1 2017-03-14 Martin Liska <mliska@suse.cz>
3 * Makefile.in: Install gcov-dump.
5 2017-03-14 Martin Liska <mliska@suse.cz>
7 * multiple_target.c (expand_target_clones): Bail out for
10 2017-03-14 Richard Biener <rguenther@suse.de>
12 * alias.c (struct alias_set_entry): Pack properly.
13 * cfgloop.h (struct loop): Likewise.
14 * cse.c (struct set): Likewise.
15 * ipa-utils.c (struct searchc_env): Likewise.
16 * loop-invariant.c (struct invariant): Likewise.
17 * lra-remat.c (struct cand): Likewise.
18 * recog.c (struct change_t): Likewise.
19 * rtl.h (struct address_info): Likewise.
20 * symbol-summary.h (function_summary): Likewise.
21 * tree-loop-distribution.c (struct partition): Likewise.
22 * tree-object-size.c (struct object_size_info): Likewise.
23 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
24 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
25 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
26 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
27 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
28 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29 (struct _stmt_vec_info): Likewise.
31 2017-03-14 Martin Liska <mliska@suse.cz>
34 * multiple_target.c (create_dispatcher_calls): Check that
35 a target can create a function dispatcher.
37 2017-03-14 Martin Liska <mliska@suse.cz>
40 * multiple_target.c (expand_target_clones): Drop local.local
41 flag for default implementation.
43 2017-03-14 Richard Biener <rguenther@suse.de>
45 PR tree-optimization/80030
46 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
48 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
50 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
51 gcc_fallthrough() instead of __attribute__((fallthrough));
53 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
55 * doc/gcc.texi: Remove "up" link to (DIR).
56 * doc/gccint.texi: Ditto.
58 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
60 * doc/install.texi (Specific) <avr>: Remove reference to
63 2017-03-13 Jeff Law <law@redhat.com>
65 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
66 attribute rather than comments.
68 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
69 match_scratch operand is highest.
71 2017-03-13 Martin Liska <mliska@suse.cz>
74 * ipa-pure-const.c (warn_function_noreturn): If the declarations
75 is a CHKP clone, use original declaration.
77 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
79 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
80 (arc_conditional_register_usage): Use a different allocation order
81 when optimizing for size.
82 * common/config/arc/arc-common.c (arc_option_optimization_table):
83 Section anchors default on when optimizing for size.
85 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
87 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
89 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
91 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
92 * config/arc/arc.md (cpu_facility): Add cd variant.
93 (*movqi_insn): Add code density variant.
94 (*movhi_insn): Likewise.
95 (*movqi_insn): Likewise.
96 (*addsi3_mixed): Likewise.
97 (subsi3_insn): Likewise.
99 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
101 * config/arc/arc.md (movsi_cond_exec): Update constraint.
103 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
105 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
106 expressions with MINUS and UNARY ops.
108 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
111 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
113 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
114 between vec_select and vector argument.
115 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
116 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
117 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
118 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
119 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
120 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
122 2017-03-13 Richard Biener <rguenther@suse.de>
125 * params.def (vect-max-peeling-for-alignment): Fix typo.
127 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
129 * doc/install.texi (Specific) <mips-*-*>: Remove description of
130 issue that only occurred with binutils below 2.18.
132 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
134 * doc/install.texi (Specific) <cris-axis-elf>: No longer
135 refer to binutils 2.11/2.12 minimum.
137 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
139 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
140 ftp.kernel.org and simplify binutils requirement.
142 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
144 * doc/invoke.texi (Warning Options): Fix spelling of link-time
146 (Optimize Options): Ditto. Also remove redundancy.
148 2017-03-10 David Malcolm <dmalcolm@redhat.com>
151 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
153 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
154 to G_ to avoid double translation.
156 2017-03-10 David Malcolm <dmalcolm@redhat.com>
159 * auto-profile.c (get_combined_location): Convert leading
160 character of diagnostics to lower case and remove trailing period.
161 (read_profile): Likewise for various diagnostics.
162 * config/arm/arm.c (arm_option_override): Remove trailing period
163 from various diagnostics.
164 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
165 (msp430_expand_delay_cycles): Likewise.
167 2017-03-10 David Malcolm <dmalcolm@redhat.com>
170 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
171 full command-line argument, rather than just "str".
172 (aarch64_validate_march): Likewise.
173 (aarch64_validate_mtune): Likewise.
175 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
177 PR rtl-optimization/78911
178 * lra-assigns.c (must_not_spill_p): New function.
181 2017-03-10 Jakub Jelinek <jakub@redhat.com>
183 PR tree-optimization/79981
184 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
185 ATOMIC_COMPARE_EXCHANGE ifn result.
186 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
187 IFN_ATOMIC_COMPARE_EXCHANGE.
189 2017-03-10 David Malcolm <dmalcolm@redhat.com>
192 * opts.c (parse_sanitizer_options): Add missing question mark to
193 "did you mean" message.
195 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
197 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
199 (VMULEUH_UNS): Likewise.
200 (VMULOUB_UNS): Likewise.
201 (VMULOUH_UNS): Likewise.
202 * config/rs6000/rs6000.c (builtin_function_type): Remove
203 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
205 2017-03-10 David Malcolm <dmalcolm@redhat.com>
208 * read-rtl-function.c (function_reader::read_rtx_operand): Update
209 x with result of extra_parsing_for_operand_code_0.
210 (function_reader::extra_parsing_for_operand_code_0): Convert
211 return type from void to rtx, returning x. When reading
212 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
213 larger size containing struct block_symbol.
215 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
217 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
218 -mfloat128-hardware without -m64.
220 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
223 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
224 entries to the case statement that marks unsigned arguments to
225 overloaded functions.
227 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
229 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
230 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
232 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
235 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
236 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
238 2017-03-10 Martin Liska <mliska@suse.cz>
242 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
243 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
246 2017-03-10 Olivier Hainque <hainque@adacore.com>
248 * tree-switch-conversion (array_value_type): Start by resetting
249 candidate type to it's main variant.
251 2017-03-10 Jakub Jelinek <jakub@redhat.com>
253 PR rtl-optimization/79909
254 * combine.c (try_combine): Use simplify_replace_rtx on individual
255 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
256 of the whole CALL_INSN_FUNCTION_USAGE.
258 PR tree-optimization/79972
259 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
260 get_range_info on SSA_NAMEs. Formatting fixes.
262 2017-03-10 Richard Biener <rguenther@suse.de>
263 Jakub Jelinek <jakub@redhat.com>
265 PR tree-optimization/77975
266 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
268 (get_val_for): For constant x return it. Formatting fix.
269 (loop_niter_by_eval): Avoid pointless looping if the next iteration
270 would use the same bases as the current one.
272 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
274 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
275 instead of vec_select for V1TImode.
276 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
278 (VSX_LE_128): Add V1TI to this mode iterator.
279 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
280 (*vsx_le_perm_store_<mode>): Likewise.
281 (pre-reload splitter for VSX stores): Likewise.
282 (post-reload splitter for VSX stores): Likewise.
283 (*vsx_xxpermdi2_le_<mode>): Likewise.
284 (*vsx_lxvd2x2_le_<mode>): Likewise.
285 (*vsx_stxvd2x2_le_<mode>): Likewise.
287 2017-03-09 Michael Eager <eager@eagercon.com>
289 Correct failures with --enable-checking=yes,rtl.
291 * config/microblaze/microblaze.c (microblaze_expand_shift):
292 Replace GET_CODE test with CONST_INT_P and INTVAL test with
294 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
295 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
297 2017-03-09 Richard Biener <rguenther@suse.de>
299 PR tree-optimization/79977
300 * graphite-scop-detection.c (scop_detection::merge_sese):
301 Handle the case of extra exits to blocks dominating the entry.
303 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
305 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
308 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
310 PR rtl-optimization/79949
311 * lra-constraints.c (process_alt_operands): Check memory when
312 trying to predict a cycle. Print about the overall increase.
314 2017-03-09 Richard Biener <rguenther@suse.de>
317 * gimple-expr.c (useless_type_conversion_p): Preserve
318 TYPE_SATURATING for fixed-point types.
320 2017-03-09 Richard Biener <rguenther@suse.de>
323 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
324 alignment of BLKmode params.
326 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
329 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
330 (VALL_NO_V2Q): Likewise.
332 * config/aarch64/aarch64-simd.md
333 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
335 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
336 VALL_NO_V2Q mode iterator.
337 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
339 2017-03-09 Martin Liska <mliska@suse.cz>
341 PR tree-optimization/79631
342 * tree-chkp-opt.c (chkp_is_constant_addr): Call
343 tree_int_cst_sign_bit just for INTEGER constants.
345 2017-03-09 Martin Liska <mliska@suse.cz>
349 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
352 2017-03-09 Marek Polacek <polacek@redhat.com>
355 * tree.c (inchash::add_expr): Handle TREE_VEC.
357 2017-03-09 Martin Liska <mliska@suse.cz>
360 (chkp_narrow_size_and_offset): New function.
361 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
362 (void chkp_parse_bit_field_ref): New function.
363 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
364 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
366 2017-03-09 Martin Liska <mliska@suse.cz>
369 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
370 (chkp_find_bounds_1): Remove gcc_unreachable.
372 2017-03-09 Jakub Jelinek <jakub@redhat.com>
375 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
376 BUILT_IN_SYNC*, determine the access type from the size suffix and
377 always build a MEM_REF with that type. Handle forgotten
378 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
381 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
382 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
383 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
384 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
385 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
386 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
387 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
388 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
389 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
390 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
391 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
392 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
393 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
394 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
395 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
396 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
397 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
398 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
399 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
400 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
401 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
402 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
403 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
404 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
405 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
406 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
407 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
408 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
409 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
410 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
411 definitions outside of __OPTIMIZE__ guarded section.
414 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
415 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
416 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
417 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
420 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
422 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
423 ("vfenez<mode>"): Add missing constraints.
425 2017-03-08 Martin Sebor <msebor@redhat.com>
428 * config/nds32/nds32.c (nds32_option_override):
430 2017-03-08 Jakub Jelinek <jakub@redhat.com>
433 * gimplify.c (gimplify_omp_for): Replace index var in outer
434 taskloop statement with an artificial variable and add
435 OMP_CLAUSE_PRIVATE clause for it.
437 2017-03-08 Richard Biener <rguenther@suse.de>
439 PR tree-optimization/79955
440 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
441 for accesses that are completely outside of the variable.
443 2017-03-08 Andrew Haley <aph@redhat.com>
445 PR tree-optimization/79943
446 * tree-ssa-loop-split.c (compute_new_first_bound): When
447 calculating the new upper bound, (END-BEG) should be added, not
450 2017-03-08 Jakub Jelinek <jakub@redhat.com>
452 * config/avr/avr.md (setmemhi): Make sure match_dup
453 operand number comes before match_scratch.
455 2017-03-08 Richard Biener <rguenther@suse.de>
457 PR tree-optimization/79920
458 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
459 with ncopies == 1 to ...
460 (vect_transform_slp_perm_load): ... here. Properly compute
461 all element loads by iterating VF times over the group. Do
462 not handle ncopies (computed in a broken way) in
463 vect_create_mask_and_perm.
465 2017-03-08 Jakub Jelinek <jakub@redhat.com>
468 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
469 is a uniform vector, use uniform_vector_p return value instead of
470 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
472 2017-03-07 Marek Polacek <polacek@redhat.com>
475 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
476 (alloca_call_type): Likewise.
478 2017-03-07 Martin Liska <mliska@suse.cz>
480 * gcov.c (process_args): Put comment to correct location.
482 2017-03-07 Martin Liska <mliska@suse.cz>
485 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
486 Use array_at_struct_end_p instead of DECL_CHAIN (field).
487 (chkp_narrow_bounds_for_field): Likewise.
488 (chkp_parse_array_and_component_ref): Pass one more argument to
491 2017-03-07 Richard Biener <rguenther@suse.de>
493 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
496 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
498 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
499 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
501 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
504 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
505 to end of description.
506 (PARAM_MAX_STORES_TO_MERGE): Likewise.
508 2017-03-07 Jakub Jelinek <jakub@redhat.com>
510 PR rtl-optimization/79901
511 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
513 (*avx512f_<code><mode>3<mask_name>): ... this.
514 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
515 iterator instead of VI8_AVX2_AVX512BW.
517 PR rtl-optimization/79901
518 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
519 min/max expander, expand it using expand_vec_cond_expr.
522 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
525 2017-03-06 Jakub Jelinek <jakub@redhat.com>
528 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
529 to void * for PCH reasons.
530 * dwarf2out.c (output_loc_operands, output_die): Cast
531 v.val_vec.array to unsigned char *.
533 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
536 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
539 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
541 PR rtl-optimization/79571
542 * lra-constraints.c (process_alt_operands): Calculate static
543 reject and subtract it from overall when only addresses will be
546 2017-03-06 Julia Koval <julia.koval@intel.com>
549 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
550 incoming stack boundary to 128 for 64-bit targets.
552 2017-03-06 Richard Biener <rguenther@suse.de>
554 PR tree-optimization/79894
555 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
556 to NULL after folding it.
558 2017-03-06 Richard Biener <rguenther@suse.de>
560 PR tree-optimization/79824
561 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
562 check disabling peeling for gaps.
564 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
566 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
567 attributes): Document gettimeofday.
569 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
571 * config/s390/s390.c (s390_option_override_internal): Set
572 PARAM_MIN_VECT_LOOP_BOUND
574 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
576 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
577 * config/s390/s390.md: Likewise.
579 2017-03-06 Jakub Jelinek <jakub@redhat.com>
582 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
583 (<avx2_avx512>_perm<mode>): Rename to ...
584 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
586 (<avx512>_perm<mode>_mask): Rename to ...
587 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
589 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
590 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
591 instead of VI8F_256_512.
592 (avx512f_perm<mode>): New define_expand.
593 (avx512f_perm<mode>_mask): Likewise.
594 (avx512f_perm<mode>_1<mask_name>): New define_insn.
595 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
597 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
599 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
600 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
602 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
604 2017-03-06 Martin Liska <mliska@suse.cz>
607 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
608 when having a SSA NAME w/o VAR_DECL assigned to it.
610 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
612 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
613 msa_dpsub_<su>_d): Fix MODE for vec_select.
615 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
617 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
619 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
621 2017-03-06 Richard Biener <rguenther@suse.de>
623 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
624 * plugin.c (register_plugin_info): Likewise.
625 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
627 2017-03-05 Jakub Jelinek <jakub@redhat.com>
629 * config/i386/sse.md (sse_storehps, sse_storelps,
630 avx_<castmode><avxsizesuffix>_<castmode>,
631 avx512f_<castmode><avxsizesuffix>_<castmode>,
632 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
633 in condition that at least one operand is not a MEM.
635 2017-03-03 Jakub Jelinek <jakub@redhat.com>
638 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
639 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
641 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
642 gimple_call_nothrow_p flag based on whether original builtin can throw.
643 If it can, emit following stmts on the fallthrough edge.
644 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
645 don't create new bb if inserting just debug stmts on the edge, try to
646 insert them on the fallthru bb or just reset debug stmts.
648 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
651 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
652 restore recog_data (including the operand rtxes inside it) around
653 the call to get_insn_template.
655 2017-03-03 Martin Sebor <msebor@redhat.com>
657 PR tree-optimization/79699
658 * context.c (context::~context): Free MPFR caches to avoid
659 a memory leak on program exit.
661 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
663 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
664 Use wide_int::ulow () instead of .elt (0).
666 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
668 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
669 (*pushxf): Limit oF constraint to 32bit targets and add oC
670 constraint for 64bit targets.
671 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
672 (*pushdf): Change rmF constraint to rmC.
674 2017-03-03 Martin Liska <mliska@suse.cz>
676 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
677 Remove unused variable.
679 2017-03-03 Jakub Jelinek <jakub@redhat.com>
682 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
683 is a memory operand, increase num_memory.
684 (ix86_expand_args_builtin): Likewise.
686 2017-03-03 Jan Hubicka <jh@suse.cz>
689 * ipa-devirt.c (maybe_record_node): Properly handle
690 __cxa_pure_virtual visibility.
692 2017-03-03 Martin Liska <mliska@suse.cz>
694 PR tree-optimization/79803
695 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
697 (pass_loop_prefetch::execute): Disabled optimization if an
698 assumption about L1 cache size is not met.
700 2017-03-03 Martin Liska <mliska@suse.cz>
702 PR rtl-optimization/79574
703 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
704 (hash_scan_set): Likewise.
705 (dump_hash_table): Likewise.
706 (hoist_code): Likewise.
708 2017-03-03 Richard Biener <rguenther@suse.de>
710 * fixed-value.c (fixed_from_string): Restore use of elt (1)
711 in place of uhigh ().
712 (fixed_convert_from_real): Likewise.
714 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
717 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
719 2017-03-03 Richard Biener <rguenther@suse.de>
722 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
723 TYPE_OVERFLOW_UNDEFINED check.
725 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
727 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
729 (vector_ae_<mode>_p): Likewise.
730 (vector_nez_<mode>_p): Likewise.
731 (vector_ne_v2di_p): Likewise.
732 (vector_ae_v2di_p): Likewise.
733 (vector_ne_<mode>_p): Likewise.
734 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
736 (vsx_tsqrt<mode>2_fe): Likewise.
738 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
741 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
743 2017-03-02 Jakub Jelinek <jakub@redhat.com>
745 PR rtl-optimization/79780
746 * cprop.c (one_cprop_pass): When second and further conditional trap
747 in a single basic block is turned into an unconditional trap, turn it
748 into a deleted note to avoid RTL verification failures.
750 2017-03-02 Richard Biener <rguenther@suse.de>
752 * fold-const.c (const_binop): Use ulow () instead of elt (0).
754 2017-03-02 Richard Biener <rguenther@suse.de>
756 PR tree-optimization/79345
758 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
759 param and abort the walk, returning -1 if it is hit.
760 (walk_aliased_vdefs): Take a limit param and pass it on.
761 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
762 defaulting to 0 and return a signed int.
763 * tree-ssa-uninit.c (struct check_defs_data): New struct.
764 (check_defs): New helper.
765 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
766 about uninitialized memory.
767 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
768 bogus uninitialized warning.
769 (fixed_convert_from_real): Likewise.
771 2017-03-02 Bin Cheng <bin.cheng@arm.com>
773 PR tree-optimization/66768
774 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
775 iv_use if base object can't be determined.
777 2017-03-02 Jakub Jelinek <jakub@redhat.com>
779 PR tree-optimization/79345
780 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
781 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
782 (get_pattern_stats): Initialize it.
783 * genemit.c (gen_expand): Verify match_scratch numbers come after
784 match_operand/match_dup numbers.
785 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
786 match_scratch numbers.
787 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
789 * config/s390/s390.md (trunctdsd2): Likewise.
791 2017-03-02 Richard Biener <rguenther@suse.de>
793 * wide-int.h (wide_int_storage::operator=): Implement in terms
796 2017-03-02 Richard Biener <rguenther@suse.de>
798 PR tree-optimization/79777
799 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
800 the to insert expression to sth existing.
802 2017-03-01 Martin Sebor <msebor@redhat.com>
805 * gimple-ssa-sprintf.c
806 (directive::known_width_and_precision): New function.
807 (format_integer): Use it.
808 (get_mpfr_format_length): Consider the full range of precision
809 when computing %g output with the # flag. Set the likely byte
810 count to 3 rather than 1 when precision is indeterminate.
811 (format_floating): Correct the lower bound of precision.
813 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
815 * doc/invoke.texi: Document default code model for 64-bit Linux.
817 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
820 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
821 udiv rather than div since input pattern is unsigned.
823 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
825 * config/i386/i386.c (print_reg): Warn for values of
826 unsupported size in integer register.
828 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
831 * config/rs6000/predicates.md (current_file_function_operand): Do
832 not allow self calls to be local if the function is replaceable.
834 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
837 * config/rs6000/altivec.h (vec_ctz and others): Change the
838 preprocessor macro that controls conditional compilation from
839 _ARCH_PWR9 to __POWER9_VECTOR__.
840 (vec_all_ne): Change parameterization of __altivec_scalar_pred
841 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
842 control (instead of _ARCH_PWR9 control) so that template
843 definition uses power9-specific function.
844 (vec_any_eq): Likewise.
845 (vec_all_ne): Change macro definition to use a power9-specific
846 expansion under #ifdef __POWER9_VECTOR__ control (instead of
848 (vec_any_eq) Likewise.
849 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
850 expansion for CMPNEF to remove support for xvcmpnesp instruction.
851 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
852 support for xvcmpnedp instruction.
853 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
854 macro expansion so that Power9 implementation of vec_all_ne does
855 not use the AltiVec predicate framework.
856 (VCMPNEH_P): Likewise.
857 (VCMPNEW_P): Likewise.
858 (VCMPNED_P): Likewise.
859 (VCMPNEFP_P): Likewise.
860 (VCMPNEDP_P): Likewise.
861 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
862 implementation of vec_any_eq to not use AltiVec predicate
864 (VCMPAEH_P): Likewise.
865 (VCMPAEW_P): Likewise.
866 (VCMPAED_P): Likewise.
867 (VCMPAEFP_P): Likewise.
868 (VCMPAEDP_P): Likewise.
869 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
870 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
871 not use the AltiVec predicate framework.
872 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
873 of vec_any_eq to not use AltiVec predicate framework.
874 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
875 support for predefined __POWER9_VECTOR__ macro to indicate that
876 Power9 instruction selection is enabled.
877 (altivec_overloaded_builtins): Remove extraneous
878 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
879 function argument types RS6000_BTI_bool_V16QI and
880 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
881 entry for overloaded function argument types RS6000_BTI_bool_V4SI
882 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
883 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
884 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
885 Power9 for implementations of vec_cmpne. Change the signature for
886 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
887 (representing vec_all_ne) to remove the previously described first
888 argument of type RS6000_BTI_INTSI, as this was an artifact of
889 reliance on the AltiVec predicate framework, which is no longer
890 used in the implementation of these functions. Add
891 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
892 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
893 since, unlike the AltiVec predicate framework implementation, we
894 do not share function descriptors between vec_alle and vec_anyeq.
895 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
896 set of modes that receive special treatment even when
897 TARGET_P9_VECTOR is true. The special treatment emits code that
898 does not depend on Power9 instructions.
899 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
900 define_expand to not rely on AltiVec predicate framework.
901 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
903 (vector_ne_v2di_p): Change this define_expand to not rely on
904 AltiVec predicate framework.
905 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
907 (vector_ne_<mode>_p): Change this define_expand to not rely on
908 AltiVec predicate framework.
909 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
911 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
912 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
914 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
915 define_insn pattern because the xvcmpne<VSs>. instruction is not
917 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
918 instruction is not supported.
920 2017-03-01 Jakub Jelinek <jakub@redhat.com>
922 * config/nvptx/nvptx.c: Include intl.h.
924 2017-03-01 Martin Jambor <mjambor@suse.cz>
927 * ipa-prop.h (ipa_bits): Removed field known.
928 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
929 to pointers. Adjusted their comments to warn about their sharing.
930 (ipcp_transformation_summary): Change bits to a vector of pointers.
931 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
932 (ipa_get_ipa_bits_for_value): Declare.
933 * tree-vrp.h (value_range): Mark as GTY((for_user)).
934 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
935 (ipa_bits_hash_table): Likewise.
936 (ipa_vr_ggc_hash_traits): Likewise.
937 (ipa_vr_hash_table): Likewise.
938 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
939 being pointers and vr_known being removed.
940 (ipa_set_jf_unknown): Likewise.
941 (ipa_get_ipa_bits_for_value): New function.
942 (ipa_set_jfunc_bits): Likewise.
943 (ipa_get_value_range): New overloaded functions.
944 (ipa_set_jfunc_vr): Likewise.
945 (ipa_compute_jump_functions_for_edge): Use the above functions to
946 construct bits and vr parts of jump functions.
947 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
948 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
950 (ipcp_grow_transformations_if_necessary): Also allocate
951 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
953 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
954 them. Fix too long lines.
955 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
956 vr_known being removed.
957 (ipa_read_jump_function): Use new setter functions to construct bits
958 and vr parts of jump functions or set them to NULL.
959 (write_ipcp_transformation_info): Adjust for bits being pointers.
960 (read_ipcp_transformation_info): Likewise.
961 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
963 Include gt-ipa-prop.h.
964 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
966 (ipcp_store_bits_results): Likewise.
967 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
968 Do not write to existing jump functions but use a temporary instead.
970 2017-03-01 Jakub Jelinek <jakub@redhat.com>
973 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
974 attempt to use its first operand as BIT_FIELD_REF base.
976 2017-03-01 Richard Biener <rguenther@suse.de>
979 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
980 interpolating formula in wrapping arithmetic.
981 (chrec_apply): Convert chrec_evaluate return value to wanted type.
983 2017-03-01 Jakub Jelinek <jakub@redhat.com>
985 PR tree-optimization/79734
986 * tree-vect-generic.c (expand_vector_condition): Optimize
987 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
988 Handle VEC_COND_EXPR where comparison has different inner width from
991 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
993 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
994 markup, and similar issues. Remove @opindex entries for things
995 that aren't options. Add missing -mmpy-option entries.
997 2017-02-28 Jakub Jelinek <jakub@redhat.com>
999 PR tree-optimization/79737
1000 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
1001 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
1002 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
1003 instead of byte_size. Formatting fix.
1004 (shift_bytes_in_array_right): Formatting fix.
1006 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
1009 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
1010 condition on optimize for the leaf function test.
1012 2017-02-28 Martin Liska <mliska@suse.cz>
1015 * read-rtl-function.c (function_reader::handle_unknown_directive):
1016 Bail out when one uses -flto.
1018 2017-02-28 Martin Liska <mliska@suse.cz>
1020 * common.opt: Replace space with tabular for options of <number>
1022 * config/i386/i386.opt: Show <number> value for
1023 -mlarge-data-threshold.
1024 * opts.c (print_filtered_help): Do not display number in hexadecimal
1027 2017-02-28 Martin Liska <mliska@suse.cz>
1029 * common.opt: Fix --help=option -Q for options which are of
1032 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
1034 * config/i386/i386.c (print_reg): Error out for values
1035 of 8-bit size in invalid integer register.
1037 2017-02-28 Martin Sebor <msebor@redhat.com>
1039 PR tree-optimization/79691
1040 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
1042 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1045 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
1046 gcc_unreachable with output_operand_lossage.
1048 2017-02-28 Richard Biener <rguenther@suse.de>
1050 PR tree-optimization/79740
1051 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
1053 (visit_nary_op): Insert the nary into the hashtable if we
1054 pattern-matched sth.
1055 * tree-ssa-pre.c (eliminate_insert): Robustify.
1057 2017-02-28 Richard Biener <rguenther@suse.de>
1060 * fold-const.c (decode_field_reference): Reject out-of-bound
1063 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1065 * config/i386/i386.c: Include intl.h.
1066 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
1067 instead of just cond ? "..." : "...".
1068 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
1069 * coverage.c (read_counts_file): Likewise.
1070 * omp-offload.c: Include intl.h.
1071 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
1072 of just cond ? "..." : "...".
1073 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
1074 of just cond ? "..." : "...".
1076 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
1079 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
1081 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
1082 'tune for' CPU name.
1083 * config/arm/arm-cpu-data.h: Regenerated.
1085 2017-02-28 Richard Biener <rguenther@suse.de>
1087 PR tree-optimization/79732
1088 * tree-inline.c (expand_call_inline): Do not shadow var.
1090 2017-02-28 Richard Biener <rguenther@suse.de>
1092 PR tree-optimization/79723
1093 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
1094 address-space properly.
1096 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
1098 * doc/optinfo.texi (Optimization groups): Fix option used for
1100 * doc/invoke.texi (-fopt-info): Document "omp".
1101 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
1102 (OPTGROUP_ALL): Add OPTGROUP_OMP.
1103 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
1104 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
1105 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
1107 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
1109 * dumpfile.c (optgroup_options): Instead of "openmp", associate
1110 OPTGROUP_OMP with "omp".
1112 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
1115 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
1116 for arithmetic shift of unsigned V2DI.
1118 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
1120 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
1121 arc/linux.h headers.
1122 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
1123 (LINK_SPEC): Likewise.
1124 (ARC_TLS_EXTRA_START_SPEC): Likewise.
1125 (EXTRA_SPECS): Likewise.
1126 (STARTFILE_SPEC): Likewise.
1127 (ENDFILE_SPEC): Likewise.
1128 (LIB_SPEC): Likewise.
1129 (TARGET_SDATA_DEFAULT): Likewise.
1130 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
1131 (MULTILIB_DEFAULTS): Likewise.
1132 (DWARF2_UNWIND_INFO): Likewise.
1133 * config/arc/big.h: New file.
1134 * config/arc/elf.h: Likewise.
1135 * config/arc/linux.h: Likewise.
1136 * config/arc/t-uClibc: Remove.
1138 2017-02-27 Bin Cheng <bin.cheng@arm.com>
1140 PR tree-optimization/77536
1141 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
1142 (tree_transform_and_unroll_loop): Use above function to compute the
1143 estimated niter of unrolled loop and use it when scaling profile.
1144 Also use count info rather than frequency if it's non-zero.
1145 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
1146 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
1147 (vect_transform_loop): Call above function.
1149 2017-02-27 Richard Biener <rguenther@suse.de>
1151 PR tree-optimization/45397
1152 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
1153 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
1154 (visit_nary_op): Add pattern matching for CSEing sign-changed
1155 or truncated operations with wider ones.
1157 2017-02-27 Richard Biener <rguenther@suse.de>
1159 PR tree-optimization/79690
1160 * tree-vect-stmts.c (vectorizable_store): Use vector type
1161 built from the DR with address-space.
1163 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
1165 * doc/invoke.texi (Optimize Options): Refine the description
1166 of asan-use-after-return.
1168 2017-02-25 Alan Modra <amodra@gmail.com>
1170 PR rtl-optimization/79584
1171 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
1172 base, not ad->base_term, the reg within base. Remove assertion
1173 that ad->base == ad->base_term. Replace gen_int_mode using
1174 bogus mode with const0_rtx.
1176 2017-02-25 Jakub Jelinek <jakub@redhat.com>
1179 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
1180 FMA_EXPR like tcc_binary or tcc_unary.
1182 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
1185 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
1187 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
1188 (output_loc_operands): Handle DW_OP_call_ref and
1189 DW_OP_GNU_variable_value.
1190 (struct variable_value_struct): New type.
1191 (struct variable_value_hasher): Likewise.
1192 (variable_value_hash): New variable.
1193 (string_types): Remove.
1194 (copy_loc_descr): New function.
1195 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
1196 (prepend_loc_descr_to_each): New function.
1197 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
1198 instead of add_loc_descr_to_each if the first argument is single
1199 location list and the second has multiple.
1200 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
1201 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
1202 when looking for variable value which doesn't have other location info.
1203 (loc_list_from_tree): Formatting fix.
1204 (gen_array_type_die): Simplify DW_AT_string_length handling.
1205 (adjust_string_types): Remove.
1206 (gen_subprogram_die): Don't call adjust_string_types nor test/set
1207 string_types. Call resolve_variable_values.
1208 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
1209 (resolve_addr_in_expr): Likewise. Add A argument.
1210 (copy_deref_exprloc): Remove deref argument. Adjust for the
1211 original expression being DW_OP_GNU_variable_value with optionally
1212 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
1213 optionally after it.
1214 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
1215 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
1216 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
1217 (variable_value_hasher::hash, variable_value_hasher::equal): New
1219 (resolve_variable_value_in_expr, resolve_variable_value,
1220 resolve_variable_values, note_variable_value_in_expr,
1221 note_variable_value): New functions.
1222 (dwarf2out_early_finish): Call note_variable_value on all toplevel
1225 2017-02-24 Jakub Jelinek <jakub@redhat.com>
1228 * opts.h (handle_generated_option): Add GENERATED_P argument.
1229 * opts-common.c (handle_option): Adjust function comment.
1230 (handle_generated_option): Add GENERATED_P argument, pass it to
1232 (control_warning_option): Pass false to handle_generated_option
1234 * opts.c (maybe_default_option): Pass true to handle_generated_option
1236 * optc-gen.awk: Likewise.
1238 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1240 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
1241 a REG, look at the REG it is a SUBREG of.
1242 (splitter for cmpeqsi_t): Ditto.
1244 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1246 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
1247 the special USEs with the pattern of the insn, not the insn itself.
1249 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
1252 * doc/invoke.texi: Document -mload-store-pairs.
1254 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1255 Sandra Loosemore <sandra@codesourcery.com>
1257 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
1258 argument isn't a CONST_INT.
1259 (nios2_alternate_compare_const): Assert op is a CONST_INT.
1260 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
1261 (nios2_validate_compare): Bypass alternate compare logic if *op2
1263 (ldstwm_operation_p): Return false if first_base is not a REG or
1264 if first_offset is not a CONST_INT.
1266 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1268 * config/cris/cris.md: Use correct operand in a define_peephole2.
1270 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1272 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
1274 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1276 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
1277 this_insn if it is an INSN or JUMP_INSN.
1278 (force_offsettable): Look at base, not at addr.
1279 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
1280 on things that aren't necessarily CONST_INTs.
1282 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
1284 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
1285 -mfpmath=sse is the default also for x86-32 targets with SSE2
1286 instruction set when @option{-ffast-math} is enabled
1288 2017-02-24 Jeff Law <law@redhat.com>
1290 PR rtl-optimizatoin/79286
1291 * ira.c (update_equiv_regs): Drop may_trap_p exception to
1294 2017-02-24 Richard Biener <rguenther@suse.de>
1296 PR tree-optimization/79389
1297 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
1300 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
1302 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
1303 function comment to reflect reality.
1304 (loop_exits_before_overflow): Fix typo in function description.
1306 2017-02-24 Richard Biener <rguenther@suse.de>
1308 PR tree-optimization/79389
1309 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
1310 properly that a threading opportunity exists. Detect conditional
1311 copy/constant propagation opportunities.
1313 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
1315 * config/visium/visium.md (type): Add trap.
1316 (b): New mode attribute.
1317 (*btst): Rename into...
1318 (*btst<mode>): ...this and adjust.
1319 (*cbranchsi4_btst_insn): Rename into...
1320 (*cbranch<mode>4_btst_insn): ...this and adjust.
1321 (trap): New define_insn.
1323 2017-02-23 Jakub Jelinek <jakub@redhat.com>
1325 PR tree-optimization/79389
1326 * ifcvt.c (struct noce_if_info): Add rev_cond field.
1327 (noce_reversed_cond_code): New function.
1328 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
1329 reversed_comparison_code. Formatting fix.
1330 (noce_try_store_flag): Test rev_cond != NULL in addition to
1331 reversed_comparison_code.
1332 (noce_try_store_flag_constants): Likewise.
1333 (noce_try_store_flag_mask): Likewise.
1334 (noce_try_addcc): Use rev_cond if non-NULL instead of
1335 reversed_comparison_code.
1336 (noce_try_cmove_arith): Likewise. Formatting fixes.
1337 (noce_try_minmax, noce_try_abs): Clear rev_cond.
1338 (noce_find_if_block): Initialize rev_cond.
1339 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
1340 instead of false as last argument never attempt to reverse it
1343 2017-02-23 Bin Cheng <bin.cheng@arm.com>
1345 PR tree-optimization/79663
1346 * tree-predcom.c (combine_chains): Process refs in reverse order
1347 only for ZERO length chains, and add explaining comment.
1349 2017-02-23 Jeff Law <law@redhat.com>
1351 PR tree-optimization/79578
1352 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
1353 in call to operand_equal_p.
1355 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
1358 * config/i386/cpuid.h: Fix another undefined behavior.
1360 2017-02-23 Richard Biener <rguenther@suse.de>
1362 PR tree-optimization/79683
1363 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
1364 vector types for data-refs.
1366 2017-02-23 Martin Liska <mliska@suse.cz>
1368 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
1370 2017-02-23 Jakub Jelinek <jakub@redhat.com>
1373 * internal-fn.c (get_range_pos_neg): Moved to ...
1374 * tree.c (get_range_pos_neg): ... here. No longer static.
1375 * tree.h (get_range_pos_neg): New prototype.
1376 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
1377 are known to be in between 0 and signed maximum inclusive, try to
1378 expand both unsigned and signed divmod and use the cheaper one from
1381 2017-02-22 Jeff Law <law@redhat.com>
1383 PR tree-optimization/79578
1384 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
1385 to compare base operands.
1387 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
1390 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
1391 gpc_reg_operand instead of fpr_reg_operand.
1393 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
1395 * config/mips/mips.c (mips_return_in_memory): Force FP
1396 vector types to be returned in memory for o32 ABI.
1398 2017-02-22 Jakub Jelinek <jakub@redhat.com>
1400 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
1401 instead of DW_TAG_member for static data member declarations and don't
1402 set no_linkage_name for static inline data members.
1403 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
1406 2017-02-22 Martin Liska <mliska@suse.cz>
1408 * doc/invoke.texi: Replace inequality signs with square brackets
1411 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1413 PR tree-optimization/68644
1414 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
1416 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
1419 * lra-constraints.c (simplify_operand_subreg): Handle
1420 WORD_REGISTER_OPERATIONS targets.
1422 2017-02-22 Jakub Jelinek <jakub@redhat.com>
1425 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
1426 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
1427 elimination by swapping fld*.
1429 2017-02-22 Richard Biener <rguenther@suse.de>
1431 PR tree-optimization/79673
1432 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
1433 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
1434 irrelevant address-space qualifiers and avoiding a
1435 ADDR_SPACE_CONVERT_EXPR from fold_convert.
1437 2017-02-22 Richard Biener <rguenther@suse.de>
1439 PR tree-optimization/79666
1440 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
1441 to not symbolically negate if that may introduce undefined
1444 2017-02-22 Martin Liska <mliska@suse.cz>
1447 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
1448 * data-streamer-out.c (streamer_write_gcov_count_stream):
1450 * value-prof.c (stream_out_histogram_value): Make assert more
1451 precise based on type of counter.
1453 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
1456 * config/i386/i386.md (standard_x87sse_constant_load splitter):
1457 Use nonimmediate_operand instead of memory_operand for operand 1.
1458 (float-extend standard_x87sse_constant_load splitter): Ditto.
1460 2017-02-21 Jeff Law <law@redhat.com>
1462 PR tree-optimization/79621
1463 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
1464 blocks with edges to themselves.
1466 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1469 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
1470 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
1471 Use gimple_call_builtin_p.
1474 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
1475 on temporarily removed DEBUG_INSNs.
1477 PR tree-optimization/79649
1478 * tree-loop-distribution.c (classify_partition): Give up on
1479 non-generic address space loads/stores.
1481 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
1483 * doc/loop.texi (Loop manipulation): Remove nonexistent
1484 tree_ssa_loop_version from the documentation.
1485 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
1487 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1490 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
1491 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
1492 * config/rs6000/rs6000.c: Include except.h.
1493 (rs6000_expand_split_stack_prologue): Call
1494 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
1496 2017-02-21 Martin Jambor <mjambor@suse.cz>
1499 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
1502 2017-02-21 Martin Jambor <mjambor@suse.cz>
1504 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
1505 for backward compatibility only.
1506 * doc/invoke.texi (Option Summary): Remove all references to
1509 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
1513 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1515 * lra-constraints.c (curr_insn_transform): Handle
1516 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
1518 2017-02-21 Martin Liska <mliska@suse.cz>
1520 * config/i386/i386.opt: Replace -masm-dialect with -masm.
1522 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
1524 PR translation/79638
1525 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
1527 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
1530 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
1531 (arm_function_ok_for_sibcall): Return false for an indirect call by
1532 descriptor if all the argument registers are used.
1533 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
1534 alignment of the function.
1536 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1538 PR tree-optimization/61441
1539 * simplify-rtx.c (simplify_const_unary_operation): For
1540 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
1541 the sNaN unmodified.
1543 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
1545 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
1546 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
1547 instead of SYSTEM_HEADER_DIR.
1549 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
1550 Martin Liška <mliska@suse.cz>
1552 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
1553 Fix typos and grammar, use active voice, and clarify.
1555 2017-02-20 Marek Polacek <polacek@redhat.com>
1558 * gimplify.c (gimplify_expr): Handle unused *&&L;.
1561 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
1563 2017-02-20 Jakub Jelinek <jakub@redhat.com>
1566 * config/i386/i386.c (ix86_expand_builtin): Handle
1567 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
1568 ix86_builtins_isa[fcode].isa as a requirement of those
1569 flags and any other flag in the bitmask.
1570 (ix86_init_mmx_sse_builtins): Use 0 instead of
1571 ~OPTION_MASK_ISA_64BIT as mask.
1572 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
1573 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
1574 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
1575 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
1577 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1580 * lra-constraints.c (split_reg): Check requested split mode
1581 is supported by the register.
1583 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1585 * lra-constraints.c (simplify_operand_subreg): Remove early
1588 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1591 * lra-constraints.c (curr_insn_transform): Tighten condition
1592 for converting SUBREG reloads from OP_OUT to OP_INOUT.
1594 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1597 * lra-constraints.c (curr_insn_transform): Handle
1598 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
1600 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
1603 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
1605 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
1607 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
1610 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
1613 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1615 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
1618 2017-02-18 Jakub Jelinek <jakub@redhat.com>
1620 * final.c (last_columnnum, override_columnnum): New variables.
1621 (final_start_function): Set last_columnnum, pass it to begin_prologue
1622 hook and pass 0 to dwarf2out_begin_prologue.
1623 (final_scan_insn): Update override_columnnum. Pass last_columnnum
1624 to source_line debug hook.
1625 (notice_source_line): Compute last_columnnum and for debug_column_info
1626 return true on column changes.
1627 * debug.h (struct gcc_debug_hooks): Add column argument to
1628 source_line and begin_prologue hooks.
1629 (debug_nothing_int_charstar_int_bool): Remove prototype.
1630 (debug_nothing_int_int_charstar,
1631 debug_nothing_int_int_charstar_int_bool): New prototypes.
1632 (dwarf2out_begin_prologue): Add column argument.
1633 * debug.c (do_nothing_debug_hooks): Adjust source_line and
1634 begin_prologue hooks.
1635 (debug_nothing_int_charstar_int_bool): Remove.
1636 (debug_nothing_int_int_charstar,
1637 debug_nothing_int_int_charstar_int_bool): New functions.
1638 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
1639 through to dwarf2out_source_line.
1640 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
1641 (dwarf2out_source_line): Add column argument, emit it if requested.
1642 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
1644 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
1645 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
1646 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
1647 through to dwarf2out_begin_prologue.
1648 (vmsdbgout_source_line): Add column argument, pass it through to
1649 dwarf2out_source_line.
1650 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
1651 dbxout_source_line caller.
1652 (dbxout_source_line): Add column argument.
1654 * common.opt (gno-column-info, gcolumn-info): New options.
1655 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
1656 (check_die): Also test for multiple DW_AT_decl_column attributes.
1657 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
1658 DW_AT_decl_column if requested.
1659 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
1661 (gen_variable_die): Likewise.
1662 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
1663 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
1666 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
1667 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
1668 (ix86_handle_option): Handle OPT_m3dnowa.
1669 * doc/invoke.texi (-m3dnowa): Document.
1670 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
1671 -m3dnowa instead of -m3dnow -march=athlon.
1674 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
1675 instead of gcc_assert for K, r and R code checks. Formatting fixes.
1677 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1680 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
1681 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
1682 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
1683 generator for vsx_xxpermdi_<mode>_be.
1684 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
1685 force big-endian semantics.
1686 (vsx_xxpermdi_<mode>_be): New define_expand with same
1687 implementation as previous version of vsx_xxpermdi_<mode>.
1689 2017-02-17 Jakub Jelinek <jakub@redhat.com>
1691 PR tree-optimization/79327
1692 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
1693 variable, its initialization and use.
1695 2017-02-17 Julia Koval <julia.koval@intel.com>
1697 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
1698 (OPTION_MASK_ISA_PKU_UNSET): New.
1699 (ix86_handle_option): Handle -mrdpid.
1700 * config/i386/cpuid.h (bit_RDPID): New.
1701 * config/i386/driver-i386.c (host_detect_local_cpu):
1702 Detect RDPID feature.
1703 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
1704 * config/i386/i386-c.c (ix86_target_macros_internal):
1706 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
1707 (ix86_valid_target_attribute_inner_p): Add "rdpid".
1708 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
1709 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
1710 * config/i386/i386.md (define_insn "rdpid"): New.
1711 * config/i386/i386.opt Add -mrdpid.
1712 * config/i386/immintrin.h (_rdpid_u32): New.
1714 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
1716 PR rtl-optimization/79541
1717 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
1718 instead of transforming it into USE.
1720 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
1722 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
1723 If HONOR_SNANS (SFmode) force the input to a register.
1724 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
1725 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
1726 an frsp or similar insn.
1728 2017-02-17 Martin Liska <mliska@suse.cz>
1730 PR rtl-optimization/79577
1731 * params.def (selsched-max-sched-times): Increase minimum to 1.
1733 2017-02-17 Martin Liska <mliska@suse.cz>
1735 PR rtl-optimization/79574
1736 * gcse.c (want_to_gcse_p): Prevent integer overflow.
1738 2017-02-17 Martin Liska <mliska@suse.cz>
1740 PR tree-optimization/79529
1741 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
1742 ssa_defined_default_def_p to handle cases which are implicitly
1744 * tree-ssa.c (ssa_defined_default_def_p): New function.
1745 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
1746 which are implicitly defined.
1747 * tree-ssa.h (ssa_defined_default_def_p): Declare.
1749 2017-02-17 Richard Biener <rguenther@suse.de>
1752 * params.def (max-ssa-name-query-depth): Limit to 10.
1754 2017-02-17 Richard Biener <rguenther@suse.de>
1756 PR tree-optimization/79552
1757 * tree-ssa-structalias.c (visit_loadstore): Properly verify
1760 2017-02-17 Richard Biener <rguenther@suse.de>
1763 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
1765 2017-02-17 Marek Polacek <polacek@redhat.com>
1768 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
1769 (fold_negate_expr): New wrapper.
1771 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
1773 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
1774 Correct terminology and de-emphasize pre-standard behavior.
1776 2017-02-16 Alan Modra <amodra@gmail.com>
1778 PR rtl-optimization/79286
1779 * ira.c (def_dominates_uses): New function.
1780 (update_equiv_regs): Don't create an equivalence for insns that
1781 may trap where the register def does not dominate the use.
1783 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
1785 PR rtl-optimization/78127
1786 * lra.c (lra): Call lra_eliminate before finish the loop after
1789 2017-02-16 Richard Biener <rguenther@suse.de>
1791 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
1793 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
1794 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
1795 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
1796 (isl_val_int_from_wi): New function.
1797 (extract_affine_gmp): Rename to ...
1798 (extract_affine_wi): ... this, take a widest_int.
1799 (extract_affine_int): Just wrap extract_affine_wi.
1800 (add_param_constraints): Use isl_val_int_from_wi.
1801 (add_loop_constraints): Likewise, and extract_affine_wi.
1803 2017-02-15 Jeff Law <law@redhat.com>
1806 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
1807 ira_init_register_move_cost_if_necessary.
1809 2017-02-15 Martin Sebor <msebor@redhat.com>
1812 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
1813 removed in a prior commit.
1815 2017-02-15 Bin Cheng <bin.cheng@arm.com>
1817 PR tree-optimization/79347
1818 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
1819 counters during peeling.
1821 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
1823 * Makefile.in (site.exp): Remove "set ISLVER".
1825 2017-02-15 Jakub Jelinek <jakub@redhat.com>
1828 * real.c (real_from_integer): Call real_convert even for decimal.
1830 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
1833 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
1835 2017-02-14 Andrew Pinski <apinski@cavium.com>
1837 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
1838 cores and change the partno/implementer to be correct.
1839 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
1840 the 'B" as the implementer.
1841 * config/aarch64/aarch64-tune.md: Regenerate.
1843 2017-02-14 Carl Love <cel@us.ibm.com>
1845 * config/rs6000/rs6000.c: Add case statement entry to make the
1846 xvcvuxdsp built-in argument unsigned.
1847 * config/rs6000/vsx.md: Fix the source and return operand types so they
1848 match the instruction definitions from the ISA document. Fix typo
1849 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
1852 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
1855 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
1856 member early_clobber_alts.
1857 * lra-lives.c (reg_early_clobber_p): New.
1858 (process_bb_lives): Use it.
1859 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
1860 (debug_operand_data): Initialize early_clobber_alts.
1861 (setup_operand_alternative): Set up early_clobber_alts.
1862 (collect_non_operand_hard_regs): Ditto. Pass early clobber
1863 alternatives to new_insn_reg.
1864 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
1866 (lra_update_insn_regno_info): Pass the new arg.
1868 2017-02-14 Jakub Jelinek <jakub@redhat.com>
1871 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
1872 (new_oacc_loop_raw): Don't clear already cleared fields.
1875 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
1876 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
1877 _mm512_prefetch_i64gather_ps): New inline functions and macros.
1879 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
1882 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
1884 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
1887 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
1888 the extra instruction to the right place to store 128-bit constant
1891 2017-02-14 Martin Sebor <msebor@redhat.com>
1894 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
1895 warning for strings of unknown length.
1897 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
1899 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
1901 2017-02-14 Jeff Law <law@redhat.com>
1904 * ira-costs.c (scan_one_insn): Initialize register move costs
1905 for pseudos seen in USE/CLOBBER insns.
1907 PR tree-optimization/79095
1908 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
1909 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
1910 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
1911 if the operands are known to be not equal, then the resulting range
1913 (intersect_ranges): If the new range is ~[0,0] and the old range is
1914 wide, then prefer ~[0,0].
1915 * tree-vrp.c (overflow_comparison_p_1): New function.
1916 (overflow_comparison_p): New function.
1917 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
1918 if NAME is used in an overflow test.
1919 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
1920 overflow check that can be expressed as an equality test, then adjust
1921 ops to be that equality test.
1923 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1925 * config/s390/s390-builtin-types.def: Remove flags argument.
1926 * config/s390/s390.c (s390_init_builtins): Likewise.
1928 2017-02-14 Martin Liska <mliska@suse.cz>
1930 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
1931 vector. Fix trailing white spaces.
1933 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
1935 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
1938 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1940 PR rtl-optimization/68664
1941 * config/arm/arm.c (arm_sched_can_speculate_insn):
1942 New function. Declare prototype.
1943 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
1945 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1947 PR rtl-optimization/68664
1948 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
1950 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
1952 2017-02-14 Amit Pawar <amit.pawar@amd.com>
1954 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
1955 max skip bytes for function, loop and jump.
1957 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1959 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
1960 ABS_EXPR for gimple dump.
1962 2017-02-14 Jakub Jelinek <jakub@redhat.com>
1965 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
1967 PR tree-optimization/79408
1968 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
1969 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
1970 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
1971 also if rhs1 is INTEGER_CST.
1973 2017-02-14 Richard Biener <rguenther@suse.de>
1976 * tree-into-ssa.c (insert_phi_nodes): When the function can
1977 have abnormal edges rewrite SSA names with broken use-def
1978 dominance out of SSA and register them for PHI insertion.
1980 2017-02-13 Martin Sebor <msebor@redhat.com>
1983 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
1984 clearing info.nowrite flag when snprintf size argument is a range.
1986 2017-02-13 Jakub Jelinek <jakub@redhat.com>
1988 * cprop.c (cprop_jump): Add missing space in string literal.
1989 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
1990 (get_constraint_for_component_ref): Likewise.
1991 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
1992 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
1993 * lra-constraints.c (process_alt_operands): Likewise.
1994 * ipa-inline.c (inline_small_functions): Likewise.
1995 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
1996 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
1997 * trans-mem.c (diagnose_tm_1_op): Likewise.
1998 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
1999 (grid_parallel_clauses_gridifiable): Likewise.
2001 * config/nvptx/mkoffload.c (process): Add space in between
2004 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
2005 "MOD4_SSE_REGS" and "ALL_REGS".
2007 * spellcheck.c (test_data): Add , in between "foo" and "food".
2009 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2012 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
2013 boundary crossing check and subsequent code generation agree.
2015 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2017 * config/aarch64/aarch64.c (has_memory_op): Delete.
2018 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
2021 2017-02-13 Jakub Jelinek <jakub@redhat.com>
2023 PR rtl-optimization/79388
2024 PR rtl-optimization/79450
2025 * combine.c (distribute_notes): When removing TEM_INSN for which
2026 corresponding dest has last value recorded, invalidate that last
2029 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2031 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
2032 of explicit '@'. Add missing assembly comment marker on branch costs
2035 2017-02-13 Nathan Sidwell <nathan@acm.org>
2037 * gengtype-lex.l (<in_struct>): Add '/'.
2039 2017-02-13 Martin Liska <mliska@suse.cz>
2042 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
2044 2017-02-13 Richard Biener <rguenther@suse.de>
2046 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
2048 * configure: Re-generate.
2049 * config.in: Likewise.
2050 * graphite-dependences.c: Simplify as if
2051 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
2052 * graphite-isl-ast-to-gimple.c: Likewise.
2053 * graphite-optimize-isl.c: Likewise.
2054 * graphite-poly.c: Likewise.
2055 * graphite-sese-to-poly.c: Likewise.
2056 * graphite.h: Likewise.
2057 * toplev.c: Include isl/version.h and use isl_version () for
2058 printing the ISL version.
2059 * doc/install.texi: Update ISL requirement.
2061 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
2063 * doc/standards.texi (Standards): Update reference to
2066 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
2068 * doc/extend.texi (Named Address Spaces): sourceware.org now
2070 * doc/install.texi (Binaries): Ditto.
2073 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
2075 * doc/cpp.texi: Replace "stringify"/"stringification" with C
2076 standard terminology "stringize"/"stringizing" throughout.
2077 * doc/cppinternals.texi: Likewise.
2079 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
2081 * doc/extend.texi: Fix some spelling mistakes and typos.
2082 * doc/invoke.texi: Likewise.
2084 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
2087 * params.def (inline-min-speedup) Change from 10 to 8.
2089 2017-02-11 Jakub Jelinek <jakub@redhat.com>
2091 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
2094 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
2097 * ipa-inline-analysis.c (get_minimal_bb): New function.
2098 (record_modified): Use it.
2099 (remap_edge_change_prob): Handle also ancestor functions.
2101 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
2103 * doc/contrib.texi (Contributors): Remove broken link into
2104 the Mauve CVS repository.
2106 2017-02-11 Jakub Jelinek <jakub@redhat.com>
2109 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
2110 result computation whenever lhs doesn't have vector mode, not
2111 just when it has BLKmode.
2113 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
2115 * doc/makefile.texi (profiledbootstrap): Refer to the
2116 installation instructions only in textual form.
2118 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2121 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
2123 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
2125 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
2126 (Specific): Update mingw-w64 reference.
2128 (Specific): Remove broken link to Renesas RX processor.
2130 2017-02-10 Richard Biener <rguenther@suse.de>
2132 * toplev.c (process_options): Do not mention obsolete graphite
2133 options when printing sorry message about missing graphite support.
2134 Mention -floop-nest-optimize.
2136 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
2138 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
2139 (vtst_p16): Likewise.
2140 (vtstq_p8): Likewise.
2141 (vtstq_p16): Likewise.
2143 (vtstq_p64): Likewise.
2144 * config/arm/arm_neon.h (vgetq_lane_p64): New.
2145 (vset_lane_p64): New.
2146 (vsetq_lane_p64): New.
2148 2017-02-10 Jakub Jelinek <jakub@redhat.com>
2150 PR tree-optimization/79411
2151 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
2152 stmt operands are SSA_NAMEs used in abnormal phis.
2153 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
2156 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
2159 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
2162 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
2164 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
2166 2017-02-09 Jakub Jelinek <jakub@redhat.com>
2168 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
2172 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
2173 not arbitrary TREE_CONSTANT.
2176 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
2177 "omp declare target link" attribute unless is_global_var.
2178 * omp-offload.c (find_link_var_op): Likewise.
2180 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
2181 Chung-Lin Tang <cltang@codesourcery.com>
2183 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
2185 (gimplify_adjust_omp_clauses): Don't delete TILE.
2186 (gimplify_omp_for): Deal with TILE.
2187 * internal-fn.c (expand_GOACC_TILE): New function.
2188 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
2190 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
2191 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
2193 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
2194 avoid DIV for outermost collapse var.
2195 (expand_oacc_for): Insert tile element loop as needed. Adjust.
2196 Remove out of date comments, fix whitespace.
2197 * omp-general.c (omp_extract_for_data): Deal with tiling.
2198 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
2199 adjust OLF_DIM_BASE value.
2200 (struct omp_for_data): Add tiling field.
2201 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
2202 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
2203 for auto loops. Remove default auto determining, moved to
2204 oacc_loop_fixed_partitions.
2205 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
2206 stmts, add e_mask field.
2207 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
2208 (oacc_thread_numbers): Use oacc_dim_call.
2209 (oacc_xform_tile): New.
2210 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
2211 (finish_oacc_loop): Adjust for ifns vector.
2212 (oacc_loop_discover_walk): Append loop abstraction sites to list,
2213 add case for GOACC_TILE fns.
2214 (oacc_loop_xform_loop): Delete.
2215 (oacc_loop_process): Iterate over call list directly, and add
2216 handling for GOACC_TILE fns.
2217 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
2219 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
2220 vector partitioning to outer loops. Assign 2 partitions to loops
2221 when available. Add TILE handling.
2222 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
2223 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
2224 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
2225 * tree.c (omp_clause_num_ops): Adjust TILE ops.
2226 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
2228 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
2230 * configure.ac (ACX_BUGURL): Update.
2231 * configure: Regenerate.
2233 2017-02-09 Richard Biener <rguenther@suse.de>
2235 PR tree-optimization/69823
2236 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
2237 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
2239 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
2241 * config/arc/arc-c.def: Add __NPS400__ definition.
2242 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
2243 (TARGET_NPS400): Define.
2245 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
2247 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
2249 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
2251 (arc_cpu_types): Fill the arch_info field with a pointer into the
2252 arc_arch_types table.
2253 (arc_selected_cpu): Declare.
2254 * config/arc/arc.c (arc_selected_cpu): Make global.
2255 (arc_selected_arch): Delete.
2256 (arc_base_cpu): Delete.
2257 (arc_override_options): Remove references to deleted variables,
2258 update access to arch information.
2259 (ARC_OPT): Update access to arch information.
2260 (ARC_OPTX): Likewise.
2261 * config/arc/arc.h (arc_base_cpu): Remove declaration.
2262 (TARGET_ARC600): Update access to arch information.
2263 (TARGET_ARC601): Likewise.
2264 (TARGET_ARC700): Likewise.
2265 (TARGET_EM): Likewise.
2266 (TARGET_HS): Likewise.
2267 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
2270 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
2273 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
2274 condition/operands for integer GE/LE/GEU/LEU operations.
2276 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
2278 PR translation/79397
2279 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
2282 2017-02-08 Martin Jambor <mjambor@suse.cz>
2285 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
2286 whether allocation happened.
2287 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
2288 nothing was allocated.
2290 2017-02-08 Jakub Jelinek <jakub@redhat.com>
2292 PR tree-optimization/79408
2293 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
2294 constant, but SSA_NAME with a known integer range, use the minimum
2295 of that range instead of op1 to determine if modulo can be replaced
2296 with its first operand.
2298 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2300 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
2302 2017-02-08 Richard Biener <rguenther@suse.de>
2304 PR tree-optimization/71824
2305 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
2306 Check all loops contained in the merged region.
2308 2017-02-07 Andrew Pinski <apinski@cavium.com>
2310 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
2312 2017-02-07 Andrew Pinski <apinski@cavium.com>
2314 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
2315 (thunderxt88): Likewise.
2316 (thunderxt81): Disable LSE and change v8.1 to v8.
2317 (thunderxt83): Likewise.
2319 2017-02-07 Jakub Jelinek <jakub@redhat.com>
2320 Richard Biener <rguenther@suse.de>
2323 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
2324 type from int to size_t.
2325 * ira-costs.c (struct_costs_size): Change type from int to size_t.
2327 2017-02-07 Jakub Jelinek <jakub@redhat.com>
2329 PR rtl-optimization/79386
2330 * cprop.c (bypass_conditional_jumps): Initialize
2331 bypass_last_basic_block already before splitting bbs after
2332 unconditional traps...
2333 (bypass_conditional_jumps): ... rather than here.
2336 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
2337 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
2338 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
2339 fix -masm=intel patterns.
2341 2017-02-07 Richard Biener <rguenther@suse.de>
2343 PR tree-optimization/79256
2345 * builtins.c (get_object_alignment_2): Use min_align_of_type
2346 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
2347 and ADJUST_FIELD_ALIGN.
2349 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
2351 * doc/tm.texi: Regenerate.
2352 * stor-layout.c (layout_decl): Adjust.
2353 (update_alignment_for_field): Likewise.
2354 (place_field): Likewise.
2355 (min_align_of_type): Likewise.
2356 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
2357 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
2358 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
2359 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
2360 * config/frv/frv.c (frv_adjust_field_align): Likewise.
2361 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
2362 * config/i386/i386.c (x86_field_alignment): Likewise.
2363 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
2364 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
2365 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
2366 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
2367 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
2368 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
2372 2017-01-30 Richard Biener <rguenther@suse.de>
2374 PR tree-optimization/79256
2375 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
2376 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
2379 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
2381 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
2382 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
2383 builtins to SImode and emit a zero-extend, if necessary.
2385 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2387 * docs/invoke.texi (RISC-V Options): Alphabetize.
2389 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2391 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
2394 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2396 * config/riscv/riscv.c: New file.
2397 * gcc/common/config/riscv/riscv-common.c: Likewise.
2398 * config.gcc: Likewise.
2399 * config/riscv/constraints.md: Likewise.
2400 * config/riscv/elf.h: Likewise.
2401 * config/riscv/generic.md: Likewise.
2402 * config/riscv/linux.h: Likewise.
2403 * config/riscv/multilib-generator: Likewise.
2404 * config/riscv/peephole.md: Likewise.
2405 * config/riscv/pic.md: Likewise.
2406 * config/riscv/predicates.md: Likewise.
2407 * config/riscv/riscv-builtins.c: Likewise.
2408 * config/riscv/riscv-c.c: Likewise.
2409 * config/riscv/riscv-ftypes.def: Likewise.
2410 * config/riscv/riscv-modes.def: Likewise.
2411 * config/riscv/riscv-opts.h: Likewise.
2412 * config/riscv/riscv-protos.h: Likewise.
2413 * config/riscv/riscv.h: Likewise.
2414 * config/riscv/riscv.md: Likewise.
2415 * config/riscv/riscv.opt: Likewise.
2416 * config/riscv/sync.md: Likewise.
2417 * config/riscv/t-elf-multilib: Likewise.
2418 * config/riscv/t-linux: Likewise.
2419 * config/riscv/t-linux-multilib: Likewise.
2420 * config/riscv/t-riscv: Likewise.
2421 * configure.ac: Likewise.
2422 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
2423 Waterman as RISC-V maintainers.
2424 * doc/install.texi: Add RISC-V entries.
2425 * doc/invoke.texi: Add RISC-V options section.
2426 * doc/md.texi: Add RISC-V constraints section.
2427 * configure: Regenerated.
2429 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2432 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
2433 false values to be constant vectors with all 0 or all 1 bits set.
2434 (vcondu<mode><mode>): Likewise.
2435 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
2437 (fpmask_comparison_operator): Update comment.
2438 (vecint_comparison_operator): New predicate.
2439 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
2440 vector conditionals when the true and false values are constant
2441 vectors with all 0 bits or all 1 bits set.
2443 2017-02-06 Martin Sebor <msebor@redhat.com>
2445 PR tree-optimization/79376
2446 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
2448 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
2450 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
2451 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
2452 to simplify split condition.
2454 2017-02-06 Jakub Jelinek <jakub@redhat.com>
2456 * omp-expand.c (oxpand_omp_atomic_fetch_op,
2457 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
2460 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
2462 PR rtl-optimization/68664
2463 * target.def (can_speculate_insn): New hook.
2464 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
2465 * doc/tm.texi: Regenerate.
2466 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
2467 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
2468 (rs6000_sched_can_speculate_insn): New function.
2470 2017-02-06 Jakub Jelinek <jakub@redhat.com>
2472 PR tree-optimization/79284
2473 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
2474 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
2475 vectorizable_mask_load_store, vectorizable_operation,
2476 vect_is_simple_cond, get_same_sized_vectype): Use it instead
2477 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
2478 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
2479 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
2480 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
2481 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
2482 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
2483 is_gimple_assign (stmt). Replace another such test with
2484 is_gimple_assign (stmt).
2486 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
2489 * config/avr/avr.c (rtl-iter.h): Include it.
2490 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
2491 (avr_legitimate_combined_insn): ...and implementation.
2493 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
2495 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
2496 * config/s390/s390.c (s390_const_operand_ok)
2497 (s390_canonicalize_comparison, s390_extract_part)
2498 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
2499 (s390_contiguous_bitmask_p, s390_rtx_costs)
2500 (legitimize_pic_address): Likewise.
2501 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
2502 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
2503 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
2504 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
2505 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
2507 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
2509 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
2510 REGNO($0) == REGNO($1).
2512 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2514 * config/s390/linux.h(SIZE_TYPE): Add comment.
2516 2017-02-06 Julian Brown <julian@codesourcery.com>
2517 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
2518 Virendra Pathak <virendra.pathak@broadcom.com>
2520 * config/aarch64/aarch64-cores.def: Change the scheduler
2522 * config/aarch64/aarch64.md: Include thunderx2t99.md.
2523 * config/aarch64/thunderx2t99.md: New file.
2525 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2527 * doc/standards.texi (Go Language): Update link to language
2530 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
2532 * tree-eh.c (lower_resx): Sanitize profile.
2533 (cleanup_empty_eh_move_lp): Likewise.
2535 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
2538 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
2540 * cfgloopmanip.h (loop_version): Update prototype.
2541 * modulo-sched.c (sms_schedule): Update call of loop_version.
2542 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
2543 * tree-parloops.c (gen_parallel_loop): Likewise.
2544 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
2545 * tree-ssa-loop-split.c (split_loop): Likewise.
2546 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
2547 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
2549 2017-02-05 Martin Liska <mliska@suse.cz>
2552 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
2554 (print_operand_address): Initialize a struct to zero.
2556 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2558 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
2559 garbage collector only in textual form.
2561 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2563 * doc/extend.texi (x86 specific memory model extensions for
2564 transactional memory): Simplify a phrase.
2566 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
2569 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
2570 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
2571 (atomic_storedi_1): Likewise.
2573 2017-02-04 Jakub Jelinek <jakub@redhat.com>
2575 PR tree-optimization/79338
2576 * tree-parloops.c (gather_scalar_reductions): Don't call
2577 vect_analyze_loop_form for loop->inner before destroying loop's
2580 2017-02-03 Martin Sebor <msebor@redhat.com>
2582 PR tree-optimization/79327
2583 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
2584 when precision has resulted in leading zeros.
2585 (format_integer): Adjust the likely counter to assume an unknown
2586 argument that may be zero is non-zero.
2588 2017-02-03 Jason Merrill <jason@redhat.com>
2591 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
2592 avoid copying non-taken branch.
2594 2017-02-03 Jakub Jelinek <jakub@redhat.com>
2596 PR tree-optimization/79340
2597 * tree-vect-loop.c (vectorizable_reduction): Release
2598 vec_defs elements after safe_splicing them into other vectors.
2601 PR tree-optimization/79327
2602 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
2603 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
2605 (format_integer): Use wide_int_to_tree instead of build_int_cst
2606 + to_?hwi. If argmin is NULL, just set argmin and argmax to
2607 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
2608 of shortest and longest sequence.
2610 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
2612 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
2613 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
2615 2017-02-03 Walter Lee <walt@tilera.com>
2618 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
2619 after initial stackframe link reg save.
2620 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
2622 2017-02-03 Jakub Jelinek <jakub@redhat.com>
2625 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
2626 wu for stxssp alternative.
2628 2017-02-03 Martin Sebor <msebor@redhat.com>
2630 PR tree-optimization/79352
2631 * gimple-fold.c (get_range_strlen): Add argument.
2632 (get_range_strlen): Change return type to bool.
2633 (get_maxval_strlen): Pass in a dummy argument.
2634 * gimple-fold.h (get_range_strlen): Change return type to bool.
2635 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
2636 * tree.h (array_at_struct_end_p): Add argument.
2637 * tree.c (array_at_struct_end_p): Handle it.
2639 2017-02-03 Martin Liska <mliska@suse.cz>
2642 * multiple_target.c (create_dispatcher_calls): Redirect edge
2643 from a caller of a dispatcher.
2644 (expand_target_clones): Make the clones local.
2645 (ipa_target_clone): Do both target clones and resolvers.
2646 (ipa_dispatcher_calls): Remove the pass.
2647 (pass_dispatcher_calls::gate): Likewise.
2648 (make_pass_dispatcher_calls): Likewise.
2649 * passes.def (pass_target_clone): Put as very first IPA early
2652 2017-02-03 Martin Liska <mliska@suse.cz>
2654 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
2655 in case of a function with ifunc attribute.
2657 2017-02-03 Martin Liska <mliska@suse.cz>
2659 * cgraph.c (cgraph_node::dump): Dump function version info.
2660 * symtab.c (symtab_node::dump_base): Add missing new line.
2662 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
2664 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
2665 (ifcombine_ifandif): Use it.
2667 2017-02-03 Martin Liska <mliska@suse.cz>
2669 * doc/invoke.texi: Document default value for
2670 use-after-scope-direct-emission-threshold.
2672 2017-02-03 Martin Liska <mliska@suse.cz>
2674 PR tree-optimization/79339
2675 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
2676 (format_floating): Likewise.
2678 2017-02-03 Martin Liska <mliska@suse.cz>
2681 * ipa-prop.c (ipa_node_params_t::insert): Remove current
2683 (ipa_node_params_t::remove): Likewise.
2684 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
2685 initialization from removed ipa_node_params_t::insert.
2686 (ipa_node_params::~ipa_node_params): Move from removed
2687 ipa_node_params_t::release.
2688 * symbol-summary.h (symbol_summary::m_released): New member.
2689 Do not release a summary twice. Do not allow to call finalizer
2690 for types of a summary that live in GGC memory.
2692 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
2694 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
2697 2017-02-02 Martin Sebor <msebor@redhat.com>
2700 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
2701 (format_string): Tighten up the range of output for non-constant
2702 strings and correct the expected range for wide non-constant strings.
2704 2017-02-02 Martin Sebor <msebor@redhat.com>
2706 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
2709 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
2711 (-fdump-tree-@var): Add to index and document how to come up
2712 with pass-specific option and dump file names.
2713 (-fdump-passes): Clarify where to look for output.
2715 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
2718 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
2719 statistics of the analyzed path; allow threading for speed when
2720 any of BBs along the path are optimized for speed.
2722 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
2725 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
2726 settings of the virtual registers.
2729 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
2731 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2732 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
2735 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2737 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
2738 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
2740 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2742 * config/s390/s390.md: Add missing comments with the expanded
2744 * config/s390/vector.md: Likewise.
2745 * config/s390/vx-builtins.md: Likewise.
2747 2017-02-02 Jakub Jelinek <jakub@redhat.com>
2750 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
2751 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
2752 conditions on a single line.
2754 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2756 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
2757 __S390_VX__ to __VX__.
2759 2017-02-01 Andrew Pinski <apinski@cavium.com>
2761 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
2762 stmt_info to record_stmt_cost.
2763 (vect_get_known_peeling_cost): Pass stmt_info if known to
2765 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
2766 cpu_vector_cost field into
2767 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
2768 field into vec_int_stmt_cost and vec_fp_stmt_cost.
2769 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
2770 splitting of scalar_stmt_cost and vec_stmt_cost.
2771 (thunderx_vector_cost): Likewise.
2772 (cortexa57_vector_cost): LIkewise.
2773 (exynosm1_vector_cost): Likewise.
2774 (xgene1_vector_cost): Likewise.
2775 (thunderx2t99_vector_cost): Improve after the splitting of the two
2777 (aarch64_builtin_vectorization_cost): Update for the splitting of
2778 scalar_stmt_cost and vec_stmt_cost.
2780 2017-02-01 Torvald Riegel <triegel@redhat.com>
2781 Richard Henderson <rth@redhat.com>
2783 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
2784 conditional on existance of a fast atomic load.
2785 * optabs-query.c (can_atomic_load_p): New function.
2786 * optabs-query.h (can_atomic_load_p): Declare it.
2787 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
2788 no fast atomic load is available for the particular size of access.
2789 (expand_atomic_compare_and_swap): Likewise.
2790 (expand_atomic_load): Likewise.
2791 (expand_atomic_store): Likewise.
2792 (expand_atomic_fetch_op): Likewise.
2793 * testsuite/lib/target-supports.exp
2794 (check_effective_target_sync_int_128): Remove x86 because it provides
2795 no fast atomic load.
2796 (check_effective_target_sync_int_128_runtime): Likewise.
2798 2017-02-01 Richard Biener <rguenther@suse.de>
2800 * graphite.c: Include tree-vectorizer.h for find_loop_location.
2801 (graphite_transform_loops): Provide opt-info for optimized nests.
2802 * tree-parloop.c (parallelize_loops): Provide opt-info for
2805 2017-02-01 Richard Biener <rguenther@suse.de>
2808 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
2811 2017-02-01 Richard Biener <rguenther@suse.de>
2813 PR tree-optimization/71824
2814 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
2815 Verify the loops are valid in the merged SESE region.
2816 (scop_detection::can_represent_loop_1): Check analyzing the
2817 evolution of the number of iterations in the region succeeds.
2819 2017-01-31 Ian Lance Taylor <iant@golang.org>
2821 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
2822 REG_ARGS_SIZE note to 32-bit push insns and call insn.
2824 2017-01-31 David Malcolm <dmalcolm@redhat.com>
2826 PR preprocessor/79210
2827 * input.c (get_substring_ranges_for_loc): Replace line_width
2828 assertion with error-handling.
2830 2017-01-31 Richard Biener <rguenther@suse.de>
2832 PR tree-optimization/77318
2833 * graphite-sese-to-poly.c (extract_affine): Fix assert.
2834 (create_pw_aff_from_tree): Take loop parameter.
2835 (add_condition_to_pbb): Pass loop of the condition to
2836 create_pw_aff_from_tree.
2838 2017-01-31 Jakub Jelinek <jakub@redhat.com>
2840 * config/s390/s390.c (s390_asan_shadow_offset): New function.
2841 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
2843 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2847 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
2849 (convert_int_to_float128): Likewise.
2850 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
2851 (convert_int_to_float128): Likewise.
2852 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
2853 (UNSPEC_IEEE128_CONVERT): Likewise.
2854 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
2855 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
2856 Use local variables for IBM extended format.
2857 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
2858 (fix_trunc<mode>si2_fprs): Likewise.
2859 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
2860 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
2861 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
2862 to know that we can now have integers of all sizes in vector
2864 (fix<uns>_<mode>di2_hw): Likewise.
2865 (float<uns>_<mode>si2_hw): Likewise.
2866 (fix_<mode>si2_hw): Likewise.
2867 (fixuns_<mode>si2_hw): Likewise.
2868 (float<uns>_<mode>di2_hw): Likewise.
2869 (float_<mode>di2_hw): Likewise.
2870 (float_<mode>si2_hw): Likewise.
2871 (floatuns_<mode>di2_hw): Likewise.
2872 (floatuns_<mode>si2_hw): Likewise.
2873 (xscvqp<su>wz_<mode>): Delete, no longer used.
2874 (xscvqp<su>dz_<mode>): Likewise.
2875 (xscv<su>dqp_<mode>): Likewise.
2876 (ieee128_mfvsrd_64bit): Likewise.
2877 (ieee128_mfvsrd_32bit): Likewise.
2878 (ieee128_mfvsrwz): Likewise.
2879 (ieee128_mtvsrw): Likewise.
2880 (ieee128_mtvsrd_64bit): Likewise.
2881 (ieee128_mtvsrd_32bit): Likewise.
2883 2017-01-31 Martin Liska <mliska@suse.cz>
2886 * ipa-prop.c (ipa_free_all_node_params): Call release method
2887 instead of ~sumbol_summary to not to trigger double times
2890 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
2892 PR tree-optimization/71691
2893 * bitmap.h (class auto_bitmap): New.
2894 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
2895 is_maybe_undefined instead of ssa_undefined_value_p.
2897 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2899 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
2900 __S390_ARCH_LEVEL__ to __ARCH__.
2902 2017-01-31 Jakub Jelinek <jakub@redhat.com>
2904 PR tree-optimization/79267
2905 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
2906 if should_remove_lhs_p is true.
2908 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
2911 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
2912 (add_alignment_attribute): New.
2913 (base_type_die): Add alignment attribute.
2914 (subrange_type_die): Likewise.
2915 (modified_type_die): Likewise.
2916 (gen_array_type_die): Likewise.
2917 (gen_descr_array_type_die: Likewise.
2918 (gen_enumeration_type_die): Likewise.
2919 (gen_subprogram_die): Likewise.
2920 (gen_variable_die): Likewise.
2921 (gen_field_die): Likewise.
2922 (gen_ptr_to_mbr_type_die): Likewise.
2923 (gen_struct_or_union_type_die): Likewise.
2924 (gen_subroutine_type_die): Likewise.
2925 (gen_typedef_die): Likewise.
2926 (base_type_cmp): Compare alignment attribute.
2928 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2931 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
2932 (setb_unsigned) New pattern for setb with CCUNS.
2933 * config/rs6000/rs6000.c (expand_block_compare): Use a different
2934 subfc./subfe sequence to avoid overflow problems. Generate a
2935 shorter sequence with cmpld/setb for power9.
2936 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
2937 for generating subfc. instruction.
2938 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
2939 now uses this instruction.
2941 2017-01-30 Ian Lance Taylor <iant@google.com>
2944 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
2945 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
2947 2017-01-30 Martin Sebor <msebor@redhat.com>
2949 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
2950 Move constant to the right of a relational operator.
2951 (get_mpfr_format_length, format_character, format_string): Ditto.
2952 (should_warn_p, maybe_warn): Same.
2954 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
2956 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
2959 * asan.c (get_translation_unit_decl): Remove function.
2960 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
2962 2017-01-30 Martin Liska <mliska@suse.cz>
2964 PR gcov-profile/79259
2965 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
2968 2017-01-30 Martin Liska <mliska@suse.cz>
2971 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
2972 Initialize variables with NULL value.
2974 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
2977 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
2979 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
2981 2017-01-30 Richard Biener <rguenther@suse.de>
2983 PR tree-optimization/79276
2984 * tree-vrp.c (process_assert_insertions): Properly adjust common
2985 when removing a duplicate.
2987 * gcc.dg/torture/pr79276.c: New testcase.
2989 2017-01-30 Richard Biener <rguenther@suse.de>
2991 PR tree-optimization/79256
2992 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
2993 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
2995 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
2997 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
3000 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
3001 ("*r<noxa>sbg_<mode>_sll_bitmask")
3002 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
3003 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
3004 Use contiguous_bitmask_nowrap_operand.
3006 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3009 * config/rs6000/altivec.h (vec_xl): Revise #define.
3010 (vec_xst): Likewise.
3012 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
3014 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
3016 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
3018 PR rtl-optimization/79194
3019 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
3020 traps before call to bypass_conditional_jumps.
3022 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
3024 PR tree-optimization/71374
3025 * lra-constraints.c (check_conflict_input_operands): New.
3026 (match_reload): Use it.
3028 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
3031 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
3032 account to calculate conflict_set.
3034 2017-01-27 Bin Cheng <bin.cheng@arm.com>
3036 PR rtl-optimization/78559
3037 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
3038 other_insn in combine.
3040 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
3042 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
3043 uint16_type_node for BT_UINT16.
3045 2017-01-27 David Malcolm <dmalcolm@redhat.com>
3047 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
3048 "RTL Tests" to menu.
3049 (GIMPLE Tests): New node.
3050 (RTL Tests): New node.
3052 2017-01-27 Richard Biener <rguenther@suse.de>
3054 PR tree-optimization/79245
3055 * tree-loop-distribution.c (distribute_loop): Apply cost
3056 modeling also to detected patterns.
3058 2017-01-27 Richard Biener <rguenther@suse.de>
3060 PR tree-optimization/71433
3061 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
3062 (compare_assert_loc): New function.
3063 (process_assert_insertions): Sort and optimize assert locations
3064 to remove duplicates and push down identical assertions on
3065 edges to their destination block.
3067 2017-01-27 Richard Biener <rguenther@suse.de>
3069 PR tree-optimization/79244
3070 * tree-vrp.c (remove_range_assertions): Forcefully propagate
3071 out SSA names even if abnormal.
3073 2017-01-27 Jakub Jelinek <jakub@redhat.com>
3075 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
3076 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
3077 instead of MPFR_RNDN.
3079 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
3082 * arm.c (arm_option_override): Don't call build_target_option_node
3083 until after doing all option overrides.
3084 (arm_valid_target_attribute_tree): Likewise.
3086 2017-01-27 Martin Liska <mliska@suse.cz>
3088 * doc/invoke.texi (-fprofile-arcs): Document profiling support
3089 for {cd}tors and C++ {cd}tors.
3091 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
3093 * config/s390/s390.md ("*setmem_long_and")
3094 ("*setmem_long_and_31z"): Use zero_extend instead of and.
3096 2017-01-26 Martin Sebor <msebor@redhat.com>
3098 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
3101 2017-01-26 Martin Sebor <msebor@redhat.com>
3103 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
3104 HAVE_DFmode before using XFmode or DFmode.
3105 (parse_directive): Avoid using the z length modifier to avoid
3106 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
3109 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
3110 to accept adjustment as an array.
3111 (get_int_range): New function.
3112 (struct directive): Make width and prec arrays.
3113 (directive::set_width, directive::set_precision): Call get_int_range.
3114 (format_integer, format_floating): Handle width and precision ranges.
3115 (format_string, parse_directive): Same.
3117 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3120 * dwarf2out.c (generate_skeleton_bottom_up): For children with
3121 comdat_type_p set, just clone them, but keep the children in the
3125 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
3126 which have direct callers with -fvar-tracking-assignments enabled
3128 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
3129 inside of type units.
3131 2017-01-26 Martin Sebor <msebor@redhat.com>
3134 * gimple-ssa-sprintf.c (struct result_range): Add likely and
3136 (struct format_result): Replace number_chars, number_chars_min,
3137 and number_chars_max with a single member of struct result_range.
3139 (format_result::operator+=): Adjust.
3140 (struct fmtresult): Remove bounded. Handle likely and unlikely
3142 (fmtresult::adjust_for_width_or_precision): New function.
3143 (fmtresult:type_max_digits): New function.
3144 (bytes_remaining): Handle likely and unlikely counters.
3145 (min_bytes_remaining): Remove.
3146 (format_percent): Simplify.
3147 (format_integer, format_floating): Set likely and unlikely counters.
3148 (get_string_length, format_character, format_string): Same.
3149 (format_plain, should_warn_p): New function.
3150 (maybe_warn): Call should_warn_p. Update diagnostic messages
3151 and handle those for all directives, including plain strings.
3152 (format_directive): Handle likely and unlikely counters.
3153 Remove unnecessary quoting from diagnostics. Add an informational
3155 (add_bytes): Remove.
3156 (pass_sprintf_length::compute_format_length): Simplify.
3157 (try_substitute_return_value): Handle likely and unlikely counters.
3159 2017-01-26 Carl Love <cel@us.ibm.com>
3161 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
3162 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
3164 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
3167 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
3168 endianess for subregs into account.
3169 * lra-constraints.c (lra_constraints): Do risky transformations
3170 always on the first iteration.
3171 * lra-lives.c (check_pseudos_live_through_calls): Add arg
3172 last_call_used_reg_set.
3173 (process_bb_lives): Define and use last_call_used_reg_set.
3174 * lra.c (lra): Always continue after lra_constraints on the first
3177 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
3179 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
3181 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
3183 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3185 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
3186 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
3187 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
3188 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
3189 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
3190 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
3191 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
3192 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
3193 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
3195 2017-01-26 Marek Polacek <polacek@redhat.com>
3198 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
3199 for the third operand.
3201 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3204 * omp-low.c (struct omp_context): Add simt_stmt field.
3205 (scan_omp_for): Return omp_context *.
3206 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
3207 context to the _simt_ SIMD stmt.
3208 (lower_omp_for): For combined SIMD with sibling _simt_
3209 SIMD, make sure to use the same decls in _looptemp_
3210 clauses as in the sibling.
3212 2017-01-26 David Sherwood <david.sherwood@arm.com>
3215 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
3218 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3221 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
3222 emit fld b; fld a; if possible.
3224 * brig-builtins.def: Update copyright years.
3225 * config/arm/arm_acle_builtins.def: Update copyright years.
3227 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
3230 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
3231 constraint instead of o for the stxsd instruction.
3233 2017-01-25 Carl Love <cel@us.ibm.com>
3235 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
3236 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
3238 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
3240 * doc/invoke.texi (C++ Dialect Options): Fix typo.
3242 2017-01-25 Richard Biener <rguenther@suse.de>
3244 PR tree-optimization/69264
3245 * target.def (vector_alignment_reachable): Improve documentation.
3246 * doc/tm.texi: Regenerate.
3247 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
3249 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
3250 earlier changes with respect to TYPE_USER_ALIGN.
3251 (vector_alignment_reachable_p): Likewise. Improve dumping.
3253 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3256 * config/arm/arm.md (xordi3): Force constant operand into a register
3259 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3261 * doc/invoke.texi (-fstore-merging): Correct default optimization
3262 levels at which it is enabled.
3263 (-O): Move -fstore-merging from list to...
3266 2017-01-25 Richard Biener <rguenther@suse.de>
3269 * omp-expand.c: Include debug.h.
3270 (expand_omp_taskreg): Make sure to generate early debug before
3271 outlining anything from a function.
3272 (expand_omp_target): Likewise.
3273 (grid_expand_target_grid_body): Likewise.
3275 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
3278 * asan.c (get_translation_unit_decl): New function.
3279 (asan_add_global): Extract modules file name from globals
3280 TRANSLATION_UNIT_DECL name.
3282 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
3285 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
3286 for long calls with APCS frame and VFP.
3288 2017-01-24 David Malcolm <dmalcolm@redhat.com>
3290 * cfg.c (original_copy_tables_initialized_p): New function.
3291 * cfg.h (original_copy_tables_initialized_p): New decl.
3292 * cfgrtl.c (relink_block_chain): Guard the call to
3293 free_original_copy_tables with a call to
3294 original_copy_tables_initialized_p.
3295 * cgraph.h (symtab_node::native_rtl_p): New decl.
3296 * cgraphunit.c (symtab_node::native_rtl_p): New function.
3297 (symtab_node::needed_p): Don't assert for early assembly output
3298 for __RTL functions.
3299 (cgraph_node::finalize_function): Set "force_output" for __RTL
3301 (cgraph_node::analyze): Bail out early for __RTL functions.
3302 (analyze_functions): Update assertion to support __RTL functions.
3303 (cgraph_node::expand): Bail out early for __RTL functions.
3304 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
3306 * function.h (struct function): Update comment for field
3308 * gimple-expr.c: Include "tree-pass.h".
3309 (gimple_has_body_p): Return false for __RTL functions.
3310 * Makefile.in (OBJS): Add run-rtl-passes.o.
3311 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
3313 (gcc::pass_manager::get_clean_slate): New accessor.
3314 * passes.c: Include "insn-addr.h".
3315 (should_skip_pass_p): Add logging. Update logic for running
3316 "expand" to be compatible with both __GIMPLE and __RTL. Guard
3317 property-provider override so it is only done for gimple passes.
3319 (skip_pass): New function.
3320 (execute_one_pass): Call skip_pass when skipping passes.
3321 * read-md.c (md_reader::read_char): Support filtering
3322 the input to a subset of line numbers.
3323 (md_reader::md_reader): Initialize fields
3324 m_first_line and m_last_line.
3325 (md_reader::read_file_fragment): New function.
3326 * read-md.h (md_reader::read_file_fragment): New decl.
3327 (md_reader::m_first_line): New field.
3328 (md_reader::m_last_line): New field.
3329 * read-rtl-function.c (function_reader::create_function): Only
3330 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
3331 curr_properties. Set DECL_INITIAL to a dummy block.
3332 (read_rtl_function_body_from_file_range): New function.
3333 * read-rtl-function.h (read_rtl_function_body_from_file_range):
3335 * run-rtl-passes.c: New file.
3336 * run-rtl-passes.h: New file.
3338 2017-01-24 Jeff Law <law@redhat.com>
3340 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
3343 2017-01-24 Bin Cheng <bin.cheng@arm.com>
3345 PR tree-optimization/79159
3346 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
3347 (record_nonwrapping_iv): Improve boundary using above function if no
3348 value range information.
3350 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
3351 Martin Jambor <mjambor@suse.cz>
3353 * brig-builtins.def: New file.
3354 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
3355 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
3356 (DEF_HSAIL_SAT_BUILTIN): Likewise.
3357 (DEF_HSAIL_INTR_BUILTIN): Likewise.
3358 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
3359 * builtin-types.def (BT_INT8): New.
3360 (BT_INT16): Likewise.
3361 (BT_UINT8): Likewise.
3362 (BT_UINT16): Likewise.
3363 (BT_FN_ULONG): Likewise.
3364 (BT_FN_UINT_INT): Likewise.
3365 (BT_FN_UINT_ULONG): Likewise.
3366 (BT_FN_UINT_LONG): Likewise.
3367 (BT_FN_UINT_PTR): Likewise.
3368 (BT_FN_ULONG_PTR): Likewise.
3369 (BT_FN_INT8_FLOAT): Likewise.
3370 (BT_FN_INT16_FLOAT): Likewise.
3371 (BT_FN_UINT32_FLOAT): Likewise.
3372 (BT_FN_UINT16_FLOAT): Likewise.
3373 (BT_FN_UINT8_FLOAT): Likewise.
3374 (BT_FN_UINT64_FLOAT): Likewise.
3375 (BT_FN_UINT16_UINT32): Likewise.
3376 (BT_FN_UINT32_UINT16): Likewise.
3377 (BT_FN_UINT16_UINT16_UINT16): Likewise.
3378 (BT_FN_INT_PTR_INT): Likewise.
3379 (BT_FN_UINT_PTR_UINT): Likewise.
3380 (BT_FN_LONG_PTR_LONG): Likewise.
3381 (BT_FN_ULONG_PTR_ULONG): Likewise.
3382 (BT_FN_VOID_UINT64_UINT64): Likewise.
3383 (BT_FN_UINT8_UINT8_UINT8): Likewise.
3384 (BT_FN_INT8_INT8_INT8): Likewise.
3385 (BT_FN_INT16_INT16_INT16): Likewise.
3386 (BT_FN_INT_INT_INT): Likewise.
3387 (BT_FN_UINT_FLOAT_UINT): Likewise.
3388 (BT_FN_FLOAT_UINT_UINT): Likewise.
3389 (BT_FN_ULONG_UINT_UINT): Likewise.
3390 (BT_FN_ULONG_UINT_PTR): Likewise.
3391 (BT_FN_ULONG_ULONG_ULONG): Likewise.
3392 (BT_FN_UINT_UINT_UINT): Likewise.
3393 (BT_FN_VOID_UINT_PTR): Likewise.
3394 (BT_FN_UINT_UINT_PTR: Likewise.
3395 (BT_FN_UINT32_UINT64_PTR): Likewise.
3396 (BT_FN_INT_INT_UINT_UINT): Likewise.
3397 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
3398 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
3399 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
3400 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
3401 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
3402 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
3403 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
3404 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
3405 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
3406 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
3407 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
3408 * doc/frontends.texi: List BRIG FE.
3409 * doc/install.texi (Testing): Add BRIG tesring requirements.
3410 * doc/invoke.texi (Overall Options): Mention BRIG.
3411 * doc/standards.texi (Standards): Doucment BRIG HSA version.
3413 2017-01-24 Richard Biener <rguenther@suse.de>
3415 PR translation/79208
3416 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
3418 2017-01-24 Martin Jambor <mjambor@suse.cz>
3421 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
3422 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
3425 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
3428 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
3429 casts from signed to unsigned really don't have a range.
3431 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
3433 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
3434 GMP_RNDx for compatiblity.
3436 2017-01-24 Martin Liska <mliska@suse.cz>
3439 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
3440 that would prevent us to call alloca with -1 as argument.
3442 2017-01-24 Jakub Jelinek <jakub@redhat.com>
3444 * dwarf2out.c (output_compilation_unit_header, output_file_names):
3445 Avoid -Wformat-security warning.
3447 2017-01-23 Andrew Pinski <apinski@cavium.com>
3449 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
3452 2017-01-23 Martin Sebor <msebor@redhat.com>
3455 * gimple-ssa-sprintf.c (warn_level): New global.
3456 (format_integer): Use it here and throughout the rest of the file.
3457 Use the same switch to compute sign as base.
3458 (maybe_warn): New function.
3459 (format_directive): Factor out warnings into maybe_warn.
3460 Add debugging output. Use warn_level.
3461 (add_bytes): Use warn_level.
3462 (pass_sprintf_length::compute_format_length): Add debugging output.
3463 (try_substitute_return_value): Same.
3464 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
3467 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
3468 (struct fmtresult, format_integer, format_floating): Adjust.
3469 (fmtresult::fmtresult): Set max correctly in two argument ctor.
3470 (get_string_length, format_string,format_directive): Same.
3471 (pass_sprintf_length::compute_format_length): Same.
3472 (try_substitute_return_value): Simplify slightly.
3475 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
3476 (fmtresult::operator+=): Outlined.
3477 (struct fmtresult): Add ctors.
3478 (struct conversion_spec): Rename...
3479 (struct directive): ...to this. Add and remove data members.
3480 (directive::set_width, directive::set_precision): New functions.
3481 (format_percent): Use fmtresult ctor.
3482 (get_width_and_precision): Remove.
3483 (format_integer): Make naming changes. Avoid computing width and
3485 (format_floating): Same. Adjust indentation.
3486 (format_character, format_none): New functions.
3487 (format_string): Moved character handling to format_character.
3488 (format_directive): Remove arguments, change return type.
3489 (parse_directive): New function.
3490 (pass_sprintf_length::compute_format_length): Move directive
3491 parsing to parse_directive.
3493 2017-01-23 Jakub Jelinek <jakub@redhat.com>
3495 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
3496 (assign_assembler_name_if_needed): ... this.
3497 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
3498 (assign_assembler_name_if_needed): ... this.
3499 (free_lang_data_in_cgraph): Adjust callers.
3500 * cgraphunit.c (cgraph_node::analyze): Likewise.
3501 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
3504 2017-01-23 Richard Biener <rguenther@suse.de>
3506 PR tree-optimization/79088
3507 PR tree-optimization/79188
3508 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
3509 resetting loop bounds after last path deletion. Reset loop
3510 bounds of the target loop, make code match the comments.
3511 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
3512 Make sure loops need no fixups.
3514 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
3516 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
3517 exponent support with double type for first argument.
3518 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
3519 type returned by __builtin_vec_extract_sig,
3520 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
3521 functions from "vector int" to "vector unsigned int" or from
3522 "vector long long int" to "vector unsigned long long int".
3523 Changed type returned by __builtin_vec_extract_exp,
3524 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
3525 functions from "vector int" to "vector unsigned int" or from
3526 "vector long long int" to "vector unsigned long long int".
3527 Changed return type of __builtin_vec_test_data_class,
3528 __builtin_vec_test_data_class_sp, and
3529 __builtin_vec_test_data_class_dp from "vector int" to
3530 "vector bool int" or from "vector long long int" to "vector bool
3531 long long int" and changed second argument type from "unsigned
3532 int" to "int". Added new overloaded function forms "vector float
3533 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
3534 "vector float __builtin_vec_insert_exp_sp (vector float, vector
3535 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
3536 double, vector unsigned long long int)" and "vector double
3537 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
3538 long int)". Changed return type of
3539 __builtin_scalar_test_data_class and
3540 __builtin_scalar_test_data_class_sp and
3541 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
3542 int" and changed second argument from "unsigned int" to "int".
3543 Changed type returned by __builtin_scalar_test_neg,
3544 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
3545 from "int" to "bool int". Added new overloaded function form
3546 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
3547 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
3548 exponent double-precision with floating point first argument.
3549 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
3550 documentation of scalar_test_data_class, scalar_test_neg,
3551 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
3552 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
3553 vec_test_data_class built-in functions to reflect refinements in
3554 their type signatures.
3556 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
3558 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
3560 (aarch64_elf_asm_destructor): Likewise.
3562 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
3564 PR rtl-optimization/78634
3565 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
3566 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
3567 * ifcvt.c (noce_try_cmove): Add missing cost check.
3569 PR rtl-optimization/71724
3570 * combine.c (if_then_else_cond): Look for situations where it is
3571 beneficial to undo the work of one of the recursive calls.
3573 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3575 PR tree-optimization/70754
3576 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3577 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3578 combined stmt before it if not NULL.
3579 (combine_chains): Process refs reversely and compute dominance point
3582 2017-01-23 Martin Liska <mliska@suse.cz>
3584 PR tree-optimization/79196
3585 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
3586 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
3588 (strlen_optimize_stmt): Call the renamed function.
3590 2017-01-23 Michael Matz <matz@suse.de>
3592 PR tree-optimization/78384
3593 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
3595 2017-01-23 Richard Biener <rguenther@suse.de>
3597 PR tree-optimization/79186
3598 * tree-vrp.c (register_new_assert_for): Make sure we've seen
3599 both incoming edges before moving an assert.
3601 2017-01-23 Martin Jambor <mjambor@suse.cz>
3603 * ipa-prop.c (load_from_param_1): Removed.
3604 (load_from_unmodified_param): Bits from load_from_param_1 put back
3606 (load_from_param): Removed.
3607 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
3608 with stmt. Reverted back to use of load_from_unmodified_param.
3610 2017-01-23 Martin Jambor <mjambor@suse.cz>
3613 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
3614 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
3615 field a pointer to garbage collected vector, mark lattices and
3616 ipcp_orig_node with GTY((skip)).
3617 (ipa_get_param_count): Adjust to descriptors being a pointer.
3618 (ipa_get_param): Likewise.
3619 (ipa_get_type): Likewise.
3620 (ipa_get_param_move_cost): Likewise.
3621 (ipa_set_param_used): Likewise.
3622 (ipa_get_controlled_uses): Likewise.
3623 (ipa_set_controlled_uses): Likewise.
3624 (ipa_is_param_used): Likewise.
3625 (ipa_node_params_t): Move into garbage collector. New methods insert
3627 (ipa_node_params_sum): Annotate wth GTY(()).
3628 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
3630 (ipa_load_from_parm_agg): Adjust declaration.
3631 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
3632 * ipa-profile.c (ipa_profile): Likewise.
3633 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
3634 (ipa_populate_param_decls): Make descriptors parameter garbage
3636 (ipa_dump_param): Adjust to descriptors being a pointer.
3637 (ipa_alloc_node_params): Likewise.
3638 (ipa_initialize_node_params): Likewise.
3639 (load_from_param_1): Make descriptors parameter garbage collected.
3640 (load_from_unmodified_param): Likewise.
3641 (load_from_param): Likewise.
3642 (ipa_load_from_parm_agg): Likewise.
3643 (ipa_node_params::~ipa_node_params): Removed.
3644 (ipa_free_all_node_params): Remove call to delete operator.
3645 (ipa_node_params_t::insert): New.
3646 (ipa_node_params_t::remove): Likewise.
3647 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
3648 copy known_csts and known_contexts vectors.
3649 (ipa_read_node_info): Adjust to descriptors being a pointer.
3650 (ipcp_modif_dom_walker): Make m_descriptors field garbage
3652 (ipcp_transform_function): Make descriptors variable garbage
3655 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
3657 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
3658 * config/i386/avx512dqintrin.h: Ditto.
3659 * config/i386/avx512fintrin.h: Ditto.
3660 * gcc/config/i386/i386.c: Handle new builtins.
3661 * config/i386/i386-builtin.def: Add new builtins.
3662 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
3663 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
3665 2017-01-23 Jakub Jelinek <jakub@redhat.com>
3666 Martin Liska <mliska@suse.cz>
3668 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
3669 * asan.c (asan_expand_poison_ifn): Support stores and use
3670 appropriate ASAN report function.
3671 * internal-fn.c (expand_ASAN_POISON_USE): New function.
3672 * internal-fn.def (ASAN_POISON_USE): Declare.
3673 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
3674 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
3675 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
3676 ASAN_POISON calls w/o LHS.
3677 * tree-ssa.c (execute_update_addresses_taken): Create clobber
3678 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
3679 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
3680 * gimplify.c (asan_poison_variables): Add attribute
3681 use_after_scope_memory to variables that really needs to live
3683 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
3684 having the attribute.
3686 2017-01-23 Martin Liska <mliska@suse.cz>
3688 * asan.c (create_asan_shadow_var): New function.
3689 (asan_expand_poison_ifn): Likewise.
3690 * asan.h (asan_expand_poison_ifn): New declaration.
3691 * internal-fn.c (expand_ASAN_POISON): Likewise.
3692 * internal-fn.def (ASAN_POISON): New builtin.
3693 * sanopt.c (pass_sanopt::execute): Expand
3694 asan_expand_poison_ifn.
3695 * tree-inline.c (copy_decl_for_dup_finish): Make function
3697 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
3698 * tree-ssa.c (is_asan_mark_p): New function.
3699 (execute_update_addresses_taken): Rewrite local variables
3700 (identified just by use-after-scope as addressable) into SSA.
3702 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
3704 * doc/install.texi (Specific): opensource.apple.com uses https
3705 now. Remove trailing slash.
3707 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
3709 * README.Portability: Remove note on an Irix compatibility issue.
3711 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
3713 * gcov.c (INCLUDE_ALGORITHM): Define.
3714 (INCLUDE_VECTOR): Define.
3715 No longer include <vector> and <algorithm> directly.
3717 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
3719 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
3721 * doc/invoke.texi (Code Gen Options): Ditto.
3723 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
3726 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
3728 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
3730 rtl-optimization/79125
3731 * cprop.c (local_cprop_pass): Handle cases where we make an
3734 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
3738 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
3739 read from, for big endian.
3741 2017-01-20 Jiong Wang <jiong.wang@arm.com>
3743 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
3744 register pauth builtins for LP64 only.
3746 2017-01-20 Marek Polacek <polacek@redhat.com>
3749 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
3752 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
3754 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
3756 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
3757 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
3758 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
3760 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3763 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
3764 in big-endian mode when they are not a single duplicated value.
3766 2017-01-20 Richard Biener <rguenther@suse.de>
3768 * BASE-VER: Bump to 7.0.1.
3770 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
3772 * omp-low.c (omplow_simd_context): New struct. Use it...
3773 (lower_rec_simd_input_clauses): ...here and...
3774 (lower_rec_input_clauses): ...here to hold common data. Adjust all
3775 references to idx, lane, max_vf, is_simt.
3777 2017-01-20 Graham Markall <graham.markall@embecosm.com>
3779 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
3782 2017-01-20 Martin Jambor <mjambor@suse.cz>
3784 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
3785 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
3787 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
3788 (GTFILES): Rename hsa.c to hsa-common.c.
3789 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
3790 * hsa-dump.c: Likewise.
3791 * hsa-gen.c: Likewise.
3792 * hsa-regalloc.c: Likewise.
3793 * ipa-hsa.c: Likewise.
3794 * omp-expand.c: Likewise.
3795 * omp-low.c: Likewise.
3796 * toplev.c: Likewise.
3798 2017-01-20 Marek Polacek <polacek@redhat.com>
3801 * doc/invoke.texi: Document -Wduplicated-branches.
3802 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
3803 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
3804 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
3805 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
3806 return 0 only when not OEP_LEXICOGRAPHIC.
3807 (fold_build_cleanup_point_expr): Use the expression
3808 location when building CLEANUP_POINT_EXPR.
3809 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
3810 * tree.c (add_expr): Handle error_mark_node.
3812 2017-01-20 Martin Liska <mliska@suse.cz>
3815 * tree-profile.c (init_ic_make_global_vars): Do not call
3817 (gimple_init_gcov_profiler): Likewise.
3819 2017-01-20 Martin Liska <mliska@suse.cz>
3822 * cgraph.h (maybe_create_reference): Remove argument and
3824 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
3826 * ipa-cp.c (create_specialized_node): Likewise.
3827 * symtab.c (symtab_node::maybe_create_reference): Handle
3828 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
3830 2017-01-20 Martin Liska <mliska@suse.cz>
3832 * read-rtl-function.c (function_reader::create_function): Use
3833 build_decl instread of build_decl_stat.
3835 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
3837 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
3838 * config/i386/avx512dqintrin.h: Ditto.
3839 * config/i386/avx512fintrin.h: Ditto.
3840 * config/i386/i386-builtin-types.def: Add new types.
3841 * gcc/config/i386/i386.c: Handle new types.
3842 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
3843 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
3844 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
3845 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
3846 (__builtin_ia32_kshiftridi): New.
3847 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
3849 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
3853 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
3854 define to rs6000_init_stack_protect_guard.
3855 (rs6000_init_stack_protect_guard): New function.
3857 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
3858 Yunqiang Su <yunqiang.su@imgtec.com>
3860 * config.gcc (supported_defaults): Add madd4.
3861 (with_madd4): Add validation.
3862 (all_defaults): Add madd4.
3863 * config/mips/mips.opt (mmadd4): New option.
3864 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
3866 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
3868 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
3869 (ISA_HAS_FUSED_MADD4): Likewise.
3870 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
3871 * gcc/doc/install.texi (--with-madd4): Document the new option.
3873 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3875 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
3876 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
3877 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
3878 (aarch64_init_pauth_hint_builtins): New.
3879 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
3880 (aarch64_expand_builtin): Expand new builtins.
3882 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3884 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
3885 * combine-stack-adj.c (no_unhandled_cfa): Handle
3886 REG_CFA_TOGGLE_RA_MANGLE.
3887 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
3888 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
3889 info for return address signing.
3890 (aarch64_expand_epilogue): Likewise.
3892 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3894 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
3895 * config/aarch64/aarch64-protos.h
3896 (aarch64_return_address_signing_enabled): New declaration.
3897 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
3899 (aarch64_expand_prologue): Sign return address before it's pushed onto
3901 (aarch64_expand_epilogue): Authenticate return address fetched from
3903 (aarch64_override_options): Sanity check for ILP32 and ISA level.
3904 (aarch64_attributes): New function attributes for "sign-return-address".
3905 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
3906 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
3907 ("*do_return"): Generate combined instructions according to key index.
3908 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
3909 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
3911 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
3912 * config/aarch64/aarch64.opt (msign-return-address=): New.
3913 * doc/extend.texi (AArch64 Function Attributes): Documents
3914 "sign-return-address=".
3915 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
3917 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
3919 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
3920 overall option summary.
3922 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3924 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
3925 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
3926 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
3927 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
3929 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
3931 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
3932 -mpower9-minmax by default for -mcpu=power9.
3933 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
3934 128-bit floating point.
3936 2017-01-20 Alan Modra <amodra@gmail.com>
3938 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
3939 optimizing for size.
3941 2017-01-20 Alan Modra <amodra@gmail.com>
3944 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
3945 for strcmp and strncmp from corresponding builtin decl.
3947 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
3949 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
3950 instead of i386/rtems-64.h.
3951 * config/i386/rtems-64.h: Remove.
3953 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
3957 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
3959 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
3961 2017-01-19 Tamar Christina <tamar.christina@arm.com>
3963 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3964 Change int to HOST_WIDE_INT.
3965 * config/aarch64/aarch64-protos.h
3966 (aarch64_simd_gen_const_vector_dup): Likewise.
3967 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
3969 2017-01-19 David Malcolm <dmalcolm@redhat.com>
3971 * langhooks-def.h (lhd_type_for_size): New decl.
3972 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
3973 * langhooks.c (lhd_type_for_size): New function, taken from
3976 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
3978 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
3979 define_bypass for CR latency.
3980 (power9-cracked-alu): Update bypass latency and remove power9-branch.
3981 (power9-alu2): Add define_bypass for CR latency.
3983 (power9-mul): Update insn latency.
3984 (power9-mul-compare): Update insn latency, bypass latency and remove
3987 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3989 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
3991 * config/aarch64/aarch64.md
3992 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
3993 aarch64_nopcrelative_literal_loads.
3994 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
3996 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
3998 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
4000 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
4002 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
4005 * config.gcc (supported_defaults): Add lxc1-sxc1.
4006 (with_lxc1_sxc1): Add validation.
4007 (all_defaults): Add lxc1-sxc1.
4008 * config/mips/mips.opt (mlxc1-sxc1): New option.
4009 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
4011 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
4012 __mips_no_lxc1_sxc1.
4013 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
4014 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
4015 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
4017 2017-01-19 Richard Biener <rguenther@suse.de>
4019 PR tree-optimization/72488
4020 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
4021 sure to restore SSA info.
4022 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
4024 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
4026 PR rtl-optimization/79121
4027 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
4028 of the inner type when shifting an extended value.
4030 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
4033 * symtab.c (symtab_node::equal_address_to): Fix comparing of
4034 interposable aliases.
4036 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
4039 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
4040 Use the evmergelohi instruction.
4041 (mov_si<mode>_e500_subreg4_2_le): Likewise.
4042 (mov_sitf_e500_subreg8_2_be): Likewise.
4043 (mov_sitf_e500_subreg12_2_le): Likewise.
4044 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
4045 (mov_si<mode>_e500_subreg4_2_be): Likewise.
4046 (mov_sitf_e500_subreg8_2_le): Likewise.
4047 (mov_sitf_e500_subreg12_2_be): Likewise.
4049 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4051 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
4052 attribute from vecsimple to vecperm.
4053 (altivec_vbpermq2): Likewise.
4055 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4058 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
4060 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4061 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
4062 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
4063 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
4064 case where N arg is SIZE_MAX.
4065 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
4066 (cmpstrsi): Add pattern.
4068 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
4070 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4071 __builtin_vec_revb builtins.
4072 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
4073 built-in functions to support generation of the ISA 3.0 XXBR<x>
4074 vector byte reverse instructions.
4075 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
4076 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
4077 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
4078 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
4079 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
4080 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
4081 (P9V_BUILTIN_VEC_REVB): Likewise.
4082 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
4083 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
4084 (p9_xxbrq_v16qi): Likewise.
4085 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
4086 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
4087 (p9_xxbrh_v8hi): Likewise.
4088 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
4089 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
4090 vec_revb built-in functions.
4092 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
4094 PR rtl-optimization/78952
4095 * config/i386/i386.md (any_extract): New code iterator.
4096 (*insvqi_2): Use any_extract for source operand.
4097 (*insvqi_3): Use any_shiftrt for source operand.
4099 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
4101 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
4103 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
4105 2017-01-18 Matthias Klose <doko@ubuntu.com>
4107 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
4109 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4111 * config/rs6000/altivec.h (vec_bperm): Change #define.
4112 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
4113 (altivec_vbpermq2): New define_insn.
4114 (altivec_vbpermd): Likewise.
4115 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
4117 (VBPERMD): Likewise.
4118 (VBPERM): New polymorphic function interface.
4119 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
4120 Add entries for P9V_BUILTIN_VEC_VBPERM.
4121 * doc/extend.texi: Add interfaces for vec_bperm.
4123 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4125 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
4126 first letter of error messages.
4127 (s390_resolve_overloaded_builtin): Likewise.
4128 * config/s390/s390.c (s390_expand_builtin): Likewise.
4129 (s390_invalid_arg_for_unprototyped_fn): Likewise.
4130 (s390_valid_target_attribute_inner_p): Likewise.
4131 * config/s390/s390.md ("tabort"): Likewise.
4133 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
4135 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
4136 (ISA_AVOID_DIV_HILO): New macro.
4137 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
4138 (ISA_HAS_DDIV): Likewise.
4140 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4142 * doc/invoke.texi (fabi-version): Correct number of occurrences.
4144 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4146 * doc/invoke.texi (fabi-version): Spelling fix.
4148 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4151 * doc/invoke.texi (fabi-version): Mention mangling fix for
4154 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4157 * doc/invoke.texi (fabi-version): Document discriminator mangling.
4159 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
4162 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
4163 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
4165 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
4167 (stack_protect_test): Ditto.
4168 * config/rs6000/rs6000.opt (mstack-protector-guard=,
4169 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
4171 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
4172 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
4173 -mstack-protector-guard-offset=.
4174 (RS/6000 and PowerPC Options): Ditto.
4176 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
4178 * config/i386/i386.h (MASK_CLASS_P): New define.
4179 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
4180 there are no registers from different register sets also when
4181 mask registers are used. Update function comment.
4182 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
4183 to (*k/*r) and (*k/*km) alternatives.
4185 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
4187 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
4188 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
4189 (EH_RETURN_HANDLER_RTX): New define.
4190 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
4191 Force frame pointer in EH return functions.
4192 (aarch64_expand_epilogue): Add barrier for eh_return.
4193 (aarch64_final_eh_return_addr): Remove.
4194 (aarch64_eh_return_handler_rtx): New function.
4195 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
4197 (aarch64_eh_return_handler_rtx): New prototype.
4199 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4201 * config/rs6000/altivec.h (vec_rlmi): New #define.
4202 (vec_vrlnm): Likewise.
4203 (vec_rlnm): Likewise.
4204 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
4205 (UNSPEC_VRLNM): Likewise.
4206 (VIlong): New mode iterator.
4207 (altivec_vrl<VI_char>mi): New define_insn.
4208 (altivec_vrl<VI_char>nm): Likewise.
4209 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
4212 (RLNM): New polymorphic function entry.
4213 (VRLWMI): New monomorphic function entry.
4215 (RLMI): New polymorphic function entry.
4216 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
4217 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
4218 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
4221 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4224 * dwarf2out.c (field_byte_offset): Restore the
4225 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
4226 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
4227 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
4230 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
4233 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
4235 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4238 * dwarf2out.c (add_data_member_location_attribute): For constant
4239 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
4240 instead of DW_AT_data_member_location, DW_AT_bit_offset and
4241 DW_AT_byte_size attributes.
4243 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
4245 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
4246 after forcing to constant memory when the code model is medium.
4248 2017-01-17 Julia Koval <julia.koval@intel.com>
4251 * config/i386/avx512fintrin.h
4252 (_mm512_i32gather_ps): Change __addr type to void const*.
4253 (_mm512_mask_i32gather_ps): Ditto.
4254 (_mm512_i32gather_pd): Ditto.
4255 (_mm512_mask_i32gather_pd): Ditto.
4256 (_mm512_i64gather_ps): Ditto.
4257 (_mm512_mask_i64gather_ps): Ditto.
4258 (_mm512_i64gather_pd): Ditto.
4259 (_mm512_mask_i64gather_pd): Ditto.
4260 (_mm512_i32gather_epi32): Ditto.
4261 (_mm512_mask_i32gather_epi32): Ditto.
4262 (_mm512_i32gather_epi64): Ditto.
4263 (_mm512_mask_i32gather_epi64): Ditto.
4264 (_mm512_i64gather_epi32): Ditto.
4265 (_mm512_mask_i64gather_epi32): Ditto.
4266 (_mm512_i64gather_epi64): Ditto.
4267 (_mm512_mask_i64gather_epi64): Ditto.
4268 (_mm512_i32scatter_ps): Change __addr type to void*.
4269 (_mm512_mask_i32scatter_ps): Ditto.
4270 (_mm512_i32scatter_pd): Ditto.
4271 (_mm512_mask_i32scatter_pd): Ditto.
4272 (_mm512_i64scatter_ps): Ditto.
4273 (_mm512_mask_i64scatter_ps): Ditto.
4274 (_mm512_i64scatter_pd): Ditto.
4275 (_mm512_mask_i64scatter_pd): Ditto.
4276 (_mm512_i32scatter_epi32): Ditto.
4277 (_mm512_mask_i32scatter_epi32): Ditto.
4278 (_mm512_i32scatter_epi64): Ditto.
4279 (_mm512_mask_i32scatter_epi64): Ditto.
4280 (_mm512_i64scatter_epi32): Ditto.
4281 (_mm512_mask_i64scatter_epi32): Ditto.
4282 (_mm512_i64scatter_epi64): Ditto.
4283 (_mm512_mask_i64scatter_epi64): Ditto.
4284 * config/i386/avx512pfintrin.h
4285 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
4286 (_mm512_mask_prefetch_i32gather_ps): Ditto.
4287 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4288 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4289 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
4290 (_mm512_prefetch_i32scatter_ps): Ditto.
4291 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4292 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4293 (_mm512_prefetch_i64scatter_pd): Ditto.
4294 (_mm512_prefetch_i64scatter_ps): Ditto.
4295 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4296 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4297 * config/i386/avx512vlintrin.h
4298 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
4299 (_mm_mmask_i32gather_ps): Ditto.
4300 (_mm256_mmask_i32gather_pd): Ditto.
4301 (_mm_mmask_i32gather_pd): Ditto.
4302 (_mm256_mmask_i64gather_ps): Ditto.
4303 (_mm_mmask_i64gather_ps): Ditto.
4304 (_mm256_mmask_i64gather_pd): Ditto.
4305 (_mm_mmask_i64gather_pd): Ditto.
4306 (_mm256_mmask_i32gather_epi32): Ditto.
4307 (_mm_mmask_i32gather_epi32): Ditto.
4308 (_mm256_mmask_i32gather_epi64): Ditto.
4309 (_mm_mmask_i32gather_epi64): Ditto.
4310 (_mm256_mmask_i64gather_epi32): Ditto.
4311 (_mm_mmask_i64gather_epi32): Ditto.
4312 (_mm256_mmask_i64gather_epi64): Ditto.
4313 (_mm_mmask_i64gather_epi64): Ditto.
4314 (_mm256_i32scatter_ps): Change __addr type to void*.
4315 (_mm256_mask_i32scatter_ps): Ditto.
4316 (_mm_i32scatter_ps): Ditto.
4317 (_mm_mask_i32scatter_ps): Ditto.
4318 (_mm256_i32scatter_pd): Ditto.
4319 (_mm256_mask_i32scatter_pd): Ditto.
4320 (_mm_i32scatter_pd): Ditto.
4321 (_mm_mask_i32scatter_pd): Ditto.
4322 (_mm256_i64scatter_ps): Ditto.
4323 (_mm256_mask_i64scatter_ps): Ditto.
4324 (_mm_i64scatter_ps): Ditto.
4325 (_mm_mask_i64scatter_ps): Ditto.
4326 (_mm256_i64scatter_pd): Ditto.
4327 (_mm256_mask_i64scatter_pd): Ditto.
4328 (_mm_i64scatter_pd): Ditto.
4329 (_mm_mask_i64scatter_pd): Ditto.
4330 (_mm256_i32scatter_epi32): Ditto.
4331 (_mm256_mask_i32scatter_epi32): Ditto.
4332 (_mm_i32scatter_epi32): Ditto.
4333 (_mm_mask_i32scatter_epi32): Ditto.
4334 (_mm256_i32scatter_epi64): Ditto.
4335 (_mm256_mask_i32scatter_epi64): Ditto.
4336 (_mm_i32scatter_epi64): Ditto.
4337 (_mm_mask_i32scatter_epi64): Ditto.
4338 (_mm256_i64scatter_epi32): Ditto.
4339 (_mm256_mask_i64scatter_epi32): Ditto.
4340 (_mm_i64scatter_epi32): Ditto.
4341 (_mm_mask_i64scatter_epi32): Ditto.
4342 (_mm256_i64scatter_epi64): Ditto.
4343 (_mm256_mask_i64scatter_epi64): Ditto.
4344 (_mm_i64scatter_epi64): Ditto.
4345 (_mm_mask_i64scatter_epi64): Ditto.
4346 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
4347 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
4348 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
4349 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
4350 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
4351 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
4352 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
4353 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
4354 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
4355 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
4356 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
4357 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
4358 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
4359 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
4360 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
4361 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
4362 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
4363 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
4364 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
4365 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
4366 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
4367 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
4368 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
4369 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
4370 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
4371 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
4372 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
4373 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
4374 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
4375 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
4376 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
4377 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
4378 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
4379 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
4380 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
4381 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
4382 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
4383 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
4384 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
4385 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
4386 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
4387 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
4388 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
4389 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
4390 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
4391 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
4392 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
4393 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
4394 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
4395 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
4396 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
4397 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
4398 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
4399 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
4400 definitions accordingly.
4402 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
4403 Kuan-Lin Chen <kuanlinchentw@gmail.com>
4406 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
4409 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
4412 * ira-conflicts.c (ira_build_conflicts): Update total conflict
4413 hard regs for inner regno.
4415 2017-01-17 Martin Liska <mliska@suse.cz>
4418 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
4419 assumption and add comment.
4421 2017-01-17 Nathan Sidwell <nathan@acm.org>
4423 * ipa-visibility.c (localize_node): New function, broken out of ...
4424 (function_and_variable_visibility): ... here. Call it.
4426 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
4429 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
4430 correctly set frequency of oudgoing edge.
4431 (duplicate_thread_path): Fix profile updating.
4433 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4436 * configure.ac: Add GCC_BASE_VER.
4437 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
4438 version from BASE-VER file.
4439 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
4440 (gcc.o): Depend on $(BASEVER).
4441 * common.opt (dumpfullversion): New option.
4442 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
4443 * doc/invoke.texi: Document -dumpfullversion.
4444 * doc/install.texi: Document --with-gcc-major-version-only.
4445 * configure: Regenerated.
4447 2017-01-17 Richard Biener <rguenther@suse.de>
4449 PR tree-optimization/71433
4450 * tree-vrp.c (register_new_assert_for): Merge same asserts
4451 on all incoming edges.
4452 (process_assert_insertions_for): Handle insertions at the
4455 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
4457 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
4458 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
4460 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
4463 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
4466 2017-01-17 Alan Modra <amodra@gmail.com>
4469 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
4470 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
4471 symbolic stack limit when pic.
4473 2017-01-16 Martin Sebor <msebor@redhat.com>
4475 PR tree-optimization/78608
4476 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
4478 2017-01-16 Jeff Law <law@redhat.com>
4481 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
4482 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
4483 for several include directories that may be relative to sysroot.
4484 * config/i386/x-mingw32 (gplus_includedir): Define.
4485 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
4486 (native_system_includedir): Likewise.
4487 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
4488 override if TARGET_SYSTEM_ROOT is defined.
4489 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
4491 PR tree-optimization/79090
4492 PR tree-optimization/33562
4493 PR tree-optimization/61912
4494 PR tree-optimization/77485
4495 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
4496 and computed trims into the dump file.
4498 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
4500 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
4502 2017-01-16 Jakub Jelinek <jakub@redhat.com>
4505 * gimplify.c (gimplify_init_constructor): If want_value and
4506 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
4510 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
4511 sequence. Formatting fixes.
4512 (doloop_optimize): Formatting fixes.
4515 * gcc.c (debug_level_greater_than_spec_func): New function.
4516 (static_spec_functions): Add debug-level-gt spec function.
4517 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
4519 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
4520 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
4521 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
4522 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
4523 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
4524 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
4526 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
4528 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
4529 QImode fixups to general and mask registers only.
4531 2017-01-16 Carl Love <cel@us.ibm.com>
4533 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
4534 for built-in functions
4535 vector signed char vec_nabs (vector signed char)
4536 vector signed short vec_nabs (vector signed short)
4537 vector signed int vec_nabs (vector signed int)
4538 vector signed long long vec_nabs (vector signed long long)
4539 vector float vec_nabs (vector float)
4540 vector double vec_nabs (vector double)
4541 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
4543 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
4544 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
4545 * doc/extend.texi: Update the documentation file for the new built-in
4548 2017-01-16 Martin Sebor <msebor@redhat.com>
4550 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
4553 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4555 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
4556 UNSPEC_VSX__XXSPLTD to require special splat handling.
4558 2017-01-16 David Malcolm <dmalcolm@redhat.com>
4561 * system.h: Poison strndup.
4563 2017-01-16 Alan Modra <amodra@gmail.com>
4566 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
4569 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
4571 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
4573 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
4575 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
4576 call recog here. Assert that INSN_CODE (insn) is non-negative.
4578 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
4581 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
4583 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
4584 in the currently scheduled RTL fragment.
4586 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
4588 PR rtl-optimization/78751
4589 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
4592 2017-01-14 Jeff Law <law@redhat.com>
4594 PR tree-optimization/79090
4595 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
4596 variable length stores.
4597 (compute_trims): Delete dead assignment to *trim_tail.
4598 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
4601 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
4603 PR rtl-optimization/78626
4604 PR rtl-optimization/78727
4605 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
4606 of a block, and split such blocks after everything else is finished.
4608 2017-01-14 Alan Modra <amodra@gmail.com>
4611 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
4612 target legitimate_combined_insn.
4613 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
4614 (rs6000_legitimate_combined_insn): New function.
4615 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
4617 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
4618 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
4619 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
4621 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
4623 * doc/frontends.texi (G++ and GCC): Remove references to Java.
4625 2017-01-13 Jeff Law <law@redhat.com>
4627 PR tree-optimization/33562
4628 PR tree-optimization/61912
4629 PR tree-optimization/77485
4630 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
4632 (delete_dead_assignment): Likewise.
4633 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
4634 statement to delete_dead_call and delete_dead_assignment.
4636 2017-01-13 David Malcolm <dmalcolm@redhat.com>
4639 * substring-locations.c (format_warning_va): Strengthen case 1 so
4640 that both endpoints of the substring must be within the format
4641 range for just the substring to be printed.
4643 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
4645 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
4646 * config/i386/i386.c (ix86_target_string): Add missing options
4647 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
4648 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
4649 flags_other and ix86_target_other to flags2_other. Display unknown
4651 (ix86_valid_target_attribute_inner_p): Add missing options and
4652 reorder options by implied ISAs, as in ix86_target_string.
4654 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
4656 * hash-table.h (hash_table::too_empty_p): New function.
4657 (hash_table::expand): Use it.
4658 (hash_table::traverse): Likewise.
4659 (hash_table::empty_slot): Use sizeof (value_type) instead of
4660 sizeof (PTR) to convert bytes to elements. Shrink the table
4661 if the current size is excessive for the current number of
4664 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
4666 * ira-costs.c (record_reg_classes): Break from the inner loop
4667 early once alt_fail is known to be true. Update outer loop
4668 handling accordingly.
4670 2017-01-13 Jeff Law <law@redhat.com>
4672 * tree-ssa-dse.c (decrement_count): New function.
4673 (increment_start_addr, maybe_trim_memstar_call): Likewise.
4674 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
4675 when we know the partially dead statement is a mem* function.
4677 PR tree-optimization/61912
4678 PR tree-optimization/77485
4679 * tree-ssa-dse.c: Include expr.h.
4680 (maybe_trim_constructor_store): New function.
4681 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
4683 PR tree-optimization/33562
4684 PR tree-optimization/61912
4685 PR tree-optimization/77485
4686 * doc/invoke.texi: Document new dse-max-object-size param.
4687 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
4688 * tree-ssa-dse.c: Include params.h.
4689 (dse_store_status): New enum.
4690 (initialize_ao_ref_for_dse): New, partially extracted from
4692 (valid_ao_ref_for_dse, normalize_ref): New.
4693 (setup_live_bytes_from_ref, compute_trims): Likewise.
4694 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
4695 (maybe_trim_partially_dead_store): Likewise.
4696 (maybe_trim_complex_store): Likewise.
4697 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
4698 Track what bytes live from the original store. Return tri-state
4699 for dead, partially dead or live.
4700 (dse_dom_walker): Add constructor, destructor and new private members.
4701 (delete_dead_call, delete_dead_assignment): New extracted from
4703 (dse_optimize_stmt): Make a member of dse_dom_walker.
4704 Use initialize_ao_ref_for_dse.
4706 PR tree-optimization/33562
4707 PR tree-optimization/61912
4708 PR tree-optimization/77485
4709 * sbitmap.h (bitmap_count_bits): Prototype.
4710 (bitmap_clear_range, bitmap_set_range): Likewise.
4711 * sbitmap.c (bitmap_clear_range): New function.
4712 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
4714 2017-01-13 Martin Liska <mliska@suse.cz>
4717 * function.c (set_cfun): Add new argument force.
4718 * function.h (set_cfun): Likewise.
4719 * ipa-inline-transform.c (inline_call): Use the function when
4720 strict alising from is dropped for function we inline to.
4722 2017-01-13 Richard Biener <rguenther@suse.de>
4724 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
4725 for dumping GIMPLE INTEGER_CSTs.
4727 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4729 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
4730 to 201112L since C++17.
4732 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
4735 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
4736 if -fsanitize=kernel-address is present.
4738 2017-01-13 Richard Biener <rguenther@suse.de>
4740 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
4741 as _Literal ( type ) number in case usual suffixes do not
4742 preserve all information.
4744 2017-01-13 Richard Biener <rguenther@suse.de>
4746 PR tree-optimization/77283
4747 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
4748 and ssa-iterators.h.
4749 (is_feasible_trace): Implement a cost model based on joiner
4752 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4755 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
4756 char or short to __float128/_Float128 directly.
4758 2017-01-12 Martin Sebor <msebor@redhat.com>
4760 to -Wformat-overflow.
4761 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
4762 (min_bytes_remaining): Same.
4763 (get_string_length): Same.
4764 (format_string): Same.
4765 (format_directive): Same.
4767 (pass_sprintf_length::handle_gimple_call): Same.
4769 2017-01-12 Jakub Jelinek <jakub@redhat.com>
4771 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
4772 info.nowrite calls with no lhs that can't throw. Return bool
4773 whether gsi_remove has been called or not.
4774 (pass_sprintf_length::handle_gimple_call): Return bool whether
4775 try_substitute_return_value called gsi_remove. Formatting fix.
4776 (pass_sprintf_length::execute): Don't use gsi_remove if
4777 handle_gimple_call returned true.
4780 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
4781 be removed due to side-effects, don't remove following barrier nor
4782 turn the successor edge into fallthru edge.
4784 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4787 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
4788 element-reversing loads and stores as not swappable.
4790 2017-01-12 Nathan Sidwell <nathan@acm.org>
4791 Nicolai Stange <nicstange@gmail.com>
4793 * combine.c (try_combine): Don't ignore result of overlap checking
4794 loop. Combine overlap & asm check into single loop.
4796 2017-01-12 Richard Biener <rguenther@suse.de>
4798 * tree-pretty-print.c (dump_generic_node): Provide -gimple
4799 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
4801 2017-01-12 Richard Biener <rguenther@suse.de>
4803 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
4804 and TS_TARGET_OPTION directly derive from TS_BASE.
4805 * tree-core.h (tree_optimization_option): Derive from tree_base.
4806 (tree_target_option): Likewise.
4808 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
4810 * config/i386/i386.c (memory_address_length): Increase len
4811 only when rip_relative_addr_p returns false.
4813 2017-01-11 Julia Koval <julia.koval@intel.com>
4815 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
4816 (OPTION_MASK_ISA_SGX_SET): New.
4817 (ix86_handle_option): Handle OPT_msgx.
4818 * config.gcc: Added sgxintrin.h.
4819 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
4820 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
4821 * config/i386/i386.c (ix86_target_string): Add -msgx.
4823 (ix86_option_override_internal): Handle new options.
4824 (ix86_valid_target_attribute_inner_p): Add sgx.
4825 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
4826 * config/i386/i386.opt: Add msgx.
4827 * config/i386/sgxintrin.h: New file.
4828 * config/i386/x86intrin.h: Add sgxintrin.h.
4830 2017-01-11 Jakub Jelinek <jakub@redhat.com>
4833 * fold-const.c (maybe_nonzero_address): Return 1 for function
4835 (tree_single_nonzero_warnv_p): Don't handle function local objects
4839 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
4842 2017-01-11 David Malcolm <dmalcolm@redhat.com>
4845 * opts.c: Include "spellcheck.h"
4846 (struct string_fragment): New struct.
4847 (struct edit_distance_traits<const string_fragment &>): New
4849 (get_closest_sanitizer_option): New function.
4850 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
4852 2017-01-11 Jakub Jelinek <jakub@redhat.com>
4854 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
4856 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
4857 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
4858 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
4859 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
4860 for initial die_offset if dwarf_split_debug_info.
4861 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
4862 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
4864 (output_skeleton_debug_sections): Formatting fix. Use
4865 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
4866 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
4868 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
4870 * config/arm/cortex-a53.md: Add bypasses for
4872 (cortex_a53_r2f): Only use for transfers.
4873 (cortex_a53_f2r): Likewise.
4874 (cortex_a53_r2f_cvt): Add reservation for conversions.
4875 (cortex_a53_f2r_cvt): Likewise.
4877 2017-01-11 Tamar Christina <tamar.christina@arm.com>
4879 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
4880 to all inlined functions, change static to extern.
4882 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
4885 * config/arm/arm.c (legitimize_pic_address): Handle reference to
4887 (arm_assemble_integer): Likewise.
4889 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
4891 * config.gcc: Use new awk script to check CPU, FPU and architecture
4892 parameters for --with-... options.
4893 * config/arm/parsecpu.awk: New file
4894 * config/arm/arm-cpus.in: New file.
4895 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
4897 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
4899 * config/arm/t-arm: Update dependency rules.
4900 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
4901 of processing .def files.
4902 * config/arm/genopt.sh: Deleted.
4903 * config/arm/gentune.sh: Deleted.
4904 * config/arm/arm-cores.def: Deleted.
4905 * config/arm/arm-arches.def: Deleted.
4906 * config/arm/arm-fpus.def: Deleted.
4907 * config/arm/arm-tune.md: Regenerated.
4908 * config/arm/arm-tables.opt: Regenerated.
4909 * config/arm/arm-cpu.h: New generated file.
4910 * config/arm/arm-cpu-data.h: New generated file.
4911 * config/arm/arm-cpu-cdata.h: New generated file.
4913 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
4916 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
4918 (input_varpool_node): Unpack dynamically_initialized bit.
4920 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
4922 PR rtl-optimization/79032
4923 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
4924 the alignment of the adjusted memory reference against that of MODE,
4925 instead of the alignment of the original memory reference.
4927 2017-01-11 Martin Jambor <mjambor@suse.cz>
4929 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
4931 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
4932 decorated functions.
4934 2017-01-11 Richard Biener <rguenther@suse.de>
4936 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
4937 set range/nonnull info for PHI results. Do not set it on
4938 stmts marked for removal.
4940 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
4942 * expr.c (store_field): In the bitfield case, fetch the return value
4943 from the registers before applying a single big-endian adjustment.
4944 Always do a final load for a BLKmode value not larger than a word.
4946 2017-01-10 David Malcolm <dmalcolm@redhat.com>
4949 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
4950 that we correctly handle column numbers greater than
4951 LINE_MAP_MAX_COLUMN_NUMBER.
4953 2017-01-10 Martin Sebor <msebor@redhat.com>
4956 * gimple-ssa-sprintf.c (get_destination_size): Call
4957 {init,fini}object_sizes.
4958 * tree-object-size.c (addr_object_size): Adjust.
4959 (pass_through_call): Adjust.
4960 (pass_object_sizes::execute): Adjust.
4961 * tree-object-size.h (fini_object_sizes): Declare.
4963 2017-01-10 Martin Sebor <msebor@redhat.com>
4965 PR tree-optimization/78775
4966 * builtins.c (get_size_range): Move...
4967 * calls.c: ...to here.
4968 (alloc_max_size): Accept zero argument.
4969 (operand_signed_p): Remove.
4970 (maybe_warn_alloc_args_overflow): Call get_size_range.
4971 * calls.h (get_size_range): Declare.
4973 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
4975 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
4976 from TI's devices.csv file as of September 2016.
4977 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
4979 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
4981 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
4982 * doc/invoke.texi: Likewise.
4983 * doc/md.texi: Likewise.
4984 * doc/objc.texi: Likewise.
4986 2017-01-10 Joshua Conner <joshconner@google.com>
4988 * config/arm/fuchsia-elf.h: New file.
4989 * config/fuchsia.h: New file.
4990 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
4991 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
4993 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
4995 2016-01-10 Richard Biener <rguenther@suse.de>
4997 PR tree-optimization/79034
4998 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
4999 Propagate out degenerate PHIs in the joiner.
5001 2017-01-10 Martin Liska <mliska@suse.cz>
5003 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
5004 (sort_congruence_classes_by_decl_uid): Likewise.
5005 (sort_congruence_class_groups_by_decl_uid): Likewise.
5006 (sem_item_optimizer::merge_classes): Sort class, groups in these
5007 classes and members in the groups by DECL_UID of declarations.
5008 This would make merge operations stable.
5010 2017-01-10 Martin Liska <mliska@suse.cz>
5012 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
5013 usage of m_classes_vec.
5014 (sem_item_optimizer::~sem_item_optimizer): Likewise.
5015 (sem_item_optimizer::get_group_by_hash): Likewise.
5016 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
5017 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
5018 (sem_item_optimizer::verify_classes): Likewise.
5019 (sem_item_optimizer::process_cong_reduction): Likewise.
5020 (sem_item_optimizer::dump_cong_classes): Likewise.
5021 (sem_item_optimizer::merge_classes): Likewise.
5022 * ipa-icf.h (congruence_class_hash): Rename from
5023 congruence_class_group_hash. Remove declaration of m_classes_vec.
5025 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
5027 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
5028 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
5029 * config.gcc: Add avx512vpopcntdqintrin.h.
5030 * config/i386/avx512vpopcntdqintrin.h: New.
5031 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
5032 * config/i386/i386-builtin-types.def: Add new types.
5033 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
5034 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
5035 __builtin_ia32_vpopcountq_v8di_mask): New.
5036 * config/i386/i386-c.c (ix86_target_macros_internal): Define
5037 __AVX512VPOPCNTDQ__.
5038 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
5039 (PTA_AVX512VPOPCNTDQ): Define.
5040 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
5041 TARGET_AVX512VPOPCNTDQ_P): Define.
5042 * config/i386/i386.opt: Add mavx512vpopcntdq.
5043 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
5044 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
5046 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5049 * predict.def (PRED_CALL): Set to 67.
5051 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
5053 * expr.c (store_field): In the bitfield case, if the value comes from
5054 a function call and is of an aggregate type returned in registers, do
5055 not modify the field mode; extract the value in all cases if the mode
5056 is BLKmode and the size is not larger than a word.
5058 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
5061 * config/i386/cpuid.h: Fix undefined behavior.
5063 2017-01-04 Jeff Law <law@redhat.com>
5065 PR tree-optimization/79007
5066 PR tree-optimization/67955
5067 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
5068 conservative for pt.null when flag_non_call_exceptions is on.
5070 2017-01-09 Jakub Jelinek <jakub@redhat.com>
5072 PR translation/79019
5073 PR translation/79020
5074 * params.def (PARAM_INLINE_MIN_SPEEDUP,
5075 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
5076 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
5078 * config/avr/avr.opt (maccumulate-args): Likewise.
5079 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
5080 * common.opt (freport-bug): Likewise.
5081 * cif-code.def (CIF_FINAL_ERROR): Likewise.
5082 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
5083 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
5084 translatable string.
5085 * config/i386/i386.c (function_value_32): Likewise.
5086 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
5087 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
5089 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
5090 * common/config/msp430/msp430-common.c (msp430_handle_option):
5092 * symtab.c (symtab_node::verify_base): Likewise.
5093 * opts.c (set_debug_level): Likewise.
5094 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
5095 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
5096 missing whitespace to translatable strings.
5097 * config/avr/avr.md (bswapsi2): Fix typo in comment.
5098 * config/sh/superh.h: Likewise.
5099 * config/i386/xopintrin.h: Likewise.
5100 * config/i386/znver1.md: Likewise.
5101 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
5102 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
5103 * double-int.h (struct double_int): Likewise.
5104 * double-int.c (div_and_round_double): Likewise.
5105 * wide-int.cc: Likewise.
5106 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
5107 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
5108 * cfgcleanup.c (crossjumps_occured): Renamed to ...
5109 (crossjumps_occurred): ... this.
5110 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
5113 PR tree-optimization/78899
5114 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
5115 returning bool return struct loop *, NULL for failure and the new
5117 (versionable_outer_loop_p): Don't version outer loop if it has
5118 dont_vectorized bit set.
5119 (tree_if_conversion): When versioning outer loop, ensure
5120 tree_if_conversion is performed also on the inner loop of the
5121 non-vectorizable outer loop copy.
5122 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
5123 LOOP_VECTORIZED in inner loop of the scalar outer loop and
5124 prevent vectorization of it.
5125 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
5126 the outer loop vectorization of the non-scalar version is attempted
5127 before vectorization of the inner loop in scalar version. If
5128 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
5129 vectorization of its inner loop.
5130 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
5131 has 2 inner loops, rename also on edges from bb whose single pred
5132 is outer_loop->header. Fix typo in function comment.
5134 2017-01-09 Martin Sebor <msebor@redhat.com>
5137 * asan.c (asan_emit_stack_protection): Increase local buffer size
5138 to avoid snprintf truncation warning.
5140 2017-01-09 Andrew Pinski <apinski@cavium.com>
5142 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
5143 to reference thunderx2t99 for the tuning structure
5144 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
5146 (thunderx2t99_extra_costs): This.
5147 * config/aarch64/aarch64-tune.md: Regenerate.
5148 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
5149 (vulcan_addrcost_table): This.
5150 (vulcan_regmove_cost): Rename to ...
5151 (thunderx2t99_regmove_cost): This.
5152 (vulcan_vector_cost): Rename to ...
5153 (thunderx2t99_vector_cost): this.
5154 (vulcan_branch_cost): Rename to ...
5155 (thunderx2t99_branch_cost): This.
5156 (vulcan_tunings): Rename to ...
5157 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
5158 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
5160 2017-01-09 Martin Jambor <mjambor@suse.cz>
5164 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
5165 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
5166 (propagate_vr_accross_jump_function): Use the above function for all
5167 value range computations for pass-through jump functions and type
5168 converasion from explicit value range values.
5169 (ipcp_propagate_stage): Do not attempt to deduce types of formal
5170 parameters from TYPE_ARG_TYPES.
5171 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
5172 (ipa_write_node_info): Stream type of the actual argument.
5173 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
5175 2017-01-09 Martin Liska <mliska@suse.cz>
5178 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
5179 (lookup_compiler): Do not show error message with have_E.
5181 2017-01-09 Jakub Jelinek <jakub@redhat.com>
5183 PR tree-optimization/78938
5184 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
5185 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
5186 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
5187 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
5190 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5192 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
5195 2017-01-09 Richard Biener <rguenther@suse.de>
5197 PR tree-optimization/78997
5198 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
5199 name condition properly.
5201 2017-01-09 Richard Biener <rguenther@suse.de>
5204 * dwarf2out.c (is_cxx): New overload with context.
5205 (is_naming_typedef_decl): Use it.
5207 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5209 * invoke.texi (Option Summary): Correct spacing in option lists
5210 and add line breaks to fix over-long lines.
5212 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5216 * extend.texi (Common Variable Attributes): Add xref to GCC
5217 Internals manual to explain mode attribute keywords.
5219 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5222 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
5223 and Preprocessor Options.
5224 (Options for Linking): Document -pthread here....
5225 (RS/6000 and PowerPC Options): ...not here.
5226 (Solaris 2 Options): ...or here.
5227 * doc/cppopts.texi: Document -pthread.
5229 2017-01-08 Martin Sebor <msebor@redhat.com>
5232 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
5233 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
5234 New member functions.
5235 (format_directive): Used them.
5237 (pass_sprintf_length::handle_gimple_call): Same.
5238 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
5239 to avoid truncation for any argument.
5240 (extract_affine_mul): Same.
5241 * tree.c (get_file_function_name): Same.
5243 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5246 * predict.def (PRED_INDIR_CALL): Set to 86.
5248 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5250 PR preprocessor/54124
5251 * doc/cppopts.texi: Reformat -d subtable to list the full name
5252 of the options. Add cross-reference to the docs for the general
5253 compiler -d options.
5254 * doc/invoke.texi (Developer Options): Add cross-reference to the
5255 preprocessor-specific -d option documentation.
5257 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5259 PR preprocessor/13498
5260 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
5261 redudant material, and reflect new command-line options.
5262 (System Headers): Likewise.
5264 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5266 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
5267 -isystem, and -idirafter. Copy-edit.
5268 * doc/cppopts.texi: Copy-edit. Remove contradiction about
5269 default for -ftrack-macro-expansion. Delete obsolete and
5270 badly-formatted implementation details about -fdebug-cpp output.
5271 * doc/cppwarnopts.texi: Copy-edit.
5273 2017-01-07 David Malcolm <dmalcolm@redhat.com>
5276 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
5277 that the transition from a max line width >= 1<<10 to narrower
5278 lines works correctly.
5280 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
5282 * doc/options.texi (PerFunction): New.
5283 * opt-functions.awk (switch_flags): Map both Optimization and
5284 PerFunction to CL_OPTIMIZATION.
5285 * opth-gen.awk: Test for PerFunction flag along with
5287 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
5288 it only when the latter is present. Skip those that don't in
5289 the hash function generator.
5290 * common.opt (fvar-tracking): Mark as PerFunction instead of
5292 (fvar-tracking-assignments): Likewise.
5293 (fvar-tracking-assignments-toggle): Likewise.
5294 (fvar-tracking-uninit): Likewise.
5296 2017-01-07 Jakub Jelinek <jakub@redhat.com>
5298 PR translation/79018
5299 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
5302 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
5305 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
5306 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
5307 TARGET_LONG_JUMP_TABLE_OFFSETS.
5308 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
5309 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
5310 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
5311 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
5312 * config/m68k/m68k.md (tablejump expander): Likewise.
5313 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
5314 TARGET_LONG_JUMP_TABLE_OFFSETS.
5315 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
5316 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
5318 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
5319 David Holsgrove <david.holsgrove@xilinx.com>
5321 * common/config/microblaze/microblaze-common.c
5322 (TARGET_EXCEPT_UNWIND_INFO): Remove.
5323 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
5325 * config/microblaze/microblaze.c (microblaze_must_save_register)
5326 (microblaze_expand_epilogue, microblaze_return_addr): Handle
5328 (microblaze_eh_return): New function.
5329 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
5330 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
5331 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
5332 * config/microblaze/microblaze.md (eh_return): New pattern.
5334 2017-01-06 Jakub Jelinek <jakub@redhat.com>
5336 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
5337 GCC_DIAGNOSTIC_STRINGIFY): Define.
5339 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
5341 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5343 * config/arm/arm.md (<mcrr>): New.
5345 * config/arm/arm.c (arm_arch5te): New.
5346 (arm_option_override): Set arm_arch5te.
5347 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
5349 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
5350 (arm_mcrr_qualifiers): ... this. New.
5351 (MRRC_QUALIFIERS): Define to...
5352 (arm_mrrc_qualifiers): ... this. New.
5353 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
5355 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
5356 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
5357 (MRRCI, mrrc, MRRC): New.
5358 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
5359 VUNSPEC_MRRC2): New.
5361 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5363 * config/arm/arm.md (<mcr>): New.
5365 * config/arm/arm.c (arm_coproc_builtin_available): Add
5366 support for mcr, mrc, mcr2 and mrc2.
5367 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
5368 (arm_mcr_qualifiers): ... this. New.
5369 (MRC_QUALIFIERS): Define to ...
5370 (arm_mrc_qualifiers): ... this. New.
5371 (MCR_QUALIFIERS): Define to ...
5372 (arm_mcr_qualifiers): ... this. New.
5373 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
5375 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
5376 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
5377 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
5380 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5382 * config/arm/arm.md (*ldc): New.
5386 * config/arm/arm.c (arm_coproc_builtin_available): Add
5387 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
5388 (arm_coproc_ldc_stc_legitimate_address): New.
5389 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
5390 'qualifier_const_pointer'.
5391 (LDC_QUALIFIERS): Define to...
5392 (arm_ldc_qualifiers): ... this. New.
5393 (STC_QUALIFIERS): Define to...
5394 (arm_stc_qualifiers): ... this. New.
5395 * config/arm/arm-protos.h
5396 (arm_coproc_ldc_stc_legitimate_address): New.
5397 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
5398 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
5399 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
5400 stc2, stcl, stc2l): New.
5401 * config/arm/constraints.md (Uz): New.
5402 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
5403 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
5404 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
5405 VUNSPEC_STC2L): New.
5407 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5409 * config/arm/arm.md (<cdp>): New.
5410 * config/arm/arm.c (neon_const_bounds): Rename this ...
5411 (arm_const_bounds): ... this.
5412 (arm_coproc_builtin_available): New.
5413 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
5414 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
5415 (CDP_QUALIFIERS): Define to...
5416 (arm_cdp_qualifiers): ... this. New.
5418 (arm_expand_builtin_args): Add case for 6 arguments.
5419 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
5420 (arm_const_bounds): ... this.
5421 (arm_coproc_builtin_available): New.
5422 * config/arm/arm_acle.h (__arm_cdp): New.
5424 * config/arm/arm_acle_builtins.def (cdp): New.
5426 * config/arm/iterators.md (CDPI,CDP,cdp): New.
5427 * config/arm/neon.md: Rename all 'neon_const_bounds' to
5429 * config/arm/types.md (coproc): New.
5430 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
5431 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
5432 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
5433 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
5435 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5437 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
5438 (UBINOP_QUALIFIERS): New.
5440 (acle_builtin_data): New. Change comment.
5441 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
5442 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
5443 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
5444 arm_acle_builtins.def.
5445 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
5446 (arm_init_acle_builtins): New.
5447 (CRC32_BUILTIN): Remove.
5448 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
5449 crc32cb, crc32ch and crc32cw.
5450 (arm_init_crc32_builtins): Remove.
5451 (arm_init_builtins): Use arm_init_acle_builtins rather
5452 than arm_init_crc32_builtins.
5453 (arm_expand_acle_builtin): New.
5454 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
5455 * config/arm/arm_acle_builtins.def: New.
5457 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5459 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
5460 (arm_builtin_datum): ... this.
5461 (arm_init_neon_builtin): Rename to ...
5462 (arm_init_builtin): ... this. Add a new parameters PREFIX
5463 and USE_SIG_IN_NAME.
5464 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
5465 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
5466 'arm_builtin_datum'.
5467 (arm_init_vfp_builtins): Likewise.
5468 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
5469 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
5470 (arm_expand_neon_args): Rename to ...
5471 (arm_expand_builtin_args): ... this. Rename builtin_arg
5472 enum values and differentiate between ARG_BUILTIN_MEMORY
5473 and ARG_BUILTIN_NEON_MEMORY.
5474 (arm_expand_neon_builtin_1): Rename to ...
5475 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
5476 values, arm_expand_builtin_args and add bool parameter NEON.
5477 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
5478 (arm_expand_vfp_builtin): Likewise.
5479 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
5481 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5484 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
5485 * predict.c (tree_estimate_probability_bb): Reverse direction of
5486 polymorphic call predictor.
5488 2017-01-06 David Malcolm <dmalcolm@redhat.com>
5490 * passes.c (execute_one_pass): Split out pass-skipping logic into...
5491 (determine_pass_name_match): ...this new function and...
5492 (should_skip_pass_p): ...this new function.
5494 2017-01-06 Nathan Sidwell <nathan@acm.org>
5496 * ipa-visibility.c (function_and_variable_visibility): Reformat
5497 comments and long lines. Remove extrneous if.
5498 * symtab.c (symtab_node::make_decl_local): Fix code format.
5499 (symtab_node::set_section_for_node): Fix comment typo.
5501 2017-01-06 Martin Liska <mliska@suse.cz>
5504 * lra-constraints.c: Rename invariant to lra_invariant.
5505 * predict.c (set_even_probabilities): Initialize e to NULL.
5507 2017-01-05 Martin Sebor <msebor@redhat.com>
5509 PR tree-optimization/78910
5510 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
5511 (format_integer): Correct off-by-one error in the handling
5512 of precision with negative numbers in signed conversions..
5514 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
5516 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
5518 2017-01-05 Jakub Jelinek <jakub@redhat.com>
5520 PR tree-optimization/71016
5521 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
5522 factor_out_conditional_conversion. Formatting fix.
5523 (factor_out_conditional_conversion): Add cond_stmt argument.
5524 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
5525 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
5528 2017-01-05 David Malcolm <dmalcolm@redhat.com>
5530 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
5531 read-rtl-function.o, and selftest-rtl.o.
5532 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
5533 (selftest::aarch64_test_loading_full_dump): New function.
5534 (selftest::aarch64_run_selftests): New function.
5535 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
5536 selftest::aarch64_run_selftests.
5537 * config/i386/i386.c
5538 (selftest::ix86_test_loading_dump_fragment_1): New function.
5539 (selftest::ix86_test_loading_call_insn): New function.
5540 (selftest::ix86_test_loading_full_dump): New function.
5541 (selftest::ix86_test_loading_unspec): New function.
5542 (selftest::ix86_run_selftests): Call the new functions.
5543 * emit-rtl.c (maybe_set_max_label_num): New function.
5544 * emit-rtl.h (maybe_set_max_label_num): New decl.
5545 * function.c (instantiate_decls): Guard call to
5546 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
5547 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
5549 * gensupport.c (gen_reader::gen_reader): Pass "false"
5550 for new "compact" param of rtx_reader.
5551 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
5552 rather than an empty string for NULL strings.
5553 * read-md.c: Potentially include config.h rather than bconfig.h.
5554 Wrap include of errors.h with #ifdef GENERATOR_FILE.
5555 (have_error): New global, copied from errors.c.
5556 (md_reader::read_name): Rename to...
5557 (md_reader::read_name_1): ...this, adding "out_loc" param,
5558 and converting "missing name or number" to returning false, rather
5560 (md_reader::read_name): Reimplement in terms of read_name_1.
5561 (md_reader::read_name_or_nil): New function.
5562 (md_reader::read_string): Handle "(nil)" by returning NULL.
5563 (md_reader::md_reader): Add new param "compact".
5564 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
5565 (md_reader::read_file): New method.
5566 * read-md.h (md_reader::md_reader): Add new param "compact".
5567 (md_reader::read_file): New method.
5568 (md_reader::is_compact): New accessor.
5569 (md_reader::read_name): Convert return type from void to file_location.
5570 (md_reader::read_name_or_nil): New decl.
5571 (md_reader::read_name_1): New decl.
5572 (md_reader::m_compact): New field.
5573 (noop_reader::noop_reader): Pass "false" for new "compact" param
5575 (rtx_reader::rtx_reader): Add new "compact" param.
5576 (rtx_reader::read_rtx_operand): Make virtual and convert return
5577 type from void to rtx.
5578 (rtx_reader::read_until): New decl.
5579 (rtx_reader::handle_any_trailing_information): New virtual function.
5580 (rtx_reader::postprocess): New virtual function.
5581 (rtx_reader::finalize_string): New virtual function.
5582 (rtx_reader::m_in_call_function_usage): New field.
5583 (rtx_reader::m_reuse_rtx_by_id): New field.
5584 * read-rtl-function.c: New file.
5585 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
5586 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
5587 (selftest::verify_three_block_rtl_cfg): New decl.
5588 * read-rtl-function.h: New file.
5589 * read-rtl.c: Potentially include config.h rather than bconfig.h.
5590 For host, include function.h, memmodel.h, and emit-rtl.h.
5591 (one_time_initialization): New function.
5592 (struct compact_insn_name): New struct.
5593 (compact_insn_names): New array.
5594 (find_code): Handle insn codes in compact dumps.
5595 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
5596 (bind_subst_iter_and_attr): Likewise.
5597 (add_condition_to_string): Likewise.
5598 (add_condition_to_rtx): Likewise.
5599 (apply_attribute_uses): Likewise.
5600 (add_current_iterators): Likewise.
5601 (apply_iterators): Likewise.
5602 (initialize_iterators): Guard usage of apply_subst_iterator with
5603 #ifdef GENERATOR_FILE.
5604 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
5605 (md_reader::read_mapping): Likewise.
5606 (add_define_attr_for_define_subst): Likewise.
5607 (add_define_subst_attr): Likewise.
5608 (read_subst_mapping): Likewise.
5609 (check_code_iterator): Likewise.
5610 (rtx_reader::read_rtx): Likewise. Move one-time initialization
5612 (one_time_initialization): New function.
5613 (rtx_reader::read_until): New method.
5614 (read_flags): New function.
5615 (parse_reg_note_name): New function.
5616 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
5617 Handle reuse_rtx ids.
5618 Wrap iterator lookup within #ifdef GENERATOR_FILE.
5619 Add parsing support for RTL dumps, mirroring the special-cases in
5620 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
5621 values, and calling handle_any_trailing_information.
5622 (rtx_reader::read_rtx_operand): Convert return type from void
5623 to rtx, returning return_rtx. Handle case 'e'. Call
5624 finalize_string on XSTR and XTMPL fields.
5625 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
5626 "(nil)" values were omitted. Call the postprocess vfunc on the
5628 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
5629 class ctor. Initialize m_in_call_function_usage. Call
5630 one_time_initialization.
5631 * rtl-tests.c (selftest::test_uncond_jump): Call
5632 set_new_first_and_last_insn.
5633 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
5634 * selftest-rtl.c: New file.
5635 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
5636 (selftest::get_insn_by_uid): New decl.
5637 * selftest-run-tests.c (selftest::run_tests): Call
5638 read_rtl_function_c_tests.
5639 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
5640 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
5643 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
5645 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
5646 operands in a special way. Assert that pos+len <= mode precision.
5648 2017-01-05 Jakub Jelinek <jakub@redhat.com>
5650 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
5651 3 argument Alias with unlimited for the negative form.
5652 (fno-vect-cost-model): Removed.
5654 2017-01-05 Martin Liska <mliska@suse.cz>
5656 * hsa-gen.c (gen_hsa_divmod): New function.
5657 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
5659 2017-01-05 Martin Liska <mliska@suse.cz>
5662 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
5665 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5667 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
5668 small constant length operands.
5670 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5672 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
5673 between loop iterations.
5675 2017-01-05 Martin Liska <mliska@suse.cz>
5678 * gimplify.c (gimplify_decl_expr): Compare to
5679 asan_poisoned_variables instread of checking flags.
5680 (gimplify_target_expr): Likewise.
5681 (gimplify_expr): Likewise.
5682 (gimplify_function_tree): Conditionally initialize
5683 asan_poisoned_variables.
5685 2017-01-04 Jeff Law <law@redhat.com>
5687 PR tree-optimizatin/78812
5688 * rtl.h (contains_mem_rtx_p): Prototype.
5689 * ifcvt.c (containts_mem_rtx_p): Move from here to...
5690 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
5691 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
5692 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
5693 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
5695 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5697 * input.c (assert_char_at_range): Default-initialize actual_range.
5699 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5701 * df-scan.c (df_ref_create_structure): Make regno unsigned,
5702 to match the caller.
5704 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5706 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
5707 insns after final jump in test to emit dummy move.
5709 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5711 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
5712 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
5714 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5716 * multiple_target.c (create_dispatcher_calls): Init e_next.
5717 * tree-ssa-loop-split.c (split_loop): Init border.
5718 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
5721 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
5726 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
5727 (altivec_register_operand): Do not return true if the operand
5728 contains a SUBREG mixing SImode and SFmode.
5729 (vsx_register_operand): Likewise.
5730 (vsx_reg_sfsubreg_ok): New predicate.
5731 (vfloat_operand): Do not return true if the operand contains a
5732 SUBREG mixing SImode and SFmode.
5733 (vint_operand): Likewise.
5734 (vlogical_operand): Likewise.
5735 (gpc_reg_operand): Likewise.
5736 (int_reg_operand): Likewise.
5737 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
5738 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
5739 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
5741 (rs6000_emit_move_si_sf_subreg): New helper function.
5742 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
5743 fixup SUBREGs involving SImode and SFmode.
5744 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
5745 numbers for the new peephole2 optimization.
5746 (peephole2 for SFmode unions): New peephole2 to optimize cases in
5747 the GLIBC math library that do AND/IOR/XOR operations on single
5748 precision floating point.
5749 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
5750 target macros to say whether we need to avoid SUBREGs mixing
5752 (TARGET_ALLOW_SF_SUBREG): Likewise.
5753 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
5754 (UNSPEC_SI_FROM_SF): Likewise.
5755 (iorxor): Change spacing.
5756 (and_ior_xor): New iterator for AND, IOR, and XOR.
5757 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
5758 (movdi_from_sf_zero_ext): Likewise.
5759 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
5760 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
5761 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
5762 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
5763 (fms<mode>4): Likewise.
5764 (fnma<mode>4): Likewise.
5765 (fnms<mode>4): Likewise.
5766 (nfma<mode>4): Likewise.
5767 (nfms<mode>4): Likewise.
5769 2017-01-04 Marek Polacek <polacek@redhat.com>
5772 * doc/invoke.texi: Document -Wpointer-compare.
5774 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5776 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
5779 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
5780 descriptions for -gdwarf-5 and emit them as uleb128 instead of
5783 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
5786 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
5787 documentation of the powerpc_popcntb_ok attribute.
5788 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
5789 code to issue warning messages if a requested CPU configuration is
5790 not supported by the binary (assembler and loader) toolchain.
5791 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
5792 made to define a built-in function that has been disabled.
5793 (paired_init_builtins): Add assertion to prevent ICE if attempt is
5794 made to define a built-in function that has been disabled.
5795 (altivec_init_builtins): Add comment explaining why definition
5796 of the DST built-in functions is not preceded by an assertion
5797 check. Add assertions to prevent ICE if attempts are made to
5798 define an altivec predicate or an abs* built-in function that has
5800 (htm_init_builtins): Add comment explaining why definition of the
5801 htm built-in functions is not preceded by an assertion check.
5803 2017-01-04 Jeff Law <law@redhat.com>
5805 PR tree-optimizatin/67955
5806 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
5807 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
5808 the points-to solution does not include pt_null. Use DECL_PT_UID
5811 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
5813 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
5814 Use gen_int_mode instead of gen_lopwart for const_int operands.
5816 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5818 PR tree-optimization/71563
5819 * match.pd: Simplify X << Y into X if Y is known to be 0 or
5820 out of range value - has low bits known to be zero.
5822 2017-01-04 Alan Modra <amodra@gmail.com>
5824 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
5825 * configure: Regenerate.
5826 * config.in: Regenerate.
5828 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5831 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
5832 a substring of the message, but strcmp with the whole message. Ifdef
5833 ENABLE_NLS, translate the message first using dgettext.
5835 2017-01-03 Jeff Law <law@redhat.com>
5837 PR tree-optimizatin/78856
5838 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
5839 (mark_threaded_blocks): Remove code to truncate thread paths that
5840 cross multiple loop headers. Instead invalidate the cached loop
5841 iteration information and handle case of a thread path walking
5842 into an irreducible region.
5844 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5847 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
5848 assertions. Add support for doing the signbit if the IEEE 128-bit
5849 floating point value is in a GPR.
5850 * config/rs6000/rs6000.md (Fsignbit): Delete.
5851 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
5852 Update the length attribute if the value is in a GPR.
5853 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
5854 the sign or zero extension instruction, since the value is always 0/1.
5855 (signbit<mode>2_dm2): Delete using <Fsignbit>.
5858 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
5859 extracting SImode to a GPR register so that we can generate a
5860 store, limit the vector to be in a traditional Altivec register
5861 for the vextuwrx instruction.
5863 2017-01-03 Ian Lance Taylor <iant@google.com>
5865 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
5867 2017-01-03 Martin Sebor <msebor@redhat.com>
5869 PR tree-optimization/78696
5870 * gimple-ssa-sprintf.c (format_floating): Correct handling of
5871 precision. Use MPFR for %f for greater fidelity. Correct handling
5873 (pass_sprintf_length::compute_format_length): Set width and precision
5874 specified by asrerisk to void_node for vararg functions.
5875 (try_substitute_return_value): Adjust dump output.
5877 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
5879 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
5881 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
5883 * doc/invoke.texi (SPARC options): Document -mlra as the default.
5884 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
5885 -mlra/-mno-lra was passed to the compiler.
5887 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
5889 PR rtl-optimization/65618
5890 * emit-rtl.c (try_split): Move initialization of "before" and
5891 "after" to just before the call to emit_insn_after_setloc.
5893 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
5895 * doc/md.texi (Standard Names): Remove reference to Java frontend.
5897 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
5899 * dwarf2out.c (gen_enumeration_type_die): When
5900 -gno-strict-dwarf, add a DW_AT_encoding attribute.
5902 2017-01-03 Jakub Jelinek <jakub@redhat.com>
5904 PR tree-optimization/78965
5905 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
5906 Change first argument from const call_info & to call_info &. For %n
5907 set info.nowrite to false.
5910 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
5911 possibly throwing calls.
5913 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
5914 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
5915 and fns handling, rather than in a separate case SSA_NAME.
5917 2017-01-02 Jeff Law <law@redhat.com>
5919 * config/darwin-driver.c (darwin_driver_init): Const-correctness
5920 fixes for first_period and second_period variables.
5922 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
5925 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
5926 (*insvqi_1): New insn pattern.
5927 (*insvqi_1_mem_rex64): Ditto.
5929 (*insvqi_3): Rename from *insvqi.
5931 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
5933 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
5935 * doc/cfg.texi (Edges): Remove reference to Java.
5936 (Maintaining the CFG): Ditto.
5938 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5941 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
5942 transparent aliases.
5944 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5947 * predict.def (PRED_CALL): Update hitrate.
5948 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
5949 * predict.c (tree_estimate_probability_bb): Split CALL predictor
5950 into direct/indirect/polymorphic variants.
5952 2017-01-01 Jakub Jelinek <jakub@redhat.com>
5954 Update copyright years.
5956 * gcc.c (process_command): Update copyright notice dates.
5957 * gcov-dump.c (print_version): Ditto.
5958 * gcov.c (print_version): Ditto.
5959 * gcov-tool.c (print_version): Ditto.
5960 * gengtype.c (create_file): Ditto.
5961 * doc/cpp.texi: Bump @copying's copyright year.
5962 * doc/cppinternals.texi: Ditto.
5963 * doc/gcc.texi: Ditto.
5964 * doc/gccint.texi: Ditto.
5965 * doc/gcov.texi: Ditto.
5966 * doc/install.texi: Ditto.
5967 * doc/invoke.texi: Ditto.
5969 Copyright (C) 2017 Free Software Foundation, Inc.
5971 Copying and distribution of this file, with or without modification,
5972 are permitted in any medium without royalty provided the copyright
5973 notice and this notice are preserved.