c (maybe_record_node): Properly handle
809 __cxa_pure_virtual visibility.
811 2017-03-03 Martin Liska <mliska@suse.cz>
813 PR tree-optimization/79803
814 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
816 (pass_loop_prefetch::execute): Disabled optimization if an
817 assumption about L1 cache size is not met.
819 2017-03-03 Martin Liska <mliska@suse.cz>
821 PR rtl-optimization/79574
822 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
823 (hash_scan_set): Likewise.
824 (dump_hash_table): Likewise.
825 (hoist_code): Likewise.
827 2017-03-03 Richard Biener <rguenther@suse.de>
829 * fixed-value.c (fixed_from_string): Restore use of elt (1)
830 in place of uhigh ().
831 (fixed_convert_from_real): Likewise.
833 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
836 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
838 2017-03-03 Richard Biener <rguenther@suse.de>
841 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
842 TYPE_OVERFLOW_UNDEFINED check.
844 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
846 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
848 (vector_ae_<mode>_p): Likewise.
849 (vector_nez_<mode>_p): Likewise.
850 (vector_ne_v2di_p): Likewise.
851 (vector_ae_v2di_p): Likewise.
852 (vector_ne_<mode>_p): Likewise.
853 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
855 (vsx_tsqrt<mode>2_fe): Likewise.
857 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
860 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
862 2017-03-02 Jakub Jelinek <jakub@redhat.com>
864 PR rtl-optimization/79780
865 * cprop.c (one_cprop_pass): When second and further conditional trap
866 in a single basic block is turned into an unconditional trap, turn it
867 into a deleted note to avoid RTL verification failures.
869 2017-03-02 Richard Biener <rguenther@suse.de>
871 * fold-const.c (const_binop): Use ulow () instead of elt (0).
873 2017-03-02 Richard Biener <rguenther@suse.de>
875 PR tree-optimization/79345
877 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
878 param and abort the walk, returning -1 if it is hit.
879 (walk_aliased_vdefs): Take a limit param and pass it on.
880 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
881 defaulting to 0 and return a signed int.
882 * tree-ssa-uninit.c (struct check_defs_data): New struct.
883 (check_defs): New helper.
884 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
885 about uninitialized memory.
886 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
887 bogus uninitialized warning.
888 (fixed_convert_from_real): Likewise.
890 2017-03-02 Bin Cheng <bin.cheng@arm.com>
892 PR tree-optimization/66768
893 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
894 iv_use if base object can't be determined.
896 2017-03-02 Jakub Jelinek <jakub@redhat.com>
898 PR tree-optimization/79345
899 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
900 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
901 (get_pattern_stats): Initialize it.
902 * genemit.c (gen_expand): Verify match_scratch numbers come after
903 match_operand/match_dup numbers.
904 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
905 match_scratch numbers.
906 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
908 * config/s390/s390.md (trunctdsd2): Likewise.
910 2017-03-02 Richard Biener <rguenther@suse.de>
912 * wide-int.h (wide_int_storage::operator=): Implement in terms
915 2017-03-02 Richard Biener <rguenther@suse.de>
917 PR tree-optimization/79777
918 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
919 the to insert expression to sth existing.
921 2017-03-01 Martin Sebor <msebor@redhat.com>
924 * gimple-ssa-sprintf.c
925 (directive::known_width_and_precision): New function.
926 (format_integer): Use it.
927 (get_mpfr_format_length): Consider the full range of precision
928 when computing %g output with the # flag. Set the likely byte
929 count to 3 rather than 1 when precision is indeterminate.
930 (format_floating): Correct the lower bound of precision.
932 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
934 * doc/invoke.texi: Document default code model for 64-bit Linux.
936 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
939 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
940 udiv rather than div since input pattern is unsigned.
942 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
944 * config/i386/i386.c (print_reg): Warn for values of
945 unsupported size in integer register.
947 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
950 * config/rs6000/predicates.md (current_file_function_operand): Do
951 not allow self calls to be local if the function is replaceable.
953 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
956 * config/rs6000/altivec.h (vec_ctz and others): Change the
957 preprocessor macro that controls conditional compilation from
958 _ARCH_PWR9 to __POWER9_VECTOR__.
959 (vec_all_ne): Change parameterization of __altivec_scalar_pred
960 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
961 control (instead of _ARCH_PWR9 control) so that template
962 definition uses power9-specific function.
963 (vec_any_eq): Likewise.
964 (vec_all_ne): Change macro definition to use a power9-specific
965 expansion under #ifdef __POWER9_VECTOR__ control (instead of
967 (vec_any_eq) Likewise.
968 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
969 expansion for CMPNEF to remove support for xvcmpnesp instruction.
970 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
971 support for xvcmpnedp instruction.
972 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
973 macro expansion so that Power9 implementation of vec_all_ne does
974 not use the AltiVec predicate framework.
975 (VCMPNEH_P): Likewise.
976 (VCMPNEW_P): Likewise.
977 (VCMPNED_P): Likewise.
978 (VCMPNEFP_P): Likewise.
979 (VCMPNEDP_P): Likewise.
980 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
981 implementation of vec_any_eq to not use AltiVec predicate
983 (VCMPAEH_P): Likewise.
984 (VCMPAEW_P): Likewise.
985 (VCMPAED_P): Likewise.
986 (VCMPAEFP_P): Likewise.
987 (VCMPAEDP_P): Likewise.
988 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
989 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
990 not use the AltiVec predicate framework.
991 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
992 of vec_any_eq to not use AltiVec predicate framework.
993 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
994 support for predefined __POWER9_VECTOR__ macro to indicate that
995 Power9 instruction selection is enabled.
996 (altivec_overloaded_builtins): Remove extraneous
997 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
998 function argument types RS6000_BTI_bool_V16QI and
999 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
1000 entry for overloaded function argument types RS6000_BTI_bool_V4SI
1001 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
1002 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
1003 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
1004 Power9 for implementations of vec_cmpne. Change the signature for
1005 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
1006 (representing vec_all_ne) to remove the previously described first
1007 argument of type RS6000_BTI_INTSI, as this was an artifact of
1008 reliance on the AltiVec predicate framework, which is no longer
1009 used in the implementation of these functions. Add
1010 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
1011 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
1012 since, unlike the AltiVec predicate framework implementation, we
1013 do not share function descriptors between vec_alle and vec_anyeq.
1014 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
1015 set of modes that receive special treatment even when
1016 TARGET_P9_VECTOR is true. The special treatment emits code that
1017 does not depend on Power9 instructions.
1018 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
1019 define_expand to not rely on AltiVec predicate framework.
1020 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
1022 (vector_ne_v2di_p): Change this define_expand to not rely on
1023 AltiVec predicate framework.
1024 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
1026 (vector_ne_<mode>_p): Change this define_expand to not rely on
1027 AltiVec predicate framework.
1028 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
1030 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
1031 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
1032 define_insn pattern.
1033 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
1034 define_insn pattern because the xvcmpne<VSs>. instruction is not
1036 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
1037 instruction is not supported.
1039 2017-03-01 Jakub Jelinek <jakub@redhat.com>
1041 * config/nvptx/nvptx.c: Include intl.h.
1043 2017-03-01 Martin Jambor <mjambor@suse.cz>
1046 * ipa-prop.h (ipa_bits): Removed field known.
1047 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
1048 to pointers. Adjusted their comments to warn about their sharing.
1049 (ipcp_transformation_summary): Change bits to a vector of pointers.
1050 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
1051 (ipa_get_ipa_bits_for_value): Declare.
1052 * tree-vrp.h (value_range): Mark as GTY((for_user)).
1053 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
1054 (ipa_bits_hash_table): Likewise.
1055 (ipa_vr_ggc_hash_traits): Likewise.
1056 (ipa_vr_hash_table): Likewise.
1057 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
1058 being pointers and vr_known being removed.
1059 (ipa_set_jf_unknown): Likewise.
1060 (ipa_get_ipa_bits_for_value): New function.
1061 (ipa_set_jfunc_bits): Likewise.
1062 (ipa_get_value_range): New overloaded functions.
1063 (ipa_set_jfunc_vr): Likewise.
1064 (ipa_compute_jump_functions_for_edge): Use the above functions to
1065 construct bits and vr parts of jump functions.
1066 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
1067 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
1069 (ipcp_grow_transformations_if_necessary): Also allocate
1070 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
1072 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
1073 them. Fix too long lines.
1074 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
1075 vr_known being removed.
1076 (ipa_read_jump_function): Use new setter functions to construct bits
1077 and vr parts of jump functions or set them to NULL.
1078 (write_ipcp_transformation_info): Adjust for bits being pointers.
1079 (read_ipcp_transformation_info): Likewise.
1080 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
1082 Include gt-ipa-prop.h.
1083 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
1085 (ipcp_store_bits_results): Likewise.
1086 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
1087 Do not write to existing jump functions but use a temporary instead.
1089 2017-03-01 Jakub Jelinek <jakub@redhat.com>
1092 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
1093 attempt to use its first operand as BIT_FIELD_REF base.
1095 2017-03-01 Richard Biener <rguenther@suse.de>
1098 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
1099 interpolating formula in wrapping arithmetic.
1100 (chrec_apply): Convert chrec_evaluate return value to wanted type.
1102 2017-03-01 Jakub Jelinek <jakub@redhat.com>
1104 PR tree-optimization/79734
1105 * tree-vect-generic.c (expand_vector_condition): Optimize
1106 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
1107 Handle VEC_COND_EXPR where comparison has different inner width from
1110 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
1112 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
1113 markup, and similar issues. Remove @opindex entries for things
1114 that aren't options. Add missing -mmpy-option entries.
1116 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1118 PR tree-optimization/79737
1119 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
1120 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
1121 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
1122 instead of byte_size. Formatting fix.
1123 (shift_bytes_in_array_right): Formatting fix.
1125 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
1128 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
1129 condition on optimize for the leaf function test.
1131 2017-02-28 Martin Liska <mliska@suse.cz>
1134 * read-rtl-function.c (function_reader::handle_unknown_directive):
1135 Bail out when one uses -flto.
1137 2017-02-28 Martin Liska <mliska@suse.cz>
1139 * common.opt: Replace space with tabular for options of <number>
1141 * config/i386/i386.opt: Show <number> value for
1142 -mlarge-data-threshold.
1143 * opts.c (print_filtered_help): Do not display number in hexadecimal
1146 2017-02-28 Martin Liska <mliska@suse.cz>
1148 * common.opt: Fix --help=option -Q for options which are of
1151 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
1153 * config/i386/i386.c (print_reg): Error out for values
1154 of 8-bit size in invalid integer register.
1156 2017-02-28 Martin Sebor <msebor@redhat.com>
1158 PR tree-optimization/79691
1159 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
1161 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1164 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
1165 gcc_unreachable with output_operand_lossage.
1167 2017-02-28 Richard Biener <rguenther@suse.de>
1169 PR tree-optimization/79740
1170 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
1172 (visit_nary_op): Insert the nary into the hashtable if we
1173 pattern-matched sth.
1174 * tree-ssa-pre.c (eliminate_insert): Robustify.
1176 2017-02-28 Richard Biener <rguenther@suse.de>
1179 * fold-const.c (decode_field_reference): Reject out-of-bound
1182 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1184 * config/i386/i386.c: Include intl.h.
1185 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
1186 instead of just cond ? "..." : "...".
1187 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
1188 * coverage.c (read_counts_file): Likewise.
1189 * omp-offload.c: Include intl.h.
1190 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
1191 of just cond ? "..." : "...".
1192 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
1193 of just cond ? "..." : "...".
1195 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
1198 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
1200 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
1201 'tune for' CPU name.
1202 * config/arm/arm-cpu-data.h: Regenerated.
1204 2017-02-28 Richard Biener <rguenther@suse.de>
1206 PR tree-optimization/79732
1207 * tree-inline.c (expand_call_inline): Do not shadow var.
1209 2017-02-28 Richard Biener <rguenther@suse.de>
1211 PR tree-optimization/79723
1212 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
1213 address-space properly.
1215 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
1217 * doc/optinfo.texi (Optimization groups): Fix option used for
1219 * doc/invoke.texi (-fopt-info): Document "omp".
1220 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
1221 (OPTGROUP_ALL): Add OPTGROUP_OMP.
1222 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
1223 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
1224 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
1226 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
1228 * dumpfile.c (optgroup_options): Instead of "openmp", associate
1229 OPTGROUP_OMP with "omp".
1231 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
1234 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
1235 for arithmetic shift of unsigned V2DI.
1237 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
1239 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
1240 arc/linux.h headers.
1241 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
1242 (LINK_SPEC): Likewise.
1243 (ARC_TLS_EXTRA_START_SPEC): Likewise.
1244 (EXTRA_SPECS): Likewise.
1245 (STARTFILE_SPEC): Likewise.
1246 (ENDFILE_SPEC): Likewise.
1247 (LIB_SPEC): Likewise.
1248 (TARGET_SDATA_DEFAULT): Likewise.
1249 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
1250 (MULTILIB_DEFAULTS): Likewise.
1251 (DWARF2_UNWIND_INFO): Likewise.
1252 * config/arc/big.h: New file.
1253 * config/arc/elf.h: Likewise.
1254 * config/arc/linux.h: Likewise.
1255 * config/arc/t-uClibc: Remove.
1257 2017-02-27 Bin Cheng <bin.cheng@arm.com>
1259 PR tree-optimization/77536
1260 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
1261 (tree_transform_and_unroll_loop): Use above function to compute the
1262 estimated niter of unrolled loop and use it when scaling profile.
1263 Also use count info rather than frequency if it's non-zero.
1264 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
1265 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
1266 (vect_transform_loop): Call above function.
1268 2017-02-27 Richard Biener <rguenther@suse.de>
1270 PR tree-optimization/45397
1271 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
1272 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
1273 (visit_nary_op): Add pattern matching for CSEing sign-changed
1274 or truncated operations with wider ones.
1276 2017-02-27 Richard Biener <rguenther@suse.de>
1278 PR tree-optimization/79690
1279 * tree-vect-stmts.c (vectorizable_store): Use vector type
1280 built from the DR with address-space.
1282 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
1284 * doc/invoke.texi (Optimize Options): Refine the description
1285 of asan-use-after-return.
1287 2017-02-25 Alan Modra <amodra@gmail.com>
1289 PR rtl-optimization/79584
1290 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
1291 base, not ad->base_term, the reg within base. Remove assertion
1292 that ad->base == ad->base_term. Replace gen_int_mode using
1293 bogus mode with const0_rtx.
1295 2017-02-25 Jakub Jelinek <jakub@redhat.com>
1298 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
1299 FMA_EXPR like tcc_binary or tcc_unary.
1301 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
1304 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
1306 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
1307 (output_loc_operands): Handle DW_OP_call_ref and
1308 DW_OP_GNU_variable_value.
1309 (struct variable_value_struct): New type.
1310 (struct variable_value_hasher): Likewise.
1311 (variable_value_hash): New variable.
1312 (string_types): Remove.
1313 (copy_loc_descr): New function.
1314 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
1315 (prepend_loc_descr_to_each): New function.
1316 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
1317 instead of add_loc_descr_to_each if the first argument is single
1318 location list and the second has multiple.
1319 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
1320 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
1321 when looking for variable value which doesn't have other location info.
1322 (loc_list_from_tree): Formatting fix.
1323 (gen_array_type_die): Simplify DW_AT_string_length handling.
1324 (adjust_string_types): Remove.
1325 (gen_subprogram_die): Don't call adjust_string_types nor test/set
1326 string_types. Call resolve_variable_values.
1327 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
1328 (resolve_addr_in_expr): Likewise. Add A argument.
1329 (copy_deref_exprloc): Remove deref argument. Adjust for the
1330 original expression being DW_OP_GNU_variable_value with optionally
1331 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
1332 optionally after it.
1333 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
1334 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
1335 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
1336 (variable_value_hasher::hash, variable_value_hasher::equal): New
1338 (resolve_variable_value_in_expr, resolve_variable_value,
1339 resolve_variable_values, note_variable_value_in_expr,
1340 note_variable_value): New functions.
1341 (dwarf2out_early_finish): Call note_variable_value on all toplevel
1344 2017-02-24 Jakub Jelinek <jakub@redhat.com>
1347 * opts.h (handle_generated_option): Add GENERATED_P argument.
1348 * opts-common.c (handle_option): Adjust function comment.
1349 (handle_generated_option): Add GENERATED_P argument, pass it to
1351 (control_warning_option): Pass false to handle_generated_option
1353 * opts.c (maybe_default_option): Pass true to handle_generated_option
1355 * optc-gen.awk: Likewise.
1357 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1359 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
1360 a REG, look at the REG it is a SUBREG of.
1361 (splitter for cmpeqsi_t): Ditto.
1363 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1365 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
1366 the special USEs with the pattern of the insn, not the insn itself.
1368 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
1371 * doc/invoke.texi: Document -mload-store-pairs.
1373 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1374 Sandra Loosemore <sandra@codesourcery.com>
1376 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
1377 argument isn't a CONST_INT.
1378 (nios2_alternate_compare_const): Assert op is a CONST_INT.
1379 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
1380 (nios2_validate_compare): Bypass alternate compare logic if *op2
1382 (ldstwm_operation_p): Return false if first_base is not a REG or
1383 if first_offset is not a CONST_INT.
1385 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1387 * config/cris/cris.md: Use correct operand in a define_peephole2.
1389 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1391 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
1393 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1395 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
1396 this_insn if it is an INSN or JUMP_INSN.
1397 (force_offsettable): Look at base, not at addr.
1398 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
1399 on things that aren't necessarily CONST_INTs.
1401 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
1403 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
1404 -mfpmath=sse is the default also for x86-32 targets with SSE2
1405 instruction set when @option{-ffast-math} is enabled
1407 2017-02-24 Jeff Law <law@redhat.com>
1409 PR rtl-optimizatoin/79286
1410 * ira.c (update_equiv_regs): Drop may_trap_p exception to
1413 2017-02-24 Richard Biener <rguenther@suse.de>
1415 PR tree-optimization/79389
1416 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
1419 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
1421 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
1422 function comment to reflect reality.
1423 (loop_exits_before_overflow): Fix typo in function description.
1425 2017-02-24 Richard Biener <rguenther@suse.de>
1427 PR tree-optimization/79389
1428 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
1429 properly that a threading opportunity exists. Detect conditional
1430 copy/constant propagation opportunities.
1432 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
1434 * config/visium/visium.md (type): Add trap.
1435 (b): New mode attribute.
1436 (*btst): Rename into...
1437 (*btst<mode>): ...this and adjust.
1438 (*cbranchsi4_btst_insn): Rename into...
1439 (*cbranch<mode>4_btst_insn): ...this and adjust.
1440 (trap): New define_insn.
1442 2017-02-23 Jakub Jelinek <jakub@redhat.com>
1444 PR tree-optimization/79389
1445 * ifcvt.c (struct noce_if_info): Add rev_cond field.
1446 (noce_reversed_cond_code): New function.
1447 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
1448 reversed_comparison_code. Formatting fix.
1449 (noce_try_store_flag): Test rev_cond != NULL in addition to
1450 reversed_comparison_code.
1451 (noce_try_store_flag_constants): Likewise.
1452 (noce_try_store_flag_mask): Likewise.
1453 (noce_try_addcc): Use rev_cond if non-NULL instead of
1454 reversed_comparison_code.
1455 (noce_try_cmove_arith): Likewise. Formatting fixes.
1456 (noce_try_minmax, noce_try_abs): Clear rev_cond.
1457 (noce_find_if_block): Initialize rev_cond.
1458 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
1459 instead of false as last argument never attempt to reverse it
1462 2017-02-23 Bin Cheng <bin.cheng@arm.com>
1464 PR tree-optimization/79663
1465 * tree-predcom.c (combine_chains): Process refs in reverse order
1466 only for ZERO length chains, and add explaining comment.
1468 2017-02-23 Jeff Law <law@redhat.com>
1470 PR tree-optimization/79578
1471 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
1472 in call to operand_equal_p.
1474 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
1477 * config/i386/cpuid.h: Fix another undefined behavior.
1479 2017-02-23 Richard Biener <rguenther@suse.de>
1481 PR tree-optimization/79683
1482 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
1483 vector types for data-refs.
1485 2017-02-23 Martin Liska <mliska@suse.cz>
1487 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
1489 2017-02-23 Jakub Jelinek <jakub@redhat.com>
1492 * internal-fn.c (get_range_pos_neg): Moved to ...
1493 * tree.c (get_range_pos_neg): ... here. No longer static.
1494 * tree.h (get_range_pos_neg): New prototype.
1495 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
1496 are known to be in between 0 and signed maximum inclusive, try to
1497 expand both unsigned and signed divmod and use the cheaper one from
1500 2017-02-22 Jeff Law <law@redhat.com>
1502 PR tree-optimization/79578
1503 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
1504 to compare base operands.
1506 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
1509 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
1510 gpc_reg_operand instead of fpr_reg_operand.
1512 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
1514 * config/mips/mips.c (mips_return_in_memory): Force FP
1515 vector types to be returned in memory for o32 ABI.
1517 2017-02-22 Jakub Jelinek <jakub@redhat.com>
1519 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
1520 instead of DW_TAG_member for static data member declarations and don't
1521 set no_linkage_name for static inline data members.
1522 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
1525 2017-02-22 Martin Liska <mliska@suse.cz>
1527 * doc/invoke.texi: Replace inequality signs with square brackets
1530 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1532 PR tree-optimization/68644
1533 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
1535 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
1538 * lra-constraints.c (simplify_operand_subreg): Handle
1539 WORD_REGISTER_OPERATIONS targets.
1541 2017-02-22 Jakub Jelinek <jakub@redhat.com>
1544 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
1545 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
1546 elimination by swapping fld*.
1548 2017-02-22 Richard Biener <rguenther@suse.de>
1550 PR tree-optimization/79673
1551 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
1552 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
1553 irrelevant address-space qualifiers and avoiding a
1554 ADDR_SPACE_CONVERT_EXPR from fold_convert.
1556 2017-02-22 Richard Biener <rguenther@suse.de>
1558 PR tree-optimization/79666
1559 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
1560 to not symbolically negate if that may introduce undefined
1563 2017-02-22 Martin Liska <mliska@suse.cz>
1566 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
1567 * data-streamer-out.c (streamer_write_gcov_count_stream):
1569 * value-prof.c (stream_out_histogram_value): Make assert more
1570 precise based on type of counter.
1572 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
1575 * config/i386/i386.md (standard_x87sse_constant_load splitter):
1576 Use nonimmediate_operand instead of memory_operand for operand 1.
1577 (float-extend standard_x87sse_constant_load splitter): Ditto.
1579 2017-02-21 Jeff Law <law@redhat.com>
1581 PR tree-optimization/79621
1582 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
1583 blocks with edges to themselves.
1585 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1588 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
1589 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
1590 Use gimple_call_builtin_p.
1593 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
1594 on temporarily removed DEBUG_INSNs.
1596 PR tree-optimization/79649
1597 * tree-loop-distribution.c (classify_partition): Give up on
1598 non-generic address space loads/stores.
1600 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
1602 * doc/loop.texi (Loop manipulation): Remove nonexistent
1603 tree_ssa_loop_version from the documentation.
1604 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
1606 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1609 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
1610 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
1611 * config/rs6000/rs6000.c: Include except.h.
1612 (rs6000_expand_split_stack_prologue): Call
1613 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
1615 2017-02-21 Martin Jambor <mjambor@suse.cz>
1618 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
1621 2017-02-21 Martin Jambor <mjambor@suse.cz>
1623 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
1624 for backward compatibility only.
1625 * doc/invoke.texi (Option Summary): Remove all references to
1628 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
1632 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1634 * lra-constraints.c (curr_insn_transform): Handle
1635 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
1637 2017-02-21 Martin Liska <mliska@suse.cz>
1639 * config/i386/i386.opt: Replace -masm-dialect with -masm.
1641 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
1643 PR translation/79638
1644 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
1646 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
1649 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
1650 (arm_function_ok_for_sibcall): Return false for an indirect call by
1651 descriptor if all the argument registers are used.
1652 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
1653 alignment of the function.
1655 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1657 PR tree-optimization/61441
1658 * simplify-rtx.c (simplify_const_unary_operation): For
1659 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
1660 the sNaN unmodified.
1662 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
1664 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
1665 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
1666 instead of SYSTEM_HEADER_DIR.
1668 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
1669 Martin Liška <mliska@suse.cz>
1671 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
1672 Fix typos and grammar, use active voice, and clarify.
1674 2017-02-20 Marek Polacek <polacek@redhat.com>
1677 * gimplify.c (gimplify_expr): Handle unused *&&L;.
1680 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
1682 2017-02-20 Jakub Jelinek <jakub@redhat.com>
1685 * config/i386/i386.c (ix86_expand_builtin): Handle
1686 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
1687 ix86_builtins_isa[fcode].isa as a requirement of those
1688 flags and any other flag in the bitmask.
1689 (ix86_init_mmx_sse_builtins): Use 0 instead of
1690 ~OPTION_MASK_ISA_64BIT as mask.
1691 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
1692 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
1693 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
1694 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
1696 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1699 * lra-constraints.c (split_reg): Check requested split mode
1700 is supported by the register.
1702 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1704 * lra-constraints.c (simplify_operand_subreg): Remove early
1707 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1710 * lra-constraints.c (curr_insn_transform): Tighten condition
1711 for converting SUBREG reloads from OP_OUT to OP_INOUT.
1713 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1716 * lra-constraints.c (curr_insn_transform): Handle
1717 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
1719 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
1722 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
1724 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
1726 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
1729 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
1732 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1734 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
1737 2017-02-18 Jakub Jelinek <jakub@redhat.com>
1739 * final.c (last_columnnum, override_columnnum): New variables.
1740 (final_start_function): Set last_columnnum, pass it to begin_prologue
1741 hook and pass 0 to dwarf2out_begin_prologue.
1742 (final_scan_insn): Update override_columnnum. Pass last_columnnum
1743 to source_line debug hook.
1744 (notice_source_line): Compute last_columnnum and for debug_column_info
1745 return true on column changes.
1746 * debug.h (struct gcc_debug_hooks): Add column argument to
1747 source_line and begin_prologue hooks.
1748 (debug_nothing_int_charstar_int_bool): Remove prototype.
1749 (debug_nothing_int_int_charstar,
1750 debug_nothing_int_int_charstar_int_bool): New prototypes.
1751 (dwarf2out_begin_prologue): Add column argument.
1752 * debug.c (do_nothing_debug_hooks): Adjust source_line and
1753 begin_prologue hooks.
1754 (debug_nothing_int_charstar_int_bool): Remove.
1755 (debug_nothing_int_int_charstar,
1756 debug_nothing_int_int_charstar_int_bool): New functions.
1757 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
1758 through to dwarf2out_source_line.
1759 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
1760 (dwarf2out_source_line): Add column argument, emit it if requested.
1761 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
1763 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
1764 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
1765 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
1766 through to dwarf2out_begin_prologue.
1767 (vmsdbgout_source_line): Add column argument, pass it through to
1768 dwarf2out_source_line.
1769 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
1770 dbxout_source_line caller.
1771 (dbxout_source_line): Add column argument.
1773 * common.opt (gno-column-info, gcolumn-info): New options.
1774 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
1775 (check_die): Also test for multiple DW_AT_decl_column attributes.
1776 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
1777 DW_AT_decl_column if requested.
1778 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
1780 (gen_variable_die): Likewise.
1781 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
1782 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
1785 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
1786 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
1787 (ix86_handle_option): Handle OPT_m3dnowa.
1788 * doc/invoke.texi (-m3dnowa): Document.
1789 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
1790 -m3dnowa instead of -m3dnow -march=athlon.
1793 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
1794 instead of gcc_assert for K, r and R code checks. Formatting fixes.
1796 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1799 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
1800 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
1801 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
1802 generator for vsx_xxpermdi_<mode>_be.
1803 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
1804 force big-endian semantics.
1805 (vsx_xxpermdi_<mode>_be): New define_expand with same
1806 implementation as previous version of vsx_xxpermdi_<mode>.
1808 2017-02-17 Jakub Jelinek <jakub@redhat.com>
1810 PR tree-optimization/79327
1811 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
1812 variable, its initialization and use.
1814 2017-02-17 Julia Koval <julia.koval@intel.com>
1816 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
1817 (OPTION_MASK_ISA_PKU_UNSET): New.
1818 (ix86_handle_option): Handle -mrdpid.
1819 * config/i386/cpuid.h (bit_RDPID): New.
1820 * config/i386/driver-i386.c (host_detect_local_cpu):
1821 Detect RDPID feature.
1822 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
1823 * config/i386/i386-c.c (ix86_target_macros_internal):
1825 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
1826 (ix86_valid_target_attribute_inner_p): Add "rdpid".
1827 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
1828 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
1829 * config/i386/i386.md (define_insn "rdpid"): New.
1830 * config/i386/i386.opt Add -mrdpid.
1831 * config/i386/immintrin.h (_rdpid_u32): New.
1833 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
1835 PR rtl-optimization/79541
1836 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
1837 instead of transforming it into USE.
1839 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
1841 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
1842 If HONOR_SNANS (SFmode) force the input to a register.
1843 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
1844 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
1845 an frsp or similar insn.
1847 2017-02-17 Martin Liska <mliska@suse.cz>
1849 PR rtl-optimization/79577
1850 * params.def (selsched-max-sched-times): Increase minimum to 1.
1852 2017-02-17 Martin Liska <mliska@suse.cz>
1854 PR rtl-optimization/79574
1855 * gcse.c (want_to_gcse_p): Prevent integer overflow.
1857 2017-02-17 Martin Liska <mliska@suse.cz>
1859 PR tree-optimization/79529
1860 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
1861 ssa_defined_default_def_p to handle cases which are implicitly
1863 * tree-ssa.c (ssa_defined_default_def_p): New function.
1864 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
1865 which are implicitly defined.
1866 * tree-ssa.h (ssa_defined_default_def_p): Declare.
1868 2017-02-17 Richard Biener <rguenther@suse.de>
1871 * params.def (max-ssa-name-query-depth): Limit to 10.
1873 2017-02-17 Richard Biener <rguenther@suse.de>
1875 PR tree-optimization/79552
1876 * tree-ssa-structalias.c (visit_loadstore): Properly verify
1879 2017-02-17 Richard Biener <rguenther@suse.de>
1882 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
1884 2017-02-17 Marek Polacek <polacek@redhat.com>
1887 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
1888 (fold_negate_expr): New wrapper.
1890 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
1892 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
1893 Correct terminology and de-emphasize pre-standard behavior.
1895 2017-02-16 Alan Modra <amodra@gmail.com>
1897 PR rtl-optimization/79286
1898 * ira.c (def_dominates_uses): New function.
1899 (update_equiv_regs): Don't create an equivalence for insns that
1900 may trap where the register def does not dominate the use.
1902 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
1904 PR rtl-optimization/78127
1905 * lra.c (lra): Call lra_eliminate before finish the loop after
1908 2017-02-16 Richard Biener <rguenther@suse.de>
1910 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
1912 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
1913 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
1914 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
1915 (isl_val_int_from_wi): New function.
1916 (extract_affine_gmp): Rename to ...
1917 (extract_affine_wi): ... this, take a widest_int.
1918 (extract_affine_int): Just wrap extract_affine_wi.
1919 (add_param_constraints): Use isl_val_int_from_wi.
1920 (add_loop_constraints): Likewise, and extract_affine_wi.
1922 2017-02-15 Jeff Law <law@redhat.com>
1925 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
1926 ira_init_register_move_cost_if_necessary.
1928 2017-02-15 Martin Sebor <msebor@redhat.com>
1931 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
1932 removed in a prior commit.
1934 2017-02-15 Bin Cheng <bin.cheng@arm.com>
1936 PR tree-optimization/79347
1937 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
1938 counters during peeling.
1940 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
1942 * Makefile.in (site.exp): Remove "set ISLVER".
1944 2017-02-15 Jakub Jelinek <jakub@redhat.com>
1947 * real.c (real_from_integer): Call real_convert even for decimal.
1949 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
1952 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
1954 2017-02-14 Andrew Pinski <apinski@cavium.com>
1956 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
1957 cores and change the partno/implementer to be correct.
1958 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
1959 the 'B" as the implementer.
1960 * config/aarch64/aarch64-tune.md: Regenerate.
1962 2017-02-14 Carl Love <cel@us.ibm.com>
1964 * config/rs6000/rs6000.c: Add case statement entry to make the
1965 xvcvuxdsp built-in argument unsigned.
1966 * config/rs6000/vsx.md: Fix the source and return operand types so they
1967 match the instruction definitions from the ISA document. Fix typo
1968 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
1971 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
1974 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
1975 member early_clobber_alts.
1976 * lra-lives.c (reg_early_clobber_p): New.
1977 (process_bb_lives): Use it.
1978 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
1979 (debug_operand_data): Initialize early_clobber_alts.
1980 (setup_operand_alternative): Set up early_clobber_alts.
1981 (collect_non_operand_hard_regs): Ditto. Pass early clobber
1982 alternatives to new_insn_reg.
1983 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
1985 (lra_update_insn_regno_info): Pass the new arg.
1987 2017-02-14 Jakub Jelinek <jakub@redhat.com>
1990 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
1991 (new_oacc_loop_raw): Don't clear already cleared fields.
1994 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
1995 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
1996 _mm512_prefetch_i64gather_ps): New inline functions and macros.
1998 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
2001 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
2003 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
2006 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
2007 the extra instruction to the right place to store 128-bit constant
2010 2017-02-14 Martin Sebor <msebor@redhat.com>
2013 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
2014 warning for strings of unknown length.
2016 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
2018 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
2020 2017-02-14 Jeff Law <law@redhat.com>
2023 * ira-costs.c (scan_one_insn): Initialize register move costs
2024 for pseudos seen in USE/CLOBBER insns.
2026 PR tree-optimization/79095
2027 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
2028 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
2029 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
2030 if the operands are known to be not equal, then the resulting range
2032 (intersect_ranges): If the new range is ~[0,0] and the old range is
2033 wide, then prefer ~[0,0].
2034 * tree-vrp.c (overflow_comparison_p_1): New function.
2035 (overflow_comparison_p): New function.
2036 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
2037 if NAME is used in an overflow test.
2038 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
2039 overflow check that can be expressed as an equality test, then adjust
2040 ops to be that equality test.
2042 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2044 * config/s390/s390-builtin-types.def: Remove flags argument.
2045 * config/s390/s390.c (s390_init_builtins): Likewise.
2047 2017-02-14 Martin Liska <mliska@suse.cz>
2049 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
2050 vector. Fix trailing white spaces.
2052 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
2054 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
2057 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2059 PR rtl-optimization/68664
2060 * config/arm/arm.c (arm_sched_can_speculate_insn):
2061 New function. Declare prototype.
2062 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
2064 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2066 PR rtl-optimization/68664
2067 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
2069 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
2071 2017-02-14 Amit Pawar <amit.pawar@amd.com>
2073 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
2074 max skip bytes for function, loop and jump.
2076 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2078 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
2079 ABS_EXPR for gimple dump.
2081 2017-02-14 Jakub Jelinek <jakub@redhat.com>
2084 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
2086 PR tree-optimization/79408
2087 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
2088 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
2089 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
2090 also if rhs1 is INTEGER_CST.
2092 2017-02-14 Richard Biener <rguenther@suse.de>
2095 * tree-into-ssa.c (insert_phi_nodes): When the function can
2096 have abnormal edges rewrite SSA names with broken use-def
2097 dominance out of SSA and register them for PHI insertion.
2099 2017-02-13 Martin Sebor <msebor@redhat.com>
2102 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
2103 clearing info.nowrite flag when snprintf size argument is a range.
2105 2017-02-13 Jakub Jelinek <jakub@redhat.com>
2107 * cprop.c (cprop_jump): Add missing space in string literal.
2108 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
2109 (get_constraint_for_component_ref): Likewise.
2110 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
2111 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
2112 * lra-constraints.c (process_alt_operands): Likewise.
2113 * ipa-inline.c (inline_small_functions): Likewise.
2114 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
2115 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
2116 * trans-mem.c (diagnose_tm_1_op): Likewise.
2117 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
2118 (grid_parallel_clauses_gridifiable): Likewise.
2120 * config/nvptx/mkoffload.c (process): Add space in between
2123 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
2124 "MOD4_SSE_REGS" and "ALL_REGS".
2126 * spellcheck.c (test_data): Add , in between "foo" and "food".
2128 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2131 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
2132 boundary crossing check and subsequent code generation agree.
2134 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2136 * config/aarch64/aarch64.c (has_memory_op): Delete.
2137 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
2140 2017-02-13 Jakub Jelinek <jakub@redhat.com>
2142 PR rtl-optimization/79388
2143 PR rtl-optimization/79450
2144 * combine.c (distribute_notes): When removing TEM_INSN for which
2145 corresponding dest has last value recorded, invalidate that last
2148 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2150 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
2151 of explicit '@'. Add missing assembly comment marker on branch costs
2154 2017-02-13 Nathan Sidwell <nathan@acm.org>
2156 * gengtype-lex.l (<in_struct>): Add '/'.
2158 2017-02-13 Martin Liska <mliska@suse.cz>
2161 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
2163 2017-02-13 Richard Biener <rguenther@suse.de>
2165 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
2167 * configure: Re-generate.
2168 * config.in: Likewise.
2169 * graphite-dependences.c: Simplify as if
2170 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
2171 * graphite-isl-ast-to-gimple.c: Likewise.
2172 * graphite-optimize-isl.c: Likewise.
2173 * graphite-poly.c: Likewise.
2174 * graphite-sese-to-poly.c: Likewise.
2175 * graphite.h: Likewise.
2176 * toplev.c: Include isl/version.h and use isl_version () for
2177 printing the ISL version.
2178 * doc/install.texi: Update ISL requirement.
2180 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
2182 * doc/standards.texi (Standards): Update reference to
2185 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
2187 * doc/extend.texi (Named Address Spaces): sourceware.org now
2189 * doc/install.texi (Binaries): Ditto.
2192 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
2194 * doc/cpp.texi: Replace "stringify"/"stringification" with C
2195 standard terminology "stringize"/"stringizing" throughout.
2196 * doc/cppinternals.texi: Likewise.
2198 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
2200 * doc/extend.texi: Fix some spelling mistakes and typos.
2201 * doc/invoke.texi: Likewise.
2203 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
2206 * params.def (inline-min-speedup) Change from 10 to 8.
2208 2017-02-11 Jakub Jelinek <jakub@redhat.com>
2210 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
2213 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
2216 * ipa-inline-analysis.c (get_minimal_bb): New function.
2217 (record_modified): Use it.
2218 (remap_edge_change_prob): Handle also ancestor functions.
2220 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
2222 * doc/contrib.texi (Contributors): Remove broken link into
2223 the Mauve CVS repository.
2225 2017-02-11 Jakub Jelinek <jakub@redhat.com>
2228 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
2229 result computation whenever lhs doesn't have vector mode, not
2230 just when it has BLKmode.
2232 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
2234 * doc/makefile.texi (profiledbootstrap): Refer to the
2235 installation instructions only in textual form.
2237 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2240 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
2242 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
2244 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
2245 (Specific): Update mingw-w64 reference.
2247 (Specific): Remove broken link to Renesas RX processor.
2249 2017-02-10 Richard Biener <rguenther@suse.de>
2251 * toplev.c (process_options): Do not mention obsolete graphite
2252 options when printing sorry message about missing graphite support.
2253 Mention -floop-nest-optimize.
2255 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
2257 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
2258 (vtst_p16): Likewise.
2259 (vtstq_p8): Likewise.
2260 (vtstq_p16): Likewise.
2262 (vtstq_p64): Likewise.
2263 * config/arm/arm_neon.h (vgetq_lane_p64): New.
2264 (vset_lane_p64): New.
2265 (vsetq_lane_p64): New.
2267 2017-02-10 Jakub Jelinek <jakub@redhat.com>
2269 PR tree-optimization/79411
2270 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
2271 stmt operands are SSA_NAMEs used in abnormal phis.
2272 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
2275 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
2278 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
2281 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
2283 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
2285 2017-02-09 Jakub Jelinek <jakub@redhat.com>
2287 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
2291 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
2292 not arbitrary TREE_CONSTANT.
2295 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
2296 "omp declare target link" attribute unless is_global_var.
2297 * omp-offload.c (find_link_var_op): Likewise.
2299 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
2300 Chung-Lin Tang <cltang@codesourcery.com>
2302 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
2304 (gimplify_adjust_omp_clauses): Don't delete TILE.
2305 (gimplify_omp_for): Deal with TILE.
2306 * internal-fn.c (expand_GOACC_TILE): New function.
2307 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
2309 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
2310 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
2312 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
2313 avoid DIV for outermost collapse var.
2314 (expand_oacc_for): Insert tile element loop as needed. Adjust.
2315 Remove out of date comments, fix whitespace.
2316 * omp-general.c (omp_extract_for_data): Deal with tiling.
2317 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
2318 adjust OLF_DIM_BASE value.
2319 (struct omp_for_data): Add tiling field.
2320 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
2321 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
2322 for auto loops. Remove default auto determining, moved to
2323 oacc_loop_fixed_partitions.
2324 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
2325 stmts, add e_mask field.
2326 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
2327 (oacc_thread_numbers): Use oacc_dim_call.
2328 (oacc_xform_tile): New.
2329 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
2330 (finish_oacc_loop): Adjust for ifns vector.
2331 (oacc_loop_discover_walk): Append loop abstraction sites to list,
2332 add case for GOACC_TILE fns.
2333 (oacc_loop_xform_loop): Delete.
2334 (oacc_loop_process): Iterate over call list directly, and add
2335 handling for GOACC_TILE fns.
2336 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
2338 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
2339 vector partitioning to outer loops. Assign 2 partitions to loops
2340 when available. Add TILE handling.
2341 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
2342 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
2343 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
2344 * tree.c (omp_clause_num_ops): Adjust TILE ops.
2345 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
2347 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
2349 * configure.ac (ACX_BUGURL): Update.
2350 * configure: Regenerate.
2352 2017-02-09 Richard Biener <rguenther@suse.de>
2354 PR tree-optimization/69823
2355 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
2356 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
2358 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
2360 * config/arc/arc-c.def: Add __NPS400__ definition.
2361 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
2362 (TARGET_NPS400): Define.
2364 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
2366 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
2368 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
2370 (arc_cpu_types): Fill the arch_info field with a pointer into the
2371 arc_arch_types table.
2372 (arc_selected_cpu): Declare.
2373 * config/arc/arc.c (arc_selected_cpu): Make global.
2374 (arc_selected_arch): Delete.
2375 (arc_base_cpu): Delete.
2376 (arc_override_options): Remove references to deleted variables,
2377 update access to arch information.
2378 (ARC_OPT): Update access to arch information.
2379 (ARC_OPTX): Likewise.
2380 * config/arc/arc.h (arc_base_cpu): Remove declaration.
2381 (TARGET_ARC600): Update access to arch information.
2382 (TARGET_ARC601): Likewise.
2383 (TARGET_ARC700): Likewise.
2384 (TARGET_EM): Likewise.
2385 (TARGET_HS): Likewise.
2386 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
2389 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
2392 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
2393 condition/operands for integer GE/LE/GEU/LEU operations.
2395 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
2397 PR translation/79397
2398 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
2401 2017-02-08 Martin Jambor <mjambor@suse.cz>
2404 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
2405 whether allocation happened.
2406 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
2407 nothing was allocated.
2409 2017-02-08 Jakub Jelinek <jakub@redhat.com>
2411 PR tree-optimization/79408
2412 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
2413 constant, but SSA_NAME with a known integer range, use the minimum
2414 of that range instead of op1 to determine if modulo can be replaced
2415 with its first operand.
2417 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2419 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
2421 2017-02-08 Richard Biener <rguenther@suse.de>
2423 PR tree-optimization/71824
2424 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
2425 Check all loops contained in the merged region.
2427 2017-02-07 Andrew Pinski <apinski@cavium.com>
2429 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
2431 2017-02-07 Andrew Pinski <apinski@cavium.com>
2433 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
2434 (thunderxt88): Likewise.
2435 (thunderxt81): Disable LSE and change v8.1 to v8.
2436 (thunderxt83): Likewise.
2438 2017-02-07 Jakub Jelinek <jakub@redhat.com>
2439 Richard Biener <rguenther@suse.de>
2442 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
2443 type from int to size_t.
2444 * ira-costs.c (struct_costs_size): Change type from int to size_t.
2446 2017-02-07 Jakub Jelinek <jakub@redhat.com>
2448 PR rtl-optimization/79386
2449 * cprop.c (bypass_conditional_jumps): Initialize
2450 bypass_last_basic_block already before splitting bbs after
2451 unconditional traps...
2452 (bypass_conditional_jumps): ... rather than here.
2455 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
2456 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
2457 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
2458 fix -masm=intel patterns.
2460 2017-02-07 Richard Biener <rguenther@suse.de>
2462 PR tree-optimization/79256
2464 * builtins.c (get_object_alignment_2): Use min_align_of_type
2465 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
2466 and ADJUST_FIELD_ALIGN.
2468 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
2470 * doc/tm.texi: Regenerate.
2471 * stor-layout.c (layout_decl): Adjust.
2472 (update_alignment_for_field): Likewise.
2473 (place_field): Likewise.
2474 (min_align_of_type): Likewise.
2475 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
2476 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
2477 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
2478 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
2479 * config/frv/frv.c (frv_adjust_field_align): Likewise.
2480 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
2481 * config/i386/i386.c (x86_field_alignment): Likewise.
2482 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
2483 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
2484 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
2485 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
2486 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
2487 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
2491 2017-01-30 Richard Biener <rguenther@suse.de>
2493 PR tree-optimization/79256
2494 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
2495 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
2498 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
2500 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
2501 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
2502 builtins to SImode and emit a zero-extend, if necessary.
2504 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2506 * docs/invoke.texi (RISC-V Options): Alphabetize.
2508 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2510 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
2513 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2515 * config/riscv/riscv.c: New file.
2516 * gcc/common/config/riscv/riscv-common.c: Likewise.
2517 * config.gcc: Likewise.
2518 * config/riscv/constraints.md: Likewise.
2519 * config/riscv/elf.h: Likewise.
2520 * config/riscv/generic.md: Likewise.
2521 * config/riscv/linux.h: Likewise.
2522 * config/riscv/multilib-generator: Likewise.
2523 * config/riscv/peephole.md: Likewise.
2524 * config/riscv/pic.md: Likewise.
2525 * config/riscv/predicates.md: Likewise.
2526 * config/riscv/riscv-builtins.c: Likewise.
2527 * config/riscv/riscv-c.c: Likewise.
2528 * config/riscv/riscv-ftypes.def: Likewise.
2529 * config/riscv/riscv-modes.def: Likewise.
2530 * config/riscv/riscv-opts.h: Likewise.
2531 * config/riscv/riscv-protos.h: Likewise.
2532 * config/riscv/riscv.h: Likewise.
2533 * config/riscv/riscv.md: Likewise.
2534 * config/riscv/riscv.opt: Likewise.
2535 * config/riscv/sync.md: Likewise.
2536 * config/riscv/t-elf-multilib: Likewise.
2537 * config/riscv/t-linux: Likewise.
2538 * config/riscv/t-linux-multilib: Likewise.
2539 * config/riscv/t-riscv: Likewise.
2540 * configure.ac: Likewise.
2541 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
2542 Waterman as RISC-V maintainers.
2543 * doc/install.texi: Add RISC-V entries.
2544 * doc/invoke.texi: Add RISC-V options section.
2545 * doc/md.texi: Add RISC-V constraints section.
2546 * configure: Regenerated.
2548 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2551 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
2552 false values to be constant vectors with all 0 or all 1 bits set.
2553 (vcondu<mode><mode>): Likewise.
2554 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
2556 (fpmask_comparison_operator): Update comment.
2557 (vecint_comparison_operator): New predicate.
2558 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
2559 vector conditionals when the true and false values are constant
2560 vectors with all 0 bits or all 1 bits set.
2562 2017-02-06 Martin Sebor <msebor@redhat.com>
2564 PR tree-optimization/79376
2565 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
2567 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
2569 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
2570 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
2571 to simplify split condition.
2573 2017-02-06 Jakub Jelinek <jakub@redhat.com>
2575 * omp-expand.c (oxpand_omp_atomic_fetch_op,
2576 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
2579 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
2581 PR rtl-optimization/68664
2582 * target.def (can_speculate_insn): New hook.
2583 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
2584 * doc/tm.texi: Regenerate.
2585 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
2586 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
2587 (rs6000_sched_can_speculate_insn): New function.
2589 2017-02-06 Jakub Jelinek <jakub@redhat.com>
2591 PR tree-optimization/79284
2592 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
2593 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
2594 vectorizable_mask_load_store, vectorizable_operation,
2595 vect_is_simple_cond, get_same_sized_vectype): Use it instead
2596 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
2597 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
2598 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
2599 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
2600 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
2601 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
2602 is_gimple_assign (stmt). Replace another such test with
2603 is_gimple_assign (stmt).
2605 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
2608 * config/avr/avr.c (rtl-iter.h): Include it.
2609 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
2610 (avr_legitimate_combined_insn): ...and implementation.
2612 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
2614 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
2615 * config/s390/s390.c (s390_const_operand_ok)
2616 (s390_canonicalize_comparison, s390_extract_part)
2617 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
2618 (s390_contiguous_bitmask_p, s390_rtx_costs)
2619 (legitimize_pic_address): Likewise.
2620 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
2621 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
2622 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
2623 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
2624 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
2626 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
2628 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
2629 REGNO($0) == REGNO($1).
2631 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2633 * config/s390/linux.h(SIZE_TYPE): Add comment.
2635 2017-02-06 Julian Brown <julian@codesourcery.com>
2636 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
2637 Virendra Pathak <virendra.pathak@broadcom.com>
2639 * config/aarch64/aarch64-cores.def: Change the scheduler
2641 * config/aarch64/aarch64.md: Include thunderx2t99.md.
2642 * config/aarch64/thunderx2t99.md: New file.
2644 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2646 * doc/standards.texi (Go Language): Update link to language
2649 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
2651 * tree-eh.c (lower_resx): Sanitize profile.
2652 (cleanup_empty_eh_move_lp): Likewise.
2654 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
2657 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
2659 * cfgloopmanip.h (loop_version): Update prototype.
2660 * modulo-sched.c (sms_schedule): Update call of loop_version.
2661 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
2662 * tree-parloops.c (gen_parallel_loop): Likewise.
2663 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
2664 * tree-ssa-loop-split.c (split_loop): Likewise.
2665 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
2666 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
2668 2017-02-05 Martin Liska <mliska@suse.cz>
2671 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
2673 (print_operand_address): Initialize a struct to zero.
2675 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2677 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
2678 garbage collector only in textual form.
2680 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2682 * doc/extend.texi (x86 specific memory model extensions for
2683 transactional memory): Simplify a phrase.
2685 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
2688 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
2689 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
2690 (atomic_storedi_1): Likewise.
2692 2017-02-04 Jakub Jelinek <jakub@redhat.com>
2694 PR tree-optimization/79338
2695 * tree-parloops.c (gather_scalar_reductions): Don't call
2696 vect_analyze_loop_form for loop->inner before destroying loop's
2699 2017-02-03 Martin Sebor <msebor@redhat.com>
2701 PR tree-optimization/79327
2702 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
2703 when precision has resulted in leading zeros.
2704 (format_integer): Adjust the likely counter to assume an unknown
2705 argument that may be zero is non-zero.
2707 2017-02-03 Jason Merrill <jason@redhat.com>
2710 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
2711 avoid copying non-taken branch.
2713 2017-02-03 Jakub Jelinek <jakub@redhat.com>
2715 PR tree-optimization/79340
2716 * tree-vect-loop.c (vectorizable_reduction): Release
2717 vec_defs elements after safe_splicing them into other vectors.
2720 PR tree-optimization/79327
2721 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
2722 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
2724 (format_integer): Use wide_int_to_tree instead of build_int_cst
2725 + to_?hwi. If argmin is NULL, just set argmin and argmax to
2726 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
2727 of shortest and longest sequence.
2729 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
2731 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
2732 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
2734 2017-02-03 Walter Lee <walt@tilera.com>
2737 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
2738 after initial stackframe link reg save.
2739 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
2741 2017-02-03 Jakub Jelinek <jakub@redhat.com>
2744 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
2745 wu for stxssp alternative.
2747 2017-02-03 Martin Sebor <msebor@redhat.com>
2749 PR tree-optimization/79352
2750 * gimple-fold.c (get_range_strlen): Add argument.
2751 (get_range_strlen): Change return type to bool.
2752 (get_maxval_strlen): Pass in a dummy argument.
2753 * gimple-fold.h (get_range_strlen): Change return type to bool.
2754 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
2755 * tree.h (array_at_struct_end_p): Add argument.
2756 * tree.c (array_at_struct_end_p): Handle it.
2758 2017-02-03 Martin Liska <mliska@suse.cz>
2761 * multiple_target.c (create_dispatcher_calls): Redirect edge
2762 from a caller of a dispatcher.
2763 (expand_target_clones): Make the clones local.
2764 (ipa_target_clone): Do both target clones and resolvers.
2765 (ipa_dispatcher_calls): Remove the pass.
2766 (pass_dispatcher_calls::gate): Likewise.
2767 (make_pass_dispatcher_calls): Likewise.
2768 * passes.def (pass_target_clone): Put as very first IPA early
2771 2017-02-03 Martin Liska <mliska@suse.cz>
2773 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
2774 in case of a function with ifunc attribute.
2776 2017-02-03 Martin Liska <mliska@suse.cz>
2778 * cgraph.c (cgraph_node::dump): Dump function version info.
2779 * symtab.c (symtab_node::dump_base): Add missing new line.
2781 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
2783 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
2784 (ifcombine_ifandif): Use it.
2786 2017-02-03 Martin Liska <mliska@suse.cz>
2788 * doc/invoke.texi: Document default value for
2789 use-after-scope-direct-emission-threshold.
2791 2017-02-03 Martin Liska <mliska@suse.cz>
2793 PR tree-optimization/79339
2794 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
2795 (format_floating): Likewise.
2797 2017-02-03 Martin Liska <mliska@suse.cz>
2800 * ipa-prop.c (ipa_node_params_t::insert): Remove current
2802 (ipa_node_params_t::remove): Likewise.
2803 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
2804 initialization from removed ipa_node_params_t::insert.
2805 (ipa_node_params::~ipa_node_params): Move from removed
2806 ipa_node_params_t::release.
2807 * symbol-summary.h (symbol_summary::m_released): New member.
2808 Do not release a summary twice. Do not allow to call finalizer
2809 for types of a summary that live in GGC memory.
2811 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
2813 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
2816 2017-02-02 Martin Sebor <msebor@redhat.com>
2819 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
2820 (format_string): Tighten up the range of output for non-constant
2821 strings and correct the expected range for wide non-constant strings.
2823 2017-02-02 Martin Sebor <msebor@redhat.com>
2825 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
2828 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
2830 (-fdump-tree-@var): Add to index and document how to come up
2831 with pass-specific option and dump file names.
2832 (-fdump-passes): Clarify where to look for output.
2834 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
2837 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
2838 statistics of the analyzed path; allow threading for speed when
2839 any of BBs along the path are optimized for speed.
2841 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
2844 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
2845 settings of the virtual registers.
2848 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
2850 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2851 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
2854 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2856 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
2857 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
2859 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2861 * config/s390/s390.md: Add missing comments with the expanded
2863 * config/s390/vector.md: Likewise.
2864 * config/s390/vx-builtins.md: Likewise.
2866 2017-02-02 Jakub Jelinek <jakub@redhat.com>
2869 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
2870 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
2871 conditions on a single line.
2873 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2875 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
2876 __S390_VX__ to __VX__.
2878 2017-02-01 Andrew Pinski <apinski@cavium.com>
2880 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
2881 stmt_info to record_stmt_cost.
2882 (vect_get_known_peeling_cost): Pass stmt_info if known to
2884 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
2885 cpu_vector_cost field into
2886 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
2887 field into vec_int_stmt_cost and vec_fp_stmt_cost.
2888 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
2889 splitting of scalar_stmt_cost and vec_stmt_cost.
2890 (thunderx_vector_cost): Likewise.
2891 (cortexa57_vector_cost): LIkewise.
2892 (exynosm1_vector_cost): Likewise.
2893 (xgene1_vector_cost): Likewise.
2894 (thunderx2t99_vector_cost): Improve after the splitting of the two
2896 (aarch64_builtin_vectorization_cost): Update for the splitting of
2897 scalar_stmt_cost and vec_stmt_cost.
2899 2017-02-01 Torvald Riegel <triegel@redhat.com>
2900 Richard Henderson <rth@redhat.com>
2902 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
2903 conditional on existance of a fast atomic load.
2904 * optabs-query.c (can_atomic_load_p): New function.
2905 * optabs-query.h (can_atomic_load_p): Declare it.
2906 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
2907 no fast atomic load is available for the particular size of access.
2908 (expand_atomic_compare_and_swap): Likewise.
2909 (expand_atomic_load): Likewise.
2910 (expand_atomic_store): Likewise.
2911 (expand_atomic_fetch_op): Likewise.
2912 * testsuite/lib/target-supports.exp
2913 (check_effective_target_sync_int_128): Remove x86 because it provides
2914 no fast atomic load.
2915 (check_effective_target_sync_int_128_runtime): Likewise.
2917 2017-02-01 Richard Biener <rguenther@suse.de>
2919 * graphite.c: Include tree-vectorizer.h for find_loop_location.
2920 (graphite_transform_loops): Provide opt-info for optimized nests.
2921 * tree-parloop.c (parallelize_loops): Provide opt-info for
2924 2017-02-01 Richard Biener <rguenther@suse.de>
2927 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
2930 2017-02-01 Richard Biener <rguenther@suse.de>
2932 PR tree-optimization/71824
2933 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
2934 Verify the loops are valid in the merged SESE region.
2935 (scop_detection::can_represent_loop_1): Check analyzing the
2936 evolution of the number of iterations in the region succeeds.
2938 2017-01-31 Ian Lance Taylor <iant@golang.org>
2940 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
2941 REG_ARGS_SIZE note to 32-bit push insns and call insn.
2943 2017-01-31 David Malcolm <dmalcolm@redhat.com>
2945 PR preprocessor/79210
2946 * input.c (get_substring_ranges_for_loc): Replace line_width
2947 assertion with error-handling.
2949 2017-01-31 Richard Biener <rguenther@suse.de>
2951 PR tree-optimization/77318
2952 * graphite-sese-to-poly.c (extract_affine): Fix assert.
2953 (create_pw_aff_from_tree): Take loop parameter.
2954 (add_condition_to_pbb): Pass loop of the condition to
2955 create_pw_aff_from_tree.
2957 2017-01-31 Jakub Jelinek <jakub@redhat.com>
2959 * config/s390/s390.c (s390_asan_shadow_offset): New function.
2960 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
2962 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2966 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
2968 (convert_int_to_float128): Likewise.
2969 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
2970 (convert_int_to_float128): Likewise.
2971 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
2972 (UNSPEC_IEEE128_CONVERT): Likewise.
2973 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
2974 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
2975 Use local variables for IBM extended format.
2976 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
2977 (fix_trunc<mode>si2_fprs): Likewise.
2978 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
2979 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
2980 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
2981 to know that we can now have integers of all sizes in vector
2983 (fix<uns>_<mode>di2_hw): Likewise.
2984 (float<uns>_<mode>si2_hw): Likewise.
2985 (fix_<mode>si2_hw): Likewise.
2986 (fixuns_<mode>si2_hw): Likewise.
2987 (float<uns>_<mode>di2_hw): Likewise.
2988 (float_<mode>di2_hw): Likewise.
2989 (float_<mode>si2_hw): Likewise.
2990 (floatuns_<mode>di2_hw): Likewise.
2991 (floatuns_<mode>si2_hw): Likewise.
2992 (xscvqp<su>wz_<mode>): Delete, no longer used.
2993 (xscvqp<su>dz_<mode>): Likewise.
2994 (xscv<su>dqp_<mode>): Likewise.
2995 (ieee128_mfvsrd_64bit): Likewise.
2996 (ieee128_mfvsrd_32bit): Likewise.
2997 (ieee128_mfvsrwz): Likewise.
2998 (ieee128_mtvsrw): Likewise.
2999 (ieee128_mtvsrd_64bit): Likewise.
3000 (ieee128_mtvsrd_32bit): Likewise.
3002 2017-01-31 Martin Liska <mliska@suse.cz>
3005 * ipa-prop.c (ipa_free_all_node_params): Call release method
3006 instead of ~sumbol_summary to not to trigger double times
3009 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
3011 PR tree-optimization/71691
3012 * bitmap.h (class auto_bitmap): New.
3013 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
3014 is_maybe_undefined instead of ssa_undefined_value_p.
3016 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3018 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
3019 __S390_ARCH_LEVEL__ to __ARCH__.
3021 2017-01-31 Jakub Jelinek <jakub@redhat.com>
3023 PR tree-optimization/79267
3024 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
3025 if should_remove_lhs_p is true.
3027 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
3030 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
3031 (add_alignment_attribute): New.
3032 (base_type_die): Add alignment attribute.
3033 (subrange_type_die): Likewise.
3034 (modified_type_die): Likewise.
3035 (gen_array_type_die): Likewise.
3036 (gen_descr_array_type_die: Likewise.
3037 (gen_enumeration_type_die): Likewise.
3038 (gen_subprogram_die): Likewise.
3039 (gen_variable_die): Likewise.
3040 (gen_field_die): Likewise.
3041 (gen_ptr_to_mbr_type_die): Likewise.
3042 (gen_struct_or_union_type_die): Likewise.
3043 (gen_subroutine_type_die): Likewise.
3044 (gen_typedef_die): Likewise.
3045 (base_type_cmp): Compare alignment attribute.
3047 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3050 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
3051 (setb_unsigned) New pattern for setb with CCUNS.
3052 * config/rs6000/rs6000.c (expand_block_compare): Use a different
3053 subfc./subfe sequence to avoid overflow problems. Generate a
3054 shorter sequence with cmpld/setb for power9.
3055 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
3056 for generating subfc. instruction.
3057 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
3058 now uses this instruction.
3060 2017-01-30 Ian Lance Taylor <iant@google.com>
3063 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
3064 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
3066 2017-01-30 Martin Sebor <msebor@redhat.com>
3068 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
3069 Move constant to the right of a relational operator.
3070 (get_mpfr_format_length, format_character, format_string): Ditto.
3071 (should_warn_p, maybe_warn): Same.
3073 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
3075 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3078 * asan.c (get_translation_unit_decl): Remove function.
3079 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
3081 2017-01-30 Martin Liska <mliska@suse.cz>
3083 PR gcov-profile/79259
3084 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
3087 2017-01-30 Martin Liska <mliska@suse.cz>
3090 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
3091 Initialize variables with NULL value.
3093 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
3096 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
3098 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
3100 2017-01-30 Richard Biener <rguenther@suse.de>
3102 PR tree-optimization/79276
3103 * tree-vrp.c (process_assert_insertions): Properly adjust common
3104 when removing a duplicate.
3106 * gcc.dg/torture/pr79276.c: New testcase.
3108 2017-01-30 Richard Biener <rguenther@suse.de>
3110 PR tree-optimization/79256
3111 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
3112 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
3114 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
3116 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
3119 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
3120 ("*r<noxa>sbg_<mode>_sll_bitmask")
3121 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
3122 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
3123 Use contiguous_bitmask_nowrap_operand.
3125 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3128 * config/rs6000/altivec.h (vec_xl): Revise #define.
3129 (vec_xst): Likewise.
3131 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
3133 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
3135 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
3137 PR rtl-optimization/79194
3138 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
3139 traps before call to bypass_conditional_jumps.
3141 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
3143 PR tree-optimization/71374
3144 * lra-constraints.c (check_conflict_input_operands): New.
3145 (match_reload): Use it.
3147 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
3150 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
3151 account to calculate conflict_set.
3153 2017-01-27 Bin Cheng <bin.cheng@arm.com>
3155 PR rtl-optimization/78559
3156 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
3157 other_insn in combine.
3159 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
3161 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
3162 uint16_type_node for BT_UINT16.
3164 2017-01-27 David Malcolm <dmalcolm@redhat.com>
3166 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
3167 "RTL Tests" to menu.
3168 (GIMPLE Tests): New node.
3169 (RTL Tests): New node.
3171 2017-01-27 Richard Biener <rguenther@suse.de>
3173 PR tree-optimization/79245
3174 * tree-loop-distribution.c (distribute_loop): Apply cost
3175 modeling also to detected patterns.
3177 2017-01-27 Richard Biener <rguenther@suse.de>
3179 PR tree-optimization/71433
3180 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
3181 (compare_assert_loc): New function.
3182 (process_assert_insertions): Sort and optimize assert locations
3183 to remove duplicates and push down identical assertions on
3184 edges to their destination block.
3186 2017-01-27 Richard Biener <rguenther@suse.de>
3188 PR tree-optimization/79244
3189 * tree-vrp.c (remove_range_assertions): Forcefully propagate
3190 out SSA names even if abnormal.
3192 2017-01-27 Jakub Jelinek <jakub@redhat.com>
3194 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
3195 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
3196 instead of MPFR_RNDN.
3198 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
3201 * arm.c (arm_option_override): Don't call build_target_option_node
3202 until after doing all option overrides.
3203 (arm_valid_target_attribute_tree): Likewise.
3205 2017-01-27 Martin Liska <mliska@suse.cz>
3207 * doc/invoke.texi (-fprofile-arcs): Document profiling support
3208 for {cd}tors and C++ {cd}tors.
3210 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
3212 * config/s390/s390.md ("*setmem_long_and")
3213 ("*setmem_long_and_31z"): Use zero_extend instead of and.
3215 2017-01-26 Martin Sebor <msebor@redhat.com>
3217 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
3220 2017-01-26 Martin Sebor <msebor@redhat.com>
3222 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
3223 HAVE_DFmode before using XFmode or DFmode.
3224 (parse_directive): Avoid using the z length modifier to avoid
3225 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
3228 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
3229 to accept adjustment as an array.
3230 (get_int_range): New function.
3231 (struct directive): Make width and prec arrays.
3232 (directive::set_width, directive::set_precision): Call get_int_range.
3233 (format_integer, format_floating): Handle width and precision ranges.
3234 (format_string, parse_directive): Same.
3236 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3239 * dwarf2out.c (generate_skeleton_bottom_up): For children with
3240 comdat_type_p set, just clone them, but keep the children in the
3244 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
3245 which have direct callers with -fvar-tracking-assignments enabled
3247 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
3248 inside of type units.
3250 2017-01-26 Martin Sebor <msebor@redhat.com>
3253 * gimple-ssa-sprintf.c (struct result_range): Add likely and
3255 (struct format_result): Replace number_chars, number_chars_min,
3256 and number_chars_max with a single member of struct result_range.
3258 (format_result::operator+=): Adjust.
3259 (struct fmtresult): Remove bounded. Handle likely and unlikely
3261 (fmtresult::adjust_for_width_or_precision): New function.
3262 (fmtresult:type_max_digits): New function.
3263 (bytes_remaining): Handle likely and unlikely counters.
3264 (min_bytes_remaining): Remove.
3265 (format_percent): Simplify.
3266 (format_integer, format_floating): Set likely and unlikely counters.
3267 (get_string_length, format_character, format_string): Same.
3268 (format_plain, should_warn_p): New function.
3269 (maybe_warn): Call should_warn_p. Update diagnostic messages
3270 and handle those for all directives, including plain strings.
3271 (format_directive): Handle likely and unlikely counters.
3272 Remove unnecessary quoting from diagnostics. Add an informational
3274 (add_bytes): Remove.
3275 (pass_sprintf_length::compute_format_length): Simplify.
3276 (try_substitute_return_value): Handle likely and unlikely counters.
3278 2017-01-26 Carl Love <cel@us.ibm.com>
3280 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
3281 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
3283 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
3286 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
3287 endianess for subregs into account.
3288 * lra-constraints.c (lra_constraints): Do risky transformations
3289 always on the first iteration.
3290 * lra-lives.c (check_pseudos_live_through_calls): Add arg
3291 last_call_used_reg_set.
3292 (process_bb_lives): Define and use last_call_used_reg_set.
3293 * lra.c (lra): Always continue after lra_constraints on the first
3296 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
3298 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
3300 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
3302 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3304 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
3305 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
3306 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
3307 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
3308 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
3309 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
3310 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
3311 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
3312 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
3314 2017-01-26 Marek Polacek <polacek@redhat.com>
3317 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
3318 for the third operand.
3320 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3323 * omp-low.c (struct omp_context): Add simt_stmt field.
3324 (scan_omp_for): Return omp_context *.
3325 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
3326 context to the _simt_ SIMD stmt.
3327 (lower_omp_for): For combined SIMD with sibling _simt_
3328 SIMD, make sure to use the same decls in _looptemp_
3329 clauses as in the sibling.
3331 2017-01-26 David Sherwood <david.sherwood@arm.com>
3334 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
3337 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3340 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
3341 emit fld b; fld a; if possible.
3343 * brig-builtins.def: Update copyright years.
3344 * config/arm/arm_acle_builtins.def: Update copyright years.
3346 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
3349 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
3350 constraint instead of o for the stxsd instruction.
3352 2017-01-25 Carl Love <cel@us.ibm.com>
3354 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
3355 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
3357 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
3359 * doc/invoke.texi (C++ Dialect Options): Fix typo.
3361 2017-01-25 Richard Biener <rguenther@suse.de>
3363 PR tree-optimization/69264
3364 * target.def (vector_alignment_reachable): Improve documentation.
3365 * doc/tm.texi: Regenerate.
3366 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
3368 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
3369 earlier changes with respect to TYPE_USER_ALIGN.
3370 (vector_alignment_reachable_p): Likewise. Improve dumping.
3372 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3375 * config/arm/arm.md (xordi3): Force constant operand into a register
3378 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3380 * doc/invoke.texi (-fstore-merging): Correct default optimization
3381 levels at which it is enabled.
3382 (-O): Move -fstore-merging from list to...
3385 2017-01-25 Richard Biener <rguenther@suse.de>
3388 * omp-expand.c: Include debug.h.
3389 (expand_omp_taskreg): Make sure to generate early debug before
3390 outlining anything from a function.
3391 (expand_omp_target): Likewise.
3392 (grid_expand_target_grid_body): Likewise.
3394 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
3397 * asan.c (get_translation_unit_decl): New function.
3398 (asan_add_global): Extract modules file name from globals
3399 TRANSLATION_UNIT_DECL name.
3401 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
3404 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
3405 for long calls with APCS frame and VFP.
3407 2017-01-24 David Malcolm <dmalcolm@redhat.com>
3409 * cfg.c (original_copy_tables_initialized_p): New function.
3410 * cfg.h (original_copy_tables_initialized_p): New decl.
3411 * cfgrtl.c (relink_block_chain): Guard the call to
3412 free_original_copy_tables with a call to
3413 original_copy_tables_initialized_p.
3414 * cgraph.h (symtab_node::native_rtl_p): New decl.
3415 * cgraphunit.c (symtab_node::native_rtl_p): New function.
3416 (symtab_node::needed_p): Don't assert for early assembly output
3417 for __RTL functions.
3418 (cgraph_node::finalize_function): Set "force_output" for __RTL
3420 (cgraph_node::analyze): Bail out early for __RTL functions.
3421 (analyze_functions): Update assertion to support __RTL functions.
3422 (cgraph_node::expand): Bail out early for __RTL functions.
3423 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
3425 * function.h (struct function): Update comment for field
3427 * gimple-expr.c: Include "tree-pass.h".
3428 (gimple_has_body_p): Return false for __RTL functions.
3429 * Makefile.in (OBJS): Add run-rtl-passes.o.
3430 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
3432 (gcc::pass_manager::get_clean_slate): New accessor.
3433 * passes.c: Include "insn-addr.h".
3434 (should_skip_pass_p): Add logging. Update logic for running
3435 "expand" to be compatible with both __GIMPLE and __RTL. Guard
3436 property-provider override so it is only done for gimple passes.
3438 (skip_pass): New function.
3439 (execute_one_pass): Call skip_pass when skipping passes.
3440 * read-md.c (md_reader::read_char): Support filtering
3441 the input to a subset of line numbers.
3442 (md_reader::md_reader): Initialize fields
3443 m_first_line and m_last_line.
3444 (md_reader::read_file_fragment): New function.
3445 * read-md.h (md_reader::read_file_fragment): New decl.
3446 (md_reader::m_first_line): New field.
3447 (md_reader::m_last_line): New field.
3448 * read-rtl-function.c (function_reader::create_function): Only
3449 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
3450 curr_properties. Set DECL_INITIAL to a dummy block.
3451 (read_rtl_function_body_from_file_range): New function.
3452 * read-rtl-function.h (read_rtl_function_body_from_file_range):
3454 * run-rtl-passes.c: New file.
3455 * run-rtl-passes.h: New file.
3457 2017-01-24 Jeff Law <law@redhat.com>
3459 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
3462 2017-01-24 Bin Cheng <bin.cheng@arm.com>
3464 PR tree-optimization/79159
3465 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
3466 (record_nonwrapping_iv): Improve boundary using above function if no
3467 value range information.
3469 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
3470 Martin Jambor <mjambor@suse.cz>
3472 * brig-builtins.def: New file.
3473 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
3474 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
3475 (DEF_HSAIL_SAT_BUILTIN): Likewise.
3476 (DEF_HSAIL_INTR_BUILTIN): Likewise.
3477 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
3478 * builtin-types.def (BT_INT8): New.
3479 (BT_INT16): Likewise.
3480 (BT_UINT8): Likewise.
3481 (BT_UINT16): Likewise.
3482 (BT_FN_ULONG): Likewise.
3483 (BT_FN_UINT_INT): Likewise.
3484 (BT_FN_UINT_ULONG): Likewise.
3485 (BT_FN_UINT_LONG): Likewise.
3486 (BT_FN_UINT_PTR): Likewise.
3487 (BT_FN_ULONG_PTR): Likewise.
3488 (BT_FN_INT8_FLOAT): Likewise.
3489 (BT_FN_INT16_FLOAT): Likewise.
3490 (BT_FN_UINT32_FLOAT): Likewise.
3491 (BT_FN_UINT16_FLOAT): Likewise.
3492 (BT_FN_UINT8_FLOAT): Likewise.
3493 (BT_FN_UINT64_FLOAT): Likewise.
3494 (BT_FN_UINT16_UINT32): Likewise.
3495 (BT_FN_UINT32_UINT16): Likewise.
3496 (BT_FN_UINT16_UINT16_UINT16): Likewise.
3497 (BT_FN_INT_PTR_INT): Likewise.
3498 (BT_FN_UINT_PTR_UINT): Likewise.
3499 (BT_FN_LONG_PTR_LONG): Likewise.
3500 (BT_FN_ULONG_PTR_ULONG): Likewise.
3501 (BT_FN_VOID_UINT64_UINT64): Likewise.
3502 (BT_FN_UINT8_UINT8_UINT8): Likewise.
3503 (BT_FN_INT8_INT8_INT8): Likewise.
3504 (BT_FN_INT16_INT16_INT16): Likewise.
3505 (BT_FN_INT_INT_INT): Likewise.
3506 (BT_FN_UINT_FLOAT_UINT): Likewise.
3507 (BT_FN_FLOAT_UINT_UINT): Likewise.
3508 (BT_FN_ULONG_UINT_UINT): Likewise.
3509 (BT_FN_ULONG_UINT_PTR): Likewise.
3510 (BT_FN_ULONG_ULONG_ULONG): Likewise.
3511 (BT_FN_UINT_UINT_UINT): Likewise.
3512 (BT_FN_VOID_UINT_PTR): Likewise.
3513 (BT_FN_UINT_UINT_PTR: Likewise.
3514 (BT_FN_UINT32_UINT64_PTR): Likewise.
3515 (BT_FN_INT_INT_UINT_UINT): Likewise.
3516 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
3517 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
3518 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
3519 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
3520 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
3521 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
3522 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
3523 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
3524 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
3525 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
3526 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
3527 * doc/frontends.texi: List BRIG FE.
3528 * doc/install.texi (Testing): Add BRIG tesring requirements.
3529 * doc/invoke.texi (Overall Options): Mention BRIG.
3530 * doc/standards.texi (Standards): Doucment BRIG HSA version.
3532 2017-01-24 Richard Biener <rguenther@suse.de>
3534 PR translation/79208
3535 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
3537 2017-01-24 Martin Jambor <mjambor@suse.cz>
3540 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
3541 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
3544 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
3547 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
3548 casts from signed to unsigned really don't have a range.
3550 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
3552 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
3553 GMP_RNDx for compatiblity.
3555 2017-01-24 Martin Liska <mliska@suse.cz>
3558 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
3559 that would prevent us to call alloca with -1 as argument.
3561 2017-01-24 Jakub Jelinek <jakub@redhat.com>
3563 * dwarf2out.c (output_compilation_unit_header, output_file_names):
3564 Avoid -Wformat-security warning.
3566 2017-01-23 Andrew Pinski <apinski@cavium.com>
3568 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
3571 2017-01-23 Martin Sebor <msebor@redhat.com>
3574 * gimple-ssa-sprintf.c (warn_level): New global.
3575 (format_integer): Use it here and throughout the rest of the file.
3576 Use the same switch to compute sign as base.
3577 (maybe_warn): New function.
3578 (format_directive): Factor out warnings into maybe_warn.
3579 Add debugging output. Use warn_level.
3580 (add_bytes): Use warn_level.
3581 (pass_sprintf_length::compute_format_length): Add debugging output.
3582 (try_substitute_return_value): Same.
3583 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
3586 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
3587 (struct fmtresult, format_integer, format_floating): Adjust.
3588 (fmtresult::fmtresult): Set max correctly in two argument ctor.
3589 (get_string_length, format_string,format_directive): Same.
3590 (pass_sprintf_length::compute_format_length): Same.
3591 (try_substitute_return_value): Simplify slightly.
3594 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
3595 (fmtresult::operator+=): Outlined.
3596 (struct fmtresult): Add ctors.
3597 (struct conversion_spec): Rename...
3598 (struct directive): ...to this. Add and remove data members.
3599 (directive::set_width, directive::set_precision): New functions.
3600 (format_percent): Use fmtresult ctor.
3601 (get_width_and_precision): Remove.
3602 (format_integer): Make naming changes. Avoid computing width and
3604 (format_floating): Same. Adjust indentation.
3605 (format_character, format_none): New functions.
3606 (format_string): Moved character handling to format_character.
3607 (format_directive): Remove arguments, change return type.
3608 (parse_directive): New function.
3609 (pass_sprintf_length::compute_format_length): Move directive
3610 parsing to parse_directive.
3612 2017-01-23 Jakub Jelinek <jakub@redhat.com>
3614 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
3615 (assign_assembler_name_if_needed): ... this.
3616 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
3617 (assign_assembler_name_if_needed): ... this.
3618 (free_lang_data_in_cgraph): Adjust callers.
3619 * cgraphunit.c (cgraph_node::analyze): Likewise.
3620 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
3623 2017-01-23 Richard Biener <rguenther@suse.de>
3625 PR tree-optimization/79088
3626 PR tree-optimization/79188
3627 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
3628 resetting loop bounds after last path deletion. Reset loop
3629 bounds of the target loop, make code match the comments.
3630 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
3631 Make sure loops need no fixups.
3633 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
3635 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
3636 exponent support with double type for first argument.
3637 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
3638 type returned by __builtin_vec_extract_sig,
3639 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
3640 functions from "vector int" to "vector unsigned int" or from
3641 "vector long long int" to "vector unsigned long long int".
3642 Changed type returned by __builtin_vec_extract_exp,
3643 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
3644 functions from "vector int" to "vector unsigned int" or from
3645 "vector long long int" to "vector unsigned long long int".
3646 Changed return type of __builtin_vec_test_data_class,
3647 __builtin_vec_test_data_class_sp, and
3648 __builtin_vec_test_data_class_dp from "vector int" to
3649 "vector bool int" or from "vector long long int" to "vector bool
3650 long long int" and changed second argument type from "unsigned
3651 int" to "int". Added new overloaded function forms "vector float
3652 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
3653 "vector float __builtin_vec_insert_exp_sp (vector float, vector
3654 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
3655 double, vector unsigned long long int)" and "vector double
3656 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
3657 long int)". Changed return type of
3658 __builtin_scalar_test_data_class and
3659 __builtin_scalar_test_data_class_sp and
3660 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
3661 int" and changed second argument from "unsigned int" to "int".
3662 Changed type returned by __builtin_scalar_test_neg,
3663 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
3664 from "int" to "bool int". Added new overloaded function form
3665 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
3666 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
3667 exponent double-precision with floating point first argument.
3668 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
3669 documentation of scalar_test_data_class, scalar_test_neg,
3670 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
3671 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
3672 vec_test_data_class built-in functions to reflect refinements in
3673 their type signatures.
3675 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
3677 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
3679 (aarch64_elf_asm_destructor): Likewise.
3681 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
3683 PR rtl-optimization/78634
3684 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
3685 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
3686 * ifcvt.c (noce_try_cmove): Add missing cost check.
3688 PR rtl-optimization/71724
3689 * combine.c (if_then_else_cond): Look for situations where it is
3690 beneficial to undo the work of one of the recursive calls.
3692 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3694 PR tree-optimization/70754
3695 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3696 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3697 combined stmt before it if not NULL.
3698 (combine_chains): Process refs reversely and compute dominance point
3701 2017-01-23 Martin Liska <mliska@suse.cz>
3703 PR tree-optimization/79196
3704 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
3705 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
3707 (strlen_optimize_stmt): Call the renamed function.
3709 2017-01-23 Michael Matz <matz@suse.de>
3711 PR tree-optimization/78384
3712 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
3714 2017-01-23 Richard Biener <rguenther@suse.de>
3716 PR tree-optimization/79186
3717 * tree-vrp.c (register_new_assert_for): Make sure we've seen
3718 both incoming edges before moving an assert.
3720 2017-01-23 Martin Jambor <mjambor@suse.cz>
3722 * ipa-prop.c (load_from_param_1): Removed.
3723 (load_from_unmodified_param): Bits from load_from_param_1 put back
3725 (load_from_param): Removed.
3726 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
3727 with stmt. Reverted back to use of load_from_unmodified_param.
3729 2017-01-23 Martin Jambor <mjambor@suse.cz>
3732 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
3733 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
3734 field a pointer to garbage collected vector, mark lattices and
3735 ipcp_orig_node with GTY((skip)).
3736 (ipa_get_param_count): Adjust to descriptors being a pointer.
3737 (ipa_get_param): Likewise.
3738 (ipa_get_type): Likewise.
3739 (ipa_get_param_move_cost): Likewise.
3740 (ipa_set_param_used): Likewise.
3741 (ipa_get_controlled_uses): Likewise.
3742 (ipa_set_controlled_uses): Likewise.
3743 (ipa_is_param_used): Likewise.
3744 (ipa_node_params_t): Move into garbage collector. New methods insert
3746 (ipa_node_params_sum): Annotate wth GTY(()).
3747 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
3749 (ipa_load_from_parm_agg): Adjust declaration.
3750 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
3751 * ipa-profile.c (ipa_profile): Likewise.
3752 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
3753 (ipa_populate_param_decls): Make descriptors parameter garbage
3755 (ipa_dump_param): Adjust to descriptors being a pointer.
3756 (ipa_alloc_node_params): Likewise.
3757 (ipa_initialize_node_params): Likewise.
3758 (load_from_param_1): Make descriptors parameter garbage collected.
3759 (load_from_unmodified_param): Likewise.
3760 (load_from_param): Likewise.
3761 (ipa_load_from_parm_agg): Likewise.
3762 (ipa_node_params::~ipa_node_params): Removed.
3763 (ipa_free_all_node_params): Remove call to delete operator.
3764 (ipa_node_params_t::insert): New.
3765 (ipa_node_params_t::remove): Likewise.
3766 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
3767 copy known_csts and known_contexts vectors.
3768 (ipa_read_node_info): Adjust to descriptors being a pointer.
3769 (ipcp_modif_dom_walker): Make m_descriptors field garbage
3771 (ipcp_transform_function): Make descriptors variable garbage
3774 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
3776 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
3777 * config/i386/avx512dqintrin.h: Ditto.
3778 * config/i386/avx512fintrin.h: Ditto.
3779 * gcc/config/i386/i386.c: Handle new builtins.
3780 * config/i386/i386-builtin.def: Add new builtins.
3781 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
3782 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
3784 2017-01-23 Jakub Jelinek <jakub@redhat.com>
3785 Martin Liska <mliska@suse.cz>
3787 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
3788 * asan.c (asan_expand_poison_ifn): Support stores and use
3789 appropriate ASAN report function.
3790 * internal-fn.c (expand_ASAN_POISON_USE): New function.
3791 * internal-fn.def (ASAN_POISON_USE): Declare.
3792 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
3793 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
3794 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
3795 ASAN_POISON calls w/o LHS.
3796 * tree-ssa.c (execute_update_addresses_taken): Create clobber
3797 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
3798 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
3799 * gimplify.c (asan_poison_variables): Add attribute
3800 use_after_scope_memory to variables that really needs to live
3802 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
3803 having the attribute.
3805 2017-01-23 Martin Liska <mliska@suse.cz>
3807 * asan.c (create_asan_shadow_var): New function.
3808 (asan_expand_poison_ifn): Likewise.
3809 * asan.h (asan_expand_poison_ifn): New declaration.
3810 * internal-fn.c (expand_ASAN_POISON): Likewise.
3811 * internal-fn.def (ASAN_POISON): New builtin.
3812 * sanopt.c (pass_sanopt::execute): Expand
3813 asan_expand_poison_ifn.
3814 * tree-inline.c (copy_decl_for_dup_finish): Make function
3816 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
3817 * tree-ssa.c (is_asan_mark_p): New function.
3818 (execute_update_addresses_taken): Rewrite local variables
3819 (identified just by use-after-scope as addressable) into SSA.
3821 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
3823 * doc/install.texi (Specific): opensource.apple.com uses https
3824 now. Remove trailing slash.
3826 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
3828 * README.Portability: Remove note on an Irix compatibility issue.
3830 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
3832 * gcov.c (INCLUDE_ALGORITHM): Define.
3833 (INCLUDE_VECTOR): Define.
3834 No longer include <vector> and <algorithm> directly.
3836 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
3838 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
3840 * doc/invoke.texi (Code Gen Options): Ditto.
3842 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
3845 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
3847 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
3849 rtl-optimization/79125
3850 * cprop.c (local_cprop_pass): Handle cases where we make an
3853 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
3857 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
3858 read from, for big endian.
3860 2017-01-20 Jiong Wang <jiong.wang@arm.com>
3862 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
3863 register pauth builtins for LP64 only.
3865 2017-01-20 Marek Polacek <polacek@redhat.com>
3868 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
3871 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
3873 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
3875 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
3876 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
3877 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
3879 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3882 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
3883 in big-endian mode when they are not a single duplicated value.
3885 2017-01-20 Richard Biener <rguenther@suse.de>
3887 * BASE-VER: Bump to 7.0.1.
3889 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
3891 * omp-low.c (omplow_simd_context): New struct. Use it...
3892 (lower_rec_simd_input_clauses): ...here and...
3893 (lower_rec_input_clauses): ...here to hold common data. Adjust all
3894 references to idx, lane, max_vf, is_simt.
3896 2017-01-20 Graham Markall <graham.markall@embecosm.com>
3898 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
3901 2017-01-20 Martin Jambor <mjambor@suse.cz>
3903 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
3904 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
3906 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
3907 (GTFILES): Rename hsa.c to hsa-common.c.
3908 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
3909 * hsa-dump.c: Likewise.
3910 * hsa-gen.c: Likewise.
3911 * hsa-regalloc.c: Likewise.
3912 * ipa-hsa.c: Likewise.
3913 * omp-expand.c: Likewise.
3914 * omp-low.c: Likewise.
3915 * toplev.c: Likewise.
3917 2017-01-20 Marek Polacek <polacek@redhat.com>
3920 * doc/invoke.texi: Document -Wduplicated-branches.
3921 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
3922 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
3923 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
3924 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
3925 return 0 only when not OEP_LEXICOGRAPHIC.
3926 (fold_build_cleanup_point_expr): Use the expression
3927 location when building CLEANUP_POINT_EXPR.
3928 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
3929 * tree.c (add_expr): Handle error_mark_node.
3931 2017-01-20 Martin Liska <mliska@suse.cz>
3934 * tree-profile.c (init_ic_make_global_vars): Do not call
3936 (gimple_init_gcov_profiler): Likewise.
3938 2017-01-20 Martin Liska <mliska@suse.cz>
3941 * cgraph.h (maybe_create_reference): Remove argument and
3943 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
3945 * ipa-cp.c (create_specialized_node): Likewise.
3946 * symtab.c (symtab_node::maybe_create_reference): Handle
3947 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
3949 2017-01-20 Martin Liska <mliska@suse.cz>
3951 * read-rtl-function.c (function_reader::create_function): Use
3952 build_decl instread of build_decl_stat.
3954 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
3956 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
3957 * config/i386/avx512dqintrin.h: Ditto.
3958 * config/i386/avx512fintrin.h: Ditto.
3959 * config/i386/i386-builtin-types.def: Add new types.
3960 * gcc/config/i386/i386.c: Handle new types.
3961 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
3962 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
3963 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
3964 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
3965 (__builtin_ia32_kshiftridi): New.
3966 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
3968 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
3972 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
3973 define to rs6000_init_stack_protect_guard.
3974 (rs6000_init_stack_protect_guard): New function.
3976 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
3977 Yunqiang Su <yunqiang.su@imgtec.com>
3979 * config.gcc (supported_defaults): Add madd4.
3980 (with_madd4): Add validation.
3981 (all_defaults): Add madd4.
3982 * config/mips/mips.opt (mmadd4): New option.
3983 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
3985 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
3987 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
3988 (ISA_HAS_FUSED_MADD4): Likewise.
3989 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
3990 * gcc/doc/install.texi (--with-madd4): Document the new option.
3992 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3994 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
3995 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
3996 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
3997 (aarch64_init_pauth_hint_builtins): New.
3998 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
3999 (aarch64_expand_builtin): Expand new builtins.
4001 2017-01-19 Jiong Wang <jiong.wang@arm.com>
4003 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
4004 * combine-stack-adj.c (no_unhandled_cfa): Handle
4005 REG_CFA_TOGGLE_RA_MANGLE.
4006 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
4007 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
4008 info for return address signing.
4009 (aarch64_expand_epilogue): Likewise.
4011 2017-01-19 Jiong Wang <jiong.wang@arm.com>
4013 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
4014 * config/aarch64/aarch64-protos.h
4015 (aarch64_return_address_signing_enabled): New declaration.
4016 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
4018 (aarch64_expand_prologue): Sign return address before it's pushed onto
4020 (aarch64_expand_epilogue): Authenticate return address fetched from
4022 (aarch64_override_options): Sanity check for ILP32 and ISA level.
4023 (aarch64_attributes): New function attributes for "sign-return-address".
4024 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
4025 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
4026 ("*do_return"): Generate combined instructions according to key index.
4027 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
4028 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
4030 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
4031 * config/aarch64/aarch64.opt (msign-return-address=): New.
4032 * doc/extend.texi (AArch64 Function Attributes): Documents
4033 "sign-return-address=".
4034 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
4036 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
4038 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
4039 overall option summary.
4041 2017-01-19 Jiong Wang <jiong.wang@arm.com>
4043 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
4044 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
4045 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
4046 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
4048 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
4050 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
4051 -mpower9-minmax by default for -mcpu=power9.
4052 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
4053 128-bit floating point.
4055 2017-01-20 Alan Modra <amodra@gmail.com>
4057 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
4058 optimizing for size.
4060 2017-01-20 Alan Modra <amodra@gmail.com>
4063 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
4064 for strcmp and strncmp from corresponding builtin decl.
4066 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
4068 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
4069 instead of i386/rtems-64.h.
4070 * config/i386/rtems-64.h: Remove.
4072 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
4076 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
4078 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
4080 2017-01-19 Tamar Christina <tamar.christina@arm.com>
4082 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
4083 Change int to HOST_WIDE_INT.
4084 * config/aarch64/aarch64-protos.h
4085 (aarch64_simd_gen_const_vector_dup): Likewise.
4086 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
4088 2017-01-19 David Malcolm <dmalcolm@redhat.com>
4090 * langhooks-def.h (lhd_type_for_size): New decl.
4091 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
4092 * langhooks.c (lhd_type_for_size): New function, taken from
4095 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
4097 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
4098 define_bypass for CR latency.
4099 (power9-cracked-alu): Update bypass latency and remove power9-branch.
4100 (power9-alu2): Add define_bypass for CR latency.
4102 (power9-mul): Update insn latency.
4103 (power9-mul-compare): Update insn latency, bypass latency and remove
4106 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4108 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
4110 * config/aarch64/aarch64.md
4111 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
4112 aarch64_nopcrelative_literal_loads.
4113 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
4115 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
4117 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
4119 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
4121 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
4124 * config.gcc (supported_defaults): Add lxc1-sxc1.
4125 (with_lxc1_sxc1): Add validation.
4126 (all_defaults): Add lxc1-sxc1.
4127 * config/mips/mips.opt (mlxc1-sxc1): New option.
4128 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
4130 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
4131 __mips_no_lxc1_sxc1.
4132 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
4133 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
4134 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
4136 2017-01-19 Richard Biener <rguenther@suse.de>
4138 PR tree-optimization/72488
4139 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
4140 sure to restore SSA info.
4141 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
4143 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
4145 PR rtl-optimization/79121
4146 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
4147 of the inner type when shifting an extended value.
4149 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
4152 * symtab.c (symtab_node::equal_address_to): Fix comparing of
4153 interposable aliases.
4155 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
4158 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
4159 Use the evmergelohi instruction.
4160 (mov_si<mode>_e500_subreg4_2_le): Likewise.
4161 (mov_sitf_e500_subreg8_2_be): Likewise.
4162 (mov_sitf_e500_subreg12_2_le): Likewise.
4163 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
4164 (mov_si<mode>_e500_subreg4_2_be): Likewise.
4165 (mov_sitf_e500_subreg8_2_le): Likewise.
4166 (mov_sitf_e500_subreg12_2_be): Likewise.
4168 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4170 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
4171 attribute from vecsimple to vecperm.
4172 (altivec_vbpermq2): Likewise.
4174 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4177 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
4179 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4180 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
4181 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
4182 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
4183 case where N arg is SIZE_MAX.
4184 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
4185 (cmpstrsi): Add pattern.
4187 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
4189 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4190 __builtin_vec_revb builtins.
4191 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
4192 built-in functions to support generation of the ISA 3.0 XXBR<x>
4193 vector byte reverse instructions.
4194 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
4195 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
4196 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
4197 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
4198 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
4199 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
4200 (P9V_BUILTIN_VEC_REVB): Likewise.
4201 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
4202 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
4203 (p9_xxbrq_v16qi): Likewise.
4204 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
4205 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
4206 (p9_xxbrh_v8hi): Likewise.
4207 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
4208 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
4209 vec_revb built-in functions.
4211 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
4213 PR rtl-optimization/78952
4214 * config/i386/i386.md (any_extract): New code iterator.
4215 (*insvqi_2): Use any_extract for source operand.
4216 (*insvqi_3): Use any_shiftrt for source operand.
4218 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
4220 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
4222 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
4224 2017-01-18 Matthias Klose <doko@ubuntu.com>
4226 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
4228 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4230 * config/rs6000/altivec.h (vec_bperm): Change #define.
4231 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
4232 (altivec_vbpermq2): New define_insn.
4233 (altivec_vbpermd): Likewise.
4234 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
4236 (VBPERMD): Likewise.
4237 (VBPERM): New polymorphic function interface.
4238 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
4239 Add entries for P9V_BUILTIN_VEC_VBPERM.
4240 * doc/extend.texi: Add interfaces for vec_bperm.
4242 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4244 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
4245 first letter of error messages.
4246 (s390_resolve_overloaded_builtin): Likewise.
4247 * config/s390/s390.c (s390_expand_builtin): Likewise.
4248 (s390_invalid_arg_for_unprototyped_fn): Likewise.
4249 (s390_valid_target_attribute_inner_p): Likewise.
4250 * config/s390/s390.md ("tabort"): Likewise.
4252 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
4254 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
4255 (ISA_AVOID_DIV_HILO): New macro.
4256 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
4257 (ISA_HAS_DDIV): Likewise.
4259 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4261 * doc/invoke.texi (fabi-version): Correct number of occurrences.
4263 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4265 * doc/invoke.texi (fabi-version): Spelling fix.
4267 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4270 * doc/invoke.texi (fabi-version): Mention mangling fix for
4273 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4276 * doc/invoke.texi (fabi-version): Document discriminator mangling.
4278 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
4281 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
4282 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
4284 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
4286 (stack_protect_test): Ditto.
4287 * config/rs6000/rs6000.opt (mstack-protector-guard=,
4288 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
4290 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
4291 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
4292 -mstack-protector-guard-offset=.
4293 (RS/6000 and PowerPC Options): Ditto.
4295 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
4297 * config/i386/i386.h (MASK_CLASS_P): New define.
4298 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
4299 there are no registers from different register sets also when
4300 mask registers are used. Update function comment.
4301 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
4302 to (*k/*r) and (*k/*km) alternatives.
4304 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
4306 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
4307 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
4308 (EH_RETURN_HANDLER_RTX): New define.
4309 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
4310 Force frame pointer in EH return functions.
4311 (aarch64_expand_epilogue): Add barrier for eh_return.
4312 (aarch64_final_eh_return_addr): Remove.
4313 (aarch64_eh_return_handler_rtx): New function.
4314 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
4316 (aarch64_eh_return_handler_rtx): New prototype.
4318 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4320 * config/rs6000/altivec.h (vec_rlmi): New #define.
4321 (vec_vrlnm): Likewise.
4322 (vec_rlnm): Likewise.
4323 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
4324 (UNSPEC_VRLNM): Likewise.
4325 (VIlong): New mode iterator.
4326 (altivec_vrl<VI_char>mi): New define_insn.
4327 (altivec_vrl<VI_char>nm): Likewise.
4328 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
4331 (RLNM): New polymorphic function entry.
4332 (VRLWMI): New monomorphic function entry.
4334 (RLMI): New polymorphic function entry.
4335 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
4336 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
4337 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
4340 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4343 * dwarf2out.c (field_byte_offset): Restore the
4344 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
4345 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
4346 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
4349 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
4352 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
4354 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4357 * dwarf2out.c (add_data_member_location_attribute): For constant
4358 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
4359 instead of DW_AT_data_member_location, DW_AT_bit_offset and
4360 DW_AT_byte_size attributes.
4362 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
4364 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
4365 after forcing to constant memory when the code model is medium.
4367 2017-01-17 Julia Koval <julia.koval@intel.com>
4370 * config/i386/avx512fintrin.h
4371 (_mm512_i32gather_ps): Change __addr type to void const*.
4372 (_mm512_mask_i32gather_ps): Ditto.
4373 (_mm512_i32gather_pd): Ditto.
4374 (_mm512_mask_i32gather_pd): Ditto.
4375 (_mm512_i64gather_ps): Ditto.
4376 (_mm512_mask_i64gather_ps): Ditto.
4377 (_mm512_i64gather_pd): Ditto.
4378 (_mm512_mask_i64gather_pd): Ditto.
4379 (_mm512_i32gather_epi32): Ditto.
4380 (_mm512_mask_i32gather_epi32): Ditto.
4381 (_mm512_i32gather_epi64): Ditto.
4382 (_mm512_mask_i32gather_epi64): Ditto.
4383 (_mm512_i64gather_epi32): Ditto.
4384 (_mm512_mask_i64gather_epi32): Ditto.
4385 (_mm512_i64gather_epi64): Ditto.
4386 (_mm512_mask_i64gather_epi64): Ditto.
4387 (_mm512_i32scatter_ps): Change __addr type to void*.
4388 (_mm512_mask_i32scatter_ps): Ditto.
4389 (_mm512_i32scatter_pd): Ditto.
4390 (_mm512_mask_i32scatter_pd): Ditto.
4391 (_mm512_i64scatter_ps): Ditto.
4392 (_mm512_mask_i64scatter_ps): Ditto.
4393 (_mm512_i64scatter_pd): Ditto.
4394 (_mm512_mask_i64scatter_pd): Ditto.
4395 (_mm512_i32scatter_epi32): Ditto.
4396 (_mm512_mask_i32scatter_epi32): Ditto.
4397 (_mm512_i32scatter_epi64): Ditto.
4398 (_mm512_mask_i32scatter_epi64): Ditto.
4399 (_mm512_i64scatter_epi32): Ditto.
4400 (_mm512_mask_i64scatter_epi32): Ditto.
4401 (_mm512_i64scatter_epi64): Ditto.
4402 (_mm512_mask_i64scatter_epi64): Ditto.
4403 * config/i386/avx512pfintrin.h
4404 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
4405 (_mm512_mask_prefetch_i32gather_ps): Ditto.
4406 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4407 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4408 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
4409 (_mm512_prefetch_i32scatter_ps): Ditto.
4410 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4411 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4412 (_mm512_prefetch_i64scatter_pd): Ditto.
4413 (_mm512_prefetch_i64scatter_ps): Ditto.
4414 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4415 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4416 * config/i386/avx512vlintrin.h
4417 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
4418 (_mm_mmask_i32gather_ps): Ditto.
4419 (_mm256_mmask_i32gather_pd): Ditto.
4420 (_mm_mmask_i32gather_pd): Ditto.
4421 (_mm256_mmask_i64gather_ps): Ditto.
4422 (_mm_mmask_i64gather_ps): Ditto.
4423 (_mm256_mmask_i64gather_pd): Ditto.
4424 (_mm_mmask_i64gather_pd): Ditto.
4425 (_mm256_mmask_i32gather_epi32): Ditto.
4426 (_mm_mmask_i32gather_epi32): Ditto.
4427 (_mm256_mmask_i32gather_epi64): Ditto.
4428 (_mm_mmask_i32gather_epi64): Ditto.
4429 (_mm256_mmask_i64gather_epi32): Ditto.
4430 (_mm_mmask_i64gather_epi32): Ditto.
4431 (_mm256_mmask_i64gather_epi64): Ditto.
4432 (_mm_mmask_i64gather_epi64): Ditto.
4433 (_mm256_i32scatter_ps): Change __addr type to void*.
4434 (_mm256_mask_i32scatter_ps): Ditto.
4435 (_mm_i32scatter_ps): Ditto.
4436 (_mm_mask_i32scatter_ps): Ditto.
4437 (_mm256_i32scatter_pd): Ditto.
4438 (_mm256_mask_i32scatter_pd): Ditto.
4439 (_mm_i32scatter_pd): Ditto.
4440 (_mm_mask_i32scatter_pd): Ditto.
4441 (_mm256_i64scatter_ps): Ditto.
4442 (_mm256_mask_i64scatter_ps): Ditto.
4443 (_mm_i64scatter_ps): Ditto.
4444 (_mm_mask_i64scatter_ps): Ditto.
4445 (_mm256_i64scatter_pd): Ditto.
4446 (_mm256_mask_i64scatter_pd): Ditto.
4447 (_mm_i64scatter_pd): Ditto.
4448 (_mm_mask_i64scatter_pd): Ditto.
4449 (_mm256_i32scatter_epi32): Ditto.
4450 (_mm256_mask_i32scatter_epi32): Ditto.
4451 (_mm_i32scatter_epi32): Ditto.
4452 (_mm_mask_i32scatter_epi32): Ditto.
4453 (_mm256_i32scatter_epi64): Ditto.
4454 (_mm256_mask_i32scatter_epi64): Ditto.
4455 (_mm_i32scatter_epi64): Ditto.
4456 (_mm_mask_i32scatter_epi64): Ditto.
4457 (_mm256_i64scatter_epi32): Ditto.
4458 (_mm256_mask_i64scatter_epi32): Ditto.
4459 (_mm_i64scatter_epi32): Ditto.
4460 (_mm_mask_i64scatter_epi32): Ditto.
4461 (_mm256_i64scatter_epi64): Ditto.
4462 (_mm256_mask_i64scatter_epi64): Ditto.
4463 (_mm_i64scatter_epi64): Ditto.
4464 (_mm_mask_i64scatter_epi64): Ditto.
4465 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
4466 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
4467 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
4468 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
4469 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
4470 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
4471 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
4472 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
4473 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
4474 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
4475 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
4476 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
4477 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
4478 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
4479 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
4480 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
4481 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
4482 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
4483 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
4484 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
4485 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
4486 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
4487 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
4488 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
4489 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
4490 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
4491 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
4492 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
4493 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
4494 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
4495 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
4496 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
4497 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
4498 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
4499 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
4500 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
4501 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
4502 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
4503 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
4504 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
4505 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
4506 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
4507 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
4508 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
4509 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
4510 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
4511 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
4512 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
4513 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
4514 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
4515 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
4516 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
4517 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
4518 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
4519 definitions accordingly.
4521 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
4522 Kuan-Lin Chen <kuanlinchentw@gmail.com>
4525 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
4528 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
4531 * ira-conflicts.c (ira_build_conflicts): Update total conflict
4532 hard regs for inner regno.
4534 2017-01-17 Martin Liska <mliska@suse.cz>
4537 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
4538 assumption and add comment.
4540 2017-01-17 Nathan Sidwell <nathan@acm.org>
4542 * ipa-visibility.c (localize_node): New function, broken out of ...
4543 (function_and_variable_visibility): ... here. Call it.
4545 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
4548 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
4549 correctly set frequency of oudgoing edge.
4550 (duplicate_thread_path): Fix profile updating.
4552 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4555 * configure.ac: Add GCC_BASE_VER.
4556 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
4557 version from BASE-VER file.
4558 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
4559 (gcc.o): Depend on $(BASEVER).
4560 * common.opt (dumpfullversion): New option.
4561 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
4562 * doc/invoke.texi: Document -dumpfullversion.
4563 * doc/install.texi: Document --with-gcc-major-version-only.
4564 * configure: Regenerated.
4566 2017-01-17 Richard Biener <rguenther@suse.de>
4568 PR tree-optimization/71433
4569 * tree-vrp.c (register_new_assert_for): Merge same asserts
4570 on all incoming edges.
4571 (process_assert_insertions_for): Handle insertions at the
4574 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
4576 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
4577 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
4579 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
4582 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
4585 2017-01-17 Alan Modra <amodra@gmail.com>
4588 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
4589 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
4590 symbolic stack limit when pic.
4592 2017-01-16 Martin Sebor <msebor@redhat.com>
4594 PR tree-optimization/78608
4595 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
4597 2017-01-16 Jeff Law <law@redhat.com>
4600 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
4601 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
4602 for several include directories that may be relative to sysroot.
4603 * config/i386/x-mingw32 (gplus_includedir): Define.
4604 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
4605 (native_system_includedir): Likewise.
4606 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
4607 override if TARGET_SYSTEM_ROOT is defined.
4608 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
4610 PR tree-optimization/79090
4611 PR tree-optimization/33562
4612 PR tree-optimization/61912
4613 PR tree-optimization/77485
4614 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
4615 and computed trims into the dump file.
4617 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
4619 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
4621 2017-01-16 Jakub Jelinek <jakub@redhat.com>
4624 * gimplify.c (gimplify_init_constructor): If want_value and
4625 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
4629 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
4630 sequence. Formatting fixes.
4631 (doloop_optimize): Formatting fixes.
4634 * gcc.c (debug_level_greater_than_spec_func): New function.
4635 (static_spec_functions): Add debug-level-gt spec function.
4636 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
4638 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
4639 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
4640 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
4641 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
4642 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
4643 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
4645 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
4647 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
4648 QImode fixups to general and mask registers only.
4650 2017-01-16 Carl Love <cel@us.ibm.com>
4652 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
4653 for built-in functions
4654 vector signed char vec_nabs (vector signed char)
4655 vector signed short vec_nabs (vector signed short)
4656 vector signed int vec_nabs (vector signed int)
4657 vector signed long long vec_nabs (vector signed long long)
4658 vector float vec_nabs (vector float)
4659 vector double vec_nabs (vector double)
4660 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
4662 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
4663 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
4664 * doc/extend.texi: Update the documentation file for the new built-in
4667 2017-01-16 Martin Sebor <msebor@redhat.com>
4669 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
4672 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4674 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
4675 UNSPEC_VSX__XXSPLTD to require special splat handling.
4677 2017-01-16 David Malcolm <dmalcolm@redhat.com>
4680 * system.h: Poison strndup.
4682 2017-01-16 Alan Modra <amodra@gmail.com>
4685 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
4688 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
4690 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
4692 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
4694 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
4695 call recog here. Assert that INSN_CODE (insn) is non-negative.
4697 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
4700 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
4702 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
4703 in the currently scheduled RTL fragment.
4705 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
4707 PR rtl-optimization/78751
4708 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
4711 2017-01-14 Jeff Law <law@redhat.com>
4713 PR tree-optimization/79090
4714 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
4715 variable length stores.
4716 (compute_trims): Delete dead assignment to *trim_tail.
4717 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
4720 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
4722 PR rtl-optimization/78626
4723 PR rtl-optimization/78727
4724 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
4725 of a block, and split such blocks after everything else is finished.
4727 2017-01-14 Alan Modra <amodra@gmail.com>
4730 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
4731 target legitimate_combined_insn.
4732 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
4733 (rs6000_legitimate_combined_insn): New function.
4734 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
4736 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
4737 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
4738 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
4740 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
4742 * doc/frontends.texi (G++ and GCC): Remove references to Java.
4744 2017-01-13 Jeff Law <law@redhat.com>
4746 PR tree-optimization/33562
4747 PR tree-optimization/61912
4748 PR tree-optimization/77485
4749 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
4751 (delete_dead_assignment): Likewise.
4752 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
4753 statement to delete_dead_call and delete_dead_assignment.
4755 2017-01-13 David Malcolm <dmalcolm@redhat.com>
4758 * substring-locations.c (format_warning_va): Strengthen case 1 so
4759 that both endpoints of the substring must be within the format
4760 range for just the substring to be printed.
4762 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
4764 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
4765 * config/i386/i386.c (ix86_target_string): Add missing options
4766 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
4767 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
4768 flags_other and ix86_target_other to flags2_other. Display unknown
4770 (ix86_valid_target_attribute_inner_p): Add missing options and
4771 reorder options by implied ISAs, as in ix86_target_string.
4773 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
4775 * hash-table.h (hash_table::too_empty_p): New function.
4776 (hash_table::expand): Use it.
4777 (hash_table::traverse): Likewise.
4778 (hash_table::empty_slot): Use sizeof (value_type) instead of
4779 sizeof (PTR) to convert bytes to elements. Shrink the table
4780 if the current size is excessive for the current number of
4783 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
4785 * ira-costs.c (record_reg_classes): Break from the inner loop
4786 early once alt_fail is known to be true. Update outer loop
4787 handling accordingly.
4789 2017-01-13 Jeff Law <law@redhat.com>
4791 * tree-ssa-dse.c (decrement_count): New function.
4792 (increment_start_addr, maybe_trim_memstar_call): Likewise.
4793 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
4794 when we know the partially dead statement is a mem* function.
4796 PR tree-optimization/61912
4797 PR tree-optimization/77485
4798 * tree-ssa-dse.c: Include expr.h.
4799 (maybe_trim_constructor_store): New function.
4800 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
4802 PR tree-optimization/33562
4803 PR tree-optimization/61912
4804 PR tree-optimization/77485
4805 * doc/invoke.texi: Document new dse-max-object-size param.
4806 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
4807 * tree-ssa-dse.c: Include params.h.
4808 (dse_store_status): New enum.
4809 (initialize_ao_ref_for_dse): New, partially extracted from
4811 (valid_ao_ref_for_dse, normalize_ref): New.
4812 (setup_live_bytes_from_ref, compute_trims): Likewise.
4813 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
4814 (maybe_trim_partially_dead_store): Likewise.
4815 (maybe_trim_complex_store): Likewise.
4816 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
4817 Track what bytes live from the original store. Return tri-state
4818 for dead, partially dead or live.
4819 (dse_dom_walker): Add constructor, destructor and new private members.
4820 (delete_dead_call, delete_dead_assignment): New extracted from
4822 (dse_optimize_stmt): Make a member of dse_dom_walker.
4823 Use initialize_ao_ref_for_dse.
4825 PR tree-optimization/33562
4826 PR tree-optimization/61912
4827 PR tree-optimization/77485
4828 * sbitmap.h (bitmap_count_bits): Prototype.
4829 (bitmap_clear_range, bitmap_set_range): Likewise.
4830 * sbitmap.c (bitmap_clear_range): New function.
4831 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
4833 2017-01-13 Martin Liska <mliska@suse.cz>
4836 * function.c (set_cfun): Add new argument force.
4837 * function.h (set_cfun): Likewise.
4838 * ipa-inline-transform.c (inline_call): Use the function when
4839 strict alising from is dropped for function we inline to.
4841 2017-01-13 Richard Biener <rguenther@suse.de>
4843 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
4844 for dumping GIMPLE INTEGER_CSTs.
4846 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4848 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
4849 to 201112L since C++17.
4851 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
4854 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
4855 if -fsanitize=kernel-address is present.
4857 2017-01-13 Richard Biener <rguenther@suse.de>
4859 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
4860 as _Literal ( type ) number in case usual suffixes do not
4861 preserve all information.
4863 2017-01-13 Richard Biener <rguenther@suse.de>
4865 PR tree-optimization/77283
4866 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
4867 and ssa-iterators.h.
4868 (is_feasible_trace): Implement a cost model based on joiner
4871 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4874 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
4875 char or short to __float128/_Float128 directly.
4877 2017-01-12 Martin Sebor <msebor@redhat.com>
4879 to -Wformat-overflow.
4880 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
4881 (min_bytes_remaining): Same.
4882 (get_string_length): Same.
4883 (format_string): Same.
4884 (format_directive): Same.
4886 (pass_sprintf_length::handle_gimple_call): Same.
4888 2017-01-12 Jakub Jelinek <jakub@redhat.com>
4890 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
4891 info.nowrite calls with no lhs that can't throw. Return bool
4892 whether gsi_remove has been called or not.
4893 (pass_sprintf_length::handle_gimple_call): Return bool whether
4894 try_substitute_return_value called gsi_remove. Formatting fix.
4895 (pass_sprintf_length::execute): Don't use gsi_remove if
4896 handle_gimple_call returned true.
4899 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
4900 be removed due to side-effects, don't remove following barrier nor
4901 turn the successor edge into fallthru edge.
4903 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4906 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
4907 element-reversing loads and stores as not swappable.
4909 2017-01-12 Nathan Sidwell <nathan@acm.org>
4910 Nicolai Stange <nicstange@gmail.com>
4912 * combine.c (try_combine): Don't ignore result of overlap checking
4913 loop. Combine overlap & asm check into single loop.
4915 2017-01-12 Richard Biener <rguenther@suse.de>
4917 * tree-pretty-print.c (dump_generic_node): Provide -gimple
4918 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
4920 2017-01-12 Richard Biener <rguenther@suse.de>
4922 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
4923 and TS_TARGET_OPTION directly derive from TS_BASE.
4924 * tree-core.h (tree_optimization_option): Derive from tree_base.
4925 (tree_target_option): Likewise.
4927 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
4929 * config/i386/i386.c (memory_address_length): Increase len
4930 only when rip_relative_addr_p returns false.
4932 2017-01-11 Julia Koval <julia.koval@intel.com>
4934 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
4935 (OPTION_MASK_ISA_SGX_SET): New.
4936 (ix86_handle_option): Handle OPT_msgx.
4937 * config.gcc: Added sgxintrin.h.
4938 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
4939 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
4940 * config/i386/i386.c (ix86_target_string): Add -msgx.
4942 (ix86_option_override_internal): Handle new options.
4943 (ix86_valid_target_attribute_inner_p): Add sgx.
4944 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
4945 * config/i386/i386.opt: Add msgx.
4946 * config/i386/sgxintrin.h: New file.
4947 * config/i386/x86intrin.h: Add sgxintrin.h.
4949 2017-01-11 Jakub Jelinek <jakub@redhat.com>
4952 * fold-const.c (maybe_nonzero_address): Return 1 for function
4954 (tree_single_nonzero_warnv_p): Don't handle function local objects
4958 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
4961 2017-01-11 David Malcolm <dmalcolm@redhat.com>
4964 * opts.c: Include "spellcheck.h"
4965 (struct string_fragment): New struct.
4966 (struct edit_distance_traits<const string_fragment &>): New
4968 (get_closest_sanitizer_option): New function.
4969 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
4971 2017-01-11 Jakub Jelinek <jakub@redhat.com>
4973 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
4975 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
4976 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
4977 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
4978 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
4979 for initial die_offset if dwarf_split_debug_info.
4980 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
4981 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
4983 (output_skeleton_debug_sections): Formatting fix. Use
4984 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
4985 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
4987 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
4989 * config/arm/cortex-a53.md: Add bypasses for
4991 (cortex_a53_r2f): Only use for transfers.
4992 (cortex_a53_f2r): Likewise.
4993 (cortex_a53_r2f_cvt): Add reservation for conversions.
4994 (cortex_a53_f2r_cvt): Likewise.
4996 2017-01-11 Tamar Christina <tamar.christina@arm.com>
4998 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
4999 to all inlined functions, change static to extern.
5001 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
5004 * config/arm/arm.c (legitimize_pic_address): Handle reference to
5006 (arm_assemble_integer): Likewise.
5008 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
5010 * config.gcc: Use new awk script to check CPU, FPU and architecture
5011 parameters for --with-... options.
5012 * config/arm/parsecpu.awk: New file
5013 * config/arm/arm-cpus.in: New file.
5014 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
5016 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
5018 * config/arm/t-arm: Update dependency rules.
5019 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
5020 of processing .def files.
5021 * config/arm/genopt.sh: Deleted.
5022 * config/arm/gentune.sh: Deleted.
5023 * config/arm/arm-cores.def: Deleted.
5024 * config/arm/arm-arches.def: Deleted.
5025 * config/arm/arm-fpus.def: Deleted.
5026 * config/arm/arm-tune.md: Regenerated.
5027 * config/arm/arm-tables.opt: Regenerated.
5028 * config/arm/arm-cpu.h: New generated file.
5029 * config/arm/arm-cpu-data.h: New generated file.
5030 * config/arm/arm-cpu-cdata.h: New generated file.
5032 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
5035 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
5037 (input_varpool_node): Unpack dynamically_initialized bit.
5039 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
5041 PR rtl-optimization/79032
5042 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
5043 the alignment of the adjusted memory reference against that of MODE,
5044 instead of the alignment of the original memory reference.
5046 2017-01-11 Martin Jambor <mjambor@suse.cz>
5048 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
5050 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
5051 decorated functions.
5053 2017-01-11 Richard Biener <rguenther@suse.de>
5055 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
5056 set range/nonnull info for PHI results. Do not set it on
5057 stmts marked for removal.
5059 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
5061 * expr.c (store_field): In the bitfield case, fetch the return value
5062 from the registers before applying a single big-endian adjustment.
5063 Always do a final load for a BLKmode value not larger than a word.
5065 2017-01-10 David Malcolm <dmalcolm@redhat.com>
5068 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
5069 that we correctly handle column numbers greater than
5070 LINE_MAP_MAX_COLUMN_NUMBER.
5072 2017-01-10 Martin Sebor <msebor@redhat.com>
5075 * gimple-ssa-sprintf.c (get_destination_size): Call
5076 {init,fini}object_sizes.
5077 * tree-object-size.c (addr_object_size): Adjust.
5078 (pass_through_call): Adjust.
5079 (pass_object_sizes::execute): Adjust.
5080 * tree-object-size.h (fini_object_sizes): Declare.
5082 2017-01-10 Martin Sebor <msebor@redhat.com>
5084 PR tree-optimization/78775
5085 * builtins.c (get_size_range): Move...
5086 * calls.c: ...to here.
5087 (alloc_max_size): Accept zero argument.
5088 (operand_signed_p): Remove.
5089 (maybe_warn_alloc_args_overflow): Call get_size_range.
5090 * calls.h (get_size_range): Declare.
5092 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
5094 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
5095 from TI's devices.csv file as of September 2016.
5096 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
5098 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
5100 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
5101 * doc/invoke.texi: Likewise.
5102 * doc/md.texi: Likewise.
5103 * doc/objc.texi: Likewise.
5105 2017-01-10 Joshua Conner <joshconner@google.com>
5107 * config/arm/fuchsia-elf.h: New file.
5108 * config/fuchsia.h: New file.
5109 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
5110 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
5112 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
5114 2016-01-10 Richard Biener <rguenther@suse.de>
5116 PR tree-optimization/79034
5117 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
5118 Propagate out degenerate PHIs in the joiner.
5120 2017-01-10 Martin Liska <mliska@suse.cz>
5122 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
5123 (sort_congruence_classes_by_decl_uid): Likewise.
5124 (sort_congruence_class_groups_by_decl_uid): Likewise.
5125 (sem_item_optimizer::merge_classes): Sort class, groups in these
5126 classes and members in the groups by DECL_UID of declarations.
5127 This would make merge operations stable.
5129 2017-01-10 Martin Liska <mliska@suse.cz>
5131 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
5132 usage of m_classes_vec.
5133 (sem_item_optimizer::~sem_item_optimizer): Likewise.
5134 (sem_item_optimizer::get_group_by_hash): Likewise.
5135 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
5136 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
5137 (sem_item_optimizer::verify_classes): Likewise.
5138 (sem_item_optimizer::process_cong_reduction): Likewise.
5139 (sem_item_optimizer::dump_cong_classes): Likewise.
5140 (sem_item_optimizer::merge_classes): Likewise.
5141 * ipa-icf.h (congruence_class_hash): Rename from
5142 congruence_class_group_hash. Remove declaration of m_classes_vec.
5144 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
5146 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
5147 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
5148 * config.gcc: Add avx512vpopcntdqintrin.h.
5149 * config/i386/avx512vpopcntdqintrin.h: New.
5150 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
5151 * config/i386/i386-builtin-types.def: Add new types.
5152 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
5153 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
5154 __builtin_ia32_vpopcountq_v8di_mask): New.
5155 * config/i386/i386-c.c (ix86_target_macros_internal): Define
5156 __AVX512VPOPCNTDQ__.
5157 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
5158 (PTA_AVX512VPOPCNTDQ): Define.
5159 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
5160 TARGET_AVX512VPOPCNTDQ_P): Define.
5161 * config/i386/i386.opt: Add mavx512vpopcntdq.
5162 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
5163 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
5165 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5168 * predict.def (PRED_CALL): Set to 67.
5170 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
5172 * expr.c (store_field): In the bitfield case, if the value comes from
5173 a function call and is of an aggregate type returned in registers, do
5174 not modify the field mode; extract the value in all cases if the mode
5175 is BLKmode and the size is not larger than a word.
5177 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
5180 * config/i386/cpuid.h: Fix undefined behavior.
5182 2017-01-04 Jeff Law <law@redhat.com>
5184 PR tree-optimization/79007
5185 PR tree-optimization/67955
5186 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
5187 conservative for pt.null when flag_non_call_exceptions is on.
5189 2017-01-09 Jakub Jelinek <jakub@redhat.com>
5191 PR translation/79019
5192 PR translation/79020
5193 * params.def (PARAM_INLINE_MIN_SPEEDUP,
5194 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
5195 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
5197 * config/avr/avr.opt (maccumulate-args): Likewise.
5198 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
5199 * common.opt (freport-bug): Likewise.
5200 * cif-code.def (CIF_FINAL_ERROR): Likewise.
5201 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
5202 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
5203 translatable string.
5204 * config/i386/i386.c (function_value_32): Likewise.
5205 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
5206 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
5208 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
5209 * common/config/msp430/msp430-common.c (msp430_handle_option):
5211 * symtab.c (symtab_node::verify_base): Likewise.
5212 * opts.c (set_debug_level): Likewise.
5213 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
5214 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
5215 missing whitespace to translatable strings.
5216 * config/avr/avr.md (bswapsi2): Fix typo in comment.
5217 * config/sh/superh.h: Likewise.
5218 * config/i386/xopintrin.h: Likewise.
5219 * config/i386/znver1.md: Likewise.
5220 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
5221 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
5222 * double-int.h (struct double_int): Likewise.
5223 * double-int.c (div_and_round_double): Likewise.
5224 * wide-int.cc: Likewise.
5225 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
5226 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
5227 * cfgcleanup.c (crossjumps_occured): Renamed to ...
5228 (crossjumps_occurred): ... this.
5229 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
5232 PR tree-optimization/78899
5233 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
5234 returning bool return struct loop *, NULL for failure and the new
5236 (versionable_outer_loop_p): Don't version outer loop if it has
5237 dont_vectorized bit set.
5238 (tree_if_conversion): When versioning outer loop, ensure
5239 tree_if_conversion is performed also on the inner loop of the
5240 non-vectorizable outer loop copy.
5241 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
5242 LOOP_VECTORIZED in inner loop of the scalar outer loop and
5243 prevent vectorization of it.
5244 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
5245 the outer loop vectorization of the non-scalar version is attempted
5246 before vectorization of the inner loop in scalar version. If
5247 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
5248 vectorization of its inner loop.
5249 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
5250 has 2 inner loops, rename also on edges from bb whose single pred
5251 is outer_loop->header. Fix typo in function comment.
5253 2017-01-09 Martin Sebor <msebor@redhat.com>
5256 * asan.c (asan_emit_stack_protection): Increase local buffer size
5257 to avoid snprintf truncation warning.
5259 2017-01-09 Andrew Pinski <apinski@cavium.com>
5261 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
5262 to reference thunderx2t99 for the tuning structure
5263 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
5265 (thunderx2t99_extra_costs): This.
5266 * config/aarch64/aarch64-tune.md: Regenerate.
5267 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
5268 (vulcan_addrcost_table): This.
5269 (vulcan_regmove_cost): Rename to ...
5270 (thunderx2t99_regmove_cost): This.
5271 (vulcan_vector_cost): Rename to ...
5272 (thunderx2t99_vector_cost): this.
5273 (vulcan_branch_cost): Rename to ...
5274 (thunderx2t99_branch_cost): This.
5275 (vulcan_tunings): Rename to ...
5276 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
5277 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
5279 2017-01-09 Martin Jambor <mjambor@suse.cz>
5283 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
5284 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
5285 (propagate_vr_accross_jump_function): Use the above function for all
5286 value range computations for pass-through jump functions and type
5287 converasion from explicit value range values.
5288 (ipcp_propagate_stage): Do not attempt to deduce types of formal
5289 parameters from TYPE_ARG_TYPES.
5290 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
5291 (ipa_write_node_info): Stream type of the actual argument.
5292 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
5294 2017-01-09 Martin Liska <mliska@suse.cz>
5297 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
5298 (lookup_compiler): Do not show error message with have_E.
5300 2017-01-09 Jakub Jelinek <jakub@redhat.com>
5302 PR tree-optimization/78938
5303 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
5304 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
5305 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
5306 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
5309 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5311 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
5314 2017-01-09 Richard Biener <rguenther@suse.de>
5316 PR tree-optimization/78997
5317 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
5318 name condition properly.
5320 2017-01-09 Richard Biener <rguenther@suse.de>
5323 * dwarf2out.c (is_cxx): New overload with context.
5324 (is_naming_typedef_decl): Use it.
5326 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5328 * invoke.texi (Option Summary): Correct spacing in option lists
5329 and add line breaks to fix over-long lines.
5331 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5335 * extend.texi (Common Variable Attributes): Add xref to GCC
5336 Internals manual to explain mode attribute keywords.
5338 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5341 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
5342 and Preprocessor Options.
5343 (Options for Linking): Document -pthread here....
5344 (RS/6000 and PowerPC Options): ...not here.
5345 (Solaris 2 Options): ...or here.
5346 * doc/cppopts.texi: Document -pthread.
5348 2017-01-08 Martin Sebor <msebor@redhat.com>
5351 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
5352 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
5353 New member functions.
5354 (format_directive): Used them.
5356 (pass_sprintf_length::handle_gimple_call): Same.
5357 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
5358 to avoid truncation for any argument.
5359 (extract_affine_mul): Same.
5360 * tree.c (get_file_function_name): Same.
5362 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5365 * predict.def (PRED_INDIR_CALL): Set to 86.
5367 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5369 PR preprocessor/54124
5370 * doc/cppopts.texi: Reformat -d subtable to list the full name
5371 of the options. Add cross-reference to the docs for the general
5372 compiler -d options.
5373 * doc/invoke.texi (Developer Options): Add cross-reference to the
5374 preprocessor-specific -d option documentation.
5376 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5378 PR preprocessor/13498
5379 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
5380 redudant material, and reflect new command-line options.
5381 (System Headers): Likewise.
5383 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5385 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
5386 -isystem, and -idirafter. Copy-edit.
5387 * doc/cppopts.texi: Copy-edit. Remove contradiction about
5388 default for -ftrack-macro-expansion. Delete obsolete and
5389 badly-formatted implementation details about -fdebug-cpp output.
5390 * doc/cppwarnopts.texi: Copy-edit.
5392 2017-01-07 David Malcolm <dmalcolm@redhat.com>
5395 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
5396 that the transition from a max line width >= 1<<10 to narrower
5397 lines works correctly.
5399 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
5401 * doc/options.texi (PerFunction): New.
5402 * opt-functions.awk (switch_flags): Map both Optimization and
5403 PerFunction to CL_OPTIMIZATION.
5404 * opth-gen.awk: Test for PerFunction flag along with
5406 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
5407 it only when the latter is present. Skip those that don't in
5408 the hash function generator.
5409 * common.opt (fvar-tracking): Mark as PerFunction instead of
5411 (fvar-tracking-assignments): Likewise.
5412 (fvar-tracking-assignments-toggle): Likewise.
5413 (fvar-tracking-uninit): Likewise.
5415 2017-01-07 Jakub Jelinek <jakub@redhat.com>
5417 PR translation/79018
5418 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
5421 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
5424 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
5425 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
5426 TARGET_LONG_JUMP_TABLE_OFFSETS.
5427 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
5428 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
5429 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
5430 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
5431 * config/m68k/m68k.md (tablejump expander): Likewise.
5432 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
5433 TARGET_LONG_JUMP_TABLE_OFFSETS.
5434 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
5435 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
5437 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
5438 David Holsgrove <david.holsgrove@xilinx.com>
5440 * common/config/microblaze/microblaze-common.c
5441 (TARGET_EXCEPT_UNWIND_INFO): Remove.
5442 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
5444 * config/microblaze/microblaze.c (microblaze_must_save_register)
5445 (microblaze_expand_epilogue, microblaze_return_addr): Handle
5447 (microblaze_eh_return): New function.
5448 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
5449 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
5450 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
5451 * config/microblaze/microblaze.md (eh_return): New pattern.
5453 2017-01-06 Jakub Jelinek <jakub@redhat.com>
5455 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
5456 GCC_DIAGNOSTIC_STRINGIFY): Define.
5458 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
5460 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5462 * config/arm/arm.md (<mcrr>): New.
5464 * config/arm/arm.c (arm_arch5te): New.
5465 (arm_option_override): Set arm_arch5te.
5466 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
5468 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
5469 (arm_mcrr_qualifiers): ... this. New.
5470 (MRRC_QUALIFIERS): Define to...
5471 (arm_mrrc_qualifiers): ... this. New.
5472 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
5474 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
5475 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
5476 (MRRCI, mrrc, MRRC): New.
5477 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
5478 VUNSPEC_MRRC2): New.
5480 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5482 * config/arm/arm.md (<mcr>): New.
5484 * config/arm/arm.c (arm_coproc_builtin_available): Add
5485 support for mcr, mrc, mcr2 and mrc2.
5486 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
5487 (arm_mcr_qualifiers): ... this. New.
5488 (MRC_QUALIFIERS): Define to ...
5489 (arm_mrc_qualifiers): ... this. New.
5490 (MCR_QUALIFIERS): Define to ...
5491 (arm_mcr_qualifiers): ... this. New.
5492 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
5494 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
5495 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
5496 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
5499 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5501 * config/arm/arm.md (*ldc): New.
5505 * config/arm/arm.c (arm_coproc_builtin_available): Add
5506 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
5507 (arm_coproc_ldc_stc_legitimate_address): New.
5508 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
5509 'qualifier_const_pointer'.
5510 (LDC_QUALIFIERS): Define to...
5511 (arm_ldc_qualifiers): ... this. New.
5512 (STC_QUALIFIERS): Define to...
5513 (arm_stc_qualifiers): ... this. New.
5514 * config/arm/arm-protos.h
5515 (arm_coproc_ldc_stc_legitimate_address): New.
5516 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
5517 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
5518 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
5519 stc2, stcl, stc2l): New.
5520 * config/arm/constraints.md (Uz): New.
5521 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
5522 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
5523 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
5524 VUNSPEC_STC2L): New.
5526 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5528 * config/arm/arm.md (<cdp>): New.
5529 * config/arm/arm.c (neon_const_bounds): Rename this ...
5530 (arm_const_bounds): ... this.
5531 (arm_coproc_builtin_available): New.
5532 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
5533 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
5534 (CDP_QUALIFIERS): Define to...
5535 (arm_cdp_qualifiers): ... this. New.
5537 (arm_expand_builtin_args): Add case for 6 arguments.
5538 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
5539 (arm_const_bounds): ... this.
5540 (arm_coproc_builtin_available): New.
5541 * config/arm/arm_acle.h (__arm_cdp): New.
5543 * config/arm/arm_acle_builtins.def (cdp): New.
5545 * config/arm/iterators.md (CDPI,CDP,cdp): New.
5546 * config/arm/neon.md: Rename all 'neon_const_bounds' to
5548 * config/arm/types.md (coproc): New.
5549 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
5550 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
5551 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
5552 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
5554 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5556 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
5557 (UBINOP_QUALIFIERS): New.
5559 (acle_builtin_data): New. Change comment.
5560 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
5561 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
5562 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
5563 arm_acle_builtins.def.
5564 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
5565 (arm_init_acle_builtins): New.
5566 (CRC32_BUILTIN): Remove.
5567 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
5568 crc32cb, crc32ch and crc32cw.
5569 (arm_init_crc32_builtins): Remove.
5570 (arm_init_builtins): Use arm_init_acle_builtins rather
5571 than arm_init_crc32_builtins.
5572 (arm_expand_acle_builtin): New.
5573 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
5574 * config/arm/arm_acle_builtins.def: New.
5576 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5578 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
5579 (arm_builtin_datum): ... this.
5580 (arm_init_neon_builtin): Rename to ...
5581 (arm_init_builtin): ... this. Add a new parameters PREFIX
5582 and USE_SIG_IN_NAME.
5583 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
5584 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
5585 'arm_builtin_datum'.
5586 (arm_init_vfp_builtins): Likewise.
5587 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
5588 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
5589 (arm_expand_neon_args): Rename to ...
5590 (arm_expand_builtin_args): ... this. Rename builtin_arg
5591 enum values and differentiate between ARG_BUILTIN_MEMORY
5592 and ARG_BUILTIN_NEON_MEMORY.
5593 (arm_expand_neon_builtin_1): Rename to ...
5594 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
5595 values, arm_expand_builtin_args and add bool parameter NEON.
5596 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
5597 (arm_expand_vfp_builtin): Likewise.
5598 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
5600 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5603 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
5604 * predict.c (tree_estimate_probability_bb): Reverse direction of
5605 polymorphic call predictor.
5607 2017-01-06 David Malcolm <dmalcolm@redhat.com>
5609 * passes.c (execute_one_pass): Split out pass-skipping logic into...
5610 (determine_pass_name_match): ...this new function and...
5611 (should_skip_pass_p): ...this new function.
5613 2017-01-06 Nathan Sidwell <nathan@acm.org>
5615 * ipa-visibility.c (function_and_variable_visibility): Reformat
5616 comments and long lines. Remove extrneous if.
5617 * symtab.c (symtab_node::make_decl_local): Fix code format.
5618 (symtab_node::set_section_for_node): Fix comment typo.
5620 2017-01-06 Martin Liska <mliska@suse.cz>
5623 * lra-constraints.c: Rename invariant to lra_invariant.
5624 * predict.c (set_even_probabilities): Initialize e to NULL.
5626 2017-01-05 Martin Sebor <msebor@redhat.com>
5628 PR tree-optimization/78910
5629 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
5630 (format_integer): Correct off-by-one error in the handling
5631 of precision with negative numbers in signed conversions..
5633 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
5635 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
5637 2017-01-05 Jakub Jelinek <jakub@redhat.com>
5639 PR tree-optimization/71016
5640 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
5641 factor_out_conditional_conversion. Formatting fix.
5642 (factor_out_conditional_conversion): Add cond_stmt argument.
5643 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
5644 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
5647 2017-01-05 David Malcolm <dmalcolm@redhat.com>
5649 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
5650 read-rtl-function.o, and selftest-rtl.o.
5651 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
5652 (selftest::aarch64_test_loading_full_dump): New function.
5653 (selftest::aarch64_run_selftests): New function.
5654 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
5655 selftest::aarch64_run_selftests.
5656 * config/i386/i386.c
5657 (selftest::ix86_test_loading_dump_fragment_1): New function.
5658 (selftest::ix86_test_loading_call_insn): New function.
5659 (selftest::ix86_test_loading_full_dump): New function.
5660 (selftest::ix86_test_loading_unspec): New function.
5661 (selftest::ix86_run_selftests): Call the new functions.
5662 * emit-rtl.c (maybe_set_max_label_num): New function.
5663 * emit-rtl.h (maybe_set_max_label_num): New decl.
5664 * function.c (instantiate_decls): Guard call to
5665 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
5666 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
5668 * gensupport.c (gen_reader::gen_reader): Pass "false"
5669 for new "compact" param of rtx_reader.
5670 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
5671 rather than an empty string for NULL strings.
5672 * read-md.c: Potentially include config.h rather than bconfig.h.
5673 Wrap include of errors.h with #ifdef GENERATOR_FILE.
5674 (have_error): New global, copied from errors.c.
5675 (md_reader::read_name): Rename to...
5676 (md_reader::read_name_1): ...this, adding "out_loc" param,
5677 and converting "missing name or number" to returning false, rather
5679 (md_reader::read_name): Reimplement in terms of read_name_1.
5680 (md_reader::read_name_or_nil): New function.
5681 (md_reader::read_string): Handle "(nil)" by returning NULL.
5682 (md_reader::md_reader): Add new param "compact".
5683 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
5684 (md_reader::read_file): New method.
5685 * read-md.h (md_reader::md_reader): Add new param "compact".
5686 (md_reader::read_file): New method.
5687 (md_reader::is_compact): New accessor.
5688 (md_reader::read_name): Convert return type from void to file_location.
5689 (md_reader::read_name_or_nil): New decl.
5690 (md_reader::read_name_1): New decl.
5691 (md_reader::m_compact): New field.
5692 (noop_reader::noop_reader): Pass "false" for new "compact" param
5694 (rtx_reader::rtx_reader): Add new "compact" param.
5695 (rtx_reader::read_rtx_operand): Make virtual and convert return
5696 type from void to rtx.
5697 (rtx_reader::read_until): New decl.
5698 (rtx_reader::handle_any_trailing_information): New virtual function.
5699 (rtx_reader::postprocess): New virtual function.
5700 (rtx_reader::finalize_string): New virtual function.
5701 (rtx_reader::m_in_call_function_usage): New field.
5702 (rtx_reader::m_reuse_rtx_by_id): New field.
5703 * read-rtl-function.c: New file.
5704 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
5705 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
5706 (selftest::verify_three_block_rtl_cfg): New decl.
5707 * read-rtl-function.h: New file.
5708 * read-rtl.c: Potentially include config.h rather than bconfig.h.
5709 For host, include function.h, memmodel.h, and emit-rtl.h.
5710 (one_time_initialization): New function.
5711 (struct compact_insn_name): New struct.
5712 (compact_insn_names): New array.
5713 (find_code): Handle insn codes in compact dumps.
5714 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
5715 (bind_subst_iter_and_attr): Likewise.
5716 (add_condition_to_string): Likewise.
5717 (add_condition_to_rtx): Likewise.
5718 (apply_attribute_uses): Likewise.
5719 (add_current_iterators): Likewise.
5720 (apply_iterators): Likewise.
5721 (initialize_iterators): Guard usage of apply_subst_iterator with
5722 #ifdef GENERATOR_FILE.
5723 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
5724 (md_reader::read_mapping): Likewise.
5725 (add_define_attr_for_define_subst): Likewise.
5726 (add_define_subst_attr): Likewise.
5727 (read_subst_mapping): Likewise.
5728 (check_code_iterator): Likewise.
5729 (rtx_reader::read_rtx): Likewise. Move one-time initialization
5731 (one_time_initialization): New function.
5732 (rtx_reader::read_until): New method.
5733 (read_flags): New function.
5734 (parse_reg_note_name): New function.
5735 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
5736 Handle reuse_rtx ids.
5737 Wrap iterator lookup within #ifdef GENERATOR_FILE.
5738 Add parsing support for RTL dumps, mirroring the special-cases in
5739 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
5740 values, and calling handle_any_trailing_information.
5741 (rtx_reader::read_rtx_operand): Convert return type from void
5742 to rtx, returning return_rtx. Handle case 'e'. Call
5743 finalize_string on XSTR and XTMPL fields.
5744 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
5745 "(nil)" values were omitted. Call the postprocess vfunc on the
5747 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
5748 class ctor. Initialize m_in_call_function_usage. Call
5749 one_time_initialization.
5750 * rtl-tests.c (selftest::test_uncond_jump): Call
5751 set_new_first_and_last_insn.
5752 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
5753 * selftest-rtl.c: New file.
5754 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
5755 (selftest::get_insn_by_uid): New decl.
5756 * selftest-run-tests.c (selftest::run_tests): Call
5757 read_rtl_function_c_tests.
5758 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
5759 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
5762 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
5764 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
5765 operands in a special way. Assert that pos+len <= mode precision.
5767 2017-01-05 Jakub Jelinek <jakub@redhat.com>
5769 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
5770 3 argument Alias with unlimited for the negative form.
5771 (fno-vect-cost-model): Removed.
5773 2017-01-05 Martin Liska <mliska@suse.cz>
5775 * hsa-gen.c (gen_hsa_divmod): New function.
5776 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
5778 2017-01-05 Martin Liska <mliska@suse.cz>
5781 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
5784 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5786 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
5787 small constant length operands.
5789 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5791 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
5792 between loop iterations.
5794 2017-01-05 Martin Liska <mliska@suse.cz>
5797 * gimplify.c (gimplify_decl_expr): Compare to
5798 asan_poisoned_variables instread of checking flags.
5799 (gimplify_target_expr): Likewise.
5800 (gimplify_expr): Likewise.
5801 (gimplify_function_tree): Conditionally initialize
5802 asan_poisoned_variables.
5804 2017-01-04 Jeff Law <law@redhat.com>
5806 PR tree-optimizatin/78812
5807 * rtl.h (contains_mem_rtx_p): Prototype.
5808 * ifcvt.c (containts_mem_rtx_p): Move from here to...
5809 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
5810 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
5811 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
5812 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
5814 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5816 * input.c (assert_char_at_range): Default-initialize actual_range.
5818 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5820 * df-scan.c (df_ref_create_structure): Make regno unsigned,
5821 to match the caller.
5823 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5825 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
5826 insns after final jump in test to emit dummy move.
5828 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5830 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
5831 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
5833 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5835 * multiple_target.c (create_dispatcher_calls): Init e_next.
5836 * tree-ssa-loop-split.c (split_loop): Init border.
5837 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
5840 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
5845 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
5846 (altivec_register_operand): Do not return true if the operand
5847 contains a SUBREG mixing SImode and SFmode.
5848 (vsx_register_operand): Likewise.
5849 (vsx_reg_sfsubreg_ok): New predicate.
5850 (vfloat_operand): Do not return true if the operand contains a
5851 SUBREG mixing SImode and SFmode.
5852 (vint_operand): Likewise.
5853 (vlogical_operand): Likewise.
5854 (gpc_reg_operand): Likewise.
5855 (int_reg_operand): Likewise.
5856 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
5857 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
5858 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
5860 (rs6000_emit_move_si_sf_subreg): New helper function.
5861 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
5862 fixup SUBREGs involving SImode and SFmode.
5863 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
5864 numbers for the new peephole2 optimization.
5865 (peephole2 for SFmode unions): New peephole2 to optimize cases in
5866 the GLIBC math library that do AND/IOR/XOR operations on single
5867 precision floating point.
5868 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
5869 target macros to say whether we need to avoid SUBREGs mixing
5871 (TARGET_ALLOW_SF_SUBREG): Likewise.
5872 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
5873 (UNSPEC_SI_FROM_SF): Likewise.
5874 (iorxor): Change spacing.
5875 (and_ior_xor): New iterator for AND, IOR, and XOR.
5876 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
5877 (movdi_from_sf_zero_ext): Likewise.
5878 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
5879 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
5880 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
5881 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
5882 (fms<mode>4): Likewise.
5883 (fnma<mode>4): Likewise.
5884 (fnms<mode>4): Likewise.
5885 (nfma<mode>4): Likewise.
5886 (nfms<mode>4): Likewise.
5888 2017-01-04 Marek Polacek <polacek@redhat.com>
5891 * doc/invoke.texi: Document -Wpointer-compare.
5893 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5895 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
5898 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
5899 descriptions for -gdwarf-5 and emit them as uleb128 instead of
5902 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
5905 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
5906 documentation of the powerpc_popcntb_ok attribute.
5907 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
5908 code to issue warning messages if a requested CPU configuration is
5909 not supported by the binary (assembler and loader) toolchain.
5910 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
5911 made to define a built-in function that has been disabled.
5912 (paired_init_builtins): Add assertion to prevent ICE if attempt is
5913 made to define a built-in function that has been disabled.
5914 (altivec_init_builtins): Add comment explaining why definition
5915 of the DST built-in functions is not preceded by an assertion
5916 check. Add assertions to prevent ICE if attempts are made to
5917 define an altivec predicate or an abs* built-in function that has
5919 (htm_init_builtins): Add comment explaining why definition of the
5920 htm built-in functions is not preceded by an assertion check.
5922 2017-01-04 Jeff Law <law@redhat.com>
5924 PR tree-optimizatin/67955
5925 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
5926 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
5927 the points-to solution does not include pt_null. Use DECL_PT_UID
5930 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
5932 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
5933 Use gen_int_mode instead of gen_lopwart for const_int operands.
5935 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5937 PR tree-optimization/71563
5938 * match.pd: Simplify X << Y into X if Y is known to be 0 or
5939 out of range value - has low bits known to be zero.
5941 2017-01-04 Alan Modra <amodra@gmail.com>
5943 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
5944 * configure: Regenerate.
5945 * config.in: Regenerate.
5947 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5950 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
5951 a substring of the message, but strcmp with the whole message. Ifdef
5952 ENABLE_NLS, translate the message first using dgettext.
5954 2017-01-03 Jeff Law <law@redhat.com>
5956 PR tree-optimizatin/78856
5957 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
5958 (mark_threaded_blocks): Remove code to truncate thread paths that
5959 cross multiple loop headers. Instead invalidate the cached loop
5960 iteration information and handle case of a thread path walking
5961 into an irreducible region.
5963 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5966 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
5967 assertions. Add support for doing the signbit if the IEEE 128-bit
5968 floating point value is in a GPR.
5969 * config/rs6000/rs6000.md (Fsignbit): Delete.
5970 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
5971 Update the length attribute if the value is in a GPR.
5972 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
5973 the sign or zero extension instruction, since the value is always 0/1.
5974 (signbit<mode>2_dm2): Delete using <Fsignbit>.
5977 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
5978 extracting SImode to a GPR register so that we can generate a
5979 store, limit the vector to be in a traditional Altivec register
5980 for the vextuwrx instruction.
5982 2017-01-03 Ian Lance Taylor <iant@google.com>
5984 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
5986 2017-01-03 Martin Sebor <msebor@redhat.com>
5988 PR tree-optimization/78696
5989 * gimple-ssa-sprintf.c (format_floating): Correct handling of
5990 precision. Use MPFR for %f for greater fidelity. Correct handling
5992 (pass_sprintf_length::compute_format_length): Set width and precision
5993 specified by asrerisk to void_node for vararg functions.
5994 (try_substitute_return_value): Adjust dump output.
5996 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
5998 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
6000 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
6002 * doc/invoke.texi (SPARC options): Document -mlra as the default.
6003 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
6004 -mlra/-mno-lra was passed to the compiler.
6006 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
6008 PR rtl-optimization/65618
6009 * emit-rtl.c (try_split): Move initialization of "before" and
6010 "after" to just before the call to emit_insn_after_setloc.
6012 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
6014 * doc/md.texi (Standard Names): Remove reference to Java frontend.
6016 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
6018 * dwarf2out.c (gen_enumeration_type_die): When
6019 -gno-strict-dwarf, add a DW_AT_encoding attribute.
6021 2017-01-03 Jakub Jelinek <jakub@redhat.com>
6023 PR tree-optimization/78965
6024 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
6025 Change first argument from const call_info & to call_info &. For %n
6026 set info.nowrite to false.
6029 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
6030 possibly throwing calls.
6032 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
6033 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
6034 and fns handling, rather than in a separate case SSA_NAME.
6036 2017-01-02 Jeff Law <law@redhat.com>
6038 * config/darwin-driver.c (darwin_driver_init): Const-correctness
6039 fixes for first_period and second_period variables.
6041 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
6044 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
6045 (*insvqi_1): New insn pattern.
6046 (*insvqi_1_mem_rex64): Ditto.
6048 (*insvqi_3): Rename from *insvqi.
6050 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
6052 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
6054 * doc/cfg.texi (Edges): Remove reference to Java.
6055 (Maintaining the CFG): Ditto.
6057 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
6060 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
6061 transparent aliases.
6063 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
6066 * predict.def (PRED_CALL): Update hitrate.
6067 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
6068 * predict.c (tree_estimate_probability_bb): Split CALL predictor
6069 into direct/indirect/polymorphic variants.
6071 2017-01-01 Jakub Jelinek <jakub@redhat.com>
6073 Update copyright years.
6075 * gcc.c (process_command): Update copyright notice dates.
6076 * gcov-dump.c (print_version): Ditto.
6077 * gcov.c (print_version): Ditto.
6078 * gcov-tool.c (print_version): Ditto.
6079 * gengtype.c (create_file): Ditto.
6080 * doc/cpp.texi: Bump @copying's copyright year.
6081 * doc/cppinternals.texi: Ditto.
6082 * doc/gcc.texi: Ditto.
6083 * doc/gccint.texi: Ditto.
6084 * doc/gcov.texi: Ditto.
6085 * doc/install.texi: Ditto.
6086 * doc/invoke.texi: Ditto.
6088 Copyright (C) 2017 Free Software Foundation, Inc.
6090 Copying and distribution of this file, with or without modification,
6091 are permitted in any medium without royalty provided the copyright
6092 notice and this notice are preserved.