1 2017-03-16 Tamar Christina <tamar.christina@arm.com>
3 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
6 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
9 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
10 Use h_con constraint for operand 1.
11 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
12 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
14 2017-03-15 Jeff Law <law@redhat.com>
16 PR tree-optimization/71437
17 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
18 (record_temporary_equivalences): Use it.
20 PR tree-optimization/71437
21 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
22 tree-ssa-scopedtables.
23 (lookup_avail_expr, build_and_record_new_cond): Likewise.
24 (record_conditions, record_cond, vuse_eq): Likewise.
25 (record_edge_info): Adjust to API tweak of record_conditions.
26 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
27 (record_temporary_equivalences, optimize_stmt): Likewise.
28 (eliminate_redundant_computations): Likewise.
29 (record_equivalences_from_stmt): Likewise.
30 * tree-ssa-scopedtables.c: Include options.h and params.h.
31 (vuse_eq): New function, moved from tree-ssa-dom.c
32 (build_and_record_new_cond): Likewise.
33 (record_conditions): Likewise. Accept vector of conditions rather
34 than edge_equivalence structure for first argument.
35 for the first argument.
36 (avail_exprs_stack::lookup_avail_expr): New member function, moved
38 (avail_exprs_stack::record_cond): Likewise.
39 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
41 (avail_exprs_stack): Add new member functions lookup_avail_expr
43 (record_conditions): Declare.
45 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
48 * lra-constraints.c (process_alt_operands): Increase reject for
49 reloading an input/output operand.
51 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
54 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
55 insns to convert from signed/unsigned char/short to IEEE 128-bit
57 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
59 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
62 * config/i386/i386.c (ix86_vector_duplicate_value): Create
63 subreg of inner mode for values already in registers.
65 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
67 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
68 iteration reg is used after the loop.
70 2017-03-14 Martin Sebor <msebor@redhat.com>
72 PR tree-optimization/79800
73 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
74 precision in negative-positive range.
75 (format_floating): Call non-const overload with adjusted precision.
77 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
80 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
83 2017-03-14 Martin Sebor <msebor@redhat.com>
86 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
87 * builtins.def (aligned_alloc): Use it.
90 * Makefile.in (GTFILES): Add calls.c.
91 * calls.c: Include "gt-calls.h".
93 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
95 PR rtl-optimization/79728
96 * regs.h (struct target_regs): New field
97 x_contains_allocatable_regs_of_mode.
98 (contains_allocatable_regs_of_mode): New macro.
99 * reginfo.c (init_reg_sets_1): Initialize it, and change
100 contains_reg_of_mode so it includes global regs as well.
101 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
102 rather than contains_regs_of_mode.
104 2017-03-14 Martin Liska <mliska@suse.cz>
106 * doc/invoke.texi: Document options that can't be combined with
107 -fcheck-pointer-bounds.
109 2017-03-14 Martin Liska <mliska@suse.cz>
112 * doc/invoke.texi (-Wchkp): Document the option.
114 2017-03-14 Martin Liska <mliska@suse.cz>
116 * Makefile.in: Install gcov-dump.
118 2017-03-14 Martin Liska <mliska@suse.cz>
120 * multiple_target.c (expand_target_clones): Bail out for
121 an invalid attribute.
123 2017-03-14 Richard Biener <rguenther@suse.de>
125 * alias.c (struct alias_set_entry): Pack properly.
126 * cfgloop.h (struct loop): Likewise.
127 * cse.c (struct set): Likewise.
128 * ipa-utils.c (struct searchc_env): Likewise.
129 * loop-invariant.c (struct invariant): Likewise.
130 * lra-remat.c (struct cand): Likewise.
131 * recog.c (struct change_t): Likewise.
132 * rtl.h (struct address_info): Likewise.
133 * symbol-summary.h (function_summary): Likewise.
134 * tree-loop-distribution.c (struct partition): Likewise.
135 * tree-object-size.c (struct object_size_info): Likewise.
136 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
137 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
138 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
139 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
140 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
141 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
142 (struct _stmt_vec_info): Likewise.
144 2017-03-14 Martin Liska <mliska@suse.cz>
147 * multiple_target.c (create_dispatcher_calls): Check that
148 a target can create a function dispatcher.
150 2017-03-14 Martin Liska <mliska@suse.cz>
153 * multiple_target.c (expand_target_clones): Drop local.local
154 flag for default implementation.
156 2017-03-14 Richard Biener <rguenther@suse.de>
158 PR tree-optimization/80030
159 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
161 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
163 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
164 gcc_fallthrough() instead of __attribute__((fallthrough));
166 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
168 * doc/gcc.texi: Remove "up" link to (DIR).
169 * doc/gccint.texi: Ditto.
171 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
173 * doc/install.texi (Specific) <avr>: Remove reference to
176 2017-03-13 Jeff Law <law@redhat.com>
178 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
179 attribute rather than comments.
181 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
182 match_scratch operand is highest.
184 2017-03-13 Martin Liska <mliska@suse.cz>
187 * ipa-pure-const.c (warn_function_noreturn): If the declarations
188 is a CHKP clone, use original declaration.
190 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
192 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
193 (arc_conditional_register_usage): Use a different allocation order
194 when optimizing for size.
195 * common/config/arc/arc-common.c (arc_option_optimization_table):
196 Section anchors default on when optimizing for size.
198 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
200 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
202 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
204 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
205 * config/arc/arc.md (cpu_facility): Add cd variant.
206 (*movqi_insn): Add code density variant.
207 (*movhi_insn): Likewise.
208 (*movqi_insn): Likewise.
209 (*addsi3_mixed): Likewise.
210 (subsi3_insn): Likewise.
212 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
214 * config/arc/arc.md (movsi_cond_exec): Update constraint.
216 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
218 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
219 expressions with MINUS and UNARY ops.
221 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
224 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
226 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
227 between vec_select and vector argument.
228 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
229 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
230 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
231 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
232 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
233 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
235 2017-03-13 Richard Biener <rguenther@suse.de>
238 * params.def (vect-max-peeling-for-alignment): Fix typo.
240 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
242 * doc/install.texi (Specific) <mips-*-*>: Remove description of
243 issue that only occurred with binutils below 2.18.
245 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
247 * doc/install.texi (Specific) <cris-axis-elf>: No longer
248 refer to binutils 2.11/2.12 minimum.
250 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
252 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
253 ftp.kernel.org and simplify binutils requirement.
255 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
257 * doc/invoke.texi (Warning Options): Fix spelling of link-time
259 (Optimize Options): Ditto. Also remove redundancy.
261 2017-03-10 David Malcolm <dmalcolm@redhat.com>
264 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
266 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
267 to G_ to avoid double translation.
269 2017-03-10 David Malcolm <dmalcolm@redhat.com>
272 * auto-profile.c (get_combined_location): Convert leading
273 character of diagnostics to lower case and remove trailing period.
274 (read_profile): Likewise for various diagnostics.
275 * config/arm/arm.c (arm_option_override): Remove trailing period
276 from various diagnostics.
277 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
278 (msp430_expand_delay_cycles): Likewise.
280 2017-03-10 David Malcolm <dmalcolm@redhat.com>
283 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
284 full command-line argument, rather than just "str".
285 (aarch64_validate_march): Likewise.
286 (aarch64_validate_mtune): Likewise.
288 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
290 PR rtl-optimization/78911
291 * lra-assigns.c (must_not_spill_p): New function.
294 2017-03-10 Jakub Jelinek <jakub@redhat.com>
296 PR tree-optimization/79981
297 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
298 ATOMIC_COMPARE_EXCHANGE ifn result.
299 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
300 IFN_ATOMIC_COMPARE_EXCHANGE.
302 2017-03-10 David Malcolm <dmalcolm@redhat.com>
305 * opts.c (parse_sanitizer_options): Add missing question mark to
306 "did you mean" message.
308 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
310 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
312 (VMULEUH_UNS): Likewise.
313 (VMULOUB_UNS): Likewise.
314 (VMULOUH_UNS): Likewise.
315 * config/rs6000/rs6000.c (builtin_function_type): Remove
316 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
318 2017-03-10 David Malcolm <dmalcolm@redhat.com>
321 * read-rtl-function.c (function_reader::read_rtx_operand): Update
322 x with result of extra_parsing_for_operand_code_0.
323 (function_reader::extra_parsing_for_operand_code_0): Convert
324 return type from void to rtx, returning x. When reading
325 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
326 larger size containing struct block_symbol.
328 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
330 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
331 -mfloat128-hardware without -m64.
333 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
336 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
337 entries to the case statement that marks unsigned arguments to
338 overloaded functions.
340 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
342 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
343 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
345 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
348 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
349 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
351 2017-03-10 Martin Liska <mliska@suse.cz>
355 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
356 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
359 2017-03-10 Olivier Hainque <hainque@adacore.com>
361 * tree-switch-conversion (array_value_type): Start by resetting
362 candidate type to it's main variant.
364 2017-03-10 Jakub Jelinek <jakub@redhat.com>
366 PR rtl-optimization/79909
367 * combine.c (try_combine): Use simplify_replace_rtx on individual
368 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
369 of the whole CALL_INSN_FUNCTION_USAGE.
371 PR tree-optimization/79972
372 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
373 get_range_info on SSA_NAMEs. Formatting fixes.
375 2017-03-10 Richard Biener <rguenther@suse.de>
376 Jakub Jelinek <jakub@redhat.com>
378 PR tree-optimization/77975
379 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
381 (get_val_for): For constant x return it. Formatting fix.
382 (loop_niter_by_eval): Avoid pointless looping if the next iteration
383 would use the same bases as the current one.
385 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
387 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
388 instead of vec_select for V1TImode.
389 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
391 (VSX_LE_128): Add V1TI to this mode iterator.
392 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
393 (*vsx_le_perm_store_<mode>): Likewise.
394 (pre-reload splitter for VSX stores): Likewise.
395 (post-reload splitter for VSX stores): Likewise.
396 (*vsx_xxpermdi2_le_<mode>): Likewise.
397 (*vsx_lxvd2x2_le_<mode>): Likewise.
398 (*vsx_stxvd2x2_le_<mode>): Likewise.
400 2017-03-09 Michael Eager <eager@eagercon.com>
402 Correct failures with --enable-checking=yes,rtl.
404 * config/microblaze/microblaze.c (microblaze_expand_shift):
405 Replace GET_CODE test with CONST_INT_P and INTVAL test with
407 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
408 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
410 2017-03-09 Richard Biener <rguenther@suse.de>
412 PR tree-optimization/79977
413 * graphite-scop-detection.c (scop_detection::merge_sese):
414 Handle the case of extra exits to blocks dominating the entry.
416 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
418 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
421 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
423 PR rtl-optimization/79949
424 * lra-constraints.c (process_alt_operands): Check memory when
425 trying to predict a cycle. Print about the overall increase.
427 2017-03-09 Richard Biener <rguenther@suse.de>
430 * gimple-expr.c (useless_type_conversion_p): Preserve
431 TYPE_SATURATING for fixed-point types.
433 2017-03-09 Richard Biener <rguenther@suse.de>
436 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
437 alignment of BLKmode params.
439 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
442 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
443 (VALL_NO_V2Q): Likewise.
445 * config/aarch64/aarch64-simd.md
446 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
448 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
449 VALL_NO_V2Q mode iterator.
450 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
452 2017-03-09 Martin Liska <mliska@suse.cz>
454 PR tree-optimization/79631
455 * tree-chkp-opt.c (chkp_is_constant_addr): Call
456 tree_int_cst_sign_bit just for INTEGER constants.
458 2017-03-09 Martin Liska <mliska@suse.cz>
462 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
465 2017-03-09 Marek Polacek <polacek@redhat.com>
468 * tree.c (inchash::add_expr): Handle TREE_VEC.
470 2017-03-09 Martin Liska <mliska@suse.cz>
473 (chkp_narrow_size_and_offset): New function.
474 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
475 (void chkp_parse_bit_field_ref): New function.
476 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
477 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
479 2017-03-09 Martin Liska <mliska@suse.cz>
482 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
483 (chkp_find_bounds_1): Remove gcc_unreachable.
485 2017-03-09 Jakub Jelinek <jakub@redhat.com>
488 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
489 BUILT_IN_SYNC*, determine the access type from the size suffix and
490 always build a MEM_REF with that type. Handle forgotten
491 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
494 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
495 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
496 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
497 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
498 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
499 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
500 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
501 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
502 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
503 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
504 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
505 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
506 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
507 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
508 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
509 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
510 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
511 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
512 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
513 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
514 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
515 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
516 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
517 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
518 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
519 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
520 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
521 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
522 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
523 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
524 definitions outside of __OPTIMIZE__ guarded section.
527 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
528 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
529 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
530 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
533 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
535 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
536 ("vfenez<mode>"): Add missing constraints.
538 2017-03-08 Martin Sebor <msebor@redhat.com>
541 * config/nds32/nds32.c (nds32_option_override):
543 2017-03-08 Jakub Jelinek <jakub@redhat.com>
546 * gimplify.c (gimplify_omp_for): Replace index var in outer
547 taskloop statement with an artificial variable and add
548 OMP_CLAUSE_PRIVATE clause for it.
550 2017-03-08 Richard Biener <rguenther@suse.de>
552 PR tree-optimization/79955
553 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
554 for accesses that are completely outside of the variable.
556 2017-03-08 Andrew Haley <aph@redhat.com>
558 PR tree-optimization/79943
559 * tree-ssa-loop-split.c (compute_new_first_bound): When
560 calculating the new upper bound, (END-BEG) should be added, not
563 2017-03-08 Jakub Jelinek <jakub@redhat.com>
565 * config/avr/avr.md (setmemhi): Make sure match_dup
566 operand number comes before match_scratch.
568 2017-03-08 Richard Biener <rguenther@suse.de>
570 PR tree-optimization/79920
571 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
572 with ncopies == 1 to ...
573 (vect_transform_slp_perm_load): ... here. Properly compute
574 all element loads by iterating VF times over the group. Do
575 not handle ncopies (computed in a broken way) in
576 vect_create_mask_and_perm.
578 2017-03-08 Jakub Jelinek <jakub@redhat.com>
581 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
582 is a uniform vector, use uniform_vector_p return value instead of
583 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
585 2017-03-07 Marek Polacek <polacek@redhat.com>
588 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
589 (alloca_call_type): Likewise.
591 2017-03-07 Martin Liska <mliska@suse.cz>
593 * gcov.c (process_args): Put comment to correct location.
595 2017-03-07 Martin Liska <mliska@suse.cz>
598 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
599 Use array_at_struct_end_p instead of DECL_CHAIN (field).
600 (chkp_narrow_bounds_for_field): Likewise.
601 (chkp_parse_array_and_component_ref): Pass one more argument to
604 2017-03-07 Richard Biener <rguenther@suse.de>
606 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
609 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
611 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
612 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
614 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
617 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
618 to end of description.
619 (PARAM_MAX_STORES_TO_MERGE): Likewise.
621 2017-03-07 Jakub Jelinek <jakub@redhat.com>
623 PR rtl-optimization/79901
624 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
626 (*avx512f_<code><mode>3<mask_name>): ... this.
627 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
628 iterator instead of VI8_AVX2_AVX512BW.
630 PR rtl-optimization/79901
631 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
632 min/max expander, expand it using expand_vec_cond_expr.
635 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
638 2017-03-06 Jakub Jelinek <jakub@redhat.com>
641 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
642 to void * for PCH reasons.
643 * dwarf2out.c (output_loc_operands, output_die): Cast
644 v.val_vec.array to unsigned char *.
646 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
649 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
652 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
654 PR rtl-optimization/79571
655 * lra-constraints.c (process_alt_operands): Calculate static
656 reject and subtract it from overall when only addresses will be
659 2017-03-06 Julia Koval <julia.koval@intel.com>
662 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
663 incoming stack boundary to 128 for 64-bit targets.
665 2017-03-06 Richard Biener <rguenther@suse.de>
667 PR tree-optimization/79894
668 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
669 to NULL after folding it.
671 2017-03-06 Richard Biener <rguenther@suse.de>
673 PR tree-optimization/79824
674 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
675 check disabling peeling for gaps.
677 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
679 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
680 attributes): Document gettimeofday.
682 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
684 * config/s390/s390.c (s390_option_override_internal): Set
685 PARAM_MIN_VECT_LOOP_BOUND
687 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
689 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
690 * config/s390/s390.md: Likewise.
692 2017-03-06 Jakub Jelinek <jakub@redhat.com>
695 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
696 (<avx2_avx512>_perm<mode>): Rename to ...
697 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
699 (<avx512>_perm<mode>_mask): Rename to ...
700 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
702 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
703 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
704 instead of VI8F_256_512.
705 (avx512f_perm<mode>): New define_expand.
706 (avx512f_perm<mode>_mask): Likewise.
707 (avx512f_perm<mode>_1<mask_name>): New define_insn.
708 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
710 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
712 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
713 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
715 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
717 2017-03-06 Martin Liska <mliska@suse.cz>
720 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
721 when having a SSA NAME w/o VAR_DECL assigned to it.
723 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
725 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
726 msa_dpsub_<su>_d): Fix MODE for vec_select.
728 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
730 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
732 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
734 2017-03-06 Richard Biener <rguenther@suse.de>
736 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
737 * plugin.c (register_plugin_info): Likewise.
738 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
740 2017-03-05 Jakub Jelinek <jakub@redhat.com>
742 * config/i386/sse.md (sse_storehps, sse_storelps,
743 avx_<castmode><avxsizesuffix>_<castmode>,
744 avx512f_<castmode><avxsizesuffix>_<castmode>,
745 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
746 in condition that at least one operand is not a MEM.
748 2017-03-03 Jakub Jelinek <jakub@redhat.com>
751 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
752 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
754 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
755 gimple_call_nothrow_p flag based on whether original builtin can throw.
756 If it can, emit following stmts on the fallthrough edge.
757 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
758 don't create new bb if inserting just debug stmts on the edge, try to
759 insert them on the fallthru bb or just reset debug stmts.
761 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
764 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
765 restore recog_data (including the operand rtxes inside it) around
766 the call to get_insn_template.
768 2017-03-03 Martin Sebor <msebor@redhat.com>
770 PR tree-optimization/79699
771 * context.c (context::~context): Free MPFR caches to avoid
772 a memory leak on program exit.
774 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
776 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
777 Use wide_int::ulow () instead of .elt (0).
779 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
781 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
782 (*pushxf): Limit oF constraint to 32bit targets and add oC
783 constraint for 64bit targets.
784 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
785 (*pushdf): Change rmF constraint to rmC.
787 2017-03-03 Martin Liska <mliska@suse.cz>
789 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
790 Remove unused variable.
792 2017-03-03 Jakub Jelinek <jakub@redhat.com>
795 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
796 is a memory operand, increase num_memory.
797 (ix86_expand_args_builtin): Likewise.
799 2017-03-03 Jan Hubicka <jh@suse.cz>
802 * ipa-devirt.c (maybe_record_node): Properly handle
803 __cxa_pure_virtual visibility.
805 2017-03-03 Martin Liska <mliska@suse.cz>
807 PR tree-optimization/79803
808 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
810 (pass_loop_prefetch::execute): Disabled optimization if an
811 assumption about L1 cache size is not met.
813 2017-03-03 Martin Liska <mliska@suse.cz>
815 PR rtl-optimization/79574
816 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
817 (hash_scan_set): Likewise.
818 (dump_hash_table): Likewise.
819 (hoist_code): Likewise.
821 2017-03-03 Richard Biener <rguenther@suse.de>
823 * fixed-value.c (fixed_from_string): Restore use of elt (1)
824 in place of uhigh ().
825 (fixed_convert_from_real): Likewise.
827 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
830 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
832 2017-03-03 Richard Biener <rguenther@suse.de>
835 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
836 TYPE_OVERFLOW_UNDEFINED check.
838 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
840 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
842 (vector_ae_<mode>_p): Likewise.
843 (vector_nez_<mode>_p): Likewise.
844 (vector_ne_v2di_p): Likewise.
845 (vector_ae_v2di_p): Likewise.
846 (vector_ne_<mode>_p): Likewise.
847 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
849 (vsx_tsqrt<mode>2_fe): Likewise.
851 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
854 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
856 2017-03-02 Jakub Jelinek <jakub@redhat.com>
858 PR rtl-optimization/79780
859 * cprop.c (one_cprop_pass): When second and further conditional trap
860 in a single basic block is turned into an unconditional trap, turn it
861 into a deleted note to avoid RTL verification failures.
863 2017-03-02 Richard Biener <rguenther@suse.de>
865 * fold-const.c (const_binop): Use ulow () instead of elt (0).
867 2017-03-02 Richard Biener <rguenther@suse.de>
869 PR tree-optimization/79345
871 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
872 param and abort the walk, returning -1 if it is hit.
873 (walk_aliased_vdefs): Take a limit param and pass it on.
874 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
875 defaulting to 0 and return a signed int.
876 * tree-ssa-uninit.c (struct check_defs_data): New struct.
877 (check_defs): New helper.
878 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
879 about uninitialized memory.
880 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
881 bogus uninitialized warning.
882 (fixed_convert_from_real): Likewise.
884 2017-03-02 Bin Cheng <bin.cheng@arm.com>
886 PR tree-optimization/66768
887 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
888 iv_use if base object can't be determined.
890 2017-03-02 Jakub Jelinek <jakub@redhat.com>
892 PR tree-optimization/79345
893 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
894 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
895 (get_pattern_stats): Initialize it.
896 * genemit.c (gen_expand): Verify match_scratch numbers come after
897 match_operand/match_dup numbers.
898 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
899 match_scratch numbers.
900 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
902 * config/s390/s390.md (trunctdsd2): Likewise.
904 2017-03-02 Richard Biener <rguenther@suse.de>
906 * wide-int.h (wide_int_storage::operator=): Implement in terms
909 2017-03-02 Richard Biener <rguenther@suse.de>
911 PR tree-optimization/79777
912 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
913 the to insert expression to sth existing.
915 2017-03-01 Martin Sebor <msebor@redhat.com>
918 * gimple-ssa-sprintf.c
919 (directive::known_width_and_precision): New function.
920 (format_integer): Use it.
921 (get_mpfr_format_length): Consider the full range of precision
922 when computing %g output with the # flag. Set the likely byte
923 count to 3 rather than 1 when precision is indeterminate.
924 (format_floating): Correct the lower bound of precision.
926 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
928 * doc/invoke.texi: Document default code model for 64-bit Linux.
930 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
933 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
934 udiv rather than div since input pattern is unsigned.
936 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
938 * config/i386/i386.c (print_reg): Warn for values of
939 unsupported size in integer register.
941 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
944 * config/rs6000/predicates.md (current_file_function_operand): Do
945 not allow self calls to be local if the function is replaceable.
947 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
950 * config/rs6000/altivec.h (vec_ctz and others): Change the
951 preprocessor macro that controls conditional compilation from
952 _ARCH_PWR9 to __POWER9_VECTOR__.
953 (vec_all_ne): Change parameterization of __altivec_scalar_pred
954 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
955 control (instead of _ARCH_PWR9 control) so that template
956 definition uses power9-specific function.
957 (vec_any_eq): Likewise.
958 (vec_all_ne): Change macro definition to use a power9-specific
959 expansion under #ifdef __POWER9_VECTOR__ control (instead of
961 (vec_any_eq) Likewise.
962 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
963 expansion for CMPNEF to remove support for xvcmpnesp instruction.
964 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
965 support for xvcmpnedp instruction.
966 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
967 macro expansion so that Power9 implementation of vec_all_ne does
968 not use the AltiVec predicate framework.
969 (VCMPNEH_P): Likewise.
970 (VCMPNEW_P): Likewise.
971 (VCMPNED_P): Likewise.
972 (VCMPNEFP_P): Likewise.
973 (VCMPNEDP_P): Likewise.
974 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
975 implementation of vec_any_eq to not use AltiVec predicate
977 (VCMPAEH_P): Likewise.
978 (VCMPAEW_P): Likewise.
979 (VCMPAED_P): Likewise.
980 (VCMPAEFP_P): Likewise.
981 (VCMPAEDP_P): Likewise.
982 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
983 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
984 not use the AltiVec predicate framework.
985 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
986 of vec_any_eq to not use AltiVec predicate framework.
987 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
988 support for predefined __POWER9_VECTOR__ macro to indicate that
989 Power9 instruction selection is enabled.
990 (altivec_overloaded_builtins): Remove extraneous
991 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
992 function argument types RS6000_BTI_bool_V16QI and
993 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
994 entry for overloaded function argument types RS6000_BTI_bool_V4SI
995 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
996 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
997 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
998 Power9 for implementations of vec_cmpne. Change the signature for
999 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
1000 (representing vec_all_ne) to remove the previously described first
1001 argument of type RS6000_BTI_INTSI, as this was an artifact of
1002 reliance on the AltiVec predicate framework, which is no longer
1003 used in the implementation of these functions. Add
1004 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
1005 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
1006 since, unlike the AltiVec predicate framework implementation, we
1007 do not share function descriptors between vec_alle and vec_anyeq.
1008 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
1009 set of modes that receive special treatment even when
1010 TARGET_P9_VECTOR is true. The special treatment emits code that
1011 does not depend on Power9 instructions.
1012 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
1013 define_expand to not rely on AltiVec predicate framework.
1014 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
1016 (vector_ne_v2di_p): Change this define_expand to not rely on
1017 AltiVec predicate framework.
1018 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
1020 (vector_ne_<mode>_p): Change this define_expand to not rely on
1021 AltiVec predicate framework.
1022 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
1024 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
1025 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
1026 define_insn pattern.
1027 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
1028 define_insn pattern because the xvcmpne<VSs>. instruction is not
1030 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
1031 instruction is not supported.
1033 2017-03-01 Jakub Jelinek <jakub@redhat.com>
1035 * config/nvptx/nvptx.c: Include intl.h.
1037 2017-03-01 Martin Jambor <mjambor@suse.cz>
1040 * ipa-prop.h (ipa_bits): Removed field known.
1041 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
1042 to pointers. Adjusted their comments to warn about their sharing.
1043 (ipcp_transformation_summary): Change bits to a vector of pointers.
1044 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
1045 (ipa_get_ipa_bits_for_value): Declare.
1046 * tree-vrp.h (value_range): Mark as GTY((for_user)).
1047 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
1048 (ipa_bits_hash_table): Likewise.
1049 (ipa_vr_ggc_hash_traits): Likewise.
1050 (ipa_vr_hash_table): Likewise.
1051 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
1052 being pointers and vr_known being removed.
1053 (ipa_set_jf_unknown): Likewise.
1054 (ipa_get_ipa_bits_for_value): New function.
1055 (ipa_set_jfunc_bits): Likewise.
1056 (ipa_get_value_range): New overloaded functions.
1057 (ipa_set_jfunc_vr): Likewise.
1058 (ipa_compute_jump_functions_for_edge): Use the above functions to
1059 construct bits and vr parts of jump functions.
1060 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
1061 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
1063 (ipcp_grow_transformations_if_necessary): Also allocate
1064 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
1066 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
1067 them. Fix too long lines.
1068 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
1069 vr_known being removed.
1070 (ipa_read_jump_function): Use new setter functions to construct bits
1071 and vr parts of jump functions or set them to NULL.
1072 (write_ipcp_transformation_info): Adjust for bits being pointers.
1073 (read_ipcp_transformation_info): Likewise.
1074 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
1076 Include gt-ipa-prop.h.
1077 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
1079 (ipcp_store_bits_results): Likewise.
1080 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
1081 Do not write to existing jump functions but use a temporary instead.
1083 2017-03-01 Jakub Jelinek <jakub@redhat.com>
1086 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
1087 attempt to use its first operand as BIT_FIELD_REF base.
1089 2017-03-01 Richard Biener <rguenther@suse.de>
1092 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
1093 interpolating formula in wrapping arithmetic.
1094 (chrec_apply): Convert chrec_evaluate return value to wanted type.
1096 2017-03-01 Jakub Jelinek <jakub@redhat.com>
1098 PR tree-optimization/79734
1099 * tree-vect-generic.c (expand_vector_condition): Optimize
1100 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
1101 Handle VEC_COND_EXPR where comparison has different inner width from
1104 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
1106 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
1107 markup, and similar issues. Remove @opindex entries for things
1108 that aren't options. Add missing -mmpy-option entries.
1110 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1112 PR tree-optimization/79737
1113 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
1114 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
1115 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
1116 instead of byte_size. Formatting fix.
1117 (shift_bytes_in_array_right): Formatting fix.
1119 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
1122 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
1123 condition on optimize for the leaf function test.
1125 2017-02-28 Martin Liska <mliska@suse.cz>
1128 * read-rtl-function.c (function_reader::handle_unknown_directive):
1129 Bail out when one uses -flto.
1131 2017-02-28 Martin Liska <mliska@suse.cz>
1133 * common.opt: Replace space with tabular for options of <number>
1135 * config/i386/i386.opt: Show <number> value for
1136 -mlarge-data-threshold.
1137 * opts.c (print_filtered_help): Do not display number in hexadecimal
1140 2017-02-28 Martin Liska <mliska@suse.cz>
1142 * common.opt: Fix --help=option -Q for options which are of
1145 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
1147 * config/i386/i386.c (print_reg): Error out for values
1148 of 8-bit size in invalid integer register.
1150 2017-02-28 Martin Sebor <msebor@redhat.com>
1152 PR tree-optimization/79691
1153 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
1155 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1158 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
1159 gcc_unreachable with output_operand_lossage.
1161 2017-02-28 Richard Biener <rguenther@suse.de>
1163 PR tree-optimization/79740
1164 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
1166 (visit_nary_op): Insert the nary into the hashtable if we
1167 pattern-matched sth.
1168 * tree-ssa-pre.c (eliminate_insert): Robustify.
1170 2017-02-28 Richard Biener <rguenther@suse.de>
1173 * fold-const.c (decode_field_reference): Reject out-of-bound
1176 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1178 * config/i386/i386.c: Include intl.h.
1179 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
1180 instead of just cond ? "..." : "...".
1181 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
1182 * coverage.c (read_counts_file): Likewise.
1183 * omp-offload.c: Include intl.h.
1184 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
1185 of just cond ? "..." : "...".
1186 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
1187 of just cond ? "..." : "...".
1189 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
1192 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
1194 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
1195 'tune for' CPU name.
1196 * config/arm/arm-cpu-data.h: Regenerated.
1198 2017-02-28 Richard Biener <rguenther@suse.de>
1200 PR tree-optimization/79732
1201 * tree-inline.c (expand_call_inline): Do not shadow var.
1203 2017-02-28 Richard Biener <rguenther@suse.de>
1205 PR tree-optimization/79723
1206 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
1207 address-space properly.
1209 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
1211 * doc/optinfo.texi (Optimization groups): Fix option used for
1213 * doc/invoke.texi (-fopt-info): Document "omp".
1214 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
1215 (OPTGROUP_ALL): Add OPTGROUP_OMP.
1216 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
1217 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
1218 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
1220 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
1222 * dumpfile.c (optgroup_options): Instead of "openmp", associate
1223 OPTGROUP_OMP with "omp".
1225 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
1228 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
1229 for arithmetic shift of unsigned V2DI.
1231 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
1233 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
1234 arc/linux.h headers.
1235 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
1236 (LINK_SPEC): Likewise.
1237 (ARC_TLS_EXTRA_START_SPEC): Likewise.
1238 (EXTRA_SPECS): Likewise.
1239 (STARTFILE_SPEC): Likewise.
1240 (ENDFILE_SPEC): Likewise.
1241 (LIB_SPEC): Likewise.
1242 (TARGET_SDATA_DEFAULT): Likewise.
1243 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
1244 (MULTILIB_DEFAULTS): Likewise.
1245 (DWARF2_UNWIND_INFO): Likewise.
1246 * config/arc/big.h: New file.
1247 * config/arc/elf.h: Likewise.
1248 * config/arc/linux.h: Likewise.
1249 * config/arc/t-uClibc: Remove.
1251 2017-02-27 Bin Cheng <bin.cheng@arm.com>
1253 PR tree-optimization/77536
1254 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
1255 (tree_transform_and_unroll_loop): Use above function to compute the
1256 estimated niter of unrolled loop and use it when scaling profile.
1257 Also use count info rather than frequency if it's non-zero.
1258 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
1259 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
1260 (vect_transform_loop): Call above function.
1262 2017-02-27 Richard Biener <rguenther@suse.de>
1264 PR tree-optimization/45397
1265 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
1266 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
1267 (visit_nary_op): Add pattern matching for CSEing sign-changed
1268 or truncated operations with wider ones.
1270 2017-02-27 Richard Biener <rguenther@suse.de>
1272 PR tree-optimization/79690
1273 * tree-vect-stmts.c (vectorizable_store): Use vector type
1274 built from the DR with address-space.
1276 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
1278 * doc/invoke.texi (Optimize Options): Refine the description
1279 of asan-use-after-return.
1281 2017-02-25 Alan Modra <amodra@gmail.com>
1283 PR rtl-optimization/79584
1284 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
1285 base, not ad->base_term, the reg within base. Remove assertion
1286 that ad->base == ad->base_term. Replace gen_int_mode using
1287 bogus mode with const0_rtx.
1289 2017-02-25 Jakub Jelinek <jakub@redhat.com>
1292 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
1293 FMA_EXPR like tcc_binary or tcc_unary.
1295 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
1298 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
1300 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
1301 (output_loc_operands): Handle DW_OP_call_ref and
1302 DW_OP_GNU_variable_value.
1303 (struct variable_value_struct): New type.
1304 (struct variable_value_hasher): Likewise.
1305 (variable_value_hash): New variable.
1306 (string_types): Remove.
1307 (copy_loc_descr): New function.
1308 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
1309 (prepend_loc_descr_to_each): New function.
1310 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
1311 instead of add_loc_descr_to_each if the first argument is single
1312 location list and the second has multiple.
1313 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
1314 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
1315 when looking for variable value which doesn't have other location info.
1316 (loc_list_from_tree): Formatting fix.
1317 (gen_array_type_die): Simplify DW_AT_string_length handling.
1318 (adjust_string_types): Remove.
1319 (gen_subprogram_die): Don't call adjust_string_types nor test/set
1320 string_types. Call resolve_variable_values.
1321 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
1322 (resolve_addr_in_expr): Likewise. Add A argument.
1323 (copy_deref_exprloc): Remove deref argument. Adjust for the
1324 original expression being DW_OP_GNU_variable_value with optionally
1325 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
1326 optionally after it.
1327 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
1328 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
1329 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
1330 (variable_value_hasher::hash, variable_value_hasher::equal): New
1332 (resolve_variable_value_in_expr, resolve_variable_value,
1333 resolve_variable_values, note_variable_value_in_expr,
1334 note_variable_value): New functions.
1335 (dwarf2out_early_finish): Call note_variable_value on all toplevel
1338 2017-02-24 Jakub Jelinek <jakub@redhat.com>
1341 * opts.h (handle_generated_option): Add GENERATED_P argument.
1342 * opts-common.c (handle_option): Adjust function comment.
1343 (handle_generated_option): Add GENERATED_P argument, pass it to
1345 (control_warning_option): Pass false to handle_generated_option
1347 * opts.c (maybe_default_option): Pass true to handle_generated_option
1349 * optc-gen.awk: Likewise.
1351 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1353 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
1354 a REG, look at the REG it is a SUBREG of.
1355 (splitter for cmpeqsi_t): Ditto.
1357 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1359 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
1360 the special USEs with the pattern of the insn, not the insn itself.
1362 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
1365 * doc/invoke.texi: Document -mload-store-pairs.
1367 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1368 Sandra Loosemore <sandra@codesourcery.com>
1370 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
1371 argument isn't a CONST_INT.
1372 (nios2_alternate_compare_const): Assert op is a CONST_INT.
1373 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
1374 (nios2_validate_compare): Bypass alternate compare logic if *op2
1376 (ldstwm_operation_p): Return false if first_base is not a REG or
1377 if first_offset is not a CONST_INT.
1379 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1381 * config/cris/cris.md: Use correct operand in a define_peephole2.
1383 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1385 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
1387 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1389 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
1390 this_insn if it is an INSN or JUMP_INSN.
1391 (force_offsettable): Look at base, not at addr.
1392 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
1393 on things that aren't necessarily CONST_INTs.
1395 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
1397 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
1398 -mfpmath=sse is the default also for x86-32 targets with SSE2
1399 instruction set when @option{-ffast-math} is enabled
1401 2017-02-24 Jeff Law <law@redhat.com>
1403 PR rtl-optimizatoin/79286
1404 * ira.c (update_equiv_regs): Drop may_trap_p exception to
1407 2017-02-24 Richard Biener <rguenther@suse.de>
1409 PR tree-optimization/79389
1410 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
1413 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
1415 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
1416 function comment to reflect reality.
1417 (loop_exits_before_overflow): Fix typo in function description.
1419 2017-02-24 Richard Biener <rguenther@suse.de>
1421 PR tree-optimization/79389
1422 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
1423 properly that a threading opportunity exists. Detect conditional
1424 copy/constant propagation opportunities.
1426 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
1428 * config/visium/visium.md (type): Add trap.
1429 (b): New mode attribute.
1430 (*btst): Rename into...
1431 (*btst<mode>): ...this and adjust.
1432 (*cbranchsi4_btst_insn): Rename into...
1433 (*cbranch<mode>4_btst_insn): ...this and adjust.
1434 (trap): New define_insn.
1436 2017-02-23 Jakub Jelinek <jakub@redhat.com>
1438 PR tree-optimization/79389
1439 * ifcvt.c (struct noce_if_info): Add rev_cond field.
1440 (noce_reversed_cond_code): New function.
1441 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
1442 reversed_comparison_code. Formatting fix.
1443 (noce_try_store_flag): Test rev_cond != NULL in addition to
1444 reversed_comparison_code.
1445 (noce_try_store_flag_constants): Likewise.
1446 (noce_try_store_flag_mask): Likewise.
1447 (noce_try_addcc): Use rev_cond if non-NULL instead of
1448 reversed_comparison_code.
1449 (noce_try_cmove_arith): Likewise. Formatting fixes.
1450 (noce_try_minmax, noce_try_abs): Clear rev_cond.
1451 (noce_find_if_block): Initialize rev_cond.
1452 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
1453 instead of false as last argument never attempt to reverse it
1456 2017-02-23 Bin Cheng <bin.cheng@arm.com>
1458 PR tree-optimization/79663
1459 * tree-predcom.c (combine_chains): Process refs in reverse order
1460 only for ZERO length chains, and add explaining comment.
1462 2017-02-23 Jeff Law <law@redhat.com>
1464 PR tree-optimization/79578
1465 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
1466 in call to operand_equal_p.
1468 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
1471 * config/i386/cpuid.h: Fix another undefined behavior.
1473 2017-02-23 Richard Biener <rguenther@suse.de>
1475 PR tree-optimization/79683
1476 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
1477 vector types for data-refs.
1479 2017-02-23 Martin Liska <mliska@suse.cz>
1481 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
1483 2017-02-23 Jakub Jelinek <jakub@redhat.com>
1486 * internal-fn.c (get_range_pos_neg): Moved to ...
1487 * tree.c (get_range_pos_neg): ... here. No longer static.
1488 * tree.h (get_range_pos_neg): New prototype.
1489 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
1490 are known to be in between 0 and signed maximum inclusive, try to
1491 expand both unsigned and signed divmod and use the cheaper one from
1494 2017-02-22 Jeff Law <law@redhat.com>
1496 PR tree-optimization/79578
1497 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
1498 to compare base operands.
1500 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
1503 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
1504 gpc_reg_operand instead of fpr_reg_operand.
1506 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
1508 * config/mips/mips.c (mips_return_in_memory): Force FP
1509 vector types to be returned in memory for o32 ABI.
1511 2017-02-22 Jakub Jelinek <jakub@redhat.com>
1513 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
1514 instead of DW_TAG_member for static data member declarations and don't
1515 set no_linkage_name for static inline data members.
1516 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
1519 2017-02-22 Martin Liska <mliska@suse.cz>
1521 * doc/invoke.texi: Replace inequality signs with square brackets
1524 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1526 PR tree-optimization/68644
1527 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
1529 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
1532 * lra-constraints.c (simplify_operand_subreg): Handle
1533 WORD_REGISTER_OPERATIONS targets.
1535 2017-02-22 Jakub Jelinek <jakub@redhat.com>
1538 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
1539 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
1540 elimination by swapping fld*.
1542 2017-02-22 Richard Biener <rguenther@suse.de>
1544 PR tree-optimization/79673
1545 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
1546 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
1547 irrelevant address-space qualifiers and avoiding a
1548 ADDR_SPACE_CONVERT_EXPR from fold_convert.
1550 2017-02-22 Richard Biener <rguenther@suse.de>
1552 PR tree-optimization/79666
1553 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
1554 to not symbolically negate if that may introduce undefined
1557 2017-02-22 Martin Liska <mliska@suse.cz>
1560 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
1561 * data-streamer-out.c (streamer_write_gcov_count_stream):
1563 * value-prof.c (stream_out_histogram_value): Make assert more
1564 precise based on type of counter.
1566 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
1569 * config/i386/i386.md (standard_x87sse_constant_load splitter):
1570 Use nonimmediate_operand instead of memory_operand for operand 1.
1571 (float-extend standard_x87sse_constant_load splitter): Ditto.
1573 2017-02-21 Jeff Law <law@redhat.com>
1575 PR tree-optimization/79621
1576 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
1577 blocks with edges to themselves.
1579 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1582 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
1583 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
1584 Use gimple_call_builtin_p.
1587 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
1588 on temporarily removed DEBUG_INSNs.
1590 PR tree-optimization/79649
1591 * tree-loop-distribution.c (classify_partition): Give up on
1592 non-generic address space loads/stores.
1594 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
1596 * doc/loop.texi (Loop manipulation): Remove nonexistent
1597 tree_ssa_loop_version from the documentation.
1598 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
1600 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1603 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
1604 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
1605 * config/rs6000/rs6000.c: Include except.h.
1606 (rs6000_expand_split_stack_prologue): Call
1607 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
1609 2017-02-21 Martin Jambor <mjambor@suse.cz>
1612 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
1615 2017-02-21 Martin Jambor <mjambor@suse.cz>
1617 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
1618 for backward compatibility only.
1619 * doc/invoke.texi (Option Summary): Remove all references to
1622 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
1626 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1628 * lra-constraints.c (curr_insn_transform): Handle
1629 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
1631 2017-02-21 Martin Liska <mliska@suse.cz>
1633 * config/i386/i386.opt: Replace -masm-dialect with -masm.
1635 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
1637 PR translation/79638
1638 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
1640 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
1643 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
1644 (arm_function_ok_for_sibcall): Return false for an indirect call by
1645 descriptor if all the argument registers are used.
1646 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
1647 alignment of the function.
1649 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1651 PR tree-optimization/61441
1652 * simplify-rtx.c (simplify_const_unary_operation): For
1653 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
1654 the sNaN unmodified.
1656 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
1658 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
1659 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
1660 instead of SYSTEM_HEADER_DIR.
1662 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
1663 Martin Liška <mliska@suse.cz>
1665 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
1666 Fix typos and grammar, use active voice, and clarify.
1668 2017-02-20 Marek Polacek <polacek@redhat.com>
1671 * gimplify.c (gimplify_expr): Handle unused *&&L;.
1674 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
1676 2017-02-20 Jakub Jelinek <jakub@redhat.com>
1679 * config/i386/i386.c (ix86_expand_builtin): Handle
1680 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
1681 ix86_builtins_isa[fcode].isa as a requirement of those
1682 flags and any other flag in the bitmask.
1683 (ix86_init_mmx_sse_builtins): Use 0 instead of
1684 ~OPTION_MASK_ISA_64BIT as mask.
1685 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
1686 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
1687 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
1688 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
1690 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1693 * lra-constraints.c (split_reg): Check requested split mode
1694 is supported by the register.
1696 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1698 * lra-constraints.c (simplify_operand_subreg): Remove early
1701 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1704 * lra-constraints.c (curr_insn_transform): Tighten condition
1705 for converting SUBREG reloads from OP_OUT to OP_INOUT.
1707 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1710 * lra-constraints.c (curr_insn_transform): Handle
1711 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
1713 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
1716 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
1718 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
1720 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
1723 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
1726 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1728 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
1731 2017-02-18 Jakub Jelinek <jakub@redhat.com>
1733 * final.c (last_columnnum, override_columnnum): New variables.
1734 (final_start_function): Set last_columnnum, pass it to begin_prologue
1735 hook and pass 0 to dwarf2out_begin_prologue.
1736 (final_scan_insn): Update override_columnnum. Pass last_columnnum
1737 to source_line debug hook.
1738 (notice_source_line): Compute last_columnnum and for debug_column_info
1739 return true on column changes.
1740 * debug.h (struct gcc_debug_hooks): Add column argument to
1741 source_line and begin_prologue hooks.
1742 (debug_nothing_int_charstar_int_bool): Remove prototype.
1743 (debug_nothing_int_int_charstar,
1744 debug_nothing_int_int_charstar_int_bool): New prototypes.
1745 (dwarf2out_begin_prologue): Add column argument.
1746 * debug.c (do_nothing_debug_hooks): Adjust source_line and
1747 begin_prologue hooks.
1748 (debug_nothing_int_charstar_int_bool): Remove.
1749 (debug_nothing_int_int_charstar,
1750 debug_nothing_int_int_charstar_int_bool): New functions.
1751 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
1752 through to dwarf2out_source_line.
1753 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
1754 (dwarf2out_source_line): Add column argument, emit it if requested.
1755 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
1757 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
1758 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
1759 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
1760 through to dwarf2out_begin_prologue.
1761 (vmsdbgout_source_line): Add column argument, pass it through to
1762 dwarf2out_source_line.
1763 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
1764 dbxout_source_line caller.
1765 (dbxout_source_line): Add column argument.
1767 * common.opt (gno-column-info, gcolumn-info): New options.
1768 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
1769 (check_die): Also test for multiple DW_AT_decl_column attributes.
1770 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
1771 DW_AT_decl_column if requested.
1772 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
1774 (gen_variable_die): Likewise.
1775 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
1776 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
1779 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
1780 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
1781 (ix86_handle_option): Handle OPT_m3dnowa.
1782 * doc/invoke.texi (-m3dnowa): Document.
1783 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
1784 -m3dnowa instead of -m3dnow -march=athlon.
1787 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
1788 instead of gcc_assert for K, r and R code checks. Formatting fixes.
1790 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1793 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
1794 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
1795 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
1796 generator for vsx_xxpermdi_<mode>_be.
1797 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
1798 force big-endian semantics.
1799 (vsx_xxpermdi_<mode>_be): New define_expand with same
1800 implementation as previous version of vsx_xxpermdi_<mode>.
1802 2017-02-17 Jakub Jelinek <jakub@redhat.com>
1804 PR tree-optimization/79327
1805 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
1806 variable, its initialization and use.
1808 2017-02-17 Julia Koval <julia.koval@intel.com>
1810 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
1811 (OPTION_MASK_ISA_PKU_UNSET): New.
1812 (ix86_handle_option): Handle -mrdpid.
1813 * config/i386/cpuid.h (bit_RDPID): New.
1814 * config/i386/driver-i386.c (host_detect_local_cpu):
1815 Detect RDPID feature.
1816 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
1817 * config/i386/i386-c.c (ix86_target_macros_internal):
1819 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
1820 (ix86_valid_target_attribute_inner_p): Add "rdpid".
1821 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
1822 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
1823 * config/i386/i386.md (define_insn "rdpid"): New.
1824 * config/i386/i386.opt Add -mrdpid.
1825 * config/i386/immintrin.h (_rdpid_u32): New.
1827 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
1829 PR rtl-optimization/79541
1830 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
1831 instead of transforming it into USE.
1833 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
1835 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
1836 If HONOR_SNANS (SFmode) force the input to a register.
1837 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
1838 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
1839 an frsp or similar insn.
1841 2017-02-17 Martin Liska <mliska@suse.cz>
1843 PR rtl-optimization/79577
1844 * params.def (selsched-max-sched-times): Increase minimum to 1.
1846 2017-02-17 Martin Liska <mliska@suse.cz>
1848 PR rtl-optimization/79574
1849 * gcse.c (want_to_gcse_p): Prevent integer overflow.
1851 2017-02-17 Martin Liska <mliska@suse.cz>
1853 PR tree-optimization/79529
1854 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
1855 ssa_defined_default_def_p to handle cases which are implicitly
1857 * tree-ssa.c (ssa_defined_default_def_p): New function.
1858 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
1859 which are implicitly defined.
1860 * tree-ssa.h (ssa_defined_default_def_p): Declare.
1862 2017-02-17 Richard Biener <rguenther@suse.de>
1865 * params.def (max-ssa-name-query-depth): Limit to 10.
1867 2017-02-17 Richard Biener <rguenther@suse.de>
1869 PR tree-optimization/79552
1870 * tree-ssa-structalias.c (visit_loadstore): Properly verify
1873 2017-02-17 Richard Biener <rguenther@suse.de>
1876 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
1878 2017-02-17 Marek Polacek <polacek@redhat.com>
1881 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
1882 (fold_negate_expr): New wrapper.
1884 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
1886 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
1887 Correct terminology and de-emphasize pre-standard behavior.
1889 2017-02-16 Alan Modra <amodra@gmail.com>
1891 PR rtl-optimization/79286
1892 * ira.c (def_dominates_uses): New function.
1893 (update_equiv_regs): Don't create an equivalence for insns that
1894 may trap where the register def does not dominate the use.
1896 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
1898 PR rtl-optimization/78127
1899 * lra.c (lra): Call lra_eliminate before finish the loop after
1902 2017-02-16 Richard Biener <rguenther@suse.de>
1904 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
1906 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
1907 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
1908 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
1909 (isl_val_int_from_wi): New function.
1910 (extract_affine_gmp): Rename to ...
1911 (extract_affine_wi): ... this, take a widest_int.
1912 (extract_affine_int): Just wrap extract_affine_wi.
1913 (add_param_constraints): Use isl_val_int_from_wi.
1914 (add_loop_constraints): Likewise, and extract_affine_wi.
1916 2017-02-15 Jeff Law <law@redhat.com>
1919 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
1920 ira_init_register_move_cost_if_necessary.
1922 2017-02-15 Martin Sebor <msebor@redhat.com>
1925 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
1926 removed in a prior commit.
1928 2017-02-15 Bin Cheng <bin.cheng@arm.com>
1930 PR tree-optimization/79347
1931 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
1932 counters during peeling.
1934 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
1936 * Makefile.in (site.exp): Remove "set ISLVER".
1938 2017-02-15 Jakub Jelinek <jakub@redhat.com>
1941 * real.c (real_from_integer): Call real_convert even for decimal.
1943 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
1946 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
1948 2017-02-14 Andrew Pinski <apinski@cavium.com>
1950 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
1951 cores and change the partno/implementer to be correct.
1952 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
1953 the 'B" as the implementer.
1954 * config/aarch64/aarch64-tune.md: Regenerate.
1956 2017-02-14 Carl Love <cel@us.ibm.com>
1958 * config/rs6000/rs6000.c: Add case statement entry to make the
1959 xvcvuxdsp built-in argument unsigned.
1960 * config/rs6000/vsx.md: Fix the source and return operand types so they
1961 match the instruction definitions from the ISA document. Fix typo
1962 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
1965 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
1968 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
1969 member early_clobber_alts.
1970 * lra-lives.c (reg_early_clobber_p): New.
1971 (process_bb_lives): Use it.
1972 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
1973 (debug_operand_data): Initialize early_clobber_alts.
1974 (setup_operand_alternative): Set up early_clobber_alts.
1975 (collect_non_operand_hard_regs): Ditto. Pass early clobber
1976 alternatives to new_insn_reg.
1977 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
1979 (lra_update_insn_regno_info): Pass the new arg.
1981 2017-02-14 Jakub Jelinek <jakub@redhat.com>
1984 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
1985 (new_oacc_loop_raw): Don't clear already cleared fields.
1988 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
1989 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
1990 _mm512_prefetch_i64gather_ps): New inline functions and macros.
1992 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
1995 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
1997 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
2000 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
2001 the extra instruction to the right place to store 128-bit constant
2004 2017-02-14 Martin Sebor <msebor@redhat.com>
2007 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
2008 warning for strings of unknown length.
2010 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
2012 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
2014 2017-02-14 Jeff Law <law@redhat.com>
2017 * ira-costs.c (scan_one_insn): Initialize register move costs
2018 for pseudos seen in USE/CLOBBER insns.
2020 PR tree-optimization/79095
2021 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
2022 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
2023 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
2024 if the operands are known to be not equal, then the resulting range
2026 (intersect_ranges): If the new range is ~[0,0] and the old range is
2027 wide, then prefer ~[0,0].
2028 * tree-vrp.c (overflow_comparison_p_1): New function.
2029 (overflow_comparison_p): New function.
2030 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
2031 if NAME is used in an overflow test.
2032 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
2033 overflow check that can be expressed as an equality test, then adjust
2034 ops to be that equality test.
2036 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2038 * config/s390/s390-builtin-types.def: Remove flags argument.
2039 * config/s390/s390.c (s390_init_builtins): Likewise.
2041 2017-02-14 Martin Liska <mliska@suse.cz>
2043 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
2044 vector. Fix trailing white spaces.
2046 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
2048 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
2051 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2053 PR rtl-optimization/68664
2054 * config/arm/arm.c (arm_sched_can_speculate_insn):
2055 New function. Declare prototype.
2056 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
2058 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2060 PR rtl-optimization/68664
2061 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
2063 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
2065 2017-02-14 Amit Pawar <amit.pawar@amd.com>
2067 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
2068 max skip bytes for function, loop and jump.
2070 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2072 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
2073 ABS_EXPR for gimple dump.
2075 2017-02-14 Jakub Jelinek <jakub@redhat.com>
2078 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
2080 PR tree-optimization/79408
2081 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
2082 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
2083 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
2084 also if rhs1 is INTEGER_CST.
2086 2017-02-14 Richard Biener <rguenther@suse.de>
2089 * tree-into-ssa.c (insert_phi_nodes): When the function can
2090 have abnormal edges rewrite SSA names with broken use-def
2091 dominance out of SSA and register them for PHI insertion.
2093 2017-02-13 Martin Sebor <msebor@redhat.com>
2096 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
2097 clearing info.nowrite flag when snprintf size argument is a range.
2099 2017-02-13 Jakub Jelinek <jakub@redhat.com>
2101 * cprop.c (cprop_jump): Add missing space in string literal.
2102 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
2103 (get_constraint_for_component_ref): Likewise.
2104 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
2105 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
2106 * lra-constraints.c (process_alt_operands): Likewise.
2107 * ipa-inline.c (inline_small_functions): Likewise.
2108 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
2109 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
2110 * trans-mem.c (diagnose_tm_1_op): Likewise.
2111 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
2112 (grid_parallel_clauses_gridifiable): Likewise.
2114 * config/nvptx/mkoffload.c (process): Add space in between
2117 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
2118 "MOD4_SSE_REGS" and "ALL_REGS".
2120 * spellcheck.c (test_data): Add , in between "foo" and "food".
2122 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2125 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
2126 boundary crossing check and subsequent code generation agree.
2128 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2130 * config/aarch64/aarch64.c (has_memory_op): Delete.
2131 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
2134 2017-02-13 Jakub Jelinek <jakub@redhat.com>
2136 PR rtl-optimization/79388
2137 PR rtl-optimization/79450
2138 * combine.c (distribute_notes): When removing TEM_INSN for which
2139 corresponding dest has last value recorded, invalidate that last
2142 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2144 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
2145 of explicit '@'. Add missing assembly comment marker on branch costs
2148 2017-02-13 Nathan Sidwell <nathan@acm.org>
2150 * gengtype-lex.l (<in_struct>): Add '/'.
2152 2017-02-13 Martin Liska <mliska@suse.cz>
2155 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
2157 2017-02-13 Richard Biener <rguenther@suse.de>
2159 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
2161 * configure: Re-generate.
2162 * config.in: Likewise.
2163 * graphite-dependences.c: Simplify as if
2164 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
2165 * graphite-isl-ast-to-gimple.c: Likewise.
2166 * graphite-optimize-isl.c: Likewise.
2167 * graphite-poly.c: Likewise.
2168 * graphite-sese-to-poly.c: Likewise.
2169 * graphite.h: Likewise.
2170 * toplev.c: Include isl/version.h and use isl_version () for
2171 printing the ISL version.
2172 * doc/install.texi: Update ISL requirement.
2174 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
2176 * doc/standards.texi (Standards): Update reference to
2179 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
2181 * doc/extend.texi (Named Address Spaces): sourceware.org now
2183 * doc/install.texi (Binaries): Ditto.
2186 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
2188 * doc/cpp.texi: Replace "stringify"/"stringification" with C
2189 standard terminology "stringize"/"stringizing" throughout.
2190 * doc/cppinternals.texi: Likewise.
2192 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
2194 * doc/extend.texi: Fix some spelling mistakes and typos.
2195 * doc/invoke.texi: Likewise.
2197 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
2200 * params.def (inline-min-speedup) Change from 10 to 8.
2202 2017-02-11 Jakub Jelinek <jakub@redhat.com>
2204 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
2207 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
2210 * ipa-inline-analysis.c (get_minimal_bb): New function.
2211 (record_modified): Use it.
2212 (remap_edge_change_prob): Handle also ancestor functions.
2214 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
2216 * doc/contrib.texi (Contributors): Remove broken link into
2217 the Mauve CVS repository.
2219 2017-02-11 Jakub Jelinek <jakub@redhat.com>
2222 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
2223 result computation whenever lhs doesn't have vector mode, not
2224 just when it has BLKmode.
2226 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
2228 * doc/makefile.texi (profiledbootstrap): Refer to the
2229 installation instructions only in textual form.
2231 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2234 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
2236 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
2238 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
2239 (Specific): Update mingw-w64 reference.
2241 (Specific): Remove broken link to Renesas RX processor.
2243 2017-02-10 Richard Biener <rguenther@suse.de>
2245 * toplev.c (process_options): Do not mention obsolete graphite
2246 options when printing sorry message about missing graphite support.
2247 Mention -floop-nest-optimize.
2249 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
2251 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
2252 (vtst_p16): Likewise.
2253 (vtstq_p8): Likewise.
2254 (vtstq_p16): Likewise.
2256 (vtstq_p64): Likewise.
2257 * config/arm/arm_neon.h (vgetq_lane_p64): New.
2258 (vset_lane_p64): New.
2259 (vsetq_lane_p64): New.
2261 2017-02-10 Jakub Jelinek <jakub@redhat.com>
2263 PR tree-optimization/79411
2264 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
2265 stmt operands are SSA_NAMEs used in abnormal phis.
2266 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
2269 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
2272 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
2275 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
2277 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
2279 2017-02-09 Jakub Jelinek <jakub@redhat.com>
2281 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
2285 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
2286 not arbitrary TREE_CONSTANT.
2289 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
2290 "omp declare target link" attribute unless is_global_var.
2291 * omp-offload.c (find_link_var_op): Likewise.
2293 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
2294 Chung-Lin Tang <cltang@codesourcery.com>
2296 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
2298 (gimplify_adjust_omp_clauses): Don't delete TILE.
2299 (gimplify_omp_for): Deal with TILE.
2300 * internal-fn.c (expand_GOACC_TILE): New function.
2301 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
2303 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
2304 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
2306 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
2307 avoid DIV for outermost collapse var.
2308 (expand_oacc_for): Insert tile element loop as needed. Adjust.
2309 Remove out of date comments, fix whitespace.
2310 * omp-general.c (omp_extract_for_data): Deal with tiling.
2311 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
2312 adjust OLF_DIM_BASE value.
2313 (struct omp_for_data): Add tiling field.
2314 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
2315 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
2316 for auto loops. Remove default auto determining, moved to
2317 oacc_loop_fixed_partitions.
2318 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
2319 stmts, add e_mask field.
2320 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
2321 (oacc_thread_numbers): Use oacc_dim_call.
2322 (oacc_xform_tile): New.
2323 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
2324 (finish_oacc_loop): Adjust for ifns vector.
2325 (oacc_loop_discover_walk): Append loop abstraction sites to list,
2326 add case for GOACC_TILE fns.
2327 (oacc_loop_xform_loop): Delete.
2328 (oacc_loop_process): Iterate over call list directly, and add
2329 handling for GOACC_TILE fns.
2330 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
2332 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
2333 vector partitioning to outer loops. Assign 2 partitions to loops
2334 when available. Add TILE handling.
2335 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
2336 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
2337 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
2338 * tree.c (omp_clause_num_ops): Adjust TILE ops.
2339 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
2341 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
2343 * configure.ac (ACX_BUGURL): Update.
2344 * configure: Regenerate.
2346 2017-02-09 Richard Biener <rguenther@suse.de>
2348 PR tree-optimization/69823
2349 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
2350 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
2352 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
2354 * config/arc/arc-c.def: Add __NPS400__ definition.
2355 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
2356 (TARGET_NPS400): Define.
2358 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
2360 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
2362 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
2364 (arc_cpu_types): Fill the arch_info field with a pointer into the
2365 arc_arch_types table.
2366 (arc_selected_cpu): Declare.
2367 * config/arc/arc.c (arc_selected_cpu): Make global.
2368 (arc_selected_arch): Delete.
2369 (arc_base_cpu): Delete.
2370 (arc_override_options): Remove references to deleted variables,
2371 update access to arch information.
2372 (ARC_OPT): Update access to arch information.
2373 (ARC_OPTX): Likewise.
2374 * config/arc/arc.h (arc_base_cpu): Remove declaration.
2375 (TARGET_ARC600): Update access to arch information.
2376 (TARGET_ARC601): Likewise.
2377 (TARGET_ARC700): Likewise.
2378 (TARGET_EM): Likewise.
2379 (TARGET_HS): Likewise.
2380 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
2383 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
2386 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
2387 condition/operands for integer GE/LE/GEU/LEU operations.
2389 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
2391 PR translation/79397
2392 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
2395 2017-02-08 Martin Jambor <mjambor@suse.cz>
2398 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
2399 whether allocation happened.
2400 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
2401 nothing was allocated.
2403 2017-02-08 Jakub Jelinek <jakub@redhat.com>
2405 PR tree-optimization/79408
2406 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
2407 constant, but SSA_NAME with a known integer range, use the minimum
2408 of that range instead of op1 to determine if modulo can be replaced
2409 with its first operand.
2411 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2413 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
2415 2017-02-08 Richard Biener <rguenther@suse.de>
2417 PR tree-optimization/71824
2418 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
2419 Check all loops contained in the merged region.
2421 2017-02-07 Andrew Pinski <apinski@cavium.com>
2423 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
2425 2017-02-07 Andrew Pinski <apinski@cavium.com>
2427 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
2428 (thunderxt88): Likewise.
2429 (thunderxt81): Disable LSE and change v8.1 to v8.
2430 (thunderxt83): Likewise.
2432 2017-02-07 Jakub Jelinek <jakub@redhat.com>
2433 Richard Biener <rguenther@suse.de>
2436 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
2437 type from int to size_t.
2438 * ira-costs.c (struct_costs_size): Change type from int to size_t.
2440 2017-02-07 Jakub Jelinek <jakub@redhat.com>
2442 PR rtl-optimization/79386
2443 * cprop.c (bypass_conditional_jumps): Initialize
2444 bypass_last_basic_block already before splitting bbs after
2445 unconditional traps...
2446 (bypass_conditional_jumps): ... rather than here.
2449 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
2450 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
2451 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
2452 fix -masm=intel patterns.
2454 2017-02-07 Richard Biener <rguenther@suse.de>
2456 PR tree-optimization/79256
2458 * builtins.c (get_object_alignment_2): Use min_align_of_type
2459 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
2460 and ADJUST_FIELD_ALIGN.
2462 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
2464 * doc/tm.texi: Regenerate.
2465 * stor-layout.c (layout_decl): Adjust.
2466 (update_alignment_for_field): Likewise.
2467 (place_field): Likewise.
2468 (min_align_of_type): Likewise.
2469 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
2470 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
2471 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
2472 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
2473 * config/frv/frv.c (frv_adjust_field_align): Likewise.
2474 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
2475 * config/i386/i386.c (x86_field_alignment): Likewise.
2476 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
2477 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
2478 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
2479 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
2480 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
2481 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
2485 2017-01-30 Richard Biener <rguenther@suse.de>
2487 PR tree-optimization/79256
2488 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
2489 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
2492 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
2494 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
2495 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
2496 builtins to SImode and emit a zero-extend, if necessary.
2498 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2500 * docs/invoke.texi (RISC-V Options): Alphabetize.
2502 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2504 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
2507 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2509 * config/riscv/riscv.c: New file.
2510 * gcc/common/config/riscv/riscv-common.c: Likewise.
2511 * config.gcc: Likewise.
2512 * config/riscv/constraints.md: Likewise.
2513 * config/riscv/elf.h: Likewise.
2514 * config/riscv/generic.md: Likewise.
2515 * config/riscv/linux.h: Likewise.
2516 * config/riscv/multilib-generator: Likewise.
2517 * config/riscv/peephole.md: Likewise.
2518 * config/riscv/pic.md: Likewise.
2519 * config/riscv/predicates.md: Likewise.
2520 * config/riscv/riscv-builtins.c: Likewise.
2521 * config/riscv/riscv-c.c: Likewise.
2522 * config/riscv/riscv-ftypes.def: Likewise.
2523 * config/riscv/riscv-modes.def: Likewise.
2524 * config/riscv/riscv-opts.h: Likewise.
2525 * config/riscv/riscv-protos.h: Likewise.
2526 * config/riscv/riscv.h: Likewise.
2527 * config/riscv/riscv.md: Likewise.
2528 * config/riscv/riscv.opt: Likewise.
2529 * config/riscv/sync.md: Likewise.
2530 * config/riscv/t-elf-multilib: Likewise.
2531 * config/riscv/t-linux: Likewise.
2532 * config/riscv/t-linux-multilib: Likewise.
2533 * config/riscv/t-riscv: Likewise.
2534 * configure.ac: Likewise.
2535 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
2536 Waterman as RISC-V maintainers.
2537 * doc/install.texi: Add RISC-V entries.
2538 * doc/invoke.texi: Add RISC-V options section.
2539 * doc/md.texi: Add RISC-V constraints section.
2540 * configure: Regenerated.
2542 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2545 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
2546 false values to be constant vectors with all 0 or all 1 bits set.
2547 (vcondu<mode><mode>): Likewise.
2548 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
2550 (fpmask_comparison_operator): Update comment.
2551 (vecint_comparison_operator): New predicate.
2552 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
2553 vector conditionals when the true and false values are constant
2554 vectors with all 0 bits or all 1 bits set.
2556 2017-02-06 Martin Sebor <msebor@redhat.com>
2558 PR tree-optimization/79376
2559 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
2561 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
2563 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
2564 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
2565 to simplify split condition.
2567 2017-02-06 Jakub Jelinek <jakub@redhat.com>
2569 * omp-expand.c (oxpand_omp_atomic_fetch_op,
2570 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
2573 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
2575 PR rtl-optimization/68664
2576 * target.def (can_speculate_insn): New hook.
2577 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
2578 * doc/tm.texi: Regenerate.
2579 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
2580 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
2581 (rs6000_sched_can_speculate_insn): New function.
2583 2017-02-06 Jakub Jelinek <jakub@redhat.com>
2585 PR tree-optimization/79284
2586 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
2587 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
2588 vectorizable_mask_load_store, vectorizable_operation,
2589 vect_is_simple_cond, get_same_sized_vectype): Use it instead
2590 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
2591 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
2592 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
2593 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
2594 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
2595 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
2596 is_gimple_assign (stmt). Replace another such test with
2597 is_gimple_assign (stmt).
2599 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
2602 * config/avr/avr.c (rtl-iter.h): Include it.
2603 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
2604 (avr_legitimate_combined_insn): ...and implementation.
2606 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
2608 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
2609 * config/s390/s390.c (s390_const_operand_ok)
2610 (s390_canonicalize_comparison, s390_extract_part)
2611 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
2612 (s390_contiguous_bitmask_p, s390_rtx_costs)
2613 (legitimize_pic_address): Likewise.
2614 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
2615 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
2616 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
2617 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
2618 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
2620 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
2622 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
2623 REGNO($0) == REGNO($1).
2625 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2627 * config/s390/linux.h(SIZE_TYPE): Add comment.
2629 2017-02-06 Julian Brown <julian@codesourcery.com>
2630 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
2631 Virendra Pathak <virendra.pathak@broadcom.com>
2633 * config/aarch64/aarch64-cores.def: Change the scheduler
2635 * config/aarch64/aarch64.md: Include thunderx2t99.md.
2636 * config/aarch64/thunderx2t99.md: New file.
2638 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2640 * doc/standards.texi (Go Language): Update link to language
2643 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
2645 * tree-eh.c (lower_resx): Sanitize profile.
2646 (cleanup_empty_eh_move_lp): Likewise.
2648 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
2651 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
2653 * cfgloopmanip.h (loop_version): Update prototype.
2654 * modulo-sched.c (sms_schedule): Update call of loop_version.
2655 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
2656 * tree-parloops.c (gen_parallel_loop): Likewise.
2657 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
2658 * tree-ssa-loop-split.c (split_loop): Likewise.
2659 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
2660 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
2662 2017-02-05 Martin Liska <mliska@suse.cz>
2665 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
2667 (print_operand_address): Initialize a struct to zero.
2669 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2671 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
2672 garbage collector only in textual form.
2674 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2676 * doc/extend.texi (x86 specific memory model extensions for
2677 transactional memory): Simplify a phrase.
2679 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
2682 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
2683 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
2684 (atomic_storedi_1): Likewise.
2686 2017-02-04 Jakub Jelinek <jakub@redhat.com>
2688 PR tree-optimization/79338
2689 * tree-parloops.c (gather_scalar_reductions): Don't call
2690 vect_analyze_loop_form for loop->inner before destroying loop's
2693 2017-02-03 Martin Sebor <msebor@redhat.com>
2695 PR tree-optimization/79327
2696 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
2697 when precision has resulted in leading zeros.
2698 (format_integer): Adjust the likely counter to assume an unknown
2699 argument that may be zero is non-zero.
2701 2017-02-03 Jason Merrill <jason@redhat.com>
2704 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
2705 avoid copying non-taken branch.
2707 2017-02-03 Jakub Jelinek <jakub@redhat.com>
2709 PR tree-optimization/79340
2710 * tree-vect-loop.c (vectorizable_reduction): Release
2711 vec_defs elements after safe_splicing them into other vectors.
2714 PR tree-optimization/79327
2715 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
2716 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
2718 (format_integer): Use wide_int_to_tree instead of build_int_cst
2719 + to_?hwi. If argmin is NULL, just set argmin and argmax to
2720 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
2721 of shortest and longest sequence.
2723 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
2725 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
2726 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
2728 2017-02-03 Walter Lee <walt@tilera.com>
2731 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
2732 after initial stackframe link reg save.
2733 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
2735 2017-02-03 Jakub Jelinek <jakub@redhat.com>
2738 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
2739 wu for stxssp alternative.
2741 2017-02-03 Martin Sebor <msebor@redhat.com>
2743 PR tree-optimization/79352
2744 * gimple-fold.c (get_range_strlen): Add argument.
2745 (get_range_strlen): Change return type to bool.
2746 (get_maxval_strlen): Pass in a dummy argument.
2747 * gimple-fold.h (get_range_strlen): Change return type to bool.
2748 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
2749 * tree.h (array_at_struct_end_p): Add argument.
2750 * tree.c (array_at_struct_end_p): Handle it.
2752 2017-02-03 Martin Liska <mliska@suse.cz>
2755 * multiple_target.c (create_dispatcher_calls): Redirect edge
2756 from a caller of a dispatcher.
2757 (expand_target_clones): Make the clones local.
2758 (ipa_target_clone): Do both target clones and resolvers.
2759 (ipa_dispatcher_calls): Remove the pass.
2760 (pass_dispatcher_calls::gate): Likewise.
2761 (make_pass_dispatcher_calls): Likewise.
2762 * passes.def (pass_target_clone): Put as very first IPA early
2765 2017-02-03 Martin Liska <mliska@suse.cz>
2767 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
2768 in case of a function with ifunc attribute.
2770 2017-02-03 Martin Liska <mliska@suse.cz>
2772 * cgraph.c (cgraph_node::dump): Dump function version info.
2773 * symtab.c (symtab_node::dump_base): Add missing new line.
2775 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
2777 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
2778 (ifcombine_ifandif): Use it.
2780 2017-02-03 Martin Liska <mliska@suse.cz>
2782 * doc/invoke.texi: Document default value for
2783 use-after-scope-direct-emission-threshold.
2785 2017-02-03 Martin Liska <mliska@suse.cz>
2787 PR tree-optimization/79339
2788 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
2789 (format_floating): Likewise.
2791 2017-02-03 Martin Liska <mliska@suse.cz>
2794 * ipa-prop.c (ipa_node_params_t::insert): Remove current
2796 (ipa_node_params_t::remove): Likewise.
2797 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
2798 initialization from removed ipa_node_params_t::insert.
2799 (ipa_node_params::~ipa_node_params): Move from removed
2800 ipa_node_params_t::release.
2801 * symbol-summary.h (symbol_summary::m_released): New member.
2802 Do not release a summary twice. Do not allow to call finalizer
2803 for types of a summary that live in GGC memory.
2805 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
2807 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
2810 2017-02-02 Martin Sebor <msebor@redhat.com>
2813 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
2814 (format_string): Tighten up the range of output for non-constant
2815 strings and correct the expected range for wide non-constant strings.
2817 2017-02-02 Martin Sebor <msebor@redhat.com>
2819 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
2822 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
2824 (-fdump-tree-@var): Add to index and document how to come up
2825 with pass-specific option and dump file names.
2826 (-fdump-passes): Clarify where to look for output.
2828 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
2831 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
2832 statistics of the analyzed path; allow threading for speed when
2833 any of BBs along the path are optimized for speed.
2835 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
2838 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
2839 settings of the virtual registers.
2842 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
2844 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2845 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
2848 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2850 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
2851 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
2853 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2855 * config/s390/s390.md: Add missing comments with the expanded
2857 * config/s390/vector.md: Likewise.
2858 * config/s390/vx-builtins.md: Likewise.
2860 2017-02-02 Jakub Jelinek <jakub@redhat.com>
2863 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
2864 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
2865 conditions on a single line.
2867 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2869 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
2870 __S390_VX__ to __VX__.
2872 2017-02-01 Andrew Pinski <apinski@cavium.com>
2874 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
2875 stmt_info to record_stmt_cost.
2876 (vect_get_known_peeling_cost): Pass stmt_info if known to
2878 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
2879 cpu_vector_cost field into
2880 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
2881 field into vec_int_stmt_cost and vec_fp_stmt_cost.
2882 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
2883 splitting of scalar_stmt_cost and vec_stmt_cost.
2884 (thunderx_vector_cost): Likewise.
2885 (cortexa57_vector_cost): LIkewise.
2886 (exynosm1_vector_cost): Likewise.
2887 (xgene1_vector_cost): Likewise.
2888 (thunderx2t99_vector_cost): Improve after the splitting of the two
2890 (aarch64_builtin_vectorization_cost): Update for the splitting of
2891 scalar_stmt_cost and vec_stmt_cost.
2893 2017-02-01 Torvald Riegel <triegel@redhat.com>
2894 Richard Henderson <rth@redhat.com>
2896 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
2897 conditional on existance of a fast atomic load.
2898 * optabs-query.c (can_atomic_load_p): New function.
2899 * optabs-query.h (can_atomic_load_p): Declare it.
2900 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
2901 no fast atomic load is available for the particular size of access.
2902 (expand_atomic_compare_and_swap): Likewise.
2903 (expand_atomic_load): Likewise.
2904 (expand_atomic_store): Likewise.
2905 (expand_atomic_fetch_op): Likewise.
2906 * testsuite/lib/target-supports.exp
2907 (check_effective_target_sync_int_128): Remove x86 because it provides
2908 no fast atomic load.
2909 (check_effective_target_sync_int_128_runtime): Likewise.
2911 2017-02-01 Richard Biener <rguenther@suse.de>
2913 * graphite.c: Include tree-vectorizer.h for find_loop_location.
2914 (graphite_transform_loops): Provide opt-info for optimized nests.
2915 * tree-parloop.c (parallelize_loops): Provide opt-info for
2918 2017-02-01 Richard Biener <rguenther@suse.de>
2921 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
2924 2017-02-01 Richard Biener <rguenther@suse.de>
2926 PR tree-optimization/71824
2927 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
2928 Verify the loops are valid in the merged SESE region.
2929 (scop_detection::can_represent_loop_1): Check analyzing the
2930 evolution of the number of iterations in the region succeeds.
2932 2017-01-31 Ian Lance Taylor <iant@golang.org>
2934 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
2935 REG_ARGS_SIZE note to 32-bit push insns and call insn.
2937 2017-01-31 David Malcolm <dmalcolm@redhat.com>
2939 PR preprocessor/79210
2940 * input.c (get_substring_ranges_for_loc): Replace line_width
2941 assertion with error-handling.
2943 2017-01-31 Richard Biener <rguenther@suse.de>
2945 PR tree-optimization/77318
2946 * graphite-sese-to-poly.c (extract_affine): Fix assert.
2947 (create_pw_aff_from_tree): Take loop parameter.
2948 (add_condition_to_pbb): Pass loop of the condition to
2949 create_pw_aff_from_tree.
2951 2017-01-31 Jakub Jelinek <jakub@redhat.com>
2953 * config/s390/s390.c (s390_asan_shadow_offset): New function.
2954 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
2956 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2960 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
2962 (convert_int_to_float128): Likewise.
2963 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
2964 (convert_int_to_float128): Likewise.
2965 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
2966 (UNSPEC_IEEE128_CONVERT): Likewise.
2967 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
2968 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
2969 Use local variables for IBM extended format.
2970 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
2971 (fix_trunc<mode>si2_fprs): Likewise.
2972 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
2973 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
2974 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
2975 to know that we can now have integers of all sizes in vector
2977 (fix<uns>_<mode>di2_hw): Likewise.
2978 (float<uns>_<mode>si2_hw): Likewise.
2979 (fix_<mode>si2_hw): Likewise.
2980 (fixuns_<mode>si2_hw): Likewise.
2981 (float<uns>_<mode>di2_hw): Likewise.
2982 (float_<mode>di2_hw): Likewise.
2983 (float_<mode>si2_hw): Likewise.
2984 (floatuns_<mode>di2_hw): Likewise.
2985 (floatuns_<mode>si2_hw): Likewise.
2986 (xscvqp<su>wz_<mode>): Delete, no longer used.
2987 (xscvqp<su>dz_<mode>): Likewise.
2988 (xscv<su>dqp_<mode>): Likewise.
2989 (ieee128_mfvsrd_64bit): Likewise.
2990 (ieee128_mfvsrd_32bit): Likewise.
2991 (ieee128_mfvsrwz): Likewise.
2992 (ieee128_mtvsrw): Likewise.
2993 (ieee128_mtvsrd_64bit): Likewise.
2994 (ieee128_mtvsrd_32bit): Likewise.
2996 2017-01-31 Martin Liska <mliska@suse.cz>
2999 * ipa-prop.c (ipa_free_all_node_params): Call release method
3000 instead of ~sumbol_summary to not to trigger double times
3003 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
3005 PR tree-optimization/71691
3006 * bitmap.h (class auto_bitmap): New.
3007 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
3008 is_maybe_undefined instead of ssa_undefined_value_p.
3010 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3012 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
3013 __S390_ARCH_LEVEL__ to __ARCH__.
3015 2017-01-31 Jakub Jelinek <jakub@redhat.com>
3017 PR tree-optimization/79267
3018 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
3019 if should_remove_lhs_p is true.
3021 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
3024 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
3025 (add_alignment_attribute): New.
3026 (base_type_die): Add alignment attribute.
3027 (subrange_type_die): Likewise.
3028 (modified_type_die): Likewise.
3029 (gen_array_type_die): Likewise.
3030 (gen_descr_array_type_die: Likewise.
3031 (gen_enumeration_type_die): Likewise.
3032 (gen_subprogram_die): Likewise.
3033 (gen_variable_die): Likewise.
3034 (gen_field_die): Likewise.
3035 (gen_ptr_to_mbr_type_die): Likewise.
3036 (gen_struct_or_union_type_die): Likewise.
3037 (gen_subroutine_type_die): Likewise.
3038 (gen_typedef_die): Likewise.
3039 (base_type_cmp): Compare alignment attribute.
3041 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3044 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
3045 (setb_unsigned) New pattern for setb with CCUNS.
3046 * config/rs6000/rs6000.c (expand_block_compare): Use a different
3047 subfc./subfe sequence to avoid overflow problems. Generate a
3048 shorter sequence with cmpld/setb for power9.
3049 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
3050 for generating subfc. instruction.
3051 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
3052 now uses this instruction.
3054 2017-01-30 Ian Lance Taylor <iant@google.com>
3057 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
3058 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
3060 2017-01-30 Martin Sebor <msebor@redhat.com>
3062 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
3063 Move constant to the right of a relational operator.
3064 (get_mpfr_format_length, format_character, format_string): Ditto.
3065 (should_warn_p, maybe_warn): Same.
3067 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
3069 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3072 * asan.c (get_translation_unit_decl): Remove function.
3073 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
3075 2017-01-30 Martin Liska <mliska@suse.cz>
3077 PR gcov-profile/79259
3078 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
3081 2017-01-30 Martin Liska <mliska@suse.cz>
3084 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
3085 Initialize variables with NULL value.
3087 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
3090 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
3092 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
3094 2017-01-30 Richard Biener <rguenther@suse.de>
3096 PR tree-optimization/79276
3097 * tree-vrp.c (process_assert_insertions): Properly adjust common
3098 when removing a duplicate.
3100 * gcc.dg/torture/pr79276.c: New testcase.
3102 2017-01-30 Richard Biener <rguenther@suse.de>
3104 PR tree-optimization/79256
3105 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
3106 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
3108 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
3110 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
3113 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
3114 ("*r<noxa>sbg_<mode>_sll_bitmask")
3115 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
3116 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
3117 Use contiguous_bitmask_nowrap_operand.
3119 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3122 * config/rs6000/altivec.h (vec_xl): Revise #define.
3123 (vec_xst): Likewise.
3125 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
3127 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
3129 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
3131 PR rtl-optimization/79194
3132 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
3133 traps before call to bypass_conditional_jumps.
3135 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
3137 PR tree-optimization/71374
3138 * lra-constraints.c (check_conflict_input_operands): New.
3139 (match_reload): Use it.
3141 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
3144 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
3145 account to calculate conflict_set.
3147 2017-01-27 Bin Cheng <bin.cheng@arm.com>
3149 PR rtl-optimization/78559
3150 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
3151 other_insn in combine.
3153 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
3155 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
3156 uint16_type_node for BT_UINT16.
3158 2017-01-27 David Malcolm <dmalcolm@redhat.com>
3160 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
3161 "RTL Tests" to menu.
3162 (GIMPLE Tests): New node.
3163 (RTL Tests): New node.
3165 2017-01-27 Richard Biener <rguenther@suse.de>
3167 PR tree-optimization/79245
3168 * tree-loop-distribution.c (distribute_loop): Apply cost
3169 modeling also to detected patterns.
3171 2017-01-27 Richard Biener <rguenther@suse.de>
3173 PR tree-optimization/71433
3174 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
3175 (compare_assert_loc): New function.
3176 (process_assert_insertions): Sort and optimize assert locations
3177 to remove duplicates and push down identical assertions on
3178 edges to their destination block.
3180 2017-01-27 Richard Biener <rguenther@suse.de>
3182 PR tree-optimization/79244
3183 * tree-vrp.c (remove_range_assertions): Forcefully propagate
3184 out SSA names even if abnormal.
3186 2017-01-27 Jakub Jelinek <jakub@redhat.com>
3188 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
3189 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
3190 instead of MPFR_RNDN.
3192 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
3195 * arm.c (arm_option_override): Don't call build_target_option_node
3196 until after doing all option overrides.
3197 (arm_valid_target_attribute_tree): Likewise.
3199 2017-01-27 Martin Liska <mliska@suse.cz>
3201 * doc/invoke.texi (-fprofile-arcs): Document profiling support
3202 for {cd}tors and C++ {cd}tors.
3204 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
3206 * config/s390/s390.md ("*setmem_long_and")
3207 ("*setmem_long_and_31z"): Use zero_extend instead of and.
3209 2017-01-26 Martin Sebor <msebor@redhat.com>
3211 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
3214 2017-01-26 Martin Sebor <msebor@redhat.com>
3216 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
3217 HAVE_DFmode before using XFmode or DFmode.
3218 (parse_directive): Avoid using the z length modifier to avoid
3219 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
3222 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
3223 to accept adjustment as an array.
3224 (get_int_range): New function.
3225 (struct directive): Make width and prec arrays.
3226 (directive::set_width, directive::set_precision): Call get_int_range.
3227 (format_integer, format_floating): Handle width and precision ranges.
3228 (format_string, parse_directive): Same.
3230 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3233 * dwarf2out.c (generate_skeleton_bottom_up): For children with
3234 comdat_type_p set, just clone them, but keep the children in the
3238 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
3239 which have direct callers with -fvar-tracking-assignments enabled
3241 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
3242 inside of type units.
3244 2017-01-26 Martin Sebor <msebor@redhat.com>
3247 * gimple-ssa-sprintf.c (struct result_range): Add likely and
3249 (struct format_result): Replace number_chars, number_chars_min,
3250 and number_chars_max with a single member of struct result_range.
3252 (format_result::operator+=): Adjust.
3253 (struct fmtresult): Remove bounded. Handle likely and unlikely
3255 (fmtresult::adjust_for_width_or_precision): New function.
3256 (fmtresult:type_max_digits): New function.
3257 (bytes_remaining): Handle likely and unlikely counters.
3258 (min_bytes_remaining): Remove.
3259 (format_percent): Simplify.
3260 (format_integer, format_floating): Set likely and unlikely counters.
3261 (get_string_length, format_character, format_string): Same.
3262 (format_plain, should_warn_p): New function.
3263 (maybe_warn): Call should_warn_p. Update diagnostic messages
3264 and handle those for all directives, including plain strings.
3265 (format_directive): Handle likely and unlikely counters.
3266 Remove unnecessary quoting from diagnostics. Add an informational
3268 (add_bytes): Remove.
3269 (pass_sprintf_length::compute_format_length): Simplify.
3270 (try_substitute_return_value): Handle likely and unlikely counters.
3272 2017-01-26 Carl Love <cel@us.ibm.com>
3274 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
3275 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
3277 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
3280 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
3281 endianess for subregs into account.
3282 * lra-constraints.c (lra_constraints): Do risky transformations
3283 always on the first iteration.
3284 * lra-lives.c (check_pseudos_live_through_calls): Add arg
3285 last_call_used_reg_set.
3286 (process_bb_lives): Define and use last_call_used_reg_set.
3287 * lra.c (lra): Always continue after lra_constraints on the first
3290 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
3292 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
3294 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
3296 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3298 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
3299 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
3300 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
3301 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
3302 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
3303 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
3304 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
3305 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
3306 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
3308 2017-01-26 Marek Polacek <polacek@redhat.com>
3311 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
3312 for the third operand.
3314 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3317 * omp-low.c (struct omp_context): Add simt_stmt field.
3318 (scan_omp_for): Return omp_context *.
3319 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
3320 context to the _simt_ SIMD stmt.
3321 (lower_omp_for): For combined SIMD with sibling _simt_
3322 SIMD, make sure to use the same decls in _looptemp_
3323 clauses as in the sibling.
3325 2017-01-26 David Sherwood <david.sherwood@arm.com>
3328 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
3331 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3334 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
3335 emit fld b; fld a; if possible.
3337 * brig-builtins.def: Update copyright years.
3338 * config/arm/arm_acle_builtins.def: Update copyright years.
3340 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
3343 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
3344 constraint instead of o for the stxsd instruction.
3346 2017-01-25 Carl Love <cel@us.ibm.com>
3348 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
3349 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
3351 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
3353 * doc/invoke.texi (C++ Dialect Options): Fix typo.
3355 2017-01-25 Richard Biener <rguenther@suse.de>
3357 PR tree-optimization/69264
3358 * target.def (vector_alignment_reachable): Improve documentation.
3359 * doc/tm.texi: Regenerate.
3360 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
3362 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
3363 earlier changes with respect to TYPE_USER_ALIGN.
3364 (vector_alignment_reachable_p): Likewise. Improve dumping.
3366 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3369 * config/arm/arm.md (xordi3): Force constant operand into a register
3372 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3374 * doc/invoke.texi (-fstore-merging): Correct default optimization
3375 levels at which it is enabled.
3376 (-O): Move -fstore-merging from list to...
3379 2017-01-25 Richard Biener <rguenther@suse.de>
3382 * omp-expand.c: Include debug.h.
3383 (expand_omp_taskreg): Make sure to generate early debug before
3384 outlining anything from a function.
3385 (expand_omp_target): Likewise.
3386 (grid_expand_target_grid_body): Likewise.
3388 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
3391 * asan.c (get_translation_unit_decl): New function.
3392 (asan_add_global): Extract modules file name from globals
3393 TRANSLATION_UNIT_DECL name.
3395 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
3398 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
3399 for long calls with APCS frame and VFP.
3401 2017-01-24 David Malcolm <dmalcolm@redhat.com>
3403 * cfg.c (original_copy_tables_initialized_p): New function.
3404 * cfg.h (original_copy_tables_initialized_p): New decl.
3405 * cfgrtl.c (relink_block_chain): Guard the call to
3406 free_original_copy_tables with a call to
3407 original_copy_tables_initialized_p.
3408 * cgraph.h (symtab_node::native_rtl_p): New decl.
3409 * cgraphunit.c (symtab_node::native_rtl_p): New function.
3410 (symtab_node::needed_p): Don't assert for early assembly output
3411 for __RTL functions.
3412 (cgraph_node::finalize_function): Set "force_output" for __RTL
3414 (cgraph_node::analyze): Bail out early for __RTL functions.
3415 (analyze_functions): Update assertion to support __RTL functions.
3416 (cgraph_node::expand): Bail out early for __RTL functions.
3417 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
3419 * function.h (struct function): Update comment for field
3421 * gimple-expr.c: Include "tree-pass.h".
3422 (gimple_has_body_p): Return false for __RTL functions.
3423 * Makefile.in (OBJS): Add run-rtl-passes.o.
3424 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
3426 (gcc::pass_manager::get_clean_slate): New accessor.
3427 * passes.c: Include "insn-addr.h".
3428 (should_skip_pass_p): Add logging. Update logic for running
3429 "expand" to be compatible with both __GIMPLE and __RTL. Guard
3430 property-provider override so it is only done for gimple passes.
3432 (skip_pass): New function.
3433 (execute_one_pass): Call skip_pass when skipping passes.
3434 * read-md.c (md_reader::read_char): Support filtering
3435 the input to a subset of line numbers.
3436 (md_reader::md_reader): Initialize fields
3437 m_first_line and m_last_line.
3438 (md_reader::read_file_fragment): New function.
3439 * read-md.h (md_reader::read_file_fragment): New decl.
3440 (md_reader::m_first_line): New field.
3441 (md_reader::m_last_line): New field.
3442 * read-rtl-function.c (function_reader::create_function): Only
3443 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
3444 curr_properties. Set DECL_INITIAL to a dummy block.
3445 (read_rtl_function_body_from_file_range): New function.
3446 * read-rtl-function.h (read_rtl_function_body_from_file_range):
3448 * run-rtl-passes.c: New file.
3449 * run-rtl-passes.h: New file.
3451 2017-01-24 Jeff Law <law@redhat.com>
3453 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
3456 2017-01-24 Bin Cheng <bin.cheng@arm.com>
3458 PR tree-optimization/79159
3459 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
3460 (record_nonwrapping_iv): Improve boundary using above function if no
3461 value range information.
3463 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
3464 Martin Jambor <mjambor@suse.cz>
3466 * brig-builtins.def: New file.
3467 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
3468 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
3469 (DEF_HSAIL_SAT_BUILTIN): Likewise.
3470 (DEF_HSAIL_INTR_BUILTIN): Likewise.
3471 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
3472 * builtin-types.def (BT_INT8): New.
3473 (BT_INT16): Likewise.
3474 (BT_UINT8): Likewise.
3475 (BT_UINT16): Likewise.
3476 (BT_FN_ULONG): Likewise.
3477 (BT_FN_UINT_INT): Likewise.
3478 (BT_FN_UINT_ULONG): Likewise.
3479 (BT_FN_UINT_LONG): Likewise.
3480 (BT_FN_UINT_PTR): Likewise.
3481 (BT_FN_ULONG_PTR): Likewise.
3482 (BT_FN_INT8_FLOAT): Likewise.
3483 (BT_FN_INT16_FLOAT): Likewise.
3484 (BT_FN_UINT32_FLOAT): Likewise.
3485 (BT_FN_UINT16_FLOAT): Likewise.
3486 (BT_FN_UINT8_FLOAT): Likewise.
3487 (BT_FN_UINT64_FLOAT): Likewise.
3488 (BT_FN_UINT16_UINT32): Likewise.
3489 (BT_FN_UINT32_UINT16): Likewise.
3490 (BT_FN_UINT16_UINT16_UINT16): Likewise.
3491 (BT_FN_INT_PTR_INT): Likewise.
3492 (BT_FN_UINT_PTR_UINT): Likewise.
3493 (BT_FN_LONG_PTR_LONG): Likewise.
3494 (BT_FN_ULONG_PTR_ULONG): Likewise.
3495 (BT_FN_VOID_UINT64_UINT64): Likewise.
3496 (BT_FN_UINT8_UINT8_UINT8): Likewise.
3497 (BT_FN_INT8_INT8_INT8): Likewise.
3498 (BT_FN_INT16_INT16_INT16): Likewise.
3499 (BT_FN_INT_INT_INT): Likewise.
3500 (BT_FN_UINT_FLOAT_UINT): Likewise.
3501 (BT_FN_FLOAT_UINT_UINT): Likewise.
3502 (BT_FN_ULONG_UINT_UINT): Likewise.
3503 (BT_FN_ULONG_UINT_PTR): Likewise.
3504 (BT_FN_ULONG_ULONG_ULONG): Likewise.
3505 (BT_FN_UINT_UINT_UINT): Likewise.
3506 (BT_FN_VOID_UINT_PTR): Likewise.
3507 (BT_FN_UINT_UINT_PTR: Likewise.
3508 (BT_FN_UINT32_UINT64_PTR): Likewise.
3509 (BT_FN_INT_INT_UINT_UINT): Likewise.
3510 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
3511 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
3512 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
3513 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
3514 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
3515 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
3516 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
3517 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
3518 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
3519 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
3520 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
3521 * doc/frontends.texi: List BRIG FE.
3522 * doc/install.texi (Testing): Add BRIG tesring requirements.
3523 * doc/invoke.texi (Overall Options): Mention BRIG.
3524 * doc/standards.texi (Standards): Doucment BRIG HSA version.
3526 2017-01-24 Richard Biener <rguenther@suse.de>
3528 PR translation/79208
3529 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
3531 2017-01-24 Martin Jambor <mjambor@suse.cz>
3534 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
3535 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
3538 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
3541 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
3542 casts from signed to unsigned really don't have a range.
3544 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
3546 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
3547 GMP_RNDx for compatiblity.
3549 2017-01-24 Martin Liska <mliska@suse.cz>
3552 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
3553 that would prevent us to call alloca with -1 as argument.
3555 2017-01-24 Jakub Jelinek <jakub@redhat.com>
3557 * dwarf2out.c (output_compilation_unit_header, output_file_names):
3558 Avoid -Wformat-security warning.
3560 2017-01-23 Andrew Pinski <apinski@cavium.com>
3562 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
3565 2017-01-23 Martin Sebor <msebor@redhat.com>
3568 * gimple-ssa-sprintf.c (warn_level): New global.
3569 (format_integer): Use it here and throughout the rest of the file.
3570 Use the same switch to compute sign as base.
3571 (maybe_warn): New function.
3572 (format_directive): Factor out warnings into maybe_warn.
3573 Add debugging output. Use warn_level.
3574 (add_bytes): Use warn_level.
3575 (pass_sprintf_length::compute_format_length): Add debugging output.
3576 (try_substitute_return_value): Same.
3577 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
3580 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
3581 (struct fmtresult, format_integer, format_floating): Adjust.
3582 (fmtresult::fmtresult): Set max correctly in two argument ctor.
3583 (get_string_length, format_string,format_directive): Same.
3584 (pass_sprintf_length::compute_format_length): Same.
3585 (try_substitute_return_value): Simplify slightly.
3588 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
3589 (fmtresult::operator+=): Outlined.
3590 (struct fmtresult): Add ctors.
3591 (struct conversion_spec): Rename...
3592 (struct directive): ...to this. Add and remove data members.
3593 (directive::set_width, directive::set_precision): New functions.
3594 (format_percent): Use fmtresult ctor.
3595 (get_width_and_precision): Remove.
3596 (format_integer): Make naming changes. Avoid computing width and
3598 (format_floating): Same. Adjust indentation.
3599 (format_character, format_none): New functions.
3600 (format_string): Moved character handling to format_character.
3601 (format_directive): Remove arguments, change return type.
3602 (parse_directive): New function.
3603 (pass_sprintf_length::compute_format_length): Move directive
3604 parsing to parse_directive.
3606 2017-01-23 Jakub Jelinek <jakub@redhat.com>
3608 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
3609 (assign_assembler_name_if_needed): ... this.
3610 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
3611 (assign_assembler_name_if_needed): ... this.
3612 (free_lang_data_in_cgraph): Adjust callers.
3613 * cgraphunit.c (cgraph_node::analyze): Likewise.
3614 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
3617 2017-01-23 Richard Biener <rguenther@suse.de>
3619 PR tree-optimization/79088
3620 PR tree-optimization/79188
3621 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
3622 resetting loop bounds after last path deletion. Reset loop
3623 bounds of the target loop, make code match the comments.
3624 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
3625 Make sure loops need no fixups.
3627 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
3629 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
3630 exponent support with double type for first argument.
3631 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
3632 type returned by __builtin_vec_extract_sig,
3633 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
3634 functions from "vector int" to "vector unsigned int" or from
3635 "vector long long int" to "vector unsigned long long int".
3636 Changed type returned by __builtin_vec_extract_exp,
3637 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
3638 functions from "vector int" to "vector unsigned int" or from
3639 "vector long long int" to "vector unsigned long long int".
3640 Changed return type of __builtin_vec_test_data_class,
3641 __builtin_vec_test_data_class_sp, and
3642 __builtin_vec_test_data_class_dp from "vector int" to
3643 "vector bool int" or from "vector long long int" to "vector bool
3644 long long int" and changed second argument type from "unsigned
3645 int" to "int". Added new overloaded function forms "vector float
3646 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
3647 "vector float __builtin_vec_insert_exp_sp (vector float, vector
3648 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
3649 double, vector unsigned long long int)" and "vector double
3650 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
3651 long int)". Changed return type of
3652 __builtin_scalar_test_data_class and
3653 __builtin_scalar_test_data_class_sp and
3654 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
3655 int" and changed second argument from "unsigned int" to "int".
3656 Changed type returned by __builtin_scalar_test_neg,
3657 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
3658 from "int" to "bool int". Added new overloaded function form
3659 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
3660 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
3661 exponent double-precision with floating point first argument.
3662 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
3663 documentation of scalar_test_data_class, scalar_test_neg,
3664 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
3665 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
3666 vec_test_data_class built-in functions to reflect refinements in
3667 their type signatures.
3669 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
3671 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
3673 (aarch64_elf_asm_destructor): Likewise.
3675 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
3677 PR rtl-optimization/78634
3678 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
3679 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
3680 * ifcvt.c (noce_try_cmove): Add missing cost check.
3682 PR rtl-optimization/71724
3683 * combine.c (if_then_else_cond): Look for situations where it is
3684 beneficial to undo the work of one of the recursive calls.
3686 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3688 PR tree-optimization/70754
3689 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3690 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3691 combined stmt before it if not NULL.
3692 (combine_chains): Process refs reversely and compute dominance point
3695 2017-01-23 Martin Liska <mliska@suse.cz>
3697 PR tree-optimization/79196
3698 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
3699 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
3701 (strlen_optimize_stmt): Call the renamed function.
3703 2017-01-23 Michael Matz <matz@suse.de>
3705 PR tree-optimization/78384
3706 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
3708 2017-01-23 Richard Biener <rguenther@suse.de>
3710 PR tree-optimization/79186
3711 * tree-vrp.c (register_new_assert_for): Make sure we've seen
3712 both incoming edges before moving an assert.
3714 2017-01-23 Martin Jambor <mjambor@suse.cz>
3716 * ipa-prop.c (load_from_param_1): Removed.
3717 (load_from_unmodified_param): Bits from load_from_param_1 put back
3719 (load_from_param): Removed.
3720 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
3721 with stmt. Reverted back to use of load_from_unmodified_param.
3723 2017-01-23 Martin Jambor <mjambor@suse.cz>
3726 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
3727 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
3728 field a pointer to garbage collected vector, mark lattices and
3729 ipcp_orig_node with GTY((skip)).
3730 (ipa_get_param_count): Adjust to descriptors being a pointer.
3731 (ipa_get_param): Likewise.
3732 (ipa_get_type): Likewise.
3733 (ipa_get_param_move_cost): Likewise.
3734 (ipa_set_param_used): Likewise.
3735 (ipa_get_controlled_uses): Likewise.
3736 (ipa_set_controlled_uses): Likewise.
3737 (ipa_is_param_used): Likewise.
3738 (ipa_node_params_t): Move into garbage collector. New methods insert
3740 (ipa_node_params_sum): Annotate wth GTY(()).
3741 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
3743 (ipa_load_from_parm_agg): Adjust declaration.
3744 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
3745 * ipa-profile.c (ipa_profile): Likewise.
3746 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
3747 (ipa_populate_param_decls): Make descriptors parameter garbage
3749 (ipa_dump_param): Adjust to descriptors being a pointer.
3750 (ipa_alloc_node_params): Likewise.
3751 (ipa_initialize_node_params): Likewise.
3752 (load_from_param_1): Make descriptors parameter garbage collected.
3753 (load_from_unmodified_param): Likewise.
3754 (load_from_param): Likewise.
3755 (ipa_load_from_parm_agg): Likewise.
3756 (ipa_node_params::~ipa_node_params): Removed.
3757 (ipa_free_all_node_params): Remove call to delete operator.
3758 (ipa_node_params_t::insert): New.
3759 (ipa_node_params_t::remove): Likewise.
3760 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
3761 copy known_csts and known_contexts vectors.
3762 (ipa_read_node_info): Adjust to descriptors being a pointer.
3763 (ipcp_modif_dom_walker): Make m_descriptors field garbage
3765 (ipcp_transform_function): Make descriptors variable garbage
3768 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
3770 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
3771 * config/i386/avx512dqintrin.h: Ditto.
3772 * config/i386/avx512fintrin.h: Ditto.
3773 * gcc/config/i386/i386.c: Handle new builtins.
3774 * config/i386/i386-builtin.def: Add new builtins.
3775 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
3776 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
3778 2017-01-23 Jakub Jelinek <jakub@redhat.com>
3779 Martin Liska <mliska@suse.cz>
3781 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
3782 * asan.c (asan_expand_poison_ifn): Support stores and use
3783 appropriate ASAN report function.
3784 * internal-fn.c (expand_ASAN_POISON_USE): New function.
3785 * internal-fn.def (ASAN_POISON_USE): Declare.
3786 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
3787 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
3788 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
3789 ASAN_POISON calls w/o LHS.
3790 * tree-ssa.c (execute_update_addresses_taken): Create clobber
3791 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
3792 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
3793 * gimplify.c (asan_poison_variables): Add attribute
3794 use_after_scope_memory to variables that really needs to live
3796 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
3797 having the attribute.
3799 2017-01-23 Martin Liska <mliska@suse.cz>
3801 * asan.c (create_asan_shadow_var): New function.
3802 (asan_expand_poison_ifn): Likewise.
3803 * asan.h (asan_expand_poison_ifn): New declaration.
3804 * internal-fn.c (expand_ASAN_POISON): Likewise.
3805 * internal-fn.def (ASAN_POISON): New builtin.
3806 * sanopt.c (pass_sanopt::execute): Expand
3807 asan_expand_poison_ifn.
3808 * tree-inline.c (copy_decl_for_dup_finish): Make function
3810 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
3811 * tree-ssa.c (is_asan_mark_p): New function.
3812 (execute_update_addresses_taken): Rewrite local variables
3813 (identified just by use-after-scope as addressable) into SSA.
3815 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
3817 * doc/install.texi (Specific): opensource.apple.com uses https
3818 now. Remove trailing slash.
3820 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
3822 * README.Portability: Remove note on an Irix compatibility issue.
3824 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
3826 * gcov.c (INCLUDE_ALGORITHM): Define.
3827 (INCLUDE_VECTOR): Define.
3828 No longer include <vector> and <algorithm> directly.
3830 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
3832 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
3834 * doc/invoke.texi (Code Gen Options): Ditto.
3836 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
3839 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
3841 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
3843 rtl-optimization/79125
3844 * cprop.c (local_cprop_pass): Handle cases where we make an
3847 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
3851 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
3852 read from, for big endian.
3854 2017-01-20 Jiong Wang <jiong.wang@arm.com>
3856 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
3857 register pauth builtins for LP64 only.
3859 2017-01-20 Marek Polacek <polacek@redhat.com>
3862 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
3865 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
3867 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
3869 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
3870 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
3871 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
3873 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3876 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
3877 in big-endian mode when they are not a single duplicated value.
3879 2017-01-20 Richard Biener <rguenther@suse.de>
3881 * BASE-VER: Bump to 7.0.1.
3883 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
3885 * omp-low.c (omplow_simd_context): New struct. Use it...
3886 (lower_rec_simd_input_clauses): ...here and...
3887 (lower_rec_input_clauses): ...here to hold common data. Adjust all
3888 references to idx, lane, max_vf, is_simt.
3890 2017-01-20 Graham Markall <graham.markall@embecosm.com>
3892 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
3895 2017-01-20 Martin Jambor <mjambor@suse.cz>
3897 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
3898 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
3900 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
3901 (GTFILES): Rename hsa.c to hsa-common.c.
3902 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
3903 * hsa-dump.c: Likewise.
3904 * hsa-gen.c: Likewise.
3905 * hsa-regalloc.c: Likewise.
3906 * ipa-hsa.c: Likewise.
3907 * omp-expand.c: Likewise.
3908 * omp-low.c: Likewise.
3909 * toplev.c: Likewise.
3911 2017-01-20 Marek Polacek <polacek@redhat.com>
3914 * doc/invoke.texi: Document -Wduplicated-branches.
3915 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
3916 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
3917 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
3918 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
3919 return 0 only when not OEP_LEXICOGRAPHIC.
3920 (fold_build_cleanup_point_expr): Use the expression
3921 location when building CLEANUP_POINT_EXPR.
3922 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
3923 * tree.c (add_expr): Handle error_mark_node.
3925 2017-01-20 Martin Liska <mliska@suse.cz>
3928 * tree-profile.c (init_ic_make_global_vars): Do not call
3930 (gimple_init_gcov_profiler): Likewise.
3932 2017-01-20 Martin Liska <mliska@suse.cz>
3935 * cgraph.h (maybe_create_reference): Remove argument and
3937 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
3939 * ipa-cp.c (create_specialized_node): Likewise.
3940 * symtab.c (symtab_node::maybe_create_reference): Handle
3941 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
3943 2017-01-20 Martin Liska <mliska@suse.cz>
3945 * read-rtl-function.c (function_reader::create_function): Use
3946 build_decl instread of build_decl_stat.
3948 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
3950 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
3951 * config/i386/avx512dqintrin.h: Ditto.
3952 * config/i386/avx512fintrin.h: Ditto.
3953 * config/i386/i386-builtin-types.def: Add new types.
3954 * gcc/config/i386/i386.c: Handle new types.
3955 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
3956 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
3957 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
3958 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
3959 (__builtin_ia32_kshiftridi): New.
3960 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
3962 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
3966 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
3967 define to rs6000_init_stack_protect_guard.
3968 (rs6000_init_stack_protect_guard): New function.
3970 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
3971 Yunqiang Su <yunqiang.su@imgtec.com>
3973 * config.gcc (supported_defaults): Add madd4.
3974 (with_madd4): Add validation.
3975 (all_defaults): Add madd4.
3976 * config/mips/mips.opt (mmadd4): New option.
3977 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
3979 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
3981 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
3982 (ISA_HAS_FUSED_MADD4): Likewise.
3983 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
3984 * gcc/doc/install.texi (--with-madd4): Document the new option.
3986 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3988 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
3989 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
3990 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
3991 (aarch64_init_pauth_hint_builtins): New.
3992 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
3993 (aarch64_expand_builtin): Expand new builtins.
3995 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3997 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
3998 * combine-stack-adj.c (no_unhandled_cfa): Handle
3999 REG_CFA_TOGGLE_RA_MANGLE.
4000 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
4001 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
4002 info for return address signing.
4003 (aarch64_expand_epilogue): Likewise.
4005 2017-01-19 Jiong Wang <jiong.wang@arm.com>
4007 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
4008 * config/aarch64/aarch64-protos.h
4009 (aarch64_return_address_signing_enabled): New declaration.
4010 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
4012 (aarch64_expand_prologue): Sign return address before it's pushed onto
4014 (aarch64_expand_epilogue): Authenticate return address fetched from
4016 (aarch64_override_options): Sanity check for ILP32 and ISA level.
4017 (aarch64_attributes): New function attributes for "sign-return-address".
4018 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
4019 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
4020 ("*do_return"): Generate combined instructions according to key index.
4021 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
4022 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
4024 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
4025 * config/aarch64/aarch64.opt (msign-return-address=): New.
4026 * doc/extend.texi (AArch64 Function Attributes): Documents
4027 "sign-return-address=".
4028 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
4030 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
4032 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
4033 overall option summary.
4035 2017-01-19 Jiong Wang <jiong.wang@arm.com>
4037 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
4038 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
4039 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
4040 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
4042 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
4044 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
4045 -mpower9-minmax by default for -mcpu=power9.
4046 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
4047 128-bit floating point.
4049 2017-01-20 Alan Modra <amodra@gmail.com>
4051 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
4052 optimizing for size.
4054 2017-01-20 Alan Modra <amodra@gmail.com>
4057 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
4058 for strcmp and strncmp from corresponding builtin decl.
4060 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
4062 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
4063 instead of i386/rtems-64.h.
4064 * config/i386/rtems-64.h: Remove.
4066 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
4070 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
4072 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
4074 2017-01-19 Tamar Christina <tamar.christina@arm.com>
4076 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
4077 Change int to HOST_WIDE_INT.
4078 * config/aarch64/aarch64-protos.h
4079 (aarch64_simd_gen_const_vector_dup): Likewise.
4080 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
4082 2017-01-19 David Malcolm <dmalcolm@redhat.com>
4084 * langhooks-def.h (lhd_type_for_size): New decl.
4085 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
4086 * langhooks.c (lhd_type_for_size): New function, taken from
4089 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
4091 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
4092 define_bypass for CR latency.
4093 (power9-cracked-alu): Update bypass latency and remove power9-branch.
4094 (power9-alu2): Add define_bypass for CR latency.
4096 (power9-mul): Update insn latency.
4097 (power9-mul-compare): Update insn latency, bypass latency and remove
4100 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4102 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
4104 * config/aarch64/aarch64.md
4105 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
4106 aarch64_nopcrelative_literal_loads.
4107 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
4109 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
4111 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
4113 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
4115 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
4118 * config.gcc (supported_defaults): Add lxc1-sxc1.
4119 (with_lxc1_sxc1): Add validation.
4120 (all_defaults): Add lxc1-sxc1.
4121 * config/mips/mips.opt (mlxc1-sxc1): New option.
4122 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
4124 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
4125 __mips_no_lxc1_sxc1.
4126 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
4127 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
4128 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
4130 2017-01-19 Richard Biener <rguenther@suse.de>
4132 PR tree-optimization/72488
4133 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
4134 sure to restore SSA info.
4135 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
4137 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
4139 PR rtl-optimization/79121
4140 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
4141 of the inner type when shifting an extended value.
4143 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
4146 * symtab.c (symtab_node::equal_address_to): Fix comparing of
4147 interposable aliases.
4149 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
4152 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
4153 Use the evmergelohi instruction.
4154 (mov_si<mode>_e500_subreg4_2_le): Likewise.
4155 (mov_sitf_e500_subreg8_2_be): Likewise.
4156 (mov_sitf_e500_subreg12_2_le): Likewise.
4157 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
4158 (mov_si<mode>_e500_subreg4_2_be): Likewise.
4159 (mov_sitf_e500_subreg8_2_le): Likewise.
4160 (mov_sitf_e500_subreg12_2_be): Likewise.
4162 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4164 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
4165 attribute from vecsimple to vecperm.
4166 (altivec_vbpermq2): Likewise.
4168 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4171 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
4173 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4174 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
4175 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
4176 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
4177 case where N arg is SIZE_MAX.
4178 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
4179 (cmpstrsi): Add pattern.
4181 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
4183 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4184 __builtin_vec_revb builtins.
4185 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
4186 built-in functions to support generation of the ISA 3.0 XXBR<x>
4187 vector byte reverse instructions.
4188 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
4189 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
4190 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
4191 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
4192 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
4193 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
4194 (P9V_BUILTIN_VEC_REVB): Likewise.
4195 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
4196 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
4197 (p9_xxbrq_v16qi): Likewise.
4198 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
4199 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
4200 (p9_xxbrh_v8hi): Likewise.
4201 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
4202 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
4203 vec_revb built-in functions.
4205 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
4207 PR rtl-optimization/78952
4208 * config/i386/i386.md (any_extract): New code iterator.
4209 (*insvqi_2): Use any_extract for source operand.
4210 (*insvqi_3): Use any_shiftrt for source operand.
4212 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
4214 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
4216 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
4218 2017-01-18 Matthias Klose <doko@ubuntu.com>
4220 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
4222 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4224 * config/rs6000/altivec.h (vec_bperm): Change #define.
4225 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
4226 (altivec_vbpermq2): New define_insn.
4227 (altivec_vbpermd): Likewise.
4228 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
4230 (VBPERMD): Likewise.
4231 (VBPERM): New polymorphic function interface.
4232 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
4233 Add entries for P9V_BUILTIN_VEC_VBPERM.
4234 * doc/extend.texi: Add interfaces for vec_bperm.
4236 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4238 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
4239 first letter of error messages.
4240 (s390_resolve_overloaded_builtin): Likewise.
4241 * config/s390/s390.c (s390_expand_builtin): Likewise.
4242 (s390_invalid_arg_for_unprototyped_fn): Likewise.
4243 (s390_valid_target_attribute_inner_p): Likewise.
4244 * config/s390/s390.md ("tabort"): Likewise.
4246 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
4248 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
4249 (ISA_AVOID_DIV_HILO): New macro.
4250 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
4251 (ISA_HAS_DDIV): Likewise.
4253 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4255 * doc/invoke.texi (fabi-version): Correct number of occurrences.
4257 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4259 * doc/invoke.texi (fabi-version): Spelling fix.
4261 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4264 * doc/invoke.texi (fabi-version): Mention mangling fix for
4267 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4270 * doc/invoke.texi (fabi-version): Document discriminator mangling.
4272 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
4275 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
4276 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
4278 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
4280 (stack_protect_test): Ditto.
4281 * config/rs6000/rs6000.opt (mstack-protector-guard=,
4282 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
4284 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
4285 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
4286 -mstack-protector-guard-offset=.
4287 (RS/6000 and PowerPC Options): Ditto.
4289 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
4291 * config/i386/i386.h (MASK_CLASS_P): New define.
4292 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
4293 there are no registers from different register sets also when
4294 mask registers are used. Update function comment.
4295 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
4296 to (*k/*r) and (*k/*km) alternatives.
4298 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
4300 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
4301 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
4302 (EH_RETURN_HANDLER_RTX): New define.
4303 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
4304 Force frame pointer in EH return functions.
4305 (aarch64_expand_epilogue): Add barrier for eh_return.
4306 (aarch64_final_eh_return_addr): Remove.
4307 (aarch64_eh_return_handler_rtx): New function.
4308 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
4310 (aarch64_eh_return_handler_rtx): New prototype.
4312 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4314 * config/rs6000/altivec.h (vec_rlmi): New #define.
4315 (vec_vrlnm): Likewise.
4316 (vec_rlnm): Likewise.
4317 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
4318 (UNSPEC_VRLNM): Likewise.
4319 (VIlong): New mode iterator.
4320 (altivec_vrl<VI_char>mi): New define_insn.
4321 (altivec_vrl<VI_char>nm): Likewise.
4322 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
4325 (RLNM): New polymorphic function entry.
4326 (VRLWMI): New monomorphic function entry.
4328 (RLMI): New polymorphic function entry.
4329 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
4330 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
4331 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
4334 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4337 * dwarf2out.c (field_byte_offset): Restore the
4338 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
4339 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
4340 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
4343 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
4346 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
4348 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4351 * dwarf2out.c (add_data_member_location_attribute): For constant
4352 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
4353 instead of DW_AT_data_member_location, DW_AT_bit_offset and
4354 DW_AT_byte_size attributes.
4356 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
4358 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
4359 after forcing to constant memory when the code model is medium.
4361 2017-01-17 Julia Koval <julia.koval@intel.com>
4364 * config/i386/avx512fintrin.h
4365 (_mm512_i32gather_ps): Change __addr type to void const*.
4366 (_mm512_mask_i32gather_ps): Ditto.
4367 (_mm512_i32gather_pd): Ditto.
4368 (_mm512_mask_i32gather_pd): Ditto.
4369 (_mm512_i64gather_ps): Ditto.
4370 (_mm512_mask_i64gather_ps): Ditto.
4371 (_mm512_i64gather_pd): Ditto.
4372 (_mm512_mask_i64gather_pd): Ditto.
4373 (_mm512_i32gather_epi32): Ditto.
4374 (_mm512_mask_i32gather_epi32): Ditto.
4375 (_mm512_i32gather_epi64): Ditto.
4376 (_mm512_mask_i32gather_epi64): Ditto.
4377 (_mm512_i64gather_epi32): Ditto.
4378 (_mm512_mask_i64gather_epi32): Ditto.
4379 (_mm512_i64gather_epi64): Ditto.
4380 (_mm512_mask_i64gather_epi64): Ditto.
4381 (_mm512_i32scatter_ps): Change __addr type to void*.
4382 (_mm512_mask_i32scatter_ps): Ditto.
4383 (_mm512_i32scatter_pd): Ditto.
4384 (_mm512_mask_i32scatter_pd): Ditto.
4385 (_mm512_i64scatter_ps): Ditto.
4386 (_mm512_mask_i64scatter_ps): Ditto.
4387 (_mm512_i64scatter_pd): Ditto.
4388 (_mm512_mask_i64scatter_pd): Ditto.
4389 (_mm512_i32scatter_epi32): Ditto.
4390 (_mm512_mask_i32scatter_epi32): Ditto.
4391 (_mm512_i32scatter_epi64): Ditto.
4392 (_mm512_mask_i32scatter_epi64): Ditto.
4393 (_mm512_i64scatter_epi32): Ditto.
4394 (_mm512_mask_i64scatter_epi32): Ditto.
4395 (_mm512_i64scatter_epi64): Ditto.
4396 (_mm512_mask_i64scatter_epi64): Ditto.
4397 * config/i386/avx512pfintrin.h
4398 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
4399 (_mm512_mask_prefetch_i32gather_ps): Ditto.
4400 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4401 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4402 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
4403 (_mm512_prefetch_i32scatter_ps): Ditto.
4404 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4405 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4406 (_mm512_prefetch_i64scatter_pd): Ditto.
4407 (_mm512_prefetch_i64scatter_ps): Ditto.
4408 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4409 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4410 * config/i386/avx512vlintrin.h
4411 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
4412 (_mm_mmask_i32gather_ps): Ditto.
4413 (_mm256_mmask_i32gather_pd): Ditto.
4414 (_mm_mmask_i32gather_pd): Ditto.
4415 (_mm256_mmask_i64gather_ps): Ditto.
4416 (_mm_mmask_i64gather_ps): Ditto.
4417 (_mm256_mmask_i64gather_pd): Ditto.
4418 (_mm_mmask_i64gather_pd): Ditto.
4419 (_mm256_mmask_i32gather_epi32): Ditto.
4420 (_mm_mmask_i32gather_epi32): Ditto.
4421 (_mm256_mmask_i32gather_epi64): Ditto.
4422 (_mm_mmask_i32gather_epi64): Ditto.
4423 (_mm256_mmask_i64gather_epi32): Ditto.
4424 (_mm_mmask_i64gather_epi32): Ditto.
4425 (_mm256_mmask_i64gather_epi64): Ditto.
4426 (_mm_mmask_i64gather_epi64): Ditto.
4427 (_mm256_i32scatter_ps): Change __addr type to void*.
4428 (_mm256_mask_i32scatter_ps): Ditto.
4429 (_mm_i32scatter_ps): Ditto.
4430 (_mm_mask_i32scatter_ps): Ditto.
4431 (_mm256_i32scatter_pd): Ditto.
4432 (_mm256_mask_i32scatter_pd): Ditto.
4433 (_mm_i32scatter_pd): Ditto.
4434 (_mm_mask_i32scatter_pd): Ditto.
4435 (_mm256_i64scatter_ps): Ditto.
4436 (_mm256_mask_i64scatter_ps): Ditto.
4437 (_mm_i64scatter_ps): Ditto.
4438 (_mm_mask_i64scatter_ps): Ditto.
4439 (_mm256_i64scatter_pd): Ditto.
4440 (_mm256_mask_i64scatter_pd): Ditto.
4441 (_mm_i64scatter_pd): Ditto.
4442 (_mm_mask_i64scatter_pd): Ditto.
4443 (_mm256_i32scatter_epi32): Ditto.
4444 (_mm256_mask_i32scatter_epi32): Ditto.
4445 (_mm_i32scatter_epi32): Ditto.
4446 (_mm_mask_i32scatter_epi32): Ditto.
4447 (_mm256_i32scatter_epi64): Ditto.
4448 (_mm256_mask_i32scatter_epi64): Ditto.
4449 (_mm_i32scatter_epi64): Ditto.
4450 (_mm_mask_i32scatter_epi64): Ditto.
4451 (_mm256_i64scatter_epi32): Ditto.
4452 (_mm256_mask_i64scatter_epi32): Ditto.
4453 (_mm_i64scatter_epi32): Ditto.
4454 (_mm_mask_i64scatter_epi32): Ditto.
4455 (_mm256_i64scatter_epi64): Ditto.
4456 (_mm256_mask_i64scatter_epi64): Ditto.
4457 (_mm_i64scatter_epi64): Ditto.
4458 (_mm_mask_i64scatter_epi64): Ditto.
4459 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
4460 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
4461 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
4462 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
4463 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
4464 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
4465 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
4466 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
4467 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
4468 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
4469 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
4470 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
4471 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
4472 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
4473 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
4474 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
4475 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
4476 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
4477 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
4478 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
4479 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
4480 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
4481 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
4482 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
4483 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
4484 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
4485 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
4486 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
4487 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
4488 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
4489 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
4490 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
4491 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
4492 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
4493 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
4494 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
4495 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
4496 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
4497 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
4498 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
4499 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
4500 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
4501 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
4502 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
4503 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
4504 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
4505 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
4506 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
4507 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
4508 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
4509 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
4510 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
4511 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
4512 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
4513 definitions accordingly.
4515 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
4516 Kuan-Lin Chen <kuanlinchentw@gmail.com>
4519 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
4522 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
4525 * ira-conflicts.c (ira_build_conflicts): Update total conflict
4526 hard regs for inner regno.
4528 2017-01-17 Martin Liska <mliska@suse.cz>
4531 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
4532 assumption and add comment.
4534 2017-01-17 Nathan Sidwell <nathan@acm.org>
4536 * ipa-visibility.c (localize_node): New function, broken out of ...
4537 (function_and_variable_visibility): ... here. Call it.
4539 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
4542 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
4543 correctly set frequency of oudgoing edge.
4544 (duplicate_thread_path): Fix profile updating.
4546 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4549 * configure.ac: Add GCC_BASE_VER.
4550 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
4551 version from BASE-VER file.
4552 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
4553 (gcc.o): Depend on $(BASEVER).
4554 * common.opt (dumpfullversion): New option.
4555 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
4556 * doc/invoke.texi: Document -dumpfullversion.
4557 * doc/install.texi: Document --with-gcc-major-version-only.
4558 * configure: Regenerated.
4560 2017-01-17 Richard Biener <rguenther@suse.de>
4562 PR tree-optimization/71433
4563 * tree-vrp.c (register_new_assert_for): Merge same asserts
4564 on all incoming edges.
4565 (process_assert_insertions_for): Handle insertions at the
4568 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
4570 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
4571 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
4573 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
4576 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
4579 2017-01-17 Alan Modra <amodra@gmail.com>
4582 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
4583 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
4584 symbolic stack limit when pic.
4586 2017-01-16 Martin Sebor <msebor@redhat.com>
4588 PR tree-optimization/78608
4589 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
4591 2017-01-16 Jeff Law <law@redhat.com>
4594 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
4595 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
4596 for several include directories that may be relative to sysroot.
4597 * config/i386/x-mingw32 (gplus_includedir): Define.
4598 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
4599 (native_system_includedir): Likewise.
4600 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
4601 override if TARGET_SYSTEM_ROOT is defined.
4602 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
4604 PR tree-optimization/79090
4605 PR tree-optimization/33562
4606 PR tree-optimization/61912
4607 PR tree-optimization/77485
4608 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
4609 and computed trims into the dump file.
4611 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
4613 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
4615 2017-01-16 Jakub Jelinek <jakub@redhat.com>
4618 * gimplify.c (gimplify_init_constructor): If want_value and
4619 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
4623 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
4624 sequence. Formatting fixes.
4625 (doloop_optimize): Formatting fixes.
4628 * gcc.c (debug_level_greater_than_spec_func): New function.
4629 (static_spec_functions): Add debug-level-gt spec function.
4630 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
4632 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
4633 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
4634 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
4635 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
4636 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
4637 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
4639 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
4641 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
4642 QImode fixups to general and mask registers only.
4644 2017-01-16 Carl Love <cel@us.ibm.com>
4646 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
4647 for built-in functions
4648 vector signed char vec_nabs (vector signed char)
4649 vector signed short vec_nabs (vector signed short)
4650 vector signed int vec_nabs (vector signed int)
4651 vector signed long long vec_nabs (vector signed long long)
4652 vector float vec_nabs (vector float)
4653 vector double vec_nabs (vector double)
4654 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
4656 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
4657 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
4658 * doc/extend.texi: Update the documentation file for the new built-in
4661 2017-01-16 Martin Sebor <msebor@redhat.com>
4663 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
4666 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4668 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
4669 UNSPEC_VSX__XXSPLTD to require special splat handling.
4671 2017-01-16 David Malcolm <dmalcolm@redhat.com>
4674 * system.h: Poison strndup.
4676 2017-01-16 Alan Modra <amodra@gmail.com>
4679 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
4682 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
4684 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
4686 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
4688 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
4689 call recog here. Assert that INSN_CODE (insn) is non-negative.
4691 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
4694 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
4696 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
4697 in the currently scheduled RTL fragment.
4699 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
4701 PR rtl-optimization/78751
4702 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
4705 2017-01-14 Jeff Law <law@redhat.com>
4707 PR tree-optimization/79090
4708 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
4709 variable length stores.
4710 (compute_trims): Delete dead assignment to *trim_tail.
4711 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
4714 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
4716 PR rtl-optimization/78626
4717 PR rtl-optimization/78727
4718 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
4719 of a block, and split such blocks after everything else is finished.
4721 2017-01-14 Alan Modra <amodra@gmail.com>
4724 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
4725 target legitimate_combined_insn.
4726 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
4727 (rs6000_legitimate_combined_insn): New function.
4728 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
4730 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
4731 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
4732 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
4734 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
4736 * doc/frontends.texi (G++ and GCC): Remove references to Java.
4738 2017-01-13 Jeff Law <law@redhat.com>
4740 PR tree-optimization/33562
4741 PR tree-optimization/61912
4742 PR tree-optimization/77485
4743 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
4745 (delete_dead_assignment): Likewise.
4746 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
4747 statement to delete_dead_call and delete_dead_assignment.
4749 2017-01-13 David Malcolm <dmalcolm@redhat.com>
4752 * substring-locations.c (format_warning_va): Strengthen case 1 so
4753 that both endpoints of the substring must be within the format
4754 range for just the substring to be printed.
4756 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
4758 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
4759 * config/i386/i386.c (ix86_target_string): Add missing options
4760 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
4761 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
4762 flags_other and ix86_target_other to flags2_other. Display unknown
4764 (ix86_valid_target_attribute_inner_p): Add missing options and
4765 reorder options by implied ISAs, as in ix86_target_string.
4767 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
4769 * hash-table.h (hash_table::too_empty_p): New function.
4770 (hash_table::expand): Use it.
4771 (hash_table::traverse): Likewise.
4772 (hash_table::empty_slot): Use sizeof (value_type) instead of
4773 sizeof (PTR) to convert bytes to elements. Shrink the table
4774 if the current size is excessive for the current number of
4777 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
4779 * ira-costs.c (record_reg_classes): Break from the inner loop
4780 early once alt_fail is known to be true. Update outer loop
4781 handling accordingly.
4783 2017-01-13 Jeff Law <law@redhat.com>
4785 * tree-ssa-dse.c (decrement_count): New function.
4786 (increment_start_addr, maybe_trim_memstar_call): Likewise.
4787 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
4788 when we know the partially dead statement is a mem* function.
4790 PR tree-optimization/61912
4791 PR tree-optimization/77485
4792 * tree-ssa-dse.c: Include expr.h.
4793 (maybe_trim_constructor_store): New function.
4794 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
4796 PR tree-optimization/33562
4797 PR tree-optimization/61912
4798 PR tree-optimization/77485
4799 * doc/invoke.texi: Document new dse-max-object-size param.
4800 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
4801 * tree-ssa-dse.c: Include params.h.
4802 (dse_store_status): New enum.
4803 (initialize_ao_ref_for_dse): New, partially extracted from
4805 (valid_ao_ref_for_dse, normalize_ref): New.
4806 (setup_live_bytes_from_ref, compute_trims): Likewise.
4807 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
4808 (maybe_trim_partially_dead_store): Likewise.
4809 (maybe_trim_complex_store): Likewise.
4810 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
4811 Track what bytes live from the original store. Return tri-state
4812 for dead, partially dead or live.
4813 (dse_dom_walker): Add constructor, destructor and new private members.
4814 (delete_dead_call, delete_dead_assignment): New extracted from
4816 (dse_optimize_stmt): Make a member of dse_dom_walker.
4817 Use initialize_ao_ref_for_dse.
4819 PR tree-optimization/33562
4820 PR tree-optimization/61912
4821 PR tree-optimization/77485
4822 * sbitmap.h (bitmap_count_bits): Prototype.
4823 (bitmap_clear_range, bitmap_set_range): Likewise.
4824 * sbitmap.c (bitmap_clear_range): New function.
4825 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
4827 2017-01-13 Martin Liska <mliska@suse.cz>
4830 * function.c (set_cfun): Add new argument force.
4831 * function.h (set_cfun): Likewise.
4832 * ipa-inline-transform.c (inline_call): Use the function when
4833 strict alising from is dropped for function we inline to.
4835 2017-01-13 Richard Biener <rguenther@suse.de>
4837 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
4838 for dumping GIMPLE INTEGER_CSTs.
4840 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4842 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
4843 to 201112L since C++17.
4845 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
4848 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
4849 if -fsanitize=kernel-address is present.
4851 2017-01-13 Richard Biener <rguenther@suse.de>
4853 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
4854 as _Literal ( type ) number in case usual suffixes do not
4855 preserve all information.
4857 2017-01-13 Richard Biener <rguenther@suse.de>
4859 PR tree-optimization/77283
4860 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
4861 and ssa-iterators.h.
4862 (is_feasible_trace): Implement a cost model based on joiner
4865 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4868 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
4869 char or short to __float128/_Float128 directly.
4871 2017-01-12 Martin Sebor <msebor@redhat.com>
4873 to -Wformat-overflow.
4874 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
4875 (min_bytes_remaining): Same.
4876 (get_string_length): Same.
4877 (format_string): Same.
4878 (format_directive): Same.
4880 (pass_sprintf_length::handle_gimple_call): Same.
4882 2017-01-12 Jakub Jelinek <jakub@redhat.com>
4884 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
4885 info.nowrite calls with no lhs that can't throw. Return bool
4886 whether gsi_remove has been called or not.
4887 (pass_sprintf_length::handle_gimple_call): Return bool whether
4888 try_substitute_return_value called gsi_remove. Formatting fix.
4889 (pass_sprintf_length::execute): Don't use gsi_remove if
4890 handle_gimple_call returned true.
4893 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
4894 be removed due to side-effects, don't remove following barrier nor
4895 turn the successor edge into fallthru edge.
4897 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4900 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
4901 element-reversing loads and stores as not swappable.
4903 2017-01-12 Nathan Sidwell <nathan@acm.org>
4904 Nicolai Stange <nicstange@gmail.com>
4906 * combine.c (try_combine): Don't ignore result of overlap checking
4907 loop. Combine overlap & asm check into single loop.
4909 2017-01-12 Richard Biener <rguenther@suse.de>
4911 * tree-pretty-print.c (dump_generic_node): Provide -gimple
4912 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
4914 2017-01-12 Richard Biener <rguenther@suse.de>
4916 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
4917 and TS_TARGET_OPTION directly derive from TS_BASE.
4918 * tree-core.h (tree_optimization_option): Derive from tree_base.
4919 (tree_target_option): Likewise.
4921 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
4923 * config/i386/i386.c (memory_address_length): Increase len
4924 only when rip_relative_addr_p returns false.
4926 2017-01-11 Julia Koval <julia.koval@intel.com>
4928 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
4929 (OPTION_MASK_ISA_SGX_SET): New.
4930 (ix86_handle_option): Handle OPT_msgx.
4931 * config.gcc: Added sgxintrin.h.
4932 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
4933 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
4934 * config/i386/i386.c (ix86_target_string): Add -msgx.
4936 (ix86_option_override_internal): Handle new options.
4937 (ix86_valid_target_attribute_inner_p): Add sgx.
4938 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
4939 * config/i386/i386.opt: Add msgx.
4940 * config/i386/sgxintrin.h: New file.
4941 * config/i386/x86intrin.h: Add sgxintrin.h.
4943 2017-01-11 Jakub Jelinek <jakub@redhat.com>
4946 * fold-const.c (maybe_nonzero_address): Return 1 for function
4948 (tree_single_nonzero_warnv_p): Don't handle function local objects
4952 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
4955 2017-01-11 David Malcolm <dmalcolm@redhat.com>
4958 * opts.c: Include "spellcheck.h"
4959 (struct string_fragment): New struct.
4960 (struct edit_distance_traits<const string_fragment &>): New
4962 (get_closest_sanitizer_option): New function.
4963 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
4965 2017-01-11 Jakub Jelinek <jakub@redhat.com>
4967 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
4969 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
4970 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
4971 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
4972 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
4973 for initial die_offset if dwarf_split_debug_info.
4974 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
4975 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
4977 (output_skeleton_debug_sections): Formatting fix. Use
4978 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
4979 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
4981 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
4983 * config/arm/cortex-a53.md: Add bypasses for
4985 (cortex_a53_r2f): Only use for transfers.
4986 (cortex_a53_f2r): Likewise.
4987 (cortex_a53_r2f_cvt): Add reservation for conversions.
4988 (cortex_a53_f2r_cvt): Likewise.
4990 2017-01-11 Tamar Christina <tamar.christina@arm.com>
4992 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
4993 to all inlined functions, change static to extern.
4995 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
4998 * config/arm/arm.c (legitimize_pic_address): Handle reference to
5000 (arm_assemble_integer): Likewise.
5002 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
5004 * config.gcc: Use new awk script to check CPU, FPU and architecture
5005 parameters for --with-... options.
5006 * config/arm/parsecpu.awk: New file
5007 * config/arm/arm-cpus.in: New file.
5008 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
5010 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
5012 * config/arm/t-arm: Update dependency rules.
5013 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
5014 of processing .def files.
5015 * config/arm/genopt.sh: Deleted.
5016 * config/arm/gentune.sh: Deleted.
5017 * config/arm/arm-cores.def: Deleted.
5018 * config/arm/arm-arches.def: Deleted.
5019 * config/arm/arm-fpus.def: Deleted.
5020 * config/arm/arm-tune.md: Regenerated.
5021 * config/arm/arm-tables.opt: Regenerated.
5022 * config/arm/arm-cpu.h: New generated file.
5023 * config/arm/arm-cpu-data.h: New generated file.
5024 * config/arm/arm-cpu-cdata.h: New generated file.
5026 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
5029 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
5031 (input_varpool_node): Unpack dynamically_initialized bit.
5033 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
5035 PR rtl-optimization/79032
5036 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
5037 the alignment of the adjusted memory reference against that of MODE,
5038 instead of the alignment of the original memory reference.
5040 2017-01-11 Martin Jambor <mjambor@suse.cz>
5042 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
5044 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
5045 decorated functions.
5047 2017-01-11 Richard Biener <rguenther@suse.de>
5049 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
5050 set range/nonnull info for PHI results. Do not set it on
5051 stmts marked for removal.
5053 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
5055 * expr.c (store_field): In the bitfield case, fetch the return value
5056 from the registers before applying a single big-endian adjustment.
5057 Always do a final load for a BLKmode value not larger than a word.
5059 2017-01-10 David Malcolm <dmalcolm@redhat.com>
5062 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
5063 that we correctly handle column numbers greater than
5064 LINE_MAP_MAX_COLUMN_NUMBER.
5066 2017-01-10 Martin Sebor <msebor@redhat.com>
5069 * gimple-ssa-sprintf.c (get_destination_size): Call
5070 {init,fini}object_sizes.
5071 * tree-object-size.c (addr_object_size): Adjust.
5072 (pass_through_call): Adjust.
5073 (pass_object_sizes::execute): Adjust.
5074 * tree-object-size.h (fini_object_sizes): Declare.
5076 2017-01-10 Martin Sebor <msebor@redhat.com>
5078 PR tree-optimization/78775
5079 * builtins.c (get_size_range): Move...
5080 * calls.c: ...to here.
5081 (alloc_max_size): Accept zero argument.
5082 (operand_signed_p): Remove.
5083 (maybe_warn_alloc_args_overflow): Call get_size_range.
5084 * calls.h (get_size_range): Declare.
5086 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
5088 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
5089 from TI's devices.csv file as of September 2016.
5090 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
5092 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
5094 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
5095 * doc/invoke.texi: Likewise.
5096 * doc/md.texi: Likewise.
5097 * doc/objc.texi: Likewise.
5099 2017-01-10 Joshua Conner <joshconner@google.com>
5101 * config/arm/fuchsia-elf.h: New file.
5102 * config/fuchsia.h: New file.
5103 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
5104 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
5106 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
5108 2016-01-10 Richard Biener <rguenther@suse.de>
5110 PR tree-optimization/79034
5111 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
5112 Propagate out degenerate PHIs in the joiner.
5114 2017-01-10 Martin Liska <mliska@suse.cz>
5116 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
5117 (sort_congruence_classes_by_decl_uid): Likewise.
5118 (sort_congruence_class_groups_by_decl_uid): Likewise.
5119 (sem_item_optimizer::merge_classes): Sort class, groups in these
5120 classes and members in the groups by DECL_UID of declarations.
5121 This would make merge operations stable.
5123 2017-01-10 Martin Liska <mliska@suse.cz>
5125 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
5126 usage of m_classes_vec.
5127 (sem_item_optimizer::~sem_item_optimizer): Likewise.
5128 (sem_item_optimizer::get_group_by_hash): Likewise.
5129 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
5130 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
5131 (sem_item_optimizer::verify_classes): Likewise.
5132 (sem_item_optimizer::process_cong_reduction): Likewise.
5133 (sem_item_optimizer::dump_cong_classes): Likewise.
5134 (sem_item_optimizer::merge_classes): Likewise.
5135 * ipa-icf.h (congruence_class_hash): Rename from
5136 congruence_class_group_hash. Remove declaration of m_classes_vec.
5138 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
5140 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
5141 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
5142 * config.gcc: Add avx512vpopcntdqintrin.h.
5143 * config/i386/avx512vpopcntdqintrin.h: New.
5144 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
5145 * config/i386/i386-builtin-types.def: Add new types.
5146 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
5147 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
5148 __builtin_ia32_vpopcountq_v8di_mask): New.
5149 * config/i386/i386-c.c (ix86_target_macros_internal): Define
5150 __AVX512VPOPCNTDQ__.
5151 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
5152 (PTA_AVX512VPOPCNTDQ): Define.
5153 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
5154 TARGET_AVX512VPOPCNTDQ_P): Define.
5155 * config/i386/i386.opt: Add mavx512vpopcntdq.
5156 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
5157 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
5159 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5162 * predict.def (PRED_CALL): Set to 67.
5164 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
5166 * expr.c (store_field): In the bitfield case, if the value comes from
5167 a function call and is of an aggregate type returned in registers, do
5168 not modify the field mode; extract the value in all cases if the mode
5169 is BLKmode and the size is not larger than a word.
5171 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
5174 * config/i386/cpuid.h: Fix undefined behavior.
5176 2017-01-04 Jeff Law <law@redhat.com>
5178 PR tree-optimization/79007
5179 PR tree-optimization/67955
5180 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
5181 conservative for pt.null when flag_non_call_exceptions is on.
5183 2017-01-09 Jakub Jelinek <jakub@redhat.com>
5185 PR translation/79019
5186 PR translation/79020
5187 * params.def (PARAM_INLINE_MIN_SPEEDUP,
5188 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
5189 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
5191 * config/avr/avr.opt (maccumulate-args): Likewise.
5192 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
5193 * common.opt (freport-bug): Likewise.
5194 * cif-code.def (CIF_FINAL_ERROR): Likewise.
5195 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
5196 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
5197 translatable string.
5198 * config/i386/i386.c (function_value_32): Likewise.
5199 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
5200 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
5202 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
5203 * common/config/msp430/msp430-common.c (msp430_handle_option):
5205 * symtab.c (symtab_node::verify_base): Likewise.
5206 * opts.c (set_debug_level): Likewise.
5207 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
5208 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
5209 missing whitespace to translatable strings.
5210 * config/avr/avr.md (bswapsi2): Fix typo in comment.
5211 * config/sh/superh.h: Likewise.
5212 * config/i386/xopintrin.h: Likewise.
5213 * config/i386/znver1.md: Likewise.
5214 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
5215 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
5216 * double-int.h (struct double_int): Likewise.
5217 * double-int.c (div_and_round_double): Likewise.
5218 * wide-int.cc: Likewise.
5219 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
5220 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
5221 * cfgcleanup.c (crossjumps_occured): Renamed to ...
5222 (crossjumps_occurred): ... this.
5223 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
5226 PR tree-optimization/78899
5227 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
5228 returning bool return struct loop *, NULL for failure and the new
5230 (versionable_outer_loop_p): Don't version outer loop if it has
5231 dont_vectorized bit set.
5232 (tree_if_conversion): When versioning outer loop, ensure
5233 tree_if_conversion is performed also on the inner loop of the
5234 non-vectorizable outer loop copy.
5235 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
5236 LOOP_VECTORIZED in inner loop of the scalar outer loop and
5237 prevent vectorization of it.
5238 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
5239 the outer loop vectorization of the non-scalar version is attempted
5240 before vectorization of the inner loop in scalar version. If
5241 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
5242 vectorization of its inner loop.
5243 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
5244 has 2 inner loops, rename also on edges from bb whose single pred
5245 is outer_loop->header. Fix typo in function comment.
5247 2017-01-09 Martin Sebor <msebor@redhat.com>
5250 * asan.c (asan_emit_stack_protection): Increase local buffer size
5251 to avoid snprintf truncation warning.
5253 2017-01-09 Andrew Pinski <apinski@cavium.com>
5255 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
5256 to reference thunderx2t99 for the tuning structure
5257 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
5259 (thunderx2t99_extra_costs): This.
5260 * config/aarch64/aarch64-tune.md: Regenerate.
5261 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
5262 (vulcan_addrcost_table): This.
5263 (vulcan_regmove_cost): Rename to ...
5264 (thunderx2t99_regmove_cost): This.
5265 (vulcan_vector_cost): Rename to ...
5266 (thunderx2t99_vector_cost): this.
5267 (vulcan_branch_cost): Rename to ...
5268 (thunderx2t99_branch_cost): This.
5269 (vulcan_tunings): Rename to ...
5270 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
5271 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
5273 2017-01-09 Martin Jambor <mjambor@suse.cz>
5277 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
5278 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
5279 (propagate_vr_accross_jump_function): Use the above function for all
5280 value range computations for pass-through jump functions and type
5281 converasion from explicit value range values.
5282 (ipcp_propagate_stage): Do not attempt to deduce types of formal
5283 parameters from TYPE_ARG_TYPES.
5284 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
5285 (ipa_write_node_info): Stream type of the actual argument.
5286 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
5288 2017-01-09 Martin Liska <mliska@suse.cz>
5291 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
5292 (lookup_compiler): Do not show error message with have_E.
5294 2017-01-09 Jakub Jelinek <jakub@redhat.com>
5296 PR tree-optimization/78938
5297 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
5298 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
5299 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
5300 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
5303 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5305 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
5308 2017-01-09 Richard Biener <rguenther@suse.de>
5310 PR tree-optimization/78997
5311 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
5312 name condition properly.
5314 2017-01-09 Richard Biener <rguenther@suse.de>
5317 * dwarf2out.c (is_cxx): New overload with context.
5318 (is_naming_typedef_decl): Use it.
5320 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5322 * invoke.texi (Option Summary): Correct spacing in option lists
5323 and add line breaks to fix over-long lines.
5325 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5329 * extend.texi (Common Variable Attributes): Add xref to GCC
5330 Internals manual to explain mode attribute keywords.
5332 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5335 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
5336 and Preprocessor Options.
5337 (Options for Linking): Document -pthread here....
5338 (RS/6000 and PowerPC Options): ...not here.
5339 (Solaris 2 Options): ...or here.
5340 * doc/cppopts.texi: Document -pthread.
5342 2017-01-08 Martin Sebor <msebor@redhat.com>
5345 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
5346 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
5347 New member functions.
5348 (format_directive): Used them.
5350 (pass_sprintf_length::handle_gimple_call): Same.
5351 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
5352 to avoid truncation for any argument.
5353 (extract_affine_mul): Same.
5354 * tree.c (get_file_function_name): Same.
5356 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5359 * predict.def (PRED_INDIR_CALL): Set to 86.
5361 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5363 PR preprocessor/54124
5364 * doc/cppopts.texi: Reformat -d subtable to list the full name
5365 of the options. Add cross-reference to the docs for the general
5366 compiler -d options.
5367 * doc/invoke.texi (Developer Options): Add cross-reference to the
5368 preprocessor-specific -d option documentation.
5370 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5372 PR preprocessor/13498
5373 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
5374 redudant material, and reflect new command-line options.
5375 (System Headers): Likewise.
5377 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5379 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
5380 -isystem, and -idirafter. Copy-edit.
5381 * doc/cppopts.texi: Copy-edit. Remove contradiction about
5382 default for -ftrack-macro-expansion. Delete obsolete and
5383 badly-formatted implementation details about -fdebug-cpp output.
5384 * doc/cppwarnopts.texi: Copy-edit.
5386 2017-01-07 David Malcolm <dmalcolm@redhat.com>
5389 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
5390 that the transition from a max line width >= 1<<10 to narrower
5391 lines works correctly.
5393 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
5395 * doc/options.texi (PerFunction): New.
5396 * opt-functions.awk (switch_flags): Map both Optimization and
5397 PerFunction to CL_OPTIMIZATION.
5398 * opth-gen.awk: Test for PerFunction flag along with
5400 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
5401 it only when the latter is present. Skip those that don't in
5402 the hash function generator.
5403 * common.opt (fvar-tracking): Mark as PerFunction instead of
5405 (fvar-tracking-assignments): Likewise.
5406 (fvar-tracking-assignments-toggle): Likewise.
5407 (fvar-tracking-uninit): Likewise.
5409 2017-01-07 Jakub Jelinek <jakub@redhat.com>
5411 PR translation/79018
5412 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
5415 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
5418 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
5419 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
5420 TARGET_LONG_JUMP_TABLE_OFFSETS.
5421 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
5422 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
5423 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
5424 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
5425 * config/m68k/m68k.md (tablejump expander): Likewise.
5426 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
5427 TARGET_LONG_JUMP_TABLE_OFFSETS.
5428 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
5429 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
5431 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
5432 David Holsgrove <david.holsgrove@xilinx.com>
5434 * common/config/microblaze/microblaze-common.c
5435 (TARGET_EXCEPT_UNWIND_INFO): Remove.
5436 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
5438 * config/microblaze/microblaze.c (microblaze_must_save_register)
5439 (microblaze_expand_epilogue, microblaze_return_addr): Handle
5441 (microblaze_eh_return): New function.
5442 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
5443 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
5444 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
5445 * config/microblaze/microblaze.md (eh_return): New pattern.
5447 2017-01-06 Jakub Jelinek <jakub@redhat.com>
5449 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
5450 GCC_DIAGNOSTIC_STRINGIFY): Define.
5452 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
5454 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5456 * config/arm/arm.md (<mcrr>): New.
5458 * config/arm/arm.c (arm_arch5te): New.
5459 (arm_option_override): Set arm_arch5te.
5460 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
5462 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
5463 (arm_mcrr_qualifiers): ... this. New.
5464 (MRRC_QUALIFIERS): Define to...
5465 (arm_mrrc_qualifiers): ... this. New.
5466 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
5468 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
5469 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
5470 (MRRCI, mrrc, MRRC): New.
5471 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
5472 VUNSPEC_MRRC2): New.
5474 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5476 * config/arm/arm.md (<mcr>): New.
5478 * config/arm/arm.c (arm_coproc_builtin_available): Add
5479 support for mcr, mrc, mcr2 and mrc2.
5480 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
5481 (arm_mcr_qualifiers): ... this. New.
5482 (MRC_QUALIFIERS): Define to ...
5483 (arm_mrc_qualifiers): ... this. New.
5484 (MCR_QUALIFIERS): Define to ...
5485 (arm_mcr_qualifiers): ... this. New.
5486 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
5488 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
5489 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
5490 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
5493 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5495 * config/arm/arm.md (*ldc): New.
5499 * config/arm/arm.c (arm_coproc_builtin_available): Add
5500 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
5501 (arm_coproc_ldc_stc_legitimate_address): New.
5502 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
5503 'qualifier_const_pointer'.
5504 (LDC_QUALIFIERS): Define to...
5505 (arm_ldc_qualifiers): ... this. New.
5506 (STC_QUALIFIERS): Define to...
5507 (arm_stc_qualifiers): ... this. New.
5508 * config/arm/arm-protos.h
5509 (arm_coproc_ldc_stc_legitimate_address): New.
5510 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
5511 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
5512 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
5513 stc2, stcl, stc2l): New.
5514 * config/arm/constraints.md (Uz): New.
5515 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
5516 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
5517 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
5518 VUNSPEC_STC2L): New.
5520 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5522 * config/arm/arm.md (<cdp>): New.
5523 * config/arm/arm.c (neon_const_bounds): Rename this ...
5524 (arm_const_bounds): ... this.
5525 (arm_coproc_builtin_available): New.
5526 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
5527 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
5528 (CDP_QUALIFIERS): Define to...
5529 (arm_cdp_qualifiers): ... this. New.
5531 (arm_expand_builtin_args): Add case for 6 arguments.
5532 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
5533 (arm_const_bounds): ... this.
5534 (arm_coproc_builtin_available): New.
5535 * config/arm/arm_acle.h (__arm_cdp): New.
5537 * config/arm/arm_acle_builtins.def (cdp): New.
5539 * config/arm/iterators.md (CDPI,CDP,cdp): New.
5540 * config/arm/neon.md: Rename all 'neon_const_bounds' to
5542 * config/arm/types.md (coproc): New.
5543 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
5544 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
5545 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
5546 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
5548 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5550 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
5551 (UBINOP_QUALIFIERS): New.
5553 (acle_builtin_data): New. Change comment.
5554 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
5555 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
5556 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
5557 arm_acle_builtins.def.
5558 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
5559 (arm_init_acle_builtins): New.
5560 (CRC32_BUILTIN): Remove.
5561 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
5562 crc32cb, crc32ch and crc32cw.
5563 (arm_init_crc32_builtins): Remove.
5564 (arm_init_builtins): Use arm_init_acle_builtins rather
5565 than arm_init_crc32_builtins.
5566 (arm_expand_acle_builtin): New.
5567 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
5568 * config/arm/arm_acle_builtins.def: New.
5570 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5572 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
5573 (arm_builtin_datum): ... this.
5574 (arm_init_neon_builtin): Rename to ...
5575 (arm_init_builtin): ... this. Add a new parameters PREFIX
5576 and USE_SIG_IN_NAME.
5577 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
5578 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
5579 'arm_builtin_datum'.
5580 (arm_init_vfp_builtins): Likewise.
5581 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
5582 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
5583 (arm_expand_neon_args): Rename to ...
5584 (arm_expand_builtin_args): ... this. Rename builtin_arg
5585 enum values and differentiate between ARG_BUILTIN_MEMORY
5586 and ARG_BUILTIN_NEON_MEMORY.
5587 (arm_expand_neon_builtin_1): Rename to ...
5588 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
5589 values, arm_expand_builtin_args and add bool parameter NEON.
5590 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
5591 (arm_expand_vfp_builtin): Likewise.
5592 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
5594 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5597 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
5598 * predict.c (tree_estimate_probability_bb): Reverse direction of
5599 polymorphic call predictor.
5601 2017-01-06 David Malcolm <dmalcolm@redhat.com>
5603 * passes.c (execute_one_pass): Split out pass-skipping logic into...
5604 (determine_pass_name_match): ...this new function and...
5605 (should_skip_pass_p): ...this new function.
5607 2017-01-06 Nathan Sidwell <nathan@acm.org>
5609 * ipa-visibility.c (function_and_variable_visibility): Reformat
5610 comments and long lines. Remove extrneous if.
5611 * symtab.c (symtab_node::make_decl_local): Fix code format.
5612 (symtab_node::set_section_for_node): Fix comment typo.
5614 2017-01-06 Martin Liska <mliska@suse.cz>
5617 * lra-constraints.c: Rename invariant to lra_invariant.
5618 * predict.c (set_even_probabilities): Initialize e to NULL.
5620 2017-01-05 Martin Sebor <msebor@redhat.com>
5622 PR tree-optimization/78910
5623 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
5624 (format_integer): Correct off-by-one error in the handling
5625 of precision with negative numbers in signed conversions..
5627 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
5629 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
5631 2017-01-05 Jakub Jelinek <jakub@redhat.com>
5633 PR tree-optimization/71016
5634 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
5635 factor_out_conditional_conversion. Formatting fix.
5636 (factor_out_conditional_conversion): Add cond_stmt argument.
5637 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
5638 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
5641 2017-01-05 David Malcolm <dmalcolm@redhat.com>
5643 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
5644 read-rtl-function.o, and selftest-rtl.o.
5645 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
5646 (selftest::aarch64_test_loading_full_dump): New function.
5647 (selftest::aarch64_run_selftests): New function.
5648 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
5649 selftest::aarch64_run_selftests.
5650 * config/i386/i386.c
5651 (selftest::ix86_test_loading_dump_fragment_1): New function.
5652 (selftest::ix86_test_loading_call_insn): New function.
5653 (selftest::ix86_test_loading_full_dump): New function.
5654 (selftest::ix86_test_loading_unspec): New function.
5655 (selftest::ix86_run_selftests): Call the new functions.
5656 * emit-rtl.c (maybe_set_max_label_num): New function.
5657 * emit-rtl.h (maybe_set_max_label_num): New decl.
5658 * function.c (instantiate_decls): Guard call to
5659 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
5660 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
5662 * gensupport.c (gen_reader::gen_reader): Pass "false"
5663 for new "compact" param of rtx_reader.
5664 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
5665 rather than an empty string for NULL strings.
5666 * read-md.c: Potentially include config.h rather than bconfig.h.
5667 Wrap include of errors.h with #ifdef GENERATOR_FILE.
5668 (have_error): New global, copied from errors.c.
5669 (md_reader::read_name): Rename to...
5670 (md_reader::read_name_1): ...this, adding "out_loc" param,
5671 and converting "missing name or number" to returning false, rather
5673 (md_reader::read_name): Reimplement in terms of read_name_1.
5674 (md_reader::read_name_or_nil): New function.
5675 (md_reader::read_string): Handle "(nil)" by returning NULL.
5676 (md_reader::md_reader): Add new param "compact".
5677 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
5678 (md_reader::read_file): New method.
5679 * read-md.h (md_reader::md_reader): Add new param "compact".
5680 (md_reader::read_file): New method.
5681 (md_reader::is_compact): New accessor.
5682 (md_reader::read_name): Convert return type from void to file_location.
5683 (md_reader::read_name_or_nil): New decl.
5684 (md_reader::read_name_1): New decl.
5685 (md_reader::m_compact): New field.
5686 (noop_reader::noop_reader): Pass "false" for new "compact" param
5688 (rtx_reader::rtx_reader): Add new "compact" param.
5689 (rtx_reader::read_rtx_operand): Make virtual and convert return
5690 type from void to rtx.
5691 (rtx_reader::read_until): New decl.
5692 (rtx_reader::handle_any_trailing_information): New virtual function.
5693 (rtx_reader::postprocess): New virtual function.
5694 (rtx_reader::finalize_string): New virtual function.
5695 (rtx_reader::m_in_call_function_usage): New field.
5696 (rtx_reader::m_reuse_rtx_by_id): New field.
5697 * read-rtl-function.c: New file.
5698 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
5699 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
5700 (selftest::verify_three_block_rtl_cfg): New decl.
5701 * read-rtl-function.h: New file.
5702 * read-rtl.c: Potentially include config.h rather than bconfig.h.
5703 For host, include function.h, memmodel.h, and emit-rtl.h.
5704 (one_time_initialization): New function.
5705 (struct compact_insn_name): New struct.
5706 (compact_insn_names): New array.
5707 (find_code): Handle insn codes in compact dumps.
5708 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
5709 (bind_subst_iter_and_attr): Likewise.
5710 (add_condition_to_string): Likewise.
5711 (add_condition_to_rtx): Likewise.
5712 (apply_attribute_uses): Likewise.
5713 (add_current_iterators): Likewise.
5714 (apply_iterators): Likewise.
5715 (initialize_iterators): Guard usage of apply_subst_iterator with
5716 #ifdef GENERATOR_FILE.
5717 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
5718 (md_reader::read_mapping): Likewise.
5719 (add_define_attr_for_define_subst): Likewise.
5720 (add_define_subst_attr): Likewise.
5721 (read_subst_mapping): Likewise.
5722 (check_code_iterator): Likewise.
5723 (rtx_reader::read_rtx): Likewise. Move one-time initialization
5725 (one_time_initialization): New function.
5726 (rtx_reader::read_until): New method.
5727 (read_flags): New function.
5728 (parse_reg_note_name): New function.
5729 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
5730 Handle reuse_rtx ids.
5731 Wrap iterator lookup within #ifdef GENERATOR_FILE.
5732 Add parsing support for RTL dumps, mirroring the special-cases in
5733 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
5734 values, and calling handle_any_trailing_information.
5735 (rtx_reader::read_rtx_operand): Convert return type from void
5736 to rtx, returning return_rtx. Handle case 'e'. Call
5737 finalize_string on XSTR and XTMPL fields.
5738 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
5739 "(nil)" values were omitted. Call the postprocess vfunc on the
5741 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
5742 class ctor. Initialize m_in_call_function_usage. Call
5743 one_time_initialization.
5744 * rtl-tests.c (selftest::test_uncond_jump): Call
5745 set_new_first_and_last_insn.
5746 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
5747 * selftest-rtl.c: New file.
5748 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
5749 (selftest::get_insn_by_uid): New decl.
5750 * selftest-run-tests.c (selftest::run_tests): Call
5751 read_rtl_function_c_tests.
5752 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
5753 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
5756 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
5758 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
5759 operands in a special way. Assert that pos+len <= mode precision.
5761 2017-01-05 Jakub Jelinek <jakub@redhat.com>
5763 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
5764 3 argument Alias with unlimited for the negative form.
5765 (fno-vect-cost-model): Removed.
5767 2017-01-05 Martin Liska <mliska@suse.cz>
5769 * hsa-gen.c (gen_hsa_divmod): New function.
5770 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
5772 2017-01-05 Martin Liska <mliska@suse.cz>
5775 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
5778 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5780 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
5781 small constant length operands.
5783 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5785 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
5786 between loop iterations.
5788 2017-01-05 Martin Liska <mliska@suse.cz>
5791 * gimplify.c (gimplify_decl_expr): Compare to
5792 asan_poisoned_variables instread of checking flags.
5793 (gimplify_target_expr): Likewise.
5794 (gimplify_expr): Likewise.
5795 (gimplify_function_tree): Conditionally initialize
5796 asan_poisoned_variables.
5798 2017-01-04 Jeff Law <law@redhat.com>
5800 PR tree-optimizatin/78812
5801 * rtl.h (contains_mem_rtx_p): Prototype.
5802 * ifcvt.c (containts_mem_rtx_p): Move from here to...
5803 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
5804 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
5805 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
5806 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
5808 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5810 * input.c (assert_char_at_range): Default-initialize actual_range.
5812 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5814 * df-scan.c (df_ref_create_structure): Make regno unsigned,
5815 to match the caller.
5817 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5819 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
5820 insns after final jump in test to emit dummy move.
5822 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5824 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
5825 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
5827 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5829 * multiple_target.c (create_dispatcher_calls): Init e_next.
5830 * tree-ssa-loop-split.c (split_loop): Init border.
5831 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
5834 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
5839 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
5840 (altivec_register_operand): Do not return true if the operand
5841 contains a SUBREG mixing SImode and SFmode.
5842 (vsx_register_operand): Likewise.
5843 (vsx_reg_sfsubreg_ok): New predicate.
5844 (vfloat_operand): Do not return true if the operand contains a
5845 SUBREG mixing SImode and SFmode.
5846 (vint_operand): Likewise.
5847 (vlogical_operand): Likewise.
5848 (gpc_reg_operand): Likewise.
5849 (int_reg_operand): Likewise.
5850 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
5851 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
5852 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
5854 (rs6000_emit_move_si_sf_subreg): New helper function.
5855 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
5856 fixup SUBREGs involving SImode and SFmode.
5857 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
5858 numbers for the new peephole2 optimization.
5859 (peephole2 for SFmode unions): New peephole2 to optimize cases in
5860 the GLIBC math library that do AND/IOR/XOR operations on single
5861 precision floating point.
5862 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
5863 target macros to say whether we need to avoid SUBREGs mixing
5865 (TARGET_ALLOW_SF_SUBREG): Likewise.
5866 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
5867 (UNSPEC_SI_FROM_SF): Likewise.
5868 (iorxor): Change spacing.
5869 (and_ior_xor): New iterator for AND, IOR, and XOR.
5870 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
5871 (movdi_from_sf_zero_ext): Likewise.
5872 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
5873 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
5874 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
5875 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
5876 (fms<mode>4): Likewise.
5877 (fnma<mode>4): Likewise.
5878 (fnms<mode>4): Likewise.
5879 (nfma<mode>4): Likewise.
5880 (nfms<mode>4): Likewise.
5882 2017-01-04 Marek Polacek <polacek@redhat.com>
5885 * doc/invoke.texi: Document -Wpointer-compare.
5887 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5889 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
5892 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
5893 descriptions for -gdwarf-5 and emit them as uleb128 instead of
5896 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
5899 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
5900 documentation of the powerpc_popcntb_ok attribute.
5901 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
5902 code to issue warning messages if a requested CPU configuration is
5903 not supported by the binary (assembler and loader) toolchain.
5904 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
5905 made to define a built-in function that has been disabled.
5906 (paired_init_builtins): Add assertion to prevent ICE if attempt is
5907 made to define a built-in function that has been disabled.
5908 (altivec_init_builtins): Add comment explaining why definition
5909 of the DST built-in functions is not preceded by an assertion
5910 check. Add assertions to prevent ICE if attempts are made to
5911 define an altivec predicate or an abs* built-in function that has
5913 (htm_init_builtins): Add comment explaining why definition of the
5914 htm built-in functions is not preceded by an assertion check.
5916 2017-01-04 Jeff Law <law@redhat.com>
5918 PR tree-optimizatin/67955
5919 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
5920 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
5921 the points-to solution does not include pt_null. Use DECL_PT_UID
5924 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
5926 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
5927 Use gen_int_mode instead of gen_lopwart for const_int operands.
5929 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5931 PR tree-optimization/71563
5932 * match.pd: Simplify X << Y into X if Y is known to be 0 or
5933 out of range value - has low bits known to be zero.
5935 2017-01-04 Alan Modra <amodra@gmail.com>
5937 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
5938 * configure: Regenerate.
5939 * config.in: Regenerate.
5941 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5944 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
5945 a substring of the message, but strcmp with the whole message. Ifdef
5946 ENABLE_NLS, translate the message first using dgettext.
5948 2017-01-03 Jeff Law <law@redhat.com>
5950 PR tree-optimizatin/78856
5951 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
5952 (mark_threaded_blocks): Remove code to truncate thread paths that
5953 cross multiple loop headers. Instead invalidate the cached loop
5954 iteration information and handle case of a thread path walking
5955 into an irreducible region.
5957 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5960 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
5961 assertions. Add support for doing the signbit if the IEEE 128-bit
5962 floating point value is in a GPR.
5963 * config/rs6000/rs6000.md (Fsignbit): Delete.
5964 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
5965 Update the length attribute if the value is in a GPR.
5966 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
5967 the sign or zero extension instruction, since the value is always 0/1.
5968 (signbit<mode>2_dm2): Delete using <Fsignbit>.
5971 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
5972 extracting SImode to a GPR register so that we can generate a
5973 store, limit the vector to be in a traditional Altivec register
5974 for the vextuwrx instruction.
5976 2017-01-03 Ian Lance Taylor <iant@google.com>
5978 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
5980 2017-01-03 Martin Sebor <msebor@redhat.com>
5982 PR tree-optimization/78696
5983 * gimple-ssa-sprintf.c (format_floating): Correct handling of
5984 precision. Use MPFR for %f for greater fidelity. Correct handling
5986 (pass_sprintf_length::compute_format_length): Set width and precision
5987 specified by asrerisk to void_node for vararg functions.
5988 (try_substitute_return_value): Adjust dump output.
5990 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
5992 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
5994 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
5996 * doc/invoke.texi (SPARC options): Document -mlra as the default.
5997 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
5998 -mlra/-mno-lra was passed to the compiler.
6000 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
6002 PR rtl-optimization/65618
6003 * emit-rtl.c (try_split): Move initialization of "before" and
6004 "after" to just before the call to emit_insn_after_setloc.
6006 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
6008 * doc/md.texi (Standard Names): Remove reference to Java frontend.
6010 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
6012 * dwarf2out.c (gen_enumeration_type_die): When
6013 -gno-strict-dwarf, add a DW_AT_encoding attribute.
6015 2017-01-03 Jakub Jelinek <jakub@redhat.com>
6017 PR tree-optimization/78965
6018 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
6019 Change first argument from const call_info & to call_info &. For %n
6020 set info.nowrite to false.
6023 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
6024 possibly throwing calls.
6026 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
6027 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
6028 and fns handling, rather than in a separate case SSA_NAME.
6030 2017-01-02 Jeff Law <law@redhat.com>
6032 * config/darwin-driver.c (darwin_driver_init): Const-correctness
6033 fixes for first_period and second_period variables.
6035 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
6038 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
6039 (*insvqi_1): New insn pattern.
6040 (*insvqi_1_mem_rex64): Ditto.
6042 (*insvqi_3): Rename from *insvqi.
6044 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
6046 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
6048 * doc/cfg.texi (Edges): Remove reference to Java.
6049 (Maintaining the CFG): Ditto.
6051 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
6054 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
6055 transparent aliases.
6057 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
6060 * predict.def (PRED_CALL): Update hitrate.
6061 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
6062 * predict.c (tree_estimate_probability_bb): Split CALL predictor
6063 into direct/indirect/polymorphic variants.
6065 2017-01-01 Jakub Jelinek <jakub@redhat.com>
6067 Update copyright years.
6069 * gcc.c (process_command): Update copyright notice dates.
6070 * gcov-dump.c (print_version): Ditto.
6071 * gcov.c (print_version): Ditto.
6072 * gcov-tool.c (print_version): Ditto.
6073 * gengtype.c (create_file): Ditto.
6074 * doc/cpp.texi: Bump @copying's copyright year.
6075 * doc/cppinternals.texi: Ditto.
6076 * doc/gcc.texi: Ditto.
6077 * doc/gccint.texi: Ditto.
6078 * doc/gcov.texi: Ditto.
6079 * doc/install.texi: Ditto.
6080 * doc/invoke.texi: Ditto.
6082 Copyright (C) 2017 Free Software Foundation, Inc.
6084 Copying and distribution of this file, with or without modification,
6085 are permitted in any medium without royalty provided the copyright
6086 notice and this notice are preserved.