1 2017-03-14 Richard Biener <rguenther@suse.de>
3 * alias.c (struct alias_set_entry): Pack properly.
4 * cfgloop.h (struct loop): Likewise.
5 * cse.c (struct set): Likewise.
6 * ipa-utils.c (struct searchc_env): Likewise.
7 * loop-invariant.c (struct invariant): Likewise.
8 * lra-remat.c (struct cand): Likewise.
9 * recog.c (struct change_t): Likewise.
10 * rtl.h (struct address_info): Likewise.
11 * symbol-summary.h (function_summary): Likewise.
12 * tree-loop-distribution.c (struct partition): Likewise.
13 * tree-object-size.c (struct object_size_info): Likewise.
14 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
15 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
16 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
17 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
18 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
19 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
20 (struct _stmt_vec_info): Likewise.
22 2017-03-14 Martin Liska <mliska@suse.cz>
25 * multiple_target.c (create_dispatcher_calls): Check that
26 a target can create a function dispatcher.
28 2017-03-14 Martin Liska <mliska@suse.cz>
31 * multiple_target.c (expand_target_clones): Drop local.local
32 flag for default implementation.
34 2017-03-14 Richard Biener <rguenther@suse.de>
36 PR tree-optimization/80030
37 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
39 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
41 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
42 gcc_fallthrough() instead of __attribute__((fallthrough));
44 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
46 * doc/gcc.texi: Remove "up" link to (DIR).
47 * doc/gccint.texi: Ditto.
49 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
51 * doc/install.texi (Specific) <avr>: Remove reference to
54 2017-03-13 Jeff Law <law@redhat.com>
56 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
57 attribute rather than comments.
59 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
60 match_scratch operand is highest.
62 2017-03-13 Martin Liska <mliska@suse.cz>
65 * ipa-pure-const.c (warn_function_noreturn): If the declarations
66 is a CHKP clone, use original declaration.
68 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
70 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
71 (arc_conditional_register_usage): Use a different allocation order
72 when optimizing for size.
73 * common/config/arc/arc-common.c (arc_option_optimization_table):
74 Section anchors default on when optimizing for size.
76 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
78 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
80 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
82 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
83 * config/arc/arc.md (cpu_facility): Add cd variant.
84 (*movqi_insn): Add code density variant.
85 (*movhi_insn): Likewise.
86 (*movqi_insn): Likewise.
87 (*addsi3_mixed): Likewise.
88 (subsi3_insn): Likewise.
90 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
92 * config/arc/arc.md (movsi_cond_exec): Update constraint.
94 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
96 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
97 expressions with MINUS and UNARY ops.
99 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
102 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
104 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
105 between vec_select and vector argument.
106 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
107 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
108 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
109 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
110 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
111 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
113 2017-03-13 Richard Biener <rguenther@suse.de>
116 * params.def (vect-max-peeling-for-alignment): Fix typo.
118 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
120 * doc/install.texi (Specific) <mips-*-*>: Remove description of
121 issue that only occurred with binutils below 2.18.
123 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
125 * doc/install.texi (Specific) <cris-axis-elf>: No longer
126 refer to binutils 2.11/2.12 minimum.
128 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
130 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
131 ftp.kernel.org and simplify binutils requirement.
133 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
135 * doc/invoke.texi (Warning Options): Fix spelling of link-time
137 (Optimize Options): Ditto. Also remove redundancy.
139 2017-03-10 David Malcolm <dmalcolm@redhat.com>
142 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
144 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
145 to G_ to avoid double translation.
147 2017-03-10 David Malcolm <dmalcolm@redhat.com>
150 * auto-profile.c (get_combined_location): Convert leading
151 character of diagnostics to lower case and remove trailing period.
152 (read_profile): Likewise for various diagnostics.
153 * config/arm/arm.c (arm_option_override): Remove trailing period
154 from various diagnostics.
155 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
156 (msp430_expand_delay_cycles): Likewise.
158 2017-03-10 David Malcolm <dmalcolm@redhat.com>
161 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
162 full command-line argument, rather than just "str".
163 (aarch64_validate_march): Likewise.
164 (aarch64_validate_mtune): Likewise.
166 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
168 PR rtl-optimization/78911
169 * lra-assigns.c (must_not_spill_p): New function.
172 2017-03-10 Jakub Jelinek <jakub@redhat.com>
174 PR tree-optimization/79981
175 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
176 ATOMIC_COMPARE_EXCHANGE ifn result.
177 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
178 IFN_ATOMIC_COMPARE_EXCHANGE.
180 2017-03-10 David Malcolm <dmalcolm@redhat.com>
183 * opts.c (parse_sanitizer_options): Add missing question mark to
184 "did you mean" message.
186 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
188 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
190 (VMULEUH_UNS): Likewise.
191 (VMULOUB_UNS): Likewise.
192 (VMULOUH_UNS): Likewise.
193 * config/rs6000/rs6000.c (builtin_function_type): Remove
194 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
196 2017-03-10 David Malcolm <dmalcolm@redhat.com>
199 * read-rtl-function.c (function_reader::read_rtx_operand): Update
200 x with result of extra_parsing_for_operand_code_0.
201 (function_reader::extra_parsing_for_operand_code_0): Convert
202 return type from void to rtx, returning x. When reading
203 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
204 larger size containing struct block_symbol.
206 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
208 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
209 -mfloat128-hardware without -m64.
211 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
214 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
215 entries to the case statement that marks unsigned arguments to
216 overloaded functions.
218 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
220 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
221 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
223 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
226 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
227 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
229 2017-03-10 Martin Liska <mliska@suse.cz>
233 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
234 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
237 2017-03-10 Olivier Hainque <hainque@adacore.com>
239 * tree-switch-conversion (array_value_type): Start by resetting
240 candidate type to it's main variant.
242 2017-03-10 Jakub Jelinek <jakub@redhat.com>
244 PR rtl-optimization/79909
245 * combine.c (try_combine): Use simplify_replace_rtx on individual
246 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
247 of the whole CALL_INSN_FUNCTION_USAGE.
249 PR tree-optimization/79972
250 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
251 get_range_info on SSA_NAMEs. Formatting fixes.
253 2017-03-10 Richard Biener <rguenther@suse.de>
254 Jakub Jelinek <jakub@redhat.com>
256 PR tree-optimization/77975
257 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
259 (get_val_for): For constant x return it. Formatting fix.
260 (loop_niter_by_eval): Avoid pointless looping if the next iteration
261 would use the same bases as the current one.
263 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
265 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
266 instead of vec_select for V1TImode.
267 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
269 (VSX_LE_128): Add V1TI to this mode iterator.
270 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
271 (*vsx_le_perm_store_<mode>): Likewise.
272 (pre-reload splitter for VSX stores): Likewise.
273 (post-reload splitter for VSX stores): Likewise.
274 (*vsx_xxpermdi2_le_<mode>): Likewise.
275 (*vsx_lxvd2x2_le_<mode>): Likewise.
276 (*vsx_stxvd2x2_le_<mode>): Likewise.
278 2017-03-09 Michael Eager <eager@eagercon.com>
280 Correct failures with --enable-checking=yes,rtl.
282 * config/microblaze/microblaze.c (microblaze_expand_shift):
283 Replace GET_CODE test with CONST_INT_P and INTVAL test with
285 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
286 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
288 2017-03-09 Richard Biener <rguenther@suse.de>
290 PR tree-optimization/79977
291 * graphite-scop-detection.c (scop_detection::merge_sese):
292 Handle the case of extra exits to blocks dominating the entry.
294 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
296 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
299 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
301 PR rtl-optimization/79949
302 * lra-constraints.c (process_alt_operands): Check memory when
303 trying to predict a cycle. Print about the overall increase.
305 2017-03-09 Richard Biener <rguenther@suse.de>
308 * gimple-expr.c (useless_type_conversion_p): Preserve
309 TYPE_SATURATING for fixed-point types.
311 2017-03-09 Richard Biener <rguenther@suse.de>
314 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
315 alignment of BLKmode params.
317 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
320 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
321 (VALL_NO_V2Q): Likewise.
323 * config/aarch64/aarch64-simd.md
324 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
326 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
327 VALL_NO_V2Q mode iterator.
328 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
330 2017-03-09 Martin Liska <mliska@suse.cz>
332 PR tree-optimization/79631
333 * tree-chkp-opt.c (chkp_is_constant_addr): Call
334 tree_int_cst_sign_bit just for INTEGER constants.
336 2017-03-09 Martin Liska <mliska@suse.cz>
340 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
343 2017-03-09 Marek Polacek <polacek@redhat.com>
346 * tree.c (inchash::add_expr): Handle TREE_VEC.
348 2017-03-09 Martin Liska <mliska@suse.cz>
351 (chkp_narrow_size_and_offset): New function.
352 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
353 (void chkp_parse_bit_field_ref): New function.
354 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
355 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
357 2017-03-09 Martin Liska <mliska@suse.cz>
360 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
361 (chkp_find_bounds_1): Remove gcc_unreachable.
363 2017-03-09 Jakub Jelinek <jakub@redhat.com>
366 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
367 BUILT_IN_SYNC*, determine the access type from the size suffix and
368 always build a MEM_REF with that type. Handle forgotten
369 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
372 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
373 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
374 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
375 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
376 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
377 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
378 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
379 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
380 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
381 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
382 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
383 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
384 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
385 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
386 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
387 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
388 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
389 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
390 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
391 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
392 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
393 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
394 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
395 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
396 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
397 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
398 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
399 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
400 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
401 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
402 definitions outside of __OPTIMIZE__ guarded section.
405 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
406 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
407 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
408 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
411 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
413 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
414 ("vfenez<mode>"): Add missing constraints.
416 2017-03-08 Martin Sebor <msebor@redhat.com>
419 * config/nds32/nds32.c (nds32_option_override):
421 2017-03-08 Jakub Jelinek <jakub@redhat.com>
424 * gimplify.c (gimplify_omp_for): Replace index var in outer
425 taskloop statement with an artificial variable and add
426 OMP_CLAUSE_PRIVATE clause for it.
428 2017-03-08 Richard Biener <rguenther@suse.de>
430 PR tree-optimization/79955
431 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
432 for accesses that are completely outside of the variable.
434 2017-03-08 Andrew Haley <aph@redhat.com>
436 PR tree-optimization/79943
437 * tree-ssa-loop-split.c (compute_new_first_bound): When
438 calculating the new upper bound, (END-BEG) should be added, not
441 2017-03-08 Jakub Jelinek <jakub@redhat.com>
443 * config/avr/avr.md (setmemhi): Make sure match_dup
444 operand number comes before match_scratch.
446 2017-03-08 Richard Biener <rguenther@suse.de>
448 PR tree-optimization/79920
449 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
450 with ncopies == 1 to ...
451 (vect_transform_slp_perm_load): ... here. Properly compute
452 all element loads by iterating VF times over the group. Do
453 not handle ncopies (computed in a broken way) in
454 vect_create_mask_and_perm.
456 2017-03-08 Jakub Jelinek <jakub@redhat.com>
459 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
460 is a uniform vector, use uniform_vector_p return value instead of
461 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
463 2017-03-07 Marek Polacek <polacek@redhat.com>
466 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
467 (alloca_call_type): Likewise.
469 2017-03-07 Martin Liska <mliska@suse.cz>
471 * gcov.c (process_args): Put comment to correct location.
473 2017-03-07 Martin Liska <mliska@suse.cz>
476 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
477 Use array_at_struct_end_p instead of DECL_CHAIN (field).
478 (chkp_narrow_bounds_for_field): Likewise.
479 (chkp_parse_array_and_component_ref): Pass one more argument to
482 2017-03-07 Richard Biener <rguenther@suse.de>
484 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
487 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
489 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
490 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
492 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
495 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
496 to end of description.
497 (PARAM_MAX_STORES_TO_MERGE): Likewise.
499 2017-03-07 Jakub Jelinek <jakub@redhat.com>
501 PR rtl-optimization/79901
502 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
504 (*avx512f_<code><mode>3<mask_name>): ... this.
505 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
506 iterator instead of VI8_AVX2_AVX512BW.
508 PR rtl-optimization/79901
509 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
510 min/max expander, expand it using expand_vec_cond_expr.
513 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
516 2017-03-06 Jakub Jelinek <jakub@redhat.com>
519 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
520 to void * for PCH reasons.
521 * dwarf2out.c (output_loc_operands, output_die): Cast
522 v.val_vec.array to unsigned char *.
524 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
527 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
530 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
532 PR rtl-optimization/79571
533 * lra-constraints.c (process_alt_operands): Calculate static
534 reject and subtract it from overall when only addresses will be
537 2017-03-06 Julia Koval <julia.koval@intel.com>
540 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
541 incoming stack boundary to 128 for 64-bit targets.
543 2017-03-06 Richard Biener <rguenther@suse.de>
545 PR tree-optimization/79894
546 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
547 to NULL after folding it.
549 2017-03-06 Richard Biener <rguenther@suse.de>
551 PR tree-optimization/79824
552 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
553 check disabling peeling for gaps.
555 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
557 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
558 attributes): Document gettimeofday.
560 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
562 * config/s390/s390.c (s390_option_override_internal): Set
563 PARAM_MIN_VECT_LOOP_BOUND
565 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
567 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
568 * config/s390/s390.md: Likewise.
570 2017-03-06 Jakub Jelinek <jakub@redhat.com>
573 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
574 (<avx2_avx512>_perm<mode>): Rename to ...
575 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
577 (<avx512>_perm<mode>_mask): Rename to ...
578 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
580 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
581 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
582 instead of VI8F_256_512.
583 (avx512f_perm<mode>): New define_expand.
584 (avx512f_perm<mode>_mask): Likewise.
585 (avx512f_perm<mode>_1<mask_name>): New define_insn.
586 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
588 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
590 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
591 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
593 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
595 2017-03-06 Martin Liska <mliska@suse.cz>
598 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
599 when having a SSA NAME w/o VAR_DECL assigned to it.
601 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
603 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
604 msa_dpsub_<su>_d): Fix MODE for vec_select.
606 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
608 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
610 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
612 2017-03-06 Richard Biener <rguenther@suse.de>
614 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
615 * plugin.c (register_plugin_info): Likewise.
616 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
618 2017-03-05 Jakub Jelinek <jakub@redhat.com>
620 * config/i386/sse.md (sse_storehps, sse_storelps,
621 avx_<castmode><avxsizesuffix>_<castmode>,
622 avx512f_<castmode><avxsizesuffix>_<castmode>,
623 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
624 in condition that at least one operand is not a MEM.
626 2017-03-03 Jakub Jelinek <jakub@redhat.com>
629 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
630 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
632 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
633 gimple_call_nothrow_p flag based on whether original builtin can throw.
634 If it can, emit following stmts on the fallthrough edge.
635 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
636 don't create new bb if inserting just debug stmts on the edge, try to
637 insert them on the fallthru bb or just reset debug stmts.
639 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
642 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
643 restore recog_data (including the operand rtxes inside it) around
644 the call to get_insn_template.
646 2017-03-03 Martin Sebor <msebor@redhat.com>
648 PR tree-optimization/79699
649 * context.c (context::~context): Free MPFR caches to avoid
650 a memory leak on program exit.
652 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
654 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
655 Use wide_int::ulow () instead of .elt (0).
657 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
659 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
660 (*pushxf): Limit oF constraint to 32bit targets and add oC
661 constraint for 64bit targets.
662 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
663 (*pushdf): Change rmF constraint to rmC.
665 2017-03-03 Martin Liska <mliska@suse.cz>
667 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
668 Remove unused variable.
670 2017-03-03 Jakub Jelinek <jakub@redhat.com>
673 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
674 is a memory operand, increase num_memory.
675 (ix86_expand_args_builtin): Likewise.
677 2017-03-03 Jan Hubicka <jh@suse.cz>
680 * ipa-devirt.c (maybe_record_node): Properly handle
681 __cxa_pure_virtual visibility.
683 2017-03-03 Martin Liska <mliska@suse.cz>
685 PR tree-optimization/79803
686 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
688 (pass_loop_prefetch::execute): Disabled optimization if an
689 assumption about L1 cache size is not met.
691 2017-03-03 Martin Liska <mliska@suse.cz>
693 PR rtl-optimization/79574
694 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
695 (hash_scan_set): Likewise.
696 (dump_hash_table): Likewise.
697 (hoist_code): Likewise.
699 2017-03-03 Richard Biener <rguenther@suse.de>
701 * fixed-value.c (fixed_from_string): Restore use of elt (1)
702 in place of uhigh ().
703 (fixed_convert_from_real): Likewise.
705 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
708 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
710 2017-03-03 Richard Biener <rguenther@suse.de>
713 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
714 TYPE_OVERFLOW_UNDEFINED check.
716 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
718 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
720 (vector_ae_<mode>_p): Likewise.
721 (vector_nez_<mode>_p): Likewise.
722 (vector_ne_v2di_p): Likewise.
723 (vector_ae_v2di_p): Likewise.
724 (vector_ne_<mode>_p): Likewise.
725 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
727 (vsx_tsqrt<mode>2_fe): Likewise.
729 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
732 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
734 2017-03-02 Jakub Jelinek <jakub@redhat.com>
736 PR rtl-optimization/79780
737 * cprop.c (one_cprop_pass): When second and further conditional trap
738 in a single basic block is turned into an unconditional trap, turn it
739 into a deleted note to avoid RTL verification failures.
741 2017-03-02 Richard Biener <rguenther@suse.de>
743 * fold-const.c (const_binop): Use ulow () instead of elt (0).
745 2017-03-02 Richard Biener <rguenther@suse.de>
747 PR tree-optimization/79345
749 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
750 param and abort the walk, returning -1 if it is hit.
751 (walk_aliased_vdefs): Take a limit param and pass it on.
752 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
753 defaulting to 0 and return a signed int.
754 * tree-ssa-uninit.c (struct check_defs_data): New struct.
755 (check_defs): New helper.
756 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
757 about uninitialized memory.
758 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
759 bogus uninitialized warning.
760 (fixed_convert_from_real): Likewise.
762 2017-03-02 Bin Cheng <bin.cheng@arm.com>
764 PR tree-optimization/66768
765 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
766 iv_use if base object can't be determined.
768 2017-03-02 Jakub Jelinek <jakub@redhat.com>
770 PR tree-optimization/79345
771 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
772 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
773 (get_pattern_stats): Initialize it.
774 * genemit.c (gen_expand): Verify match_scratch numbers come after
775 match_operand/match_dup numbers.
776 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
777 match_scratch numbers.
778 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
780 * config/s390/s390.md (trunctdsd2): Likewise.
782 2017-03-02 Richard Biener <rguenther@suse.de>
784 * wide-int.h (wide_int_storage::operator=): Implement in terms
787 2017-03-02 Richard Biener <rguenther@suse.de>
789 PR tree-optimization/79777
790 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
791 the to insert expression to sth existing.
793 2017-03-01 Martin Sebor <msebor@redhat.com>
796 * gimple-ssa-sprintf.c
797 (directive::known_width_and_precision): New function.
798 (format_integer): Use it.
799 (get_mpfr_format_length): Consider the full range of precision
800 when computing %g output with the # flag. Set the likely byte
801 count to 3 rather than 1 when precision is indeterminate.
802 (format_floating): Correct the lower bound of precision.
804 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
806 * doc/invoke.texi: Document default code model for 64-bit Linux.
808 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
811 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
812 udiv rather than div since input pattern is unsigned.
814 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
816 * config/i386/i386.c (print_reg): Warn for values of
817 unsupported size in integer register.
819 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
822 * config/rs6000/predicates.md (current_file_function_operand): Do
823 not allow self calls to be local if the function is replaceable.
825 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
828 * config/rs6000/altivec.h (vec_ctz and others): Change the
829 preprocessor macro that controls conditional compilation from
830 _ARCH_PWR9 to __POWER9_VECTOR__.
831 (vec_all_ne): Change parameterization of __altivec_scalar_pred
832 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
833 control (instead of _ARCH_PWR9 control) so that template
834 definition uses power9-specific function.
835 (vec_any_eq): Likewise.
836 (vec_all_ne): Change macro definition to use a power9-specific
837 expansion under #ifdef __POWER9_VECTOR__ control (instead of
839 (vec_any_eq) Likewise.
840 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
841 expansion for CMPNEF to remove support for xvcmpnesp instruction.
842 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
843 support for xvcmpnedp instruction.
844 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
845 macro expansion so that Power9 implementation of vec_all_ne does
846 not use the AltiVec predicate framework.
847 (VCMPNEH_P): Likewise.
848 (VCMPNEW_P): Likewise.
849 (VCMPNED_P): Likewise.
850 (VCMPNEFP_P): Likewise.
851 (VCMPNEDP_P): Likewise.
852 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
853 implementation of vec_any_eq to not use AltiVec predicate
855 (VCMPAEH_P): Likewise.
856 (VCMPAEW_P): Likewise.
857 (VCMPAED_P): Likewise.
858 (VCMPAEFP_P): Likewise.
859 (VCMPAEDP_P): Likewise.
860 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
861 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
862 not use the AltiVec predicate framework.
863 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
864 of vec_any_eq to not use AltiVec predicate framework.
865 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
866 support for predefined __POWER9_VECTOR__ macro to indicate that
867 Power9 instruction selection is enabled.
868 (altivec_overloaded_builtins): Remove extraneous
869 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
870 function argument types RS6000_BTI_bool_V16QI and
871 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
872 entry for overloaded function argument types RS6000_BTI_bool_V4SI
873 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
874 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
875 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
876 Power9 for implementations of vec_cmpne. Change the signature for
877 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
878 (representing vec_all_ne) to remove the previously described first
879 argument of type RS6000_BTI_INTSI, as this was an artifact of
880 reliance on the AltiVec predicate framework, which is no longer
881 used in the implementation of these functions. Add
882 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
883 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
884 since, unlike the AltiVec predicate framework implementation, we
885 do not share function descriptors between vec_alle and vec_anyeq.
886 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
887 set of modes that receive special treatment even when
888 TARGET_P9_VECTOR is true. The special treatment emits code that
889 does not depend on Power9 instructions.
890 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
891 define_expand to not rely on AltiVec predicate framework.
892 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
894 (vector_ne_v2di_p): Change this define_expand to not rely on
895 AltiVec predicate framework.
896 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
898 (vector_ne_<mode>_p): Change this define_expand to not rely on
899 AltiVec predicate framework.
900 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
902 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
903 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
905 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
906 define_insn pattern because the xvcmpne<VSs>. instruction is not
908 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
909 instruction is not supported.
911 2017-03-01 Jakub Jelinek <jakub@redhat.com>
913 * config/nvptx/nvptx.c: Include intl.h.
915 2017-03-01 Martin Jambor <mjambor@suse.cz>
918 * ipa-prop.h (ipa_bits): Removed field known.
919 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
920 to pointers. Adjusted their comments to warn about their sharing.
921 (ipcp_transformation_summary): Change bits to a vector of pointers.
922 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
923 (ipa_get_ipa_bits_for_value): Declare.
924 * tree-vrp.h (value_range): Mark as GTY((for_user)).
925 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
926 (ipa_bits_hash_table): Likewise.
927 (ipa_vr_ggc_hash_traits): Likewise.
928 (ipa_vr_hash_table): Likewise.
929 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
930 being pointers and vr_known being removed.
931 (ipa_set_jf_unknown): Likewise.
932 (ipa_get_ipa_bits_for_value): New function.
933 (ipa_set_jfunc_bits): Likewise.
934 (ipa_get_value_range): New overloaded functions.
935 (ipa_set_jfunc_vr): Likewise.
936 (ipa_compute_jump_functions_for_edge): Use the above functions to
937 construct bits and vr parts of jump functions.
938 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
939 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
941 (ipcp_grow_transformations_if_necessary): Also allocate
942 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
944 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
945 them. Fix too long lines.
946 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
947 vr_known being removed.
948 (ipa_read_jump_function): Use new setter functions to construct bits
949 and vr parts of jump functions or set them to NULL.
950 (write_ipcp_transformation_info): Adjust for bits being pointers.
951 (read_ipcp_transformation_info): Likewise.
952 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
954 Include gt-ipa-prop.h.
955 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
957 (ipcp_store_bits_results): Likewise.
958 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
959 Do not write to existing jump functions but use a temporary instead.
961 2017-03-01 Jakub Jelinek <jakub@redhat.com>
964 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
965 attempt to use its first operand as BIT_FIELD_REF base.
967 2017-03-01 Richard Biener <rguenther@suse.de>
970 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
971 interpolating formula in wrapping arithmetic.
972 (chrec_apply): Convert chrec_evaluate return value to wanted type.
974 2017-03-01 Jakub Jelinek <jakub@redhat.com>
976 PR tree-optimization/79734
977 * tree-vect-generic.c (expand_vector_condition): Optimize
978 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
979 Handle VEC_COND_EXPR where comparison has different inner width from
982 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
984 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
985 markup, and similar issues. Remove @opindex entries for things
986 that aren't options. Add missing -mmpy-option entries.
988 2017-02-28 Jakub Jelinek <jakub@redhat.com>
990 PR tree-optimization/79737
991 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
992 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
993 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
994 instead of byte_size. Formatting fix.
995 (shift_bytes_in_array_right): Formatting fix.
997 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
1000 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
1001 condition on optimize for the leaf function test.
1003 2017-02-28 Martin Liska <mliska@suse.cz>
1006 * read-rtl-function.c (function_reader::handle_unknown_directive):
1007 Bail out when one uses -flto.
1009 2017-02-28 Martin Liska <mliska@suse.cz>
1011 * common.opt: Replace space with tabular for options of <number>
1013 * config/i386/i386.opt: Show <number> value for
1014 -mlarge-data-threshold.
1015 * opts.c (print_filtered_help): Do not display number in hexadecimal
1018 2017-02-28 Martin Liska <mliska@suse.cz>
1020 * common.opt: Fix --help=option -Q for options which are of
1023 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
1025 * config/i386/i386.c (print_reg): Error out for values
1026 of 8-bit size in invalid integer register.
1028 2017-02-28 Martin Sebor <msebor@redhat.com>
1030 PR tree-optimization/79691
1031 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
1033 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1036 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
1037 gcc_unreachable with output_operand_lossage.
1039 2017-02-28 Richard Biener <rguenther@suse.de>
1041 PR tree-optimization/79740
1042 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
1044 (visit_nary_op): Insert the nary into the hashtable if we
1045 pattern-matched sth.
1046 * tree-ssa-pre.c (eliminate_insert): Robustify.
1048 2017-02-28 Richard Biener <rguenther@suse.de>
1051 * fold-const.c (decode_field_reference): Reject out-of-bound
1054 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1056 * config/i386/i386.c: Include intl.h.
1057 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
1058 instead of just cond ? "..." : "...".
1059 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
1060 * coverage.c (read_counts_file): Likewise.
1061 * omp-offload.c: Include intl.h.
1062 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
1063 of just cond ? "..." : "...".
1064 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
1065 of just cond ? "..." : "...".
1067 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
1070 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
1072 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
1073 'tune for' CPU name.
1074 * config/arm/arm-cpu-data.h: Regenerated.
1076 2017-02-28 Richard Biener <rguenther@suse.de>
1078 PR tree-optimization/79732
1079 * tree-inline.c (expand_call_inline): Do not shadow var.
1081 2017-02-28 Richard Biener <rguenther@suse.de>
1083 PR tree-optimization/79723
1084 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
1085 address-space properly.
1087 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
1089 * doc/optinfo.texi (Optimization groups): Fix option used for
1091 * doc/invoke.texi (-fopt-info): Document "omp".
1092 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
1093 (OPTGROUP_ALL): Add OPTGROUP_OMP.
1094 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
1095 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
1096 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
1098 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
1100 * dumpfile.c (optgroup_options): Instead of "openmp", associate
1101 OPTGROUP_OMP with "omp".
1103 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
1106 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
1107 for arithmetic shift of unsigned V2DI.
1109 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
1111 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
1112 arc/linux.h headers.
1113 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
1114 (LINK_SPEC): Likewise.
1115 (ARC_TLS_EXTRA_START_SPEC): Likewise.
1116 (EXTRA_SPECS): Likewise.
1117 (STARTFILE_SPEC): Likewise.
1118 (ENDFILE_SPEC): Likewise.
1119 (LIB_SPEC): Likewise.
1120 (TARGET_SDATA_DEFAULT): Likewise.
1121 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
1122 (MULTILIB_DEFAULTS): Likewise.
1123 (DWARF2_UNWIND_INFO): Likewise.
1124 * config/arc/big.h: New file.
1125 * config/arc/elf.h: Likewise.
1126 * config/arc/linux.h: Likewise.
1127 * config/arc/t-uClibc: Remove.
1129 2017-02-27 Bin Cheng <bin.cheng@arm.com>
1131 PR tree-optimization/77536
1132 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
1133 (tree_transform_and_unroll_loop): Use above function to compute the
1134 estimated niter of unrolled loop and use it when scaling profile.
1135 Also use count info rather than frequency if it's non-zero.
1136 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
1137 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
1138 (vect_transform_loop): Call above function.
1140 2017-02-27 Richard Biener <rguenther@suse.de>
1142 PR tree-optimization/45397
1143 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
1144 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
1145 (visit_nary_op): Add pattern matching for CSEing sign-changed
1146 or truncated operations with wider ones.
1148 2017-02-27 Richard Biener <rguenther@suse.de>
1150 PR tree-optimization/79690
1151 * tree-vect-stmts.c (vectorizable_store): Use vector type
1152 built from the DR with address-space.
1154 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
1156 * doc/invoke.texi (Optimize Options): Refine the description
1157 of asan-use-after-return.
1159 2017-02-25 Alan Modra <amodra@gmail.com>
1161 PR rtl-optimization/79584
1162 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
1163 base, not ad->base_term, the reg within base. Remove assertion
1164 that ad->base == ad->base_term. Replace gen_int_mode using
1165 bogus mode with const0_rtx.
1167 2017-02-25 Jakub Jelinek <jakub@redhat.com>
1170 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
1171 FMA_EXPR like tcc_binary or tcc_unary.
1173 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
1176 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
1178 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
1179 (output_loc_operands): Handle DW_OP_call_ref and
1180 DW_OP_GNU_variable_value.
1181 (struct variable_value_struct): New type.
1182 (struct variable_value_hasher): Likewise.
1183 (variable_value_hash): New variable.
1184 (string_types): Remove.
1185 (copy_loc_descr): New function.
1186 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
1187 (prepend_loc_descr_to_each): New function.
1188 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
1189 instead of add_loc_descr_to_each if the first argument is single
1190 location list and the second has multiple.
1191 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
1192 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
1193 when looking for variable value which doesn't have other location info.
1194 (loc_list_from_tree): Formatting fix.
1195 (gen_array_type_die): Simplify DW_AT_string_length handling.
1196 (adjust_string_types): Remove.
1197 (gen_subprogram_die): Don't call adjust_string_types nor test/set
1198 string_types. Call resolve_variable_values.
1199 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
1200 (resolve_addr_in_expr): Likewise. Add A argument.
1201 (copy_deref_exprloc): Remove deref argument. Adjust for the
1202 original expression being DW_OP_GNU_variable_value with optionally
1203 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
1204 optionally after it.
1205 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
1206 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
1207 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
1208 (variable_value_hasher::hash, variable_value_hasher::equal): New
1210 (resolve_variable_value_in_expr, resolve_variable_value,
1211 resolve_variable_values, note_variable_value_in_expr,
1212 note_variable_value): New functions.
1213 (dwarf2out_early_finish): Call note_variable_value on all toplevel
1216 2017-02-24 Jakub Jelinek <jakub@redhat.com>
1219 * opts.h (handle_generated_option): Add GENERATED_P argument.
1220 * opts-common.c (handle_option): Adjust function comment.
1221 (handle_generated_option): Add GENERATED_P argument, pass it to
1223 (control_warning_option): Pass false to handle_generated_option
1225 * opts.c (maybe_default_option): Pass true to handle_generated_option
1227 * optc-gen.awk: Likewise.
1229 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1231 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
1232 a REG, look at the REG it is a SUBREG of.
1233 (splitter for cmpeqsi_t): Ditto.
1235 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1237 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
1238 the special USEs with the pattern of the insn, not the insn itself.
1240 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
1243 * doc/invoke.texi: Document -mload-store-pairs.
1245 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1246 Sandra Loosemore <sandra@codesourcery.com>
1248 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
1249 argument isn't a CONST_INT.
1250 (nios2_alternate_compare_const): Assert op is a CONST_INT.
1251 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
1252 (nios2_validate_compare): Bypass alternate compare logic if *op2
1254 (ldstwm_operation_p): Return false if first_base is not a REG or
1255 if first_offset is not a CONST_INT.
1257 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1259 * config/cris/cris.md: Use correct operand in a define_peephole2.
1261 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1263 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
1265 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1267 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
1268 this_insn if it is an INSN or JUMP_INSN.
1269 (force_offsettable): Look at base, not at addr.
1270 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
1271 on things that aren't necessarily CONST_INTs.
1273 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
1275 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
1276 -mfpmath=sse is the default also for x86-32 targets with SSE2
1277 instruction set when @option{-ffast-math} is enabled
1279 2017-02-24 Jeff Law <law@redhat.com>
1281 PR rtl-optimizatoin/79286
1282 * ira.c (update_equiv_regs): Drop may_trap_p exception to
1285 2017-02-24 Richard Biener <rguenther@suse.de>
1287 PR tree-optimization/79389
1288 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
1291 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
1293 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
1294 function comment to reflect reality.
1295 (loop_exits_before_overflow): Fix typo in function description.
1297 2017-02-24 Richard Biener <rguenther@suse.de>
1299 PR tree-optimization/79389
1300 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
1301 properly that a threading opportunity exists. Detect conditional
1302 copy/constant propagation opportunities.
1304 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
1306 * config/visium/visium.md (type): Add trap.
1307 (b): New mode attribute.
1308 (*btst): Rename into...
1309 (*btst<mode>): ...this and adjust.
1310 (*cbranchsi4_btst_insn): Rename into...
1311 (*cbranch<mode>4_btst_insn): ...this and adjust.
1312 (trap): New define_insn.
1314 2017-02-23 Jakub Jelinek <jakub@redhat.com>
1316 PR tree-optimization/79389
1317 * ifcvt.c (struct noce_if_info): Add rev_cond field.
1318 (noce_reversed_cond_code): New function.
1319 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
1320 reversed_comparison_code. Formatting fix.
1321 (noce_try_store_flag): Test rev_cond != NULL in addition to
1322 reversed_comparison_code.
1323 (noce_try_store_flag_constants): Likewise.
1324 (noce_try_store_flag_mask): Likewise.
1325 (noce_try_addcc): Use rev_cond if non-NULL instead of
1326 reversed_comparison_code.
1327 (noce_try_cmove_arith): Likewise. Formatting fixes.
1328 (noce_try_minmax, noce_try_abs): Clear rev_cond.
1329 (noce_find_if_block): Initialize rev_cond.
1330 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
1331 instead of false as last argument never attempt to reverse it
1334 2017-02-23 Bin Cheng <bin.cheng@arm.com>
1336 PR tree-optimization/79663
1337 * tree-predcom.c (combine_chains): Process refs in reverse order
1338 only for ZERO length chains, and add explaining comment.
1340 2017-02-23 Jeff Law <law@redhat.com>
1342 PR tree-optimization/79578
1343 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
1344 in call to operand_equal_p.
1346 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
1349 * config/i386/cpuid.h: Fix another undefined behavior.
1351 2017-02-23 Richard Biener <rguenther@suse.de>
1353 PR tree-optimization/79683
1354 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
1355 vector types for data-refs.
1357 2017-02-23 Martin Liska <mliska@suse.cz>
1359 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
1361 2017-02-23 Jakub Jelinek <jakub@redhat.com>
1364 * internal-fn.c (get_range_pos_neg): Moved to ...
1365 * tree.c (get_range_pos_neg): ... here. No longer static.
1366 * tree.h (get_range_pos_neg): New prototype.
1367 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
1368 are known to be in between 0 and signed maximum inclusive, try to
1369 expand both unsigned and signed divmod and use the cheaper one from
1372 2017-02-22 Jeff Law <law@redhat.com>
1374 PR tree-optimization/79578
1375 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
1376 to compare base operands.
1378 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
1381 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
1382 gpc_reg_operand instead of fpr_reg_operand.
1384 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
1386 * config/mips/mips.c (mips_return_in_memory): Force FP
1387 vector types to be returned in memory for o32 ABI.
1389 2017-02-22 Jakub Jelinek <jakub@redhat.com>
1391 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
1392 instead of DW_TAG_member for static data member declarations and don't
1393 set no_linkage_name for static inline data members.
1394 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
1397 2017-02-22 Martin Liska <mliska@suse.cz>
1399 * doc/invoke.texi: Replace inequality signs with square brackets
1402 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1404 PR tree-optimization/68644
1405 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
1407 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
1410 * lra-constraints.c (simplify_operand_subreg): Handle
1411 WORD_REGISTER_OPERATIONS targets.
1413 2017-02-22 Jakub Jelinek <jakub@redhat.com>
1416 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
1417 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
1418 elimination by swapping fld*.
1420 2017-02-22 Richard Biener <rguenther@suse.de>
1422 PR tree-optimization/79673
1423 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
1424 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
1425 irrelevant address-space qualifiers and avoiding a
1426 ADDR_SPACE_CONVERT_EXPR from fold_convert.
1428 2017-02-22 Richard Biener <rguenther@suse.de>
1430 PR tree-optimization/79666
1431 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
1432 to not symbolically negate if that may introduce undefined
1435 2017-02-22 Martin Liska <mliska@suse.cz>
1438 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
1439 * data-streamer-out.c (streamer_write_gcov_count_stream):
1441 * value-prof.c (stream_out_histogram_value): Make assert more
1442 precise based on type of counter.
1444 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
1447 * config/i386/i386.md (standard_x87sse_constant_load splitter):
1448 Use nonimmediate_operand instead of memory_operand for operand 1.
1449 (float-extend standard_x87sse_constant_load splitter): Ditto.
1451 2017-02-21 Jeff Law <law@redhat.com>
1453 PR tree-optimization/79621
1454 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
1455 blocks with edges to themselves.
1457 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1460 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
1461 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
1462 Use gimple_call_builtin_p.
1465 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
1466 on temporarily removed DEBUG_INSNs.
1468 PR tree-optimization/79649
1469 * tree-loop-distribution.c (classify_partition): Give up on
1470 non-generic address space loads/stores.
1472 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
1474 * doc/loop.texi (Loop manipulation): Remove nonexistent
1475 tree_ssa_loop_version from the documentation.
1476 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
1478 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1481 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
1482 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
1483 * config/rs6000/rs6000.c: Include except.h.
1484 (rs6000_expand_split_stack_prologue): Call
1485 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
1487 2017-02-21 Martin Jambor <mjambor@suse.cz>
1490 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
1493 2017-02-21 Martin Jambor <mjambor@suse.cz>
1495 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
1496 for backward compatibility only.
1497 * doc/invoke.texi (Option Summary): Remove all references to
1500 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
1504 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1506 * lra-constraints.c (curr_insn_transform): Handle
1507 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
1509 2017-02-21 Martin Liska <mliska@suse.cz>
1511 * config/i386/i386.opt: Replace -masm-dialect with -masm.
1513 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
1515 PR translation/79638
1516 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
1518 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
1521 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
1522 (arm_function_ok_for_sibcall): Return false for an indirect call by
1523 descriptor if all the argument registers are used.
1524 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
1525 alignment of the function.
1527 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1529 PR tree-optimization/61441
1530 * simplify-rtx.c (simplify_const_unary_operation): For
1531 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
1532 the sNaN unmodified.
1534 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
1536 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
1537 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
1538 instead of SYSTEM_HEADER_DIR.
1540 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
1541 Martin Liška <mliska@suse.cz>
1543 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
1544 Fix typos and grammar, use active voice, and clarify.
1546 2017-02-20 Marek Polacek <polacek@redhat.com>
1549 * gimplify.c (gimplify_expr): Handle unused *&&L;.
1552 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
1554 2017-02-20 Jakub Jelinek <jakub@redhat.com>
1557 * config/i386/i386.c (ix86_expand_builtin): Handle
1558 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
1559 ix86_builtins_isa[fcode].isa as a requirement of those
1560 flags and any other flag in the bitmask.
1561 (ix86_init_mmx_sse_builtins): Use 0 instead of
1562 ~OPTION_MASK_ISA_64BIT as mask.
1563 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
1564 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
1565 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
1566 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
1568 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1571 * lra-constraints.c (split_reg): Check requested split mode
1572 is supported by the register.
1574 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1576 * lra-constraints.c (simplify_operand_subreg): Remove early
1579 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1582 * lra-constraints.c (curr_insn_transform): Tighten condition
1583 for converting SUBREG reloads from OP_OUT to OP_INOUT.
1585 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1588 * lra-constraints.c (curr_insn_transform): Handle
1589 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
1591 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
1594 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
1596 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
1598 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
1601 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
1604 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1606 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
1609 2017-02-18 Jakub Jelinek <jakub@redhat.com>
1611 * final.c (last_columnnum, override_columnnum): New variables.
1612 (final_start_function): Set last_columnnum, pass it to begin_prologue
1613 hook and pass 0 to dwarf2out_begin_prologue.
1614 (final_scan_insn): Update override_columnnum. Pass last_columnnum
1615 to source_line debug hook.
1616 (notice_source_line): Compute last_columnnum and for debug_column_info
1617 return true on column changes.
1618 * debug.h (struct gcc_debug_hooks): Add column argument to
1619 source_line and begin_prologue hooks.
1620 (debug_nothing_int_charstar_int_bool): Remove prototype.
1621 (debug_nothing_int_int_charstar,
1622 debug_nothing_int_int_charstar_int_bool): New prototypes.
1623 (dwarf2out_begin_prologue): Add column argument.
1624 * debug.c (do_nothing_debug_hooks): Adjust source_line and
1625 begin_prologue hooks.
1626 (debug_nothing_int_charstar_int_bool): Remove.
1627 (debug_nothing_int_int_charstar,
1628 debug_nothing_int_int_charstar_int_bool): New functions.
1629 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
1630 through to dwarf2out_source_line.
1631 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
1632 (dwarf2out_source_line): Add column argument, emit it if requested.
1633 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
1635 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
1636 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
1637 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
1638 through to dwarf2out_begin_prologue.
1639 (vmsdbgout_source_line): Add column argument, pass it through to
1640 dwarf2out_source_line.
1641 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
1642 dbxout_source_line caller.
1643 (dbxout_source_line): Add column argument.
1645 * common.opt (gno-column-info, gcolumn-info): New options.
1646 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
1647 (check_die): Also test for multiple DW_AT_decl_column attributes.
1648 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
1649 DW_AT_decl_column if requested.
1650 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
1652 (gen_variable_die): Likewise.
1653 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
1654 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
1657 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
1658 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
1659 (ix86_handle_option): Handle OPT_m3dnowa.
1660 * doc/invoke.texi (-m3dnowa): Document.
1661 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
1662 -m3dnowa instead of -m3dnow -march=athlon.
1665 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
1666 instead of gcc_assert for K, r and R code checks. Formatting fixes.
1668 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1671 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
1672 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
1673 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
1674 generator for vsx_xxpermdi_<mode>_be.
1675 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
1676 force big-endian semantics.
1677 (vsx_xxpermdi_<mode>_be): New define_expand with same
1678 implementation as previous version of vsx_xxpermdi_<mode>.
1680 2017-02-17 Jakub Jelinek <jakub@redhat.com>
1682 PR tree-optimization/79327
1683 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
1684 variable, its initialization and use.
1686 2017-02-17 Julia Koval <julia.koval@intel.com>
1688 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
1689 (OPTION_MASK_ISA_PKU_UNSET): New.
1690 (ix86_handle_option): Handle -mrdpid.
1691 * config/i386/cpuid.h (bit_RDPID): New.
1692 * config/i386/driver-i386.c (host_detect_local_cpu):
1693 Detect RDPID feature.
1694 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
1695 * config/i386/i386-c.c (ix86_target_macros_internal):
1697 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
1698 (ix86_valid_target_attribute_inner_p): Add "rdpid".
1699 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
1700 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
1701 * config/i386/i386.md (define_insn "rdpid"): New.
1702 * config/i386/i386.opt Add -mrdpid.
1703 * config/i386/immintrin.h (_rdpid_u32): New.
1705 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
1707 PR rtl-optimization/79541
1708 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
1709 instead of transforming it into USE.
1711 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
1713 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
1714 If HONOR_SNANS (SFmode) force the input to a register.
1715 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
1716 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
1717 an frsp or similar insn.
1719 2017-02-17 Martin Liska <mliska@suse.cz>
1721 PR rtl-optimization/79577
1722 * params.def (selsched-max-sched-times): Increase minimum to 1.
1724 2017-02-17 Martin Liska <mliska@suse.cz>
1726 PR rtl-optimization/79574
1727 * gcse.c (want_to_gcse_p): Prevent integer overflow.
1729 2017-02-17 Martin Liska <mliska@suse.cz>
1731 PR tree-optimization/79529
1732 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
1733 ssa_defined_default_def_p to handle cases which are implicitly
1735 * tree-ssa.c (ssa_defined_default_def_p): New function.
1736 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
1737 which are implicitly defined.
1738 * tree-ssa.h (ssa_defined_default_def_p): Declare.
1740 2017-02-17 Richard Biener <rguenther@suse.de>
1743 * params.def (max-ssa-name-query-depth): Limit to 10.
1745 2017-02-17 Richard Biener <rguenther@suse.de>
1747 PR tree-optimization/79552
1748 * tree-ssa-structalias.c (visit_loadstore): Properly verify
1751 2017-02-17 Richard Biener <rguenther@suse.de>
1754 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
1756 2017-02-17 Marek Polacek <polacek@redhat.com>
1759 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
1760 (fold_negate_expr): New wrapper.
1762 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
1764 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
1765 Correct terminology and de-emphasize pre-standard behavior.
1767 2017-02-16 Alan Modra <amodra@gmail.com>
1769 PR rtl-optimization/79286
1770 * ira.c (def_dominates_uses): New function.
1771 (update_equiv_regs): Don't create an equivalence for insns that
1772 may trap where the register def does not dominate the use.
1774 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
1776 PR rtl-optimization/78127
1777 * lra.c (lra): Call lra_eliminate before finish the loop after
1780 2017-02-16 Richard Biener <rguenther@suse.de>
1782 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
1784 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
1785 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
1786 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
1787 (isl_val_int_from_wi): New function.
1788 (extract_affine_gmp): Rename to ...
1789 (extract_affine_wi): ... this, take a widest_int.
1790 (extract_affine_int): Just wrap extract_affine_wi.
1791 (add_param_constraints): Use isl_val_int_from_wi.
1792 (add_loop_constraints): Likewise, and extract_affine_wi.
1794 2017-02-15 Jeff Law <law@redhat.com>
1797 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
1798 ira_init_register_move_cost_if_necessary.
1800 2017-02-15 Martin Sebor <msebor@redhat.com>
1803 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
1804 removed in a prior commit.
1806 2017-02-15 Bin Cheng <bin.cheng@arm.com>
1808 PR tree-optimization/79347
1809 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
1810 counters during peeling.
1812 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
1814 * Makefile.in (site.exp): Remove "set ISLVER".
1816 2017-02-15 Jakub Jelinek <jakub@redhat.com>
1819 * real.c (real_from_integer): Call real_convert even for decimal.
1821 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
1824 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
1826 2017-02-14 Andrew Pinski <apinski@cavium.com>
1828 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
1829 cores and change the partno/implementer to be correct.
1830 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
1831 the 'B" as the implementer.
1832 * config/aarch64/aarch64-tune.md: Regenerate.
1834 2017-02-14 Carl Love <cel@us.ibm.com>
1836 * config/rs6000/rs6000.c: Add case statement entry to make the
1837 xvcvuxdsp built-in argument unsigned.
1838 * config/rs6000/vsx.md: Fix the source and return operand types so they
1839 match the instruction definitions from the ISA document. Fix typo
1840 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
1843 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
1846 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
1847 member early_clobber_alts.
1848 * lra-lives.c (reg_early_clobber_p): New.
1849 (process_bb_lives): Use it.
1850 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
1851 (debug_operand_data): Initialize early_clobber_alts.
1852 (setup_operand_alternative): Set up early_clobber_alts.
1853 (collect_non_operand_hard_regs): Ditto. Pass early clobber
1854 alternatives to new_insn_reg.
1855 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
1857 (lra_update_insn_regno_info): Pass the new arg.
1859 2017-02-14 Jakub Jelinek <jakub@redhat.com>
1862 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
1863 (new_oacc_loop_raw): Don't clear already cleared fields.
1866 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
1867 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
1868 _mm512_prefetch_i64gather_ps): New inline functions and macros.
1870 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
1873 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
1875 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
1878 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
1879 the extra instruction to the right place to store 128-bit constant
1882 2017-02-14 Martin Sebor <msebor@redhat.com>
1885 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
1886 warning for strings of unknown length.
1888 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
1890 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
1892 2017-02-14 Jeff Law <law@redhat.com>
1895 * ira-costs.c (scan_one_insn): Initialize register move costs
1896 for pseudos seen in USE/CLOBBER insns.
1898 PR tree-optimization/79095
1899 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
1900 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
1901 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
1902 if the operands are known to be not equal, then the resulting range
1904 (intersect_ranges): If the new range is ~[0,0] and the old range is
1905 wide, then prefer ~[0,0].
1906 * tree-vrp.c (overflow_comparison_p_1): New function.
1907 (overflow_comparison_p): New function.
1908 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
1909 if NAME is used in an overflow test.
1910 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
1911 overflow check that can be expressed as an equality test, then adjust
1912 ops to be that equality test.
1914 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1916 * config/s390/s390-builtin-types.def: Remove flags argument.
1917 * config/s390/s390.c (s390_init_builtins): Likewise.
1919 2017-02-14 Martin Liska <mliska@suse.cz>
1921 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
1922 vector. Fix trailing white spaces.
1924 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
1926 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
1929 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1931 PR rtl-optimization/68664
1932 * config/arm/arm.c (arm_sched_can_speculate_insn):
1933 New function. Declare prototype.
1934 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
1936 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1938 PR rtl-optimization/68664
1939 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
1941 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
1943 2017-02-14 Amit Pawar <amit.pawar@amd.com>
1945 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
1946 max skip bytes for function, loop and jump.
1948 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1950 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
1951 ABS_EXPR for gimple dump.
1953 2017-02-14 Jakub Jelinek <jakub@redhat.com>
1956 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
1958 PR tree-optimization/79408
1959 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
1960 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
1961 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
1962 also if rhs1 is INTEGER_CST.
1964 2017-02-14 Richard Biener <rguenther@suse.de>
1967 * tree-into-ssa.c (insert_phi_nodes): When the function can
1968 have abnormal edges rewrite SSA names with broken use-def
1969 dominance out of SSA and register them for PHI insertion.
1971 2017-02-13 Martin Sebor <msebor@redhat.com>
1974 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
1975 clearing info.nowrite flag when snprintf size argument is a range.
1977 2017-02-13 Jakub Jelinek <jakub@redhat.com>
1979 * cprop.c (cprop_jump): Add missing space in string literal.
1980 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
1981 (get_constraint_for_component_ref): Likewise.
1982 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
1983 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
1984 * lra-constraints.c (process_alt_operands): Likewise.
1985 * ipa-inline.c (inline_small_functions): Likewise.
1986 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
1987 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
1988 * trans-mem.c (diagnose_tm_1_op): Likewise.
1989 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
1990 (grid_parallel_clauses_gridifiable): Likewise.
1992 * config/nvptx/mkoffload.c (process): Add space in between
1995 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
1996 "MOD4_SSE_REGS" and "ALL_REGS".
1998 * spellcheck.c (test_data): Add , in between "foo" and "food".
2000 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2003 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
2004 boundary crossing check and subsequent code generation agree.
2006 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2008 * config/aarch64/aarch64.c (has_memory_op): Delete.
2009 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
2012 2017-02-13 Jakub Jelinek <jakub@redhat.com>
2014 PR rtl-optimization/79388
2015 PR rtl-optimization/79450
2016 * combine.c (distribute_notes): When removing TEM_INSN for which
2017 corresponding dest has last value recorded, invalidate that last
2020 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2022 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
2023 of explicit '@'. Add missing assembly comment marker on branch costs
2026 2017-02-13 Nathan Sidwell <nathan@acm.org>
2028 * gengtype-lex.l (<in_struct>): Add '/'.
2030 2017-02-13 Martin Liska <mliska@suse.cz>
2033 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
2035 2017-02-13 Richard Biener <rguenther@suse.de>
2037 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
2039 * configure: Re-generate.
2040 * config.in: Likewise.
2041 * graphite-dependences.c: Simplify as if
2042 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
2043 * graphite-isl-ast-to-gimple.c: Likewise.
2044 * graphite-optimize-isl.c: Likewise.
2045 * graphite-poly.c: Likewise.
2046 * graphite-sese-to-poly.c: Likewise.
2047 * graphite.h: Likewise.
2048 * toplev.c: Include isl/version.h and use isl_version () for
2049 printing the ISL version.
2050 * doc/install.texi: Update ISL requirement.
2052 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
2054 * doc/standards.texi (Standards): Update reference to
2057 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
2059 * doc/extend.texi (Named Address Spaces): sourceware.org now
2061 * doc/install.texi (Binaries): Ditto.
2064 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
2066 * doc/cpp.texi: Replace "stringify"/"stringification" with C
2067 standard terminology "stringize"/"stringizing" throughout.
2068 * doc/cppinternals.texi: Likewise.
2070 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
2072 * doc/extend.texi: Fix some spelling mistakes and typos.
2073 * doc/invoke.texi: Likewise.
2075 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
2078 * params.def (inline-min-speedup) Change from 10 to 8.
2080 2017-02-11 Jakub Jelinek <jakub@redhat.com>
2082 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
2085 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
2088 * ipa-inline-analysis.c (get_minimal_bb): New function.
2089 (record_modified): Use it.
2090 (remap_edge_change_prob): Handle also ancestor functions.
2092 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
2094 * doc/contrib.texi (Contributors): Remove broken link into
2095 the Mauve CVS repository.
2097 2017-02-11 Jakub Jelinek <jakub@redhat.com>
2100 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
2101 result computation whenever lhs doesn't have vector mode, not
2102 just when it has BLKmode.
2104 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
2106 * doc/makefile.texi (profiledbootstrap): Refer to the
2107 installation instructions only in textual form.
2109 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2112 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
2114 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
2116 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
2117 (Specific): Update mingw-w64 reference.
2119 (Specific): Remove broken link to Renesas RX processor.
2121 2017-02-10 Richard Biener <rguenther@suse.de>
2123 * toplev.c (process_options): Do not mention obsolete graphite
2124 options when printing sorry message about missing graphite support.
2125 Mention -floop-nest-optimize.
2127 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
2129 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
2130 (vtst_p16): Likewise.
2131 (vtstq_p8): Likewise.
2132 (vtstq_p16): Likewise.
2134 (vtstq_p64): Likewise.
2135 * config/arm/arm_neon.h (vgetq_lane_p64): New.
2136 (vset_lane_p64): New.
2137 (vsetq_lane_p64): New.
2139 2017-02-10 Jakub Jelinek <jakub@redhat.com>
2141 PR tree-optimization/79411
2142 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
2143 stmt operands are SSA_NAMEs used in abnormal phis.
2144 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
2147 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
2150 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
2153 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
2155 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
2157 2017-02-09 Jakub Jelinek <jakub@redhat.com>
2159 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
2163 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
2164 not arbitrary TREE_CONSTANT.
2167 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
2168 "omp declare target link" attribute unless is_global_var.
2169 * omp-offload.c (find_link_var_op): Likewise.
2171 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
2172 Chung-Lin Tang <cltang@codesourcery.com>
2174 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
2176 (gimplify_adjust_omp_clauses): Don't delete TILE.
2177 (gimplify_omp_for): Deal with TILE.
2178 * internal-fn.c (expand_GOACC_TILE): New function.
2179 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
2181 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
2182 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
2184 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
2185 avoid DIV for outermost collapse var.
2186 (expand_oacc_for): Insert tile element loop as needed. Adjust.
2187 Remove out of date comments, fix whitespace.
2188 * omp-general.c (omp_extract_for_data): Deal with tiling.
2189 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
2190 adjust OLF_DIM_BASE value.
2191 (struct omp_for_data): Add tiling field.
2192 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
2193 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
2194 for auto loops. Remove default auto determining, moved to
2195 oacc_loop_fixed_partitions.
2196 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
2197 stmts, add e_mask field.
2198 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
2199 (oacc_thread_numbers): Use oacc_dim_call.
2200 (oacc_xform_tile): New.
2201 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
2202 (finish_oacc_loop): Adjust for ifns vector.
2203 (oacc_loop_discover_walk): Append loop abstraction sites to list,
2204 add case for GOACC_TILE fns.
2205 (oacc_loop_xform_loop): Delete.
2206 (oacc_loop_process): Iterate over call list directly, and add
2207 handling for GOACC_TILE fns.
2208 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
2210 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
2211 vector partitioning to outer loops. Assign 2 partitions to loops
2212 when available. Add TILE handling.
2213 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
2214 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
2215 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
2216 * tree.c (omp_clause_num_ops): Adjust TILE ops.
2217 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
2219 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
2221 * configure.ac (ACX_BUGURL): Update.
2222 * configure: Regenerate.
2224 2017-02-09 Richard Biener <rguenther@suse.de>
2226 PR tree-optimization/69823
2227 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
2228 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
2230 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
2232 * config/arc/arc-c.def: Add __NPS400__ definition.
2233 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
2234 (TARGET_NPS400): Define.
2236 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
2238 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
2240 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
2242 (arc_cpu_types): Fill the arch_info field with a pointer into the
2243 arc_arch_types table.
2244 (arc_selected_cpu): Declare.
2245 * config/arc/arc.c (arc_selected_cpu): Make global.
2246 (arc_selected_arch): Delete.
2247 (arc_base_cpu): Delete.
2248 (arc_override_options): Remove references to deleted variables,
2249 update access to arch information.
2250 (ARC_OPT): Update access to arch information.
2251 (ARC_OPTX): Likewise.
2252 * config/arc/arc.h (arc_base_cpu): Remove declaration.
2253 (TARGET_ARC600): Update access to arch information.
2254 (TARGET_ARC601): Likewise.
2255 (TARGET_ARC700): Likewise.
2256 (TARGET_EM): Likewise.
2257 (TARGET_HS): Likewise.
2258 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
2261 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
2264 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
2265 condition/operands for integer GE/LE/GEU/LEU operations.
2267 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
2269 PR translation/79397
2270 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
2273 2017-02-08 Martin Jambor <mjambor@suse.cz>
2276 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
2277 whether allocation happened.
2278 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
2279 nothing was allocated.
2281 2017-02-08 Jakub Jelinek <jakub@redhat.com>
2283 PR tree-optimization/79408
2284 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
2285 constant, but SSA_NAME with a known integer range, use the minimum
2286 of that range instead of op1 to determine if modulo can be replaced
2287 with its first operand.
2289 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2291 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
2293 2017-02-08 Richard Biener <rguenther@suse.de>
2295 PR tree-optimization/71824
2296 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
2297 Check all loops contained in the merged region.
2299 2017-02-07 Andrew Pinski <apinski@cavium.com>
2301 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
2303 2017-02-07 Andrew Pinski <apinski@cavium.com>
2305 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
2306 (thunderxt88): Likewise.
2307 (thunderxt81): Disable LSE and change v8.1 to v8.
2308 (thunderxt83): Likewise.
2310 2017-02-07 Jakub Jelinek <jakub@redhat.com>
2311 Richard Biener <rguenther@suse.de>
2314 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
2315 type from int to size_t.
2316 * ira-costs.c (struct_costs_size): Change type from int to size_t.
2318 2017-02-07 Jakub Jelinek <jakub@redhat.com>
2320 PR rtl-optimization/79386
2321 * cprop.c (bypass_conditional_jumps): Initialize
2322 bypass_last_basic_block already before splitting bbs after
2323 unconditional traps...
2324 (bypass_conditional_jumps): ... rather than here.
2327 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
2328 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
2329 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
2330 fix -masm=intel patterns.
2332 2017-02-07 Richard Biener <rguenther@suse.de>
2334 PR tree-optimization/79256
2336 * builtins.c (get_object_alignment_2): Use min_align_of_type
2337 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
2338 and ADJUST_FIELD_ALIGN.
2340 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
2342 * doc/tm.texi: Regenerate.
2343 * stor-layout.c (layout_decl): Adjust.
2344 (update_alignment_for_field): Likewise.
2345 (place_field): Likewise.
2346 (min_align_of_type): Likewise.
2347 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
2348 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
2349 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
2350 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
2351 * config/frv/frv.c (frv_adjust_field_align): Likewise.
2352 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
2353 * config/i386/i386.c (x86_field_alignment): Likewise.
2354 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
2355 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
2356 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
2357 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
2358 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
2359 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
2363 2017-01-30 Richard Biener <rguenther@suse.de>
2365 PR tree-optimization/79256
2366 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
2367 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
2370 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
2372 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
2373 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
2374 builtins to SImode and emit a zero-extend, if necessary.
2376 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2378 * docs/invoke.texi (RISC-V Options): Alphabetize.
2380 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2382 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
2385 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2387 * config/riscv/riscv.c: New file.
2388 * gcc/common/config/riscv/riscv-common.c: Likewise.
2389 * config.gcc: Likewise.
2390 * config/riscv/constraints.md: Likewise.
2391 * config/riscv/elf.h: Likewise.
2392 * config/riscv/generic.md: Likewise.
2393 * config/riscv/linux.h: Likewise.
2394 * config/riscv/multilib-generator: Likewise.
2395 * config/riscv/peephole.md: Likewise.
2396 * config/riscv/pic.md: Likewise.
2397 * config/riscv/predicates.md: Likewise.
2398 * config/riscv/riscv-builtins.c: Likewise.
2399 * config/riscv/riscv-c.c: Likewise.
2400 * config/riscv/riscv-ftypes.def: Likewise.
2401 * config/riscv/riscv-modes.def: Likewise.
2402 * config/riscv/riscv-opts.h: Likewise.
2403 * config/riscv/riscv-protos.h: Likewise.
2404 * config/riscv/riscv.h: Likewise.
2405 * config/riscv/riscv.md: Likewise.
2406 * config/riscv/riscv.opt: Likewise.
2407 * config/riscv/sync.md: Likewise.
2408 * config/riscv/t-elf-multilib: Likewise.
2409 * config/riscv/t-linux: Likewise.
2410 * config/riscv/t-linux-multilib: Likewise.
2411 * config/riscv/t-riscv: Likewise.
2412 * configure.ac: Likewise.
2413 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
2414 Waterman as RISC-V maintainers.
2415 * doc/install.texi: Add RISC-V entries.
2416 * doc/invoke.texi: Add RISC-V options section.
2417 * doc/md.texi: Add RISC-V constraints section.
2418 * configure: Regenerated.
2420 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2423 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
2424 false values to be constant vectors with all 0 or all 1 bits set.
2425 (vcondu<mode><mode>): Likewise.
2426 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
2428 (fpmask_comparison_operator): Update comment.
2429 (vecint_comparison_operator): New predicate.
2430 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
2431 vector conditionals when the true and false values are constant
2432 vectors with all 0 bits or all 1 bits set.
2434 2017-02-06 Martin Sebor <msebor@redhat.com>
2436 PR tree-optimization/79376
2437 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
2439 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
2441 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
2442 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
2443 to simplify split condition.
2445 2017-02-06 Jakub Jelinek <jakub@redhat.com>
2447 * omp-expand.c (oxpand_omp_atomic_fetch_op,
2448 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
2451 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
2453 PR rtl-optimization/68664
2454 * target.def (can_speculate_insn): New hook.
2455 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
2456 * doc/tm.texi: Regenerate.
2457 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
2458 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
2459 (rs6000_sched_can_speculate_insn): New function.
2461 2017-02-06 Jakub Jelinek <jakub@redhat.com>
2463 PR tree-optimization/79284
2464 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
2465 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
2466 vectorizable_mask_load_store, vectorizable_operation,
2467 vect_is_simple_cond, get_same_sized_vectype): Use it instead
2468 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
2469 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
2470 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
2471 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
2472 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
2473 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
2474 is_gimple_assign (stmt). Replace another such test with
2475 is_gimple_assign (stmt).
2477 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
2480 * config/avr/avr.c (rtl-iter.h): Include it.
2481 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
2482 (avr_legitimate_combined_insn): ...and implementation.
2484 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
2486 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
2487 * config/s390/s390.c (s390_const_operand_ok)
2488 (s390_canonicalize_comparison, s390_extract_part)
2489 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
2490 (s390_contiguous_bitmask_p, s390_rtx_costs)
2491 (legitimize_pic_address): Likewise.
2492 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
2493 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
2494 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
2495 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
2496 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
2498 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
2500 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
2501 REGNO($0) == REGNO($1).
2503 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2505 * config/s390/linux.h(SIZE_TYPE): Add comment.
2507 2017-02-06 Julian Brown <julian@codesourcery.com>
2508 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
2509 Virendra Pathak <virendra.pathak@broadcom.com>
2511 * config/aarch64/aarch64-cores.def: Change the scheduler
2513 * config/aarch64/aarch64.md: Include thunderx2t99.md.
2514 * config/aarch64/thunderx2t99.md: New file.
2516 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2518 * doc/standards.texi (Go Language): Update link to language
2521 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
2523 * tree-eh.c (lower_resx): Sanitize profile.
2524 (cleanup_empty_eh_move_lp): Likewise.
2526 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
2529 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
2531 * cfgloopmanip.h (loop_version): Update prototype.
2532 * modulo-sched.c (sms_schedule): Update call of loop_version.
2533 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
2534 * tree-parloops.c (gen_parallel_loop): Likewise.
2535 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
2536 * tree-ssa-loop-split.c (split_loop): Likewise.
2537 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
2538 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
2540 2017-02-05 Martin Liska <mliska@suse.cz>
2543 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
2545 (print_operand_address): Initialize a struct to zero.
2547 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2549 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
2550 garbage collector only in textual form.
2552 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2554 * doc/extend.texi (x86 specific memory model extensions for
2555 transactional memory): Simplify a phrase.
2557 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
2560 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
2561 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
2562 (atomic_storedi_1): Likewise.
2564 2017-02-04 Jakub Jelinek <jakub@redhat.com>
2566 PR tree-optimization/79338
2567 * tree-parloops.c (gather_scalar_reductions): Don't call
2568 vect_analyze_loop_form for loop->inner before destroying loop's
2571 2017-02-03 Martin Sebor <msebor@redhat.com>
2573 PR tree-optimization/79327
2574 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
2575 when precision has resulted in leading zeros.
2576 (format_integer): Adjust the likely counter to assume an unknown
2577 argument that may be zero is non-zero.
2579 2017-02-03 Jason Merrill <jason@redhat.com>
2582 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
2583 avoid copying non-taken branch.
2585 2017-02-03 Jakub Jelinek <jakub@redhat.com>
2587 PR tree-optimization/79340
2588 * tree-vect-loop.c (vectorizable_reduction): Release
2589 vec_defs elements after safe_splicing them into other vectors.
2592 PR tree-optimization/79327
2593 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
2594 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
2596 (format_integer): Use wide_int_to_tree instead of build_int_cst
2597 + to_?hwi. If argmin is NULL, just set argmin and argmax to
2598 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
2599 of shortest and longest sequence.
2601 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
2603 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
2604 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
2606 2017-02-03 Walter Lee <walt@tilera.com>
2609 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
2610 after initial stackframe link reg save.
2611 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
2613 2017-02-03 Jakub Jelinek <jakub@redhat.com>
2616 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
2617 wu for stxssp alternative.
2619 2017-02-03 Martin Sebor <msebor@redhat.com>
2621 PR tree-optimization/79352
2622 * gimple-fold.c (get_range_strlen): Add argument.
2623 (get_range_strlen): Change return type to bool.
2624 (get_maxval_strlen): Pass in a dummy argument.
2625 * gimple-fold.h (get_range_strlen): Change return type to bool.
2626 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
2627 * tree.h (array_at_struct_end_p): Add argument.
2628 * tree.c (array_at_struct_end_p): Handle it.
2630 2017-02-03 Martin Liska <mliska@suse.cz>
2633 * multiple_target.c (create_dispatcher_calls): Redirect edge
2634 from a caller of a dispatcher.
2635 (expand_target_clones): Make the clones local.
2636 (ipa_target_clone): Do both target clones and resolvers.
2637 (ipa_dispatcher_calls): Remove the pass.
2638 (pass_dispatcher_calls::gate): Likewise.
2639 (make_pass_dispatcher_calls): Likewise.
2640 * passes.def (pass_target_clone): Put as very first IPA early
2643 2017-02-03 Martin Liska <mliska@suse.cz>
2645 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
2646 in case of a function with ifunc attribute.
2648 2017-02-03 Martin Liska <mliska@suse.cz>
2650 * cgraph.c (cgraph_node::dump): Dump function version info.
2651 * symtab.c (symtab_node::dump_base): Add missing new line.
2653 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
2655 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
2656 (ifcombine_ifandif): Use it.
2658 2017-02-03 Martin Liska <mliska@suse.cz>
2660 * doc/invoke.texi: Document default value for
2661 use-after-scope-direct-emission-threshold.
2663 2017-02-03 Martin Liska <mliska@suse.cz>
2665 PR tree-optimization/79339
2666 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
2667 (format_floating): Likewise.
2669 2017-02-03 Martin Liska <mliska@suse.cz>
2672 * ipa-prop.c (ipa_node_params_t::insert): Remove current
2674 (ipa_node_params_t::remove): Likewise.
2675 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
2676 initialization from removed ipa_node_params_t::insert.
2677 (ipa_node_params::~ipa_node_params): Move from removed
2678 ipa_node_params_t::release.
2679 * symbol-summary.h (symbol_summary::m_released): New member.
2680 Do not release a summary twice. Do not allow to call finalizer
2681 for types of a summary that live in GGC memory.
2683 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
2685 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
2688 2017-02-02 Martin Sebor <msebor@redhat.com>
2691 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
2692 (format_string): Tighten up the range of output for non-constant
2693 strings and correct the expected range for wide non-constant strings.
2695 2017-02-02 Martin Sebor <msebor@redhat.com>
2697 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
2700 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
2702 (-fdump-tree-@var): Add to index and document how to come up
2703 with pass-specific option and dump file names.
2704 (-fdump-passes): Clarify where to look for output.
2706 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
2709 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
2710 statistics of the analyzed path; allow threading for speed when
2711 any of BBs along the path are optimized for speed.
2713 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
2716 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
2717 settings of the virtual registers.
2720 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
2722 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2723 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
2726 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2728 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
2729 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
2731 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2733 * config/s390/s390.md: Add missing comments with the expanded
2735 * config/s390/vector.md: Likewise.
2736 * config/s390/vx-builtins.md: Likewise.
2738 2017-02-02 Jakub Jelinek <jakub@redhat.com>
2741 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
2742 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
2743 conditions on a single line.
2745 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2747 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
2748 __S390_VX__ to __VX__.
2750 2017-02-01 Andrew Pinski <apinski@cavium.com>
2752 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
2753 stmt_info to record_stmt_cost.
2754 (vect_get_known_peeling_cost): Pass stmt_info if known to
2756 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
2757 cpu_vector_cost field into
2758 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
2759 field into vec_int_stmt_cost and vec_fp_stmt_cost.
2760 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
2761 splitting of scalar_stmt_cost and vec_stmt_cost.
2762 (thunderx_vector_cost): Likewise.
2763 (cortexa57_vector_cost): LIkewise.
2764 (exynosm1_vector_cost): Likewise.
2765 (xgene1_vector_cost): Likewise.
2766 (thunderx2t99_vector_cost): Improve after the splitting of the two
2768 (aarch64_builtin_vectorization_cost): Update for the splitting of
2769 scalar_stmt_cost and vec_stmt_cost.
2771 2017-02-01 Torvald Riegel <triegel@redhat.com>
2772 Richard Henderson <rth@redhat.com>
2774 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
2775 conditional on existance of a fast atomic load.
2776 * optabs-query.c (can_atomic_load_p): New function.
2777 * optabs-query.h (can_atomic_load_p): Declare it.
2778 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
2779 no fast atomic load is available for the particular size of access.
2780 (expand_atomic_compare_and_swap): Likewise.
2781 (expand_atomic_load): Likewise.
2782 (expand_atomic_store): Likewise.
2783 (expand_atomic_fetch_op): Likewise.
2784 * testsuite/lib/target-supports.exp
2785 (check_effective_target_sync_int_128): Remove x86 because it provides
2786 no fast atomic load.
2787 (check_effective_target_sync_int_128_runtime): Likewise.
2789 2017-02-01 Richard Biener <rguenther@suse.de>
2791 * graphite.c: Include tree-vectorizer.h for find_loop_location.
2792 (graphite_transform_loops): Provide opt-info for optimized nests.
2793 * tree-parloop.c (parallelize_loops): Provide opt-info for
2796 2017-02-01 Richard Biener <rguenther@suse.de>
2799 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
2802 2017-02-01 Richard Biener <rguenther@suse.de>
2804 PR tree-optimization/71824
2805 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
2806 Verify the loops are valid in the merged SESE region.
2807 (scop_detection::can_represent_loop_1): Check analyzing the
2808 evolution of the number of iterations in the region succeeds.
2810 2017-01-31 Ian Lance Taylor <iant@golang.org>
2812 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
2813 REG_ARGS_SIZE note to 32-bit push insns and call insn.
2815 2017-01-31 David Malcolm <dmalcolm@redhat.com>
2817 PR preprocessor/79210
2818 * input.c (get_substring_ranges_for_loc): Replace line_width
2819 assertion with error-handling.
2821 2017-01-31 Richard Biener <rguenther@suse.de>
2823 PR tree-optimization/77318
2824 * graphite-sese-to-poly.c (extract_affine): Fix assert.
2825 (create_pw_aff_from_tree): Take loop parameter.
2826 (add_condition_to_pbb): Pass loop of the condition to
2827 create_pw_aff_from_tree.
2829 2017-01-31 Jakub Jelinek <jakub@redhat.com>
2831 * config/s390/s390.c (s390_asan_shadow_offset): New function.
2832 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
2834 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2838 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
2840 (convert_int_to_float128): Likewise.
2841 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
2842 (convert_int_to_float128): Likewise.
2843 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
2844 (UNSPEC_IEEE128_CONVERT): Likewise.
2845 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
2846 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
2847 Use local variables for IBM extended format.
2848 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
2849 (fix_trunc<mode>si2_fprs): Likewise.
2850 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
2851 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
2852 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
2853 to know that we can now have integers of all sizes in vector
2855 (fix<uns>_<mode>di2_hw): Likewise.
2856 (float<uns>_<mode>si2_hw): Likewise.
2857 (fix_<mode>si2_hw): Likewise.
2858 (fixuns_<mode>si2_hw): Likewise.
2859 (float<uns>_<mode>di2_hw): Likewise.
2860 (float_<mode>di2_hw): Likewise.
2861 (float_<mode>si2_hw): Likewise.
2862 (floatuns_<mode>di2_hw): Likewise.
2863 (floatuns_<mode>si2_hw): Likewise.
2864 (xscvqp<su>wz_<mode>): Delete, no longer used.
2865 (xscvqp<su>dz_<mode>): Likewise.
2866 (xscv<su>dqp_<mode>): Likewise.
2867 (ieee128_mfvsrd_64bit): Likewise.
2868 (ieee128_mfvsrd_32bit): Likewise.
2869 (ieee128_mfvsrwz): Likewise.
2870 (ieee128_mtvsrw): Likewise.
2871 (ieee128_mtvsrd_64bit): Likewise.
2872 (ieee128_mtvsrd_32bit): Likewise.
2874 2017-01-31 Martin Liska <mliska@suse.cz>
2877 * ipa-prop.c (ipa_free_all_node_params): Call release method
2878 instead of ~sumbol_summary to not to trigger double times
2881 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
2883 PR tree-optimization/71691
2884 * bitmap.h (class auto_bitmap): New.
2885 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
2886 is_maybe_undefined instead of ssa_undefined_value_p.
2888 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2890 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
2891 __S390_ARCH_LEVEL__ to __ARCH__.
2893 2017-01-31 Jakub Jelinek <jakub@redhat.com>
2895 PR tree-optimization/79267
2896 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
2897 if should_remove_lhs_p is true.
2899 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
2902 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
2903 (add_alignment_attribute): New.
2904 (base_type_die): Add alignment attribute.
2905 (subrange_type_die): Likewise.
2906 (modified_type_die): Likewise.
2907 (gen_array_type_die): Likewise.
2908 (gen_descr_array_type_die: Likewise.
2909 (gen_enumeration_type_die): Likewise.
2910 (gen_subprogram_die): Likewise.
2911 (gen_variable_die): Likewise.
2912 (gen_field_die): Likewise.
2913 (gen_ptr_to_mbr_type_die): Likewise.
2914 (gen_struct_or_union_type_die): Likewise.
2915 (gen_subroutine_type_die): Likewise.
2916 (gen_typedef_die): Likewise.
2917 (base_type_cmp): Compare alignment attribute.
2919 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2922 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
2923 (setb_unsigned) New pattern for setb with CCUNS.
2924 * config/rs6000/rs6000.c (expand_block_compare): Use a different
2925 subfc./subfe sequence to avoid overflow problems. Generate a
2926 shorter sequence with cmpld/setb for power9.
2927 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
2928 for generating subfc. instruction.
2929 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
2930 now uses this instruction.
2932 2017-01-30 Ian Lance Taylor <iant@google.com>
2935 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
2936 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
2938 2017-01-30 Martin Sebor <msebor@redhat.com>
2940 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
2941 Move constant to the right of a relational operator.
2942 (get_mpfr_format_length, format_character, format_string): Ditto.
2943 (should_warn_p, maybe_warn): Same.
2945 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
2947 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
2950 * asan.c (get_translation_unit_decl): Remove function.
2951 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
2953 2017-01-30 Martin Liska <mliska@suse.cz>
2955 PR gcov-profile/79259
2956 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
2959 2017-01-30 Martin Liska <mliska@suse.cz>
2962 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
2963 Initialize variables with NULL value.
2965 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
2968 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
2970 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
2972 2017-01-30 Richard Biener <rguenther@suse.de>
2974 PR tree-optimization/79276
2975 * tree-vrp.c (process_assert_insertions): Properly adjust common
2976 when removing a duplicate.
2978 * gcc.dg/torture/pr79276.c: New testcase.
2980 2017-01-30 Richard Biener <rguenther@suse.de>
2982 PR tree-optimization/79256
2983 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
2984 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
2986 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
2988 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
2991 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
2992 ("*r<noxa>sbg_<mode>_sll_bitmask")
2993 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
2994 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
2995 Use contiguous_bitmask_nowrap_operand.
2997 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3000 * config/rs6000/altivec.h (vec_xl): Revise #define.
3001 (vec_xst): Likewise.
3003 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
3005 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
3007 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
3009 PR rtl-optimization/79194
3010 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
3011 traps before call to bypass_conditional_jumps.
3013 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
3015 PR tree-optimization/71374
3016 * lra-constraints.c (check_conflict_input_operands): New.
3017 (match_reload): Use it.
3019 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
3022 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
3023 account to calculate conflict_set.
3025 2017-01-27 Bin Cheng <bin.cheng@arm.com>
3027 PR rtl-optimization/78559
3028 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
3029 other_insn in combine.
3031 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
3033 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
3034 uint16_type_node for BT_UINT16.
3036 2017-01-27 David Malcolm <dmalcolm@redhat.com>
3038 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
3039 "RTL Tests" to menu.
3040 (GIMPLE Tests): New node.
3041 (RTL Tests): New node.
3043 2017-01-27 Richard Biener <rguenther@suse.de>
3045 PR tree-optimization/79245
3046 * tree-loop-distribution.c (distribute_loop): Apply cost
3047 modeling also to detected patterns.
3049 2017-01-27 Richard Biener <rguenther@suse.de>
3051 PR tree-optimization/71433
3052 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
3053 (compare_assert_loc): New function.
3054 (process_assert_insertions): Sort and optimize assert locations
3055 to remove duplicates and push down identical assertions on
3056 edges to their destination block.
3058 2017-01-27 Richard Biener <rguenther@suse.de>
3060 PR tree-optimization/79244
3061 * tree-vrp.c (remove_range_assertions): Forcefully propagate
3062 out SSA names even if abnormal.
3064 2017-01-27 Jakub Jelinek <jakub@redhat.com>
3066 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
3067 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
3068 instead of MPFR_RNDN.
3070 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
3073 * arm.c (arm_option_override): Don't call build_target_option_node
3074 until after doing all option overrides.
3075 (arm_valid_target_attribute_tree): Likewise.
3077 2017-01-27 Martin Liska <mliska@suse.cz>
3079 * doc/invoke.texi (-fprofile-arcs): Document profiling support
3080 for {cd}tors and C++ {cd}tors.
3082 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
3084 * config/s390/s390.md ("*setmem_long_and")
3085 ("*setmem_long_and_31z"): Use zero_extend instead of and.
3087 2017-01-26 Martin Sebor <msebor@redhat.com>
3089 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
3092 2017-01-26 Martin Sebor <msebor@redhat.com>
3094 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
3095 HAVE_DFmode before using XFmode or DFmode.
3096 (parse_directive): Avoid using the z length modifier to avoid
3097 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
3100 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
3101 to accept adjustment as an array.
3102 (get_int_range): New function.
3103 (struct directive): Make width and prec arrays.
3104 (directive::set_width, directive::set_precision): Call get_int_range.
3105 (format_integer, format_floating): Handle width and precision ranges.
3106 (format_string, parse_directive): Same.
3108 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3111 * dwarf2out.c (generate_skeleton_bottom_up): For children with
3112 comdat_type_p set, just clone them, but keep the children in the
3116 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
3117 which have direct callers with -fvar-tracking-assignments enabled
3119 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
3120 inside of type units.
3122 2017-01-26 Martin Sebor <msebor@redhat.com>
3125 * gimple-ssa-sprintf.c (struct result_range): Add likely and
3127 (struct format_result): Replace number_chars, number_chars_min,
3128 and number_chars_max with a single member of struct result_range.
3130 (format_result::operator+=): Adjust.
3131 (struct fmtresult): Remove bounded. Handle likely and unlikely
3133 (fmtresult::adjust_for_width_or_precision): New function.
3134 (fmtresult:type_max_digits): New function.
3135 (bytes_remaining): Handle likely and unlikely counters.
3136 (min_bytes_remaining): Remove.
3137 (format_percent): Simplify.
3138 (format_integer, format_floating): Set likely and unlikely counters.
3139 (get_string_length, format_character, format_string): Same.
3140 (format_plain, should_warn_p): New function.
3141 (maybe_warn): Call should_warn_p. Update diagnostic messages
3142 and handle those for all directives, including plain strings.
3143 (format_directive): Handle likely and unlikely counters.
3144 Remove unnecessary quoting from diagnostics. Add an informational
3146 (add_bytes): Remove.
3147 (pass_sprintf_length::compute_format_length): Simplify.
3148 (try_substitute_return_value): Handle likely and unlikely counters.
3150 2017-01-26 Carl Love <cel@us.ibm.com>
3152 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
3153 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
3155 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
3158 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
3159 endianess for subregs into account.
3160 * lra-constraints.c (lra_constraints): Do risky transformations
3161 always on the first iteration.
3162 * lra-lives.c (check_pseudos_live_through_calls): Add arg
3163 last_call_used_reg_set.
3164 (process_bb_lives): Define and use last_call_used_reg_set.
3165 * lra.c (lra): Always continue after lra_constraints on the first
3168 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
3170 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
3172 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
3174 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3176 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
3177 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
3178 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
3179 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
3180 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
3181 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
3182 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
3183 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
3184 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
3186 2017-01-26 Marek Polacek <polacek@redhat.com>
3189 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
3190 for the third operand.
3192 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3195 * omp-low.c (struct omp_context): Add simt_stmt field.
3196 (scan_omp_for): Return omp_context *.
3197 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
3198 context to the _simt_ SIMD stmt.
3199 (lower_omp_for): For combined SIMD with sibling _simt_
3200 SIMD, make sure to use the same decls in _looptemp_
3201 clauses as in the sibling.
3203 2017-01-26 David Sherwood <david.sherwood@arm.com>
3206 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
3209 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3212 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
3213 emit fld b; fld a; if possible.
3215 * brig-builtins.def: Update copyright years.
3216 * config/arm/arm_acle_builtins.def: Update copyright years.
3218 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
3221 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
3222 constraint instead of o for the stxsd instruction.
3224 2017-01-25 Carl Love <cel@us.ibm.com>
3226 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
3227 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
3229 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
3231 * doc/invoke.texi (C++ Dialect Options): Fix typo.
3233 2017-01-25 Richard Biener <rguenther@suse.de>
3235 PR tree-optimization/69264
3236 * target.def (vector_alignment_reachable): Improve documentation.
3237 * doc/tm.texi: Regenerate.
3238 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
3240 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
3241 earlier changes with respect to TYPE_USER_ALIGN.
3242 (vector_alignment_reachable_p): Likewise. Improve dumping.
3244 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3247 * config/arm/arm.md (xordi3): Force constant operand into a register
3250 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3252 * doc/invoke.texi (-fstore-merging): Correct default optimization
3253 levels at which it is enabled.
3254 (-O): Move -fstore-merging from list to...
3257 2017-01-25 Richard Biener <rguenther@suse.de>
3260 * omp-expand.c: Include debug.h.
3261 (expand_omp_taskreg): Make sure to generate early debug before
3262 outlining anything from a function.
3263 (expand_omp_target): Likewise.
3264 (grid_expand_target_grid_body): Likewise.
3266 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
3269 * asan.c (get_translation_unit_decl): New function.
3270 (asan_add_global): Extract modules file name from globals
3271 TRANSLATION_UNIT_DECL name.
3273 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
3276 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
3277 for long calls with APCS frame and VFP.
3279 2017-01-24 David Malcolm <dmalcolm@redhat.com>
3281 * cfg.c (original_copy_tables_initialized_p): New function.
3282 * cfg.h (original_copy_tables_initialized_p): New decl.
3283 * cfgrtl.c (relink_block_chain): Guard the call to
3284 free_original_copy_tables with a call to
3285 original_copy_tables_initialized_p.
3286 * cgraph.h (symtab_node::native_rtl_p): New decl.
3287 * cgraphunit.c (symtab_node::native_rtl_p): New function.
3288 (symtab_node::needed_p): Don't assert for early assembly output
3289 for __RTL functions.
3290 (cgraph_node::finalize_function): Set "force_output" for __RTL
3292 (cgraph_node::analyze): Bail out early for __RTL functions.
3293 (analyze_functions): Update assertion to support __RTL functions.
3294 (cgraph_node::expand): Bail out early for __RTL functions.
3295 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
3297 * function.h (struct function): Update comment for field
3299 * gimple-expr.c: Include "tree-pass.h".
3300 (gimple_has_body_p): Return false for __RTL functions.
3301 * Makefile.in (OBJS): Add run-rtl-passes.o.
3302 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
3304 (gcc::pass_manager::get_clean_slate): New accessor.
3305 * passes.c: Include "insn-addr.h".
3306 (should_skip_pass_p): Add logging. Update logic for running
3307 "expand" to be compatible with both __GIMPLE and __RTL. Guard
3308 property-provider override so it is only done for gimple passes.
3310 (skip_pass): New function.
3311 (execute_one_pass): Call skip_pass when skipping passes.
3312 * read-md.c (md_reader::read_char): Support filtering
3313 the input to a subset of line numbers.
3314 (md_reader::md_reader): Initialize fields
3315 m_first_line and m_last_line.
3316 (md_reader::read_file_fragment): New function.
3317 * read-md.h (md_reader::read_file_fragment): New decl.
3318 (md_reader::m_first_line): New field.
3319 (md_reader::m_last_line): New field.
3320 * read-rtl-function.c (function_reader::create_function): Only
3321 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
3322 curr_properties. Set DECL_INITIAL to a dummy block.
3323 (read_rtl_function_body_from_file_range): New function.
3324 * read-rtl-function.h (read_rtl_function_body_from_file_range):
3326 * run-rtl-passes.c: New file.
3327 * run-rtl-passes.h: New file.
3329 2017-01-24 Jeff Law <law@redhat.com>
3331 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
3334 2017-01-24 Bin Cheng <bin.cheng@arm.com>
3336 PR tree-optimization/79159
3337 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
3338 (record_nonwrapping_iv): Improve boundary using above function if no
3339 value range information.
3341 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
3342 Martin Jambor <mjambor@suse.cz>
3344 * brig-builtins.def: New file.
3345 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
3346 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
3347 (DEF_HSAIL_SAT_BUILTIN): Likewise.
3348 (DEF_HSAIL_INTR_BUILTIN): Likewise.
3349 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
3350 * builtin-types.def (BT_INT8): New.
3351 (BT_INT16): Likewise.
3352 (BT_UINT8): Likewise.
3353 (BT_UINT16): Likewise.
3354 (BT_FN_ULONG): Likewise.
3355 (BT_FN_UINT_INT): Likewise.
3356 (BT_FN_UINT_ULONG): Likewise.
3357 (BT_FN_UINT_LONG): Likewise.
3358 (BT_FN_UINT_PTR): Likewise.
3359 (BT_FN_ULONG_PTR): Likewise.
3360 (BT_FN_INT8_FLOAT): Likewise.
3361 (BT_FN_INT16_FLOAT): Likewise.
3362 (BT_FN_UINT32_FLOAT): Likewise.
3363 (BT_FN_UINT16_FLOAT): Likewise.
3364 (BT_FN_UINT8_FLOAT): Likewise.
3365 (BT_FN_UINT64_FLOAT): Likewise.
3366 (BT_FN_UINT16_UINT32): Likewise.
3367 (BT_FN_UINT32_UINT16): Likewise.
3368 (BT_FN_UINT16_UINT16_UINT16): Likewise.
3369 (BT_FN_INT_PTR_INT): Likewise.
3370 (BT_FN_UINT_PTR_UINT): Likewise.
3371 (BT_FN_LONG_PTR_LONG): Likewise.
3372 (BT_FN_ULONG_PTR_ULONG): Likewise.
3373 (BT_FN_VOID_UINT64_UINT64): Likewise.
3374 (BT_FN_UINT8_UINT8_UINT8): Likewise.
3375 (BT_FN_INT8_INT8_INT8): Likewise.
3376 (BT_FN_INT16_INT16_INT16): Likewise.
3377 (BT_FN_INT_INT_INT): Likewise.
3378 (BT_FN_UINT_FLOAT_UINT): Likewise.
3379 (BT_FN_FLOAT_UINT_UINT): Likewise.
3380 (BT_FN_ULONG_UINT_UINT): Likewise.
3381 (BT_FN_ULONG_UINT_PTR): Likewise.
3382 (BT_FN_ULONG_ULONG_ULONG): Likewise.
3383 (BT_FN_UINT_UINT_UINT): Likewise.
3384 (BT_FN_VOID_UINT_PTR): Likewise.
3385 (BT_FN_UINT_UINT_PTR: Likewise.
3386 (BT_FN_UINT32_UINT64_PTR): Likewise.
3387 (BT_FN_INT_INT_UINT_UINT): Likewise.
3388 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
3389 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
3390 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
3391 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
3392 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
3393 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
3394 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
3395 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
3396 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
3397 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
3398 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
3399 * doc/frontends.texi: List BRIG FE.
3400 * doc/install.texi (Testing): Add BRIG tesring requirements.
3401 * doc/invoke.texi (Overall Options): Mention BRIG.
3402 * doc/standards.texi (Standards): Doucment BRIG HSA version.
3404 2017-01-24 Richard Biener <rguenther@suse.de>
3406 PR translation/79208
3407 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
3409 2017-01-24 Martin Jambor <mjambor@suse.cz>
3412 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
3413 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
3416 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
3419 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
3420 casts from signed to unsigned really don't have a range.
3422 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
3424 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
3425 GMP_RNDx for compatiblity.
3427 2017-01-24 Martin Liska <mliska@suse.cz>
3430 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
3431 that would prevent us to call alloca with -1 as argument.
3433 2017-01-24 Jakub Jelinek <jakub@redhat.com>
3435 * dwarf2out.c (output_compilation_unit_header, output_file_names):
3436 Avoid -Wformat-security warning.
3438 2017-01-23 Andrew Pinski <apinski@cavium.com>
3440 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
3443 2017-01-23 Martin Sebor <msebor@redhat.com>
3446 * gimple-ssa-sprintf.c (warn_level): New global.
3447 (format_integer): Use it here and throughout the rest of the file.
3448 Use the same switch to compute sign as base.
3449 (maybe_warn): New function.
3450 (format_directive): Factor out warnings into maybe_warn.
3451 Add debugging output. Use warn_level.
3452 (add_bytes): Use warn_level.
3453 (pass_sprintf_length::compute_format_length): Add debugging output.
3454 (try_substitute_return_value): Same.
3455 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
3458 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
3459 (struct fmtresult, format_integer, format_floating): Adjust.
3460 (fmtresult::fmtresult): Set max correctly in two argument ctor.
3461 (get_string_length, format_string,format_directive): Same.
3462 (pass_sprintf_length::compute_format_length): Same.
3463 (try_substitute_return_value): Simplify slightly.
3466 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
3467 (fmtresult::operator+=): Outlined.
3468 (struct fmtresult): Add ctors.
3469 (struct conversion_spec): Rename...
3470 (struct directive): ...to this. Add and remove data members.
3471 (directive::set_width, directive::set_precision): New functions.
3472 (format_percent): Use fmtresult ctor.
3473 (get_width_and_precision): Remove.
3474 (format_integer): Make naming changes. Avoid computing width and
3476 (format_floating): Same. Adjust indentation.
3477 (format_character, format_none): New functions.
3478 (format_string): Moved character handling to format_character.
3479 (format_directive): Remove arguments, change return type.
3480 (parse_directive): New function.
3481 (pass_sprintf_length::compute_format_length): Move directive
3482 parsing to parse_directive.
3484 2017-01-23 Jakub Jelinek <jakub@redhat.com>
3486 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
3487 (assign_assembler_name_if_needed): ... this.
3488 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
3489 (assign_assembler_name_if_needed): ... this.
3490 (free_lang_data_in_cgraph): Adjust callers.
3491 * cgraphunit.c (cgraph_node::analyze): Likewise.
3492 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
3495 2017-01-23 Richard Biener <rguenther@suse.de>
3497 PR tree-optimization/79088
3498 PR tree-optimization/79188
3499 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
3500 resetting loop bounds after last path deletion. Reset loop
3501 bounds of the target loop, make code match the comments.
3502 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
3503 Make sure loops need no fixups.
3505 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
3507 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
3508 exponent support with double type for first argument.
3509 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
3510 type returned by __builtin_vec_extract_sig,
3511 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
3512 functions from "vector int" to "vector unsigned int" or from
3513 "vector long long int" to "vector unsigned long long int".
3514 Changed type returned by __builtin_vec_extract_exp,
3515 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
3516 functions from "vector int" to "vector unsigned int" or from
3517 "vector long long int" to "vector unsigned long long int".
3518 Changed return type of __builtin_vec_test_data_class,
3519 __builtin_vec_test_data_class_sp, and
3520 __builtin_vec_test_data_class_dp from "vector int" to
3521 "vector bool int" or from "vector long long int" to "vector bool
3522 long long int" and changed second argument type from "unsigned
3523 int" to "int". Added new overloaded function forms "vector float
3524 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
3525 "vector float __builtin_vec_insert_exp_sp (vector float, vector
3526 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
3527 double, vector unsigned long long int)" and "vector double
3528 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
3529 long int)". Changed return type of
3530 __builtin_scalar_test_data_class and
3531 __builtin_scalar_test_data_class_sp and
3532 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
3533 int" and changed second argument from "unsigned int" to "int".
3534 Changed type returned by __builtin_scalar_test_neg,
3535 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
3536 from "int" to "bool int". Added new overloaded function form
3537 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
3538 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
3539 exponent double-precision with floating point first argument.
3540 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
3541 documentation of scalar_test_data_class, scalar_test_neg,
3542 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
3543 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
3544 vec_test_data_class built-in functions to reflect refinements in
3545 their type signatures.
3547 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
3549 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
3551 (aarch64_elf_asm_destructor): Likewise.
3553 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
3555 PR rtl-optimization/78634
3556 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
3557 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
3558 * ifcvt.c (noce_try_cmove): Add missing cost check.
3560 PR rtl-optimization/71724
3561 * combine.c (if_then_else_cond): Look for situations where it is
3562 beneficial to undo the work of one of the recursive calls.
3564 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3566 PR tree-optimization/70754
3567 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3568 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3569 combined stmt before it if not NULL.
3570 (combine_chains): Process refs reversely and compute dominance point
3573 2017-01-23 Martin Liska <mliska@suse.cz>
3575 PR tree-optimization/79196
3576 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
3577 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
3579 (strlen_optimize_stmt): Call the renamed function.
3581 2017-01-23 Michael Matz <matz@suse.de>
3583 PR tree-optimization/78384
3584 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
3586 2017-01-23 Richard Biener <rguenther@suse.de>
3588 PR tree-optimization/79186
3589 * tree-vrp.c (register_new_assert_for): Make sure we've seen
3590 both incoming edges before moving an assert.
3592 2017-01-23 Martin Jambor <mjambor@suse.cz>
3594 * ipa-prop.c (load_from_param_1): Removed.
3595 (load_from_unmodified_param): Bits from load_from_param_1 put back
3597 (load_from_param): Removed.
3598 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
3599 with stmt. Reverted back to use of load_from_unmodified_param.
3601 2017-01-23 Martin Jambor <mjambor@suse.cz>
3604 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
3605 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
3606 field a pointer to garbage collected vector, mark lattices and
3607 ipcp_orig_node with GTY((skip)).
3608 (ipa_get_param_count): Adjust to descriptors being a pointer.
3609 (ipa_get_param): Likewise.
3610 (ipa_get_type): Likewise.
3611 (ipa_get_param_move_cost): Likewise.
3612 (ipa_set_param_used): Likewise.
3613 (ipa_get_controlled_uses): Likewise.
3614 (ipa_set_controlled_uses): Likewise.
3615 (ipa_is_param_used): Likewise.
3616 (ipa_node_params_t): Move into garbage collector. New methods insert
3618 (ipa_node_params_sum): Annotate wth GTY(()).
3619 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
3621 (ipa_load_from_parm_agg): Adjust declaration.
3622 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
3623 * ipa-profile.c (ipa_profile): Likewise.
3624 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
3625 (ipa_populate_param_decls): Make descriptors parameter garbage
3627 (ipa_dump_param): Adjust to descriptors being a pointer.
3628 (ipa_alloc_node_params): Likewise.
3629 (ipa_initialize_node_params): Likewise.
3630 (load_from_param_1): Make descriptors parameter garbage collected.
3631 (load_from_unmodified_param): Likewise.
3632 (load_from_param): Likewise.
3633 (ipa_load_from_parm_agg): Likewise.
3634 (ipa_node_params::~ipa_node_params): Removed.
3635 (ipa_free_all_node_params): Remove call to delete operator.
3636 (ipa_node_params_t::insert): New.
3637 (ipa_node_params_t::remove): Likewise.
3638 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
3639 copy known_csts and known_contexts vectors.
3640 (ipa_read_node_info): Adjust to descriptors being a pointer.
3641 (ipcp_modif_dom_walker): Make m_descriptors field garbage
3643 (ipcp_transform_function): Make descriptors variable garbage
3646 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
3648 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
3649 * config/i386/avx512dqintrin.h: Ditto.
3650 * config/i386/avx512fintrin.h: Ditto.
3651 * gcc/config/i386/i386.c: Handle new builtins.
3652 * config/i386/i386-builtin.def: Add new builtins.
3653 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
3654 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
3656 2017-01-23 Jakub Jelinek <jakub@redhat.com>
3657 Martin Liska <mliska@suse.cz>
3659 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
3660 * asan.c (asan_expand_poison_ifn): Support stores and use
3661 appropriate ASAN report function.
3662 * internal-fn.c (expand_ASAN_POISON_USE): New function.
3663 * internal-fn.def (ASAN_POISON_USE): Declare.
3664 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
3665 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
3666 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
3667 ASAN_POISON calls w/o LHS.
3668 * tree-ssa.c (execute_update_addresses_taken): Create clobber
3669 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
3670 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
3671 * gimplify.c (asan_poison_variables): Add attribute
3672 use_after_scope_memory to variables that really needs to live
3674 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
3675 having the attribute.
3677 2017-01-23 Martin Liska <mliska@suse.cz>
3679 * asan.c (create_asan_shadow_var): New function.
3680 (asan_expand_poison_ifn): Likewise.
3681 * asan.h (asan_expand_poison_ifn): New declaration.
3682 * internal-fn.c (expand_ASAN_POISON): Likewise.
3683 * internal-fn.def (ASAN_POISON): New builtin.
3684 * sanopt.c (pass_sanopt::execute): Expand
3685 asan_expand_poison_ifn.
3686 * tree-inline.c (copy_decl_for_dup_finish): Make function
3688 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
3689 * tree-ssa.c (is_asan_mark_p): New function.
3690 (execute_update_addresses_taken): Rewrite local variables
3691 (identified just by use-after-scope as addressable) into SSA.
3693 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
3695 * doc/install.texi (Specific): opensource.apple.com uses https
3696 now. Remove trailing slash.
3698 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
3700 * README.Portability: Remove note on an Irix compatibility issue.
3702 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
3704 * gcov.c (INCLUDE_ALGORITHM): Define.
3705 (INCLUDE_VECTOR): Define.
3706 No longer include <vector> and <algorithm> directly.
3708 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
3710 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
3712 * doc/invoke.texi (Code Gen Options): Ditto.
3714 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
3717 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
3719 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
3721 rtl-optimization/79125
3722 * cprop.c (local_cprop_pass): Handle cases where we make an
3725 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
3729 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
3730 read from, for big endian.
3732 2017-01-20 Jiong Wang <jiong.wang@arm.com>
3734 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
3735 register pauth builtins for LP64 only.
3737 2017-01-20 Marek Polacek <polacek@redhat.com>
3740 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
3743 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
3745 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
3747 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
3748 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
3749 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
3751 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3754 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
3755 in big-endian mode when they are not a single duplicated value.
3757 2017-01-20 Richard Biener <rguenther@suse.de>
3759 * BASE-VER: Bump to 7.0.1.
3761 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
3763 * omp-low.c (omplow_simd_context): New struct. Use it...
3764 (lower_rec_simd_input_clauses): ...here and...
3765 (lower_rec_input_clauses): ...here to hold common data. Adjust all
3766 references to idx, lane, max_vf, is_simt.
3768 2017-01-20 Graham Markall <graham.markall@embecosm.com>
3770 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
3773 2017-01-20 Martin Jambor <mjambor@suse.cz>
3775 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
3776 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
3778 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
3779 (GTFILES): Rename hsa.c to hsa-common.c.
3780 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
3781 * hsa-dump.c: Likewise.
3782 * hsa-gen.c: Likewise.
3783 * hsa-regalloc.c: Likewise.
3784 * ipa-hsa.c: Likewise.
3785 * omp-expand.c: Likewise.
3786 * omp-low.c: Likewise.
3787 * toplev.c: Likewise.
3789 2017-01-20 Marek Polacek <polacek@redhat.com>
3792 * doc/invoke.texi: Document -Wduplicated-branches.
3793 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
3794 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
3795 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
3796 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
3797 return 0 only when not OEP_LEXICOGRAPHIC.
3798 (fold_build_cleanup_point_expr): Use the expression
3799 location when building CLEANUP_POINT_EXPR.
3800 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
3801 * tree.c (add_expr): Handle error_mark_node.
3803 2017-01-20 Martin Liska <mliska@suse.cz>
3806 * tree-profile.c (init_ic_make_global_vars): Do not call
3808 (gimple_init_gcov_profiler): Likewise.
3810 2017-01-20 Martin Liska <mliska@suse.cz>
3813 * cgraph.h (maybe_create_reference): Remove argument and
3815 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
3817 * ipa-cp.c (create_specialized_node): Likewise.
3818 * symtab.c (symtab_node::maybe_create_reference): Handle
3819 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
3821 2017-01-20 Martin Liska <mliska@suse.cz>
3823 * read-rtl-function.c (function_reader::create_function): Use
3824 build_decl instread of build_decl_stat.
3826 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
3828 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
3829 * config/i386/avx512dqintrin.h: Ditto.
3830 * config/i386/avx512fintrin.h: Ditto.
3831 * config/i386/i386-builtin-types.def: Add new types.
3832 * gcc/config/i386/i386.c: Handle new types.
3833 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
3834 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
3835 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
3836 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
3837 (__builtin_ia32_kshiftridi): New.
3838 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
3840 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
3844 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
3845 define to rs6000_init_stack_protect_guard.
3846 (rs6000_init_stack_protect_guard): New function.
3848 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
3849 Yunqiang Su <yunqiang.su@imgtec.com>
3851 * config.gcc (supported_defaults): Add madd4.
3852 (with_madd4): Add validation.
3853 (all_defaults): Add madd4.
3854 * config/mips/mips.opt (mmadd4): New option.
3855 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
3857 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
3859 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
3860 (ISA_HAS_FUSED_MADD4): Likewise.
3861 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
3862 * gcc/doc/install.texi (--with-madd4): Document the new option.
3864 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3866 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
3867 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
3868 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
3869 (aarch64_init_pauth_hint_builtins): New.
3870 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
3871 (aarch64_expand_builtin): Expand new builtins.
3873 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3875 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
3876 * combine-stack-adj.c (no_unhandled_cfa): Handle
3877 REG_CFA_TOGGLE_RA_MANGLE.
3878 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
3879 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
3880 info for return address signing.
3881 (aarch64_expand_epilogue): Likewise.
3883 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3885 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
3886 * config/aarch64/aarch64-protos.h
3887 (aarch64_return_address_signing_enabled): New declaration.
3888 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
3890 (aarch64_expand_prologue): Sign return address before it's pushed onto
3892 (aarch64_expand_epilogue): Authenticate return address fetched from
3894 (aarch64_override_options): Sanity check for ILP32 and ISA level.
3895 (aarch64_attributes): New function attributes for "sign-return-address".
3896 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
3897 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
3898 ("*do_return"): Generate combined instructions according to key index.
3899 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
3900 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
3902 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
3903 * config/aarch64/aarch64.opt (msign-return-address=): New.
3904 * doc/extend.texi (AArch64 Function Attributes): Documents
3905 "sign-return-address=".
3906 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
3908 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
3910 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
3911 overall option summary.
3913 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3915 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
3916 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
3917 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
3918 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
3920 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
3922 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
3923 -mpower9-minmax by default for -mcpu=power9.
3924 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
3925 128-bit floating point.
3927 2017-01-20 Alan Modra <amodra@gmail.com>
3929 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
3930 optimizing for size.
3932 2017-01-20 Alan Modra <amodra@gmail.com>
3935 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
3936 for strcmp and strncmp from corresponding builtin decl.
3938 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
3940 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
3941 instead of i386/rtems-64.h.
3942 * config/i386/rtems-64.h: Remove.
3944 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
3948 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
3950 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
3952 2017-01-19 Tamar Christina <tamar.christina@arm.com>
3954 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3955 Change int to HOST_WIDE_INT.
3956 * config/aarch64/aarch64-protos.h
3957 (aarch64_simd_gen_const_vector_dup): Likewise.
3958 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
3960 2017-01-19 David Malcolm <dmalcolm@redhat.com>
3962 * langhooks-def.h (lhd_type_for_size): New decl.
3963 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
3964 * langhooks.c (lhd_type_for_size): New function, taken from
3967 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
3969 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
3970 define_bypass for CR latency.
3971 (power9-cracked-alu): Update bypass latency and remove power9-branch.
3972 (power9-alu2): Add define_bypass for CR latency.
3974 (power9-mul): Update insn latency.
3975 (power9-mul-compare): Update insn latency, bypass latency and remove
3978 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3980 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
3982 * config/aarch64/aarch64.md
3983 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
3984 aarch64_nopcrelative_literal_loads.
3985 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
3987 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
3989 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
3991 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
3993 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
3996 * config.gcc (supported_defaults): Add lxc1-sxc1.
3997 (with_lxc1_sxc1): Add validation.
3998 (all_defaults): Add lxc1-sxc1.
3999 * config/mips/mips.opt (mlxc1-sxc1): New option.
4000 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
4002 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
4003 __mips_no_lxc1_sxc1.
4004 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
4005 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
4006 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
4008 2017-01-19 Richard Biener <rguenther@suse.de>
4010 PR tree-optimization/72488
4011 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
4012 sure to restore SSA info.
4013 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
4015 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
4017 PR rtl-optimization/79121
4018 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
4019 of the inner type when shifting an extended value.
4021 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
4024 * symtab.c (symtab_node::equal_address_to): Fix comparing of
4025 interposable aliases.
4027 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
4030 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
4031 Use the evmergelohi instruction.
4032 (mov_si<mode>_e500_subreg4_2_le): Likewise.
4033 (mov_sitf_e500_subreg8_2_be): Likewise.
4034 (mov_sitf_e500_subreg12_2_le): Likewise.
4035 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
4036 (mov_si<mode>_e500_subreg4_2_be): Likewise.
4037 (mov_sitf_e500_subreg8_2_le): Likewise.
4038 (mov_sitf_e500_subreg12_2_be): Likewise.
4040 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4042 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
4043 attribute from vecsimple to vecperm.
4044 (altivec_vbpermq2): Likewise.
4046 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4049 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
4051 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4052 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
4053 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
4054 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
4055 case where N arg is SIZE_MAX.
4056 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
4057 (cmpstrsi): Add pattern.
4059 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
4061 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4062 __builtin_vec_revb builtins.
4063 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
4064 built-in functions to support generation of the ISA 3.0 XXBR<x>
4065 vector byte reverse instructions.
4066 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
4067 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
4068 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
4069 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
4070 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
4071 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
4072 (P9V_BUILTIN_VEC_REVB): Likewise.
4073 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
4074 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
4075 (p9_xxbrq_v16qi): Likewise.
4076 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
4077 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
4078 (p9_xxbrh_v8hi): Likewise.
4079 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
4080 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
4081 vec_revb built-in functions.
4083 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
4085 PR rtl-optimization/78952
4086 * config/i386/i386.md (any_extract): New code iterator.
4087 (*insvqi_2): Use any_extract for source operand.
4088 (*insvqi_3): Use any_shiftrt for source operand.
4090 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
4092 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
4094 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
4096 2017-01-18 Matthias Klose <doko@ubuntu.com>
4098 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
4100 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4102 * config/rs6000/altivec.h (vec_bperm): Change #define.
4103 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
4104 (altivec_vbpermq2): New define_insn.
4105 (altivec_vbpermd): Likewise.
4106 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
4108 (VBPERMD): Likewise.
4109 (VBPERM): New polymorphic function interface.
4110 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
4111 Add entries for P9V_BUILTIN_VEC_VBPERM.
4112 * doc/extend.texi: Add interfaces for vec_bperm.
4114 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4116 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
4117 first letter of error messages.
4118 (s390_resolve_overloaded_builtin): Likewise.
4119 * config/s390/s390.c (s390_expand_builtin): Likewise.
4120 (s390_invalid_arg_for_unprototyped_fn): Likewise.
4121 (s390_valid_target_attribute_inner_p): Likewise.
4122 * config/s390/s390.md ("tabort"): Likewise.
4124 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
4126 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
4127 (ISA_AVOID_DIV_HILO): New macro.
4128 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
4129 (ISA_HAS_DDIV): Likewise.
4131 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4133 * doc/invoke.texi (fabi-version): Correct number of occurrences.
4135 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4137 * doc/invoke.texi (fabi-version): Spelling fix.
4139 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4142 * doc/invoke.texi (fabi-version): Mention mangling fix for
4145 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4148 * doc/invoke.texi (fabi-version): Document discriminator mangling.
4150 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
4153 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
4154 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
4156 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
4158 (stack_protect_test): Ditto.
4159 * config/rs6000/rs6000.opt (mstack-protector-guard=,
4160 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
4162 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
4163 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
4164 -mstack-protector-guard-offset=.
4165 (RS/6000 and PowerPC Options): Ditto.
4167 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
4169 * config/i386/i386.h (MASK_CLASS_P): New define.
4170 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
4171 there are no registers from different register sets also when
4172 mask registers are used. Update function comment.
4173 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
4174 to (*k/*r) and (*k/*km) alternatives.
4176 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
4178 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
4179 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
4180 (EH_RETURN_HANDLER_RTX): New define.
4181 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
4182 Force frame pointer in EH return functions.
4183 (aarch64_expand_epilogue): Add barrier for eh_return.
4184 (aarch64_final_eh_return_addr): Remove.
4185 (aarch64_eh_return_handler_rtx): New function.
4186 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
4188 (aarch64_eh_return_handler_rtx): New prototype.
4190 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4192 * config/rs6000/altivec.h (vec_rlmi): New #define.
4193 (vec_vrlnm): Likewise.
4194 (vec_rlnm): Likewise.
4195 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
4196 (UNSPEC_VRLNM): Likewise.
4197 (VIlong): New mode iterator.
4198 (altivec_vrl<VI_char>mi): New define_insn.
4199 (altivec_vrl<VI_char>nm): Likewise.
4200 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
4203 (RLNM): New polymorphic function entry.
4204 (VRLWMI): New monomorphic function entry.
4206 (RLMI): New polymorphic function entry.
4207 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
4208 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
4209 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
4212 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4215 * dwarf2out.c (field_byte_offset): Restore the
4216 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
4217 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
4218 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
4221 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
4224 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
4226 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4229 * dwarf2out.c (add_data_member_location_attribute): For constant
4230 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
4231 instead of DW_AT_data_member_location, DW_AT_bit_offset and
4232 DW_AT_byte_size attributes.
4234 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
4236 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
4237 after forcing to constant memory when the code model is medium.
4239 2017-01-17 Julia Koval <julia.koval@intel.com>
4242 * config/i386/avx512fintrin.h
4243 (_mm512_i32gather_ps): Change __addr type to void const*.
4244 (_mm512_mask_i32gather_ps): Ditto.
4245 (_mm512_i32gather_pd): Ditto.
4246 (_mm512_mask_i32gather_pd): Ditto.
4247 (_mm512_i64gather_ps): Ditto.
4248 (_mm512_mask_i64gather_ps): Ditto.
4249 (_mm512_i64gather_pd): Ditto.
4250 (_mm512_mask_i64gather_pd): Ditto.
4251 (_mm512_i32gather_epi32): Ditto.
4252 (_mm512_mask_i32gather_epi32): Ditto.
4253 (_mm512_i32gather_epi64): Ditto.
4254 (_mm512_mask_i32gather_epi64): Ditto.
4255 (_mm512_i64gather_epi32): Ditto.
4256 (_mm512_mask_i64gather_epi32): Ditto.
4257 (_mm512_i64gather_epi64): Ditto.
4258 (_mm512_mask_i64gather_epi64): Ditto.
4259 (_mm512_i32scatter_ps): Change __addr type to void*.
4260 (_mm512_mask_i32scatter_ps): Ditto.
4261 (_mm512_i32scatter_pd): Ditto.
4262 (_mm512_mask_i32scatter_pd): Ditto.
4263 (_mm512_i64scatter_ps): Ditto.
4264 (_mm512_mask_i64scatter_ps): Ditto.
4265 (_mm512_i64scatter_pd): Ditto.
4266 (_mm512_mask_i64scatter_pd): Ditto.
4267 (_mm512_i32scatter_epi32): Ditto.
4268 (_mm512_mask_i32scatter_epi32): Ditto.
4269 (_mm512_i32scatter_epi64): Ditto.
4270 (_mm512_mask_i32scatter_epi64): Ditto.
4271 (_mm512_i64scatter_epi32): Ditto.
4272 (_mm512_mask_i64scatter_epi32): Ditto.
4273 (_mm512_i64scatter_epi64): Ditto.
4274 (_mm512_mask_i64scatter_epi64): Ditto.
4275 * config/i386/avx512pfintrin.h
4276 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
4277 (_mm512_mask_prefetch_i32gather_ps): Ditto.
4278 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4279 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4280 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
4281 (_mm512_prefetch_i32scatter_ps): Ditto.
4282 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4283 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4284 (_mm512_prefetch_i64scatter_pd): Ditto.
4285 (_mm512_prefetch_i64scatter_ps): Ditto.
4286 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4287 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4288 * config/i386/avx512vlintrin.h
4289 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
4290 (_mm_mmask_i32gather_ps): Ditto.
4291 (_mm256_mmask_i32gather_pd): Ditto.
4292 (_mm_mmask_i32gather_pd): Ditto.
4293 (_mm256_mmask_i64gather_ps): Ditto.
4294 (_mm_mmask_i64gather_ps): Ditto.
4295 (_mm256_mmask_i64gather_pd): Ditto.
4296 (_mm_mmask_i64gather_pd): Ditto.
4297 (_mm256_mmask_i32gather_epi32): Ditto.
4298 (_mm_mmask_i32gather_epi32): Ditto.
4299 (_mm256_mmask_i32gather_epi64): Ditto.
4300 (_mm_mmask_i32gather_epi64): Ditto.
4301 (_mm256_mmask_i64gather_epi32): Ditto.
4302 (_mm_mmask_i64gather_epi32): Ditto.
4303 (_mm256_mmask_i64gather_epi64): Ditto.
4304 (_mm_mmask_i64gather_epi64): Ditto.
4305 (_mm256_i32scatter_ps): Change __addr type to void*.
4306 (_mm256_mask_i32scatter_ps): Ditto.
4307 (_mm_i32scatter_ps): Ditto.
4308 (_mm_mask_i32scatter_ps): Ditto.
4309 (_mm256_i32scatter_pd): Ditto.
4310 (_mm256_mask_i32scatter_pd): Ditto.
4311 (_mm_i32scatter_pd): Ditto.
4312 (_mm_mask_i32scatter_pd): Ditto.
4313 (_mm256_i64scatter_ps): Ditto.
4314 (_mm256_mask_i64scatter_ps): Ditto.
4315 (_mm_i64scatter_ps): Ditto.
4316 (_mm_mask_i64scatter_ps): Ditto.
4317 (_mm256_i64scatter_pd): Ditto.
4318 (_mm256_mask_i64scatter_pd): Ditto.
4319 (_mm_i64scatter_pd): Ditto.
4320 (_mm_mask_i64scatter_pd): Ditto.
4321 (_mm256_i32scatter_epi32): Ditto.
4322 (_mm256_mask_i32scatter_epi32): Ditto.
4323 (_mm_i32scatter_epi32): Ditto.
4324 (_mm_mask_i32scatter_epi32): Ditto.
4325 (_mm256_i32scatter_epi64): Ditto.
4326 (_mm256_mask_i32scatter_epi64): Ditto.
4327 (_mm_i32scatter_epi64): Ditto.
4328 (_mm_mask_i32scatter_epi64): Ditto.
4329 (_mm256_i64scatter_epi32): Ditto.
4330 (_mm256_mask_i64scatter_epi32): Ditto.
4331 (_mm_i64scatter_epi32): Ditto.
4332 (_mm_mask_i64scatter_epi32): Ditto.
4333 (_mm256_i64scatter_epi64): Ditto.
4334 (_mm256_mask_i64scatter_epi64): Ditto.
4335 (_mm_i64scatter_epi64): Ditto.
4336 (_mm_mask_i64scatter_epi64): Ditto.
4337 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
4338 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
4339 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
4340 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
4341 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
4342 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
4343 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
4344 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
4345 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
4346 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
4347 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
4348 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
4349 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
4350 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
4351 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
4352 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
4353 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
4354 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
4355 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
4356 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
4357 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
4358 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
4359 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
4360 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
4361 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
4362 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
4363 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
4364 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
4365 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
4366 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
4367 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
4368 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
4369 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
4370 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
4371 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
4372 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
4373 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
4374 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
4375 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
4376 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
4377 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
4378 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
4379 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
4380 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
4381 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
4382 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
4383 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
4384 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
4385 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
4386 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
4387 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
4388 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
4389 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
4390 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
4391 definitions accordingly.
4393 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
4394 Kuan-Lin Chen <kuanlinchentw@gmail.com>
4397 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
4400 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
4403 * ira-conflicts.c (ira_build_conflicts): Update total conflict
4404 hard regs for inner regno.
4406 2017-01-17 Martin Liska <mliska@suse.cz>
4409 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
4410 assumption and add comment.
4412 2017-01-17 Nathan Sidwell <nathan@acm.org>
4414 * ipa-visibility.c (localize_node): New function, broken out of ...
4415 (function_and_variable_visibility): ... here. Call it.
4417 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
4420 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
4421 correctly set frequency of oudgoing edge.
4422 (duplicate_thread_path): Fix profile updating.
4424 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4427 * configure.ac: Add GCC_BASE_VER.
4428 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
4429 version from BASE-VER file.
4430 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
4431 (gcc.o): Depend on $(BASEVER).
4432 * common.opt (dumpfullversion): New option.
4433 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
4434 * doc/invoke.texi: Document -dumpfullversion.
4435 * doc/install.texi: Document --with-gcc-major-version-only.
4436 * configure: Regenerated.
4438 2017-01-17 Richard Biener <rguenther@suse.de>
4440 PR tree-optimization/71433
4441 * tree-vrp.c (register_new_assert_for): Merge same asserts
4442 on all incoming edges.
4443 (process_assert_insertions_for): Handle insertions at the
4446 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
4448 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
4449 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
4451 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
4454 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
4457 2017-01-17 Alan Modra <amodra@gmail.com>
4460 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
4461 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
4462 symbolic stack limit when pic.
4464 2017-01-16 Martin Sebor <msebor@redhat.com>
4466 PR tree-optimization/78608
4467 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
4469 2017-01-16 Jeff Law <law@redhat.com>
4472 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
4473 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
4474 for several include directories that may be relative to sysroot.
4475 * config/i386/x-mingw32 (gplus_includedir): Define.
4476 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
4477 (native_system_includedir): Likewise.
4478 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
4479 override if TARGET_SYSTEM_ROOT is defined.
4480 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
4482 PR tree-optimization/79090
4483 PR tree-optimization/33562
4484 PR tree-optimization/61912
4485 PR tree-optimization/77485
4486 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
4487 and computed trims into the dump file.
4489 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
4491 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
4493 2017-01-16 Jakub Jelinek <jakub@redhat.com>
4496 * gimplify.c (gimplify_init_constructor): If want_value and
4497 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
4501 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
4502 sequence. Formatting fixes.
4503 (doloop_optimize): Formatting fixes.
4506 * gcc.c (debug_level_greater_than_spec_func): New function.
4507 (static_spec_functions): Add debug-level-gt spec function.
4508 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
4510 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
4511 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
4512 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
4513 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
4514 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
4515 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
4517 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
4519 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
4520 QImode fixups to general and mask registers only.
4522 2017-01-16 Carl Love <cel@us.ibm.com>
4524 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
4525 for built-in functions
4526 vector signed char vec_nabs (vector signed char)
4527 vector signed short vec_nabs (vector signed short)
4528 vector signed int vec_nabs (vector signed int)
4529 vector signed long long vec_nabs (vector signed long long)
4530 vector float vec_nabs (vector float)
4531 vector double vec_nabs (vector double)
4532 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
4534 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
4535 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
4536 * doc/extend.texi: Update the documentation file for the new built-in
4539 2017-01-16 Martin Sebor <msebor@redhat.com>
4541 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
4544 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4546 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
4547 UNSPEC_VSX__XXSPLTD to require special splat handling.
4549 2017-01-16 David Malcolm <dmalcolm@redhat.com>
4552 * system.h: Poison strndup.
4554 2017-01-16 Alan Modra <amodra@gmail.com>
4557 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
4560 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
4562 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
4564 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
4566 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
4567 call recog here. Assert that INSN_CODE (insn) is non-negative.
4569 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
4572 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
4574 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
4575 in the currently scheduled RTL fragment.
4577 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
4579 PR rtl-optimization/78751
4580 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
4583 2017-01-14 Jeff Law <law@redhat.com>
4585 PR tree-optimization/79090
4586 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
4587 variable length stores.
4588 (compute_trims): Delete dead assignment to *trim_tail.
4589 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
4592 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
4594 PR rtl-optimization/78626
4595 PR rtl-optimization/78727
4596 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
4597 of a block, and split such blocks after everything else is finished.
4599 2017-01-14 Alan Modra <amodra@gmail.com>
4602 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
4603 target legitimate_combined_insn.
4604 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
4605 (rs6000_legitimate_combined_insn): New function.
4606 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
4608 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
4609 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
4610 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
4612 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
4614 * doc/frontends.texi (G++ and GCC): Remove references to Java.
4616 2017-01-13 Jeff Law <law@redhat.com>
4618 PR tree-optimization/33562
4619 PR tree-optimization/61912
4620 PR tree-optimization/77485
4621 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
4623 (delete_dead_assignment): Likewise.
4624 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
4625 statement to delete_dead_call and delete_dead_assignment.
4627 2017-01-13 David Malcolm <dmalcolm@redhat.com>
4630 * substring-locations.c (format_warning_va): Strengthen case 1 so
4631 that both endpoints of the substring must be within the format
4632 range for just the substring to be printed.
4634 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
4636 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
4637 * config/i386/i386.c (ix86_target_string): Add missing options
4638 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
4639 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
4640 flags_other and ix86_target_other to flags2_other. Display unknown
4642 (ix86_valid_target_attribute_inner_p): Add missing options and
4643 reorder options by implied ISAs, as in ix86_target_string.
4645 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
4647 * hash-table.h (hash_table::too_empty_p): New function.
4648 (hash_table::expand): Use it.
4649 (hash_table::traverse): Likewise.
4650 (hash_table::empty_slot): Use sizeof (value_type) instead of
4651 sizeof (PTR) to convert bytes to elements. Shrink the table
4652 if the current size is excessive for the current number of
4655 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
4657 * ira-costs.c (record_reg_classes): Break from the inner loop
4658 early once alt_fail is known to be true. Update outer loop
4659 handling accordingly.
4661 2017-01-13 Jeff Law <law@redhat.com>
4663 * tree-ssa-dse.c (decrement_count): New function.
4664 (increment_start_addr, maybe_trim_memstar_call): Likewise.
4665 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
4666 when we know the partially dead statement is a mem* function.
4668 PR tree-optimization/61912
4669 PR tree-optimization/77485
4670 * tree-ssa-dse.c: Include expr.h.
4671 (maybe_trim_constructor_store): New function.
4672 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
4674 PR tree-optimization/33562
4675 PR tree-optimization/61912
4676 PR tree-optimization/77485
4677 * doc/invoke.texi: Document new dse-max-object-size param.
4678 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
4679 * tree-ssa-dse.c: Include params.h.
4680 (dse_store_status): New enum.
4681 (initialize_ao_ref_for_dse): New, partially extracted from
4683 (valid_ao_ref_for_dse, normalize_ref): New.
4684 (setup_live_bytes_from_ref, compute_trims): Likewise.
4685 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
4686 (maybe_trim_partially_dead_store): Likewise.
4687 (maybe_trim_complex_store): Likewise.
4688 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
4689 Track what bytes live from the original store. Return tri-state
4690 for dead, partially dead or live.
4691 (dse_dom_walker): Add constructor, destructor and new private members.
4692 (delete_dead_call, delete_dead_assignment): New extracted from
4694 (dse_optimize_stmt): Make a member of dse_dom_walker.
4695 Use initialize_ao_ref_for_dse.
4697 PR tree-optimization/33562
4698 PR tree-optimization/61912
4699 PR tree-optimization/77485
4700 * sbitmap.h (bitmap_count_bits): Prototype.
4701 (bitmap_clear_range, bitmap_set_range): Likewise.
4702 * sbitmap.c (bitmap_clear_range): New function.
4703 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
4705 2017-01-13 Martin Liska <mliska@suse.cz>
4708 * function.c (set_cfun): Add new argument force.
4709 * function.h (set_cfun): Likewise.
4710 * ipa-inline-transform.c (inline_call): Use the function when
4711 strict alising from is dropped for function we inline to.
4713 2017-01-13 Richard Biener <rguenther@suse.de>
4715 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
4716 for dumping GIMPLE INTEGER_CSTs.
4718 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4720 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
4721 to 201112L since C++17.
4723 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
4726 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
4727 if -fsanitize=kernel-address is present.
4729 2017-01-13 Richard Biener <rguenther@suse.de>
4731 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
4732 as _Literal ( type ) number in case usual suffixes do not
4733 preserve all information.
4735 2017-01-13 Richard Biener <rguenther@suse.de>
4737 PR tree-optimization/77283
4738 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
4739 and ssa-iterators.h.
4740 (is_feasible_trace): Implement a cost model based on joiner
4743 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4746 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
4747 char or short to __float128/_Float128 directly.
4749 2017-01-12 Martin Sebor <msebor@redhat.com>
4751 to -Wformat-overflow.
4752 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
4753 (min_bytes_remaining): Same.
4754 (get_string_length): Same.
4755 (format_string): Same.
4756 (format_directive): Same.
4758 (pass_sprintf_length::handle_gimple_call): Same.
4760 2017-01-12 Jakub Jelinek <jakub@redhat.com>
4762 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
4763 info.nowrite calls with no lhs that can't throw. Return bool
4764 whether gsi_remove has been called or not.
4765 (pass_sprintf_length::handle_gimple_call): Return bool whether
4766 try_substitute_return_value called gsi_remove. Formatting fix.
4767 (pass_sprintf_length::execute): Don't use gsi_remove if
4768 handle_gimple_call returned true.
4771 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
4772 be removed due to side-effects, don't remove following barrier nor
4773 turn the successor edge into fallthru edge.
4775 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4778 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
4779 element-reversing loads and stores as not swappable.
4781 2017-01-12 Nathan Sidwell <nathan@acm.org>
4782 Nicolai Stange <nicstange@gmail.com>
4784 * combine.c (try_combine): Don't ignore result of overlap checking
4785 loop. Combine overlap & asm check into single loop.
4787 2017-01-12 Richard Biener <rguenther@suse.de>
4789 * tree-pretty-print.c (dump_generic_node): Provide -gimple
4790 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
4792 2017-01-12 Richard Biener <rguenther@suse.de>
4794 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
4795 and TS_TARGET_OPTION directly derive from TS_BASE.
4796 * tree-core.h (tree_optimization_option): Derive from tree_base.
4797 (tree_target_option): Likewise.
4799 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
4801 * config/i386/i386.c (memory_address_length): Increase len
4802 only when rip_relative_addr_p returns false.
4804 2017-01-11 Julia Koval <julia.koval@intel.com>
4806 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
4807 (OPTION_MASK_ISA_SGX_SET): New.
4808 (ix86_handle_option): Handle OPT_msgx.
4809 * config.gcc: Added sgxintrin.h.
4810 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
4811 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
4812 * config/i386/i386.c (ix86_target_string): Add -msgx.
4814 (ix86_option_override_internal): Handle new options.
4815 (ix86_valid_target_attribute_inner_p): Add sgx.
4816 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
4817 * config/i386/i386.opt: Add msgx.
4818 * config/i386/sgxintrin.h: New file.
4819 * config/i386/x86intrin.h: Add sgxintrin.h.
4821 2017-01-11 Jakub Jelinek <jakub@redhat.com>
4824 * fold-const.c (maybe_nonzero_address): Return 1 for function
4826 (tree_single_nonzero_warnv_p): Don't handle function local objects
4830 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
4833 2017-01-11 David Malcolm <dmalcolm@redhat.com>
4836 * opts.c: Include "spellcheck.h"
4837 (struct string_fragment): New struct.
4838 (struct edit_distance_traits<const string_fragment &>): New
4840 (get_closest_sanitizer_option): New function.
4841 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
4843 2017-01-11 Jakub Jelinek <jakub@redhat.com>
4845 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
4847 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
4848 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
4849 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
4850 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
4851 for initial die_offset if dwarf_split_debug_info.
4852 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
4853 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
4855 (output_skeleton_debug_sections): Formatting fix. Use
4856 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
4857 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
4859 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
4861 * config/arm/cortex-a53.md: Add bypasses for
4863 (cortex_a53_r2f): Only use for transfers.
4864 (cortex_a53_f2r): Likewise.
4865 (cortex_a53_r2f_cvt): Add reservation for conversions.
4866 (cortex_a53_f2r_cvt): Likewise.
4868 2017-01-11 Tamar Christina <tamar.christina@arm.com>
4870 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
4871 to all inlined functions, change static to extern.
4873 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
4876 * config/arm/arm.c (legitimize_pic_address): Handle reference to
4878 (arm_assemble_integer): Likewise.
4880 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
4882 * config.gcc: Use new awk script to check CPU, FPU and architecture
4883 parameters for --with-... options.
4884 * config/arm/parsecpu.awk: New file
4885 * config/arm/arm-cpus.in: New file.
4886 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
4888 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
4890 * config/arm/t-arm: Update dependency rules.
4891 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
4892 of processing .def files.
4893 * config/arm/genopt.sh: Deleted.
4894 * config/arm/gentune.sh: Deleted.
4895 * config/arm/arm-cores.def: Deleted.
4896 * config/arm/arm-arches.def: Deleted.
4897 * config/arm/arm-fpus.def: Deleted.
4898 * config/arm/arm-tune.md: Regenerated.
4899 * config/arm/arm-tables.opt: Regenerated.
4900 * config/arm/arm-cpu.h: New generated file.
4901 * config/arm/arm-cpu-data.h: New generated file.
4902 * config/arm/arm-cpu-cdata.h: New generated file.
4904 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
4907 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
4909 (input_varpool_node): Unpack dynamically_initialized bit.
4911 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
4913 PR rtl-optimization/79032
4914 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
4915 the alignment of the adjusted memory reference against that of MODE,
4916 instead of the alignment of the original memory reference.
4918 2017-01-11 Martin Jambor <mjambor@suse.cz>
4920 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
4922 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
4923 decorated functions.
4925 2017-01-11 Richard Biener <rguenther@suse.de>
4927 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
4928 set range/nonnull info for PHI results. Do not set it on
4929 stmts marked for removal.
4931 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
4933 * expr.c (store_field): In the bitfield case, fetch the return value
4934 from the registers before applying a single big-endian adjustment.
4935 Always do a final load for a BLKmode value not larger than a word.
4937 2017-01-10 David Malcolm <dmalcolm@redhat.com>
4940 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
4941 that we correctly handle column numbers greater than
4942 LINE_MAP_MAX_COLUMN_NUMBER.
4944 2017-01-10 Martin Sebor <msebor@redhat.com>
4947 * gimple-ssa-sprintf.c (get_destination_size): Call
4948 {init,fini}object_sizes.
4949 * tree-object-size.c (addr_object_size): Adjust.
4950 (pass_through_call): Adjust.
4951 (pass_object_sizes::execute): Adjust.
4952 * tree-object-size.h (fini_object_sizes): Declare.
4954 2017-01-10 Martin Sebor <msebor@redhat.com>
4956 PR tree-optimization/78775
4957 * builtins.c (get_size_range): Move...
4958 * calls.c: ...to here.
4959 (alloc_max_size): Accept zero argument.
4960 (operand_signed_p): Remove.
4961 (maybe_warn_alloc_args_overflow): Call get_size_range.
4962 * calls.h (get_size_range): Declare.
4964 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
4966 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
4967 from TI's devices.csv file as of September 2016.
4968 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
4970 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
4972 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
4973 * doc/invoke.texi: Likewise.
4974 * doc/md.texi: Likewise.
4975 * doc/objc.texi: Likewise.
4977 2017-01-10 Joshua Conner <joshconner@google.com>
4979 * config/arm/fuchsia-elf.h: New file.
4980 * config/fuchsia.h: New file.
4981 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
4982 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
4984 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
4986 2016-01-10 Richard Biener <rguenther@suse.de>
4988 PR tree-optimization/79034
4989 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
4990 Propagate out degenerate PHIs in the joiner.
4992 2017-01-10 Martin Liska <mliska@suse.cz>
4994 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
4995 (sort_congruence_classes_by_decl_uid): Likewise.
4996 (sort_congruence_class_groups_by_decl_uid): Likewise.
4997 (sem_item_optimizer::merge_classes): Sort class, groups in these
4998 classes and members in the groups by DECL_UID of declarations.
4999 This would make merge operations stable.
5001 2017-01-10 Martin Liska <mliska@suse.cz>
5003 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
5004 usage of m_classes_vec.
5005 (sem_item_optimizer::~sem_item_optimizer): Likewise.
5006 (sem_item_optimizer::get_group_by_hash): Likewise.
5007 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
5008 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
5009 (sem_item_optimizer::verify_classes): Likewise.
5010 (sem_item_optimizer::process_cong_reduction): Likewise.
5011 (sem_item_optimizer::dump_cong_classes): Likewise.
5012 (sem_item_optimizer::merge_classes): Likewise.
5013 * ipa-icf.h (congruence_class_hash): Rename from
5014 congruence_class_group_hash. Remove declaration of m_classes_vec.
5016 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
5018 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
5019 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
5020 * config.gcc: Add avx512vpopcntdqintrin.h.
5021 * config/i386/avx512vpopcntdqintrin.h: New.
5022 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
5023 * config/i386/i386-builtin-types.def: Add new types.
5024 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
5025 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
5026 __builtin_ia32_vpopcountq_v8di_mask): New.
5027 * config/i386/i386-c.c (ix86_target_macros_internal): Define
5028 __AVX512VPOPCNTDQ__.
5029 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
5030 (PTA_AVX512VPOPCNTDQ): Define.
5031 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
5032 TARGET_AVX512VPOPCNTDQ_P): Define.
5033 * config/i386/i386.opt: Add mavx512vpopcntdq.
5034 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
5035 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
5037 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5040 * predict.def (PRED_CALL): Set to 67.
5042 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
5044 * expr.c (store_field): In the bitfield case, if the value comes from
5045 a function call and is of an aggregate type returned in registers, do
5046 not modify the field mode; extract the value in all cases if the mode
5047 is BLKmode and the size is not larger than a word.
5049 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
5052 * config/i386/cpuid.h: Fix undefined behavior.
5054 2017-01-04 Jeff Law <law@redhat.com>
5056 PR tree-optimization/79007
5057 PR tree-optimization/67955
5058 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
5059 conservative for pt.null when flag_non_call_exceptions is on.
5061 2017-01-09 Jakub Jelinek <jakub@redhat.com>
5063 PR translation/79019
5064 PR translation/79020
5065 * params.def (PARAM_INLINE_MIN_SPEEDUP,
5066 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
5067 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
5069 * config/avr/avr.opt (maccumulate-args): Likewise.
5070 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
5071 * common.opt (freport-bug): Likewise.
5072 * cif-code.def (CIF_FINAL_ERROR): Likewise.
5073 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
5074 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
5075 translatable string.
5076 * config/i386/i386.c (function_value_32): Likewise.
5077 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
5078 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
5080 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
5081 * common/config/msp430/msp430-common.c (msp430_handle_option):
5083 * symtab.c (symtab_node::verify_base): Likewise.
5084 * opts.c (set_debug_level): Likewise.
5085 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
5086 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
5087 missing whitespace to translatable strings.
5088 * config/avr/avr.md (bswapsi2): Fix typo in comment.
5089 * config/sh/superh.h: Likewise.
5090 * config/i386/xopintrin.h: Likewise.
5091 * config/i386/znver1.md: Likewise.
5092 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
5093 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
5094 * double-int.h (struct double_int): Likewise.
5095 * double-int.c (div_and_round_double): Likewise.
5096 * wide-int.cc: Likewise.
5097 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
5098 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
5099 * cfgcleanup.c (crossjumps_occured): Renamed to ...
5100 (crossjumps_occurred): ... this.
5101 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
5104 PR tree-optimization/78899
5105 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
5106 returning bool return struct loop *, NULL for failure and the new
5108 (versionable_outer_loop_p): Don't version outer loop if it has
5109 dont_vectorized bit set.
5110 (tree_if_conversion): When versioning outer loop, ensure
5111 tree_if_conversion is performed also on the inner loop of the
5112 non-vectorizable outer loop copy.
5113 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
5114 LOOP_VECTORIZED in inner loop of the scalar outer loop and
5115 prevent vectorization of it.
5116 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
5117 the outer loop vectorization of the non-scalar version is attempted
5118 before vectorization of the inner loop in scalar version. If
5119 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
5120 vectorization of its inner loop.
5121 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
5122 has 2 inner loops, rename also on edges from bb whose single pred
5123 is outer_loop->header. Fix typo in function comment.
5125 2017-01-09 Martin Sebor <msebor@redhat.com>
5128 * asan.c (asan_emit_stack_protection): Increase local buffer size
5129 to avoid snprintf truncation warning.
5131 2017-01-09 Andrew Pinski <apinski@cavium.com>
5133 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
5134 to reference thunderx2t99 for the tuning structure
5135 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
5137 (thunderx2t99_extra_costs): This.
5138 * config/aarch64/aarch64-tune.md: Regenerate.
5139 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
5140 (vulcan_addrcost_table): This.
5141 (vulcan_regmove_cost): Rename to ...
5142 (thunderx2t99_regmove_cost): This.
5143 (vulcan_vector_cost): Rename to ...
5144 (thunderx2t99_vector_cost): this.
5145 (vulcan_branch_cost): Rename to ...
5146 (thunderx2t99_branch_cost): This.
5147 (vulcan_tunings): Rename to ...
5148 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
5149 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
5151 2017-01-09 Martin Jambor <mjambor@suse.cz>
5155 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
5156 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
5157 (propagate_vr_accross_jump_function): Use the above function for all
5158 value range computations for pass-through jump functions and type
5159 converasion from explicit value range values.
5160 (ipcp_propagate_stage): Do not attempt to deduce types of formal
5161 parameters from TYPE_ARG_TYPES.
5162 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
5163 (ipa_write_node_info): Stream type of the actual argument.
5164 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
5166 2017-01-09 Martin Liska <mliska@suse.cz>
5169 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
5170 (lookup_compiler): Do not show error message with have_E.
5172 2017-01-09 Jakub Jelinek <jakub@redhat.com>
5174 PR tree-optimization/78938
5175 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
5176 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
5177 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
5178 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
5181 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5183 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
5186 2017-01-09 Richard Biener <rguenther@suse.de>
5188 PR tree-optimization/78997
5189 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
5190 name condition properly.
5192 2017-01-09 Richard Biener <rguenther@suse.de>
5195 * dwarf2out.c (is_cxx): New overload with context.
5196 (is_naming_typedef_decl): Use it.
5198 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5200 * invoke.texi (Option Summary): Correct spacing in option lists
5201 and add line breaks to fix over-long lines.
5203 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5207 * extend.texi (Common Variable Attributes): Add xref to GCC
5208 Internals manual to explain mode attribute keywords.
5210 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5213 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
5214 and Preprocessor Options.
5215 (Options for Linking): Document -pthread here....
5216 (RS/6000 and PowerPC Options): ...not here.
5217 (Solaris 2 Options): ...or here.
5218 * doc/cppopts.texi: Document -pthread.
5220 2017-01-08 Martin Sebor <msebor@redhat.com>
5223 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
5224 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
5225 New member functions.
5226 (format_directive): Used them.
5228 (pass_sprintf_length::handle_gimple_call): Same.
5229 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
5230 to avoid truncation for any argument.
5231 (extract_affine_mul): Same.
5232 * tree.c (get_file_function_name): Same.
5234 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5237 * predict.def (PRED_INDIR_CALL): Set to 86.
5239 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5241 PR preprocessor/54124
5242 * doc/cppopts.texi: Reformat -d subtable to list the full name
5243 of the options. Add cross-reference to the docs for the general
5244 compiler -d options.
5245 * doc/invoke.texi (Developer Options): Add cross-reference to the
5246 preprocessor-specific -d option documentation.
5248 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5250 PR preprocessor/13498
5251 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
5252 redudant material, and reflect new command-line options.
5253 (System Headers): Likewise.
5255 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5257 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
5258 -isystem, and -idirafter. Copy-edit.
5259 * doc/cppopts.texi: Copy-edit. Remove contradiction about
5260 default for -ftrack-macro-expansion. Delete obsolete and
5261 badly-formatted implementation details about -fdebug-cpp output.
5262 * doc/cppwarnopts.texi: Copy-edit.
5264 2017-01-07 David Malcolm <dmalcolm@redhat.com>
5267 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
5268 that the transition from a max line width >= 1<<10 to narrower
5269 lines works correctly.
5271 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
5273 * doc/options.texi (PerFunction): New.
5274 * opt-functions.awk (switch_flags): Map both Optimization and
5275 PerFunction to CL_OPTIMIZATION.
5276 * opth-gen.awk: Test for PerFunction flag along with
5278 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
5279 it only when the latter is present. Skip those that don't in
5280 the hash function generator.
5281 * common.opt (fvar-tracking): Mark as PerFunction instead of
5283 (fvar-tracking-assignments): Likewise.
5284 (fvar-tracking-assignments-toggle): Likewise.
5285 (fvar-tracking-uninit): Likewise.
5287 2017-01-07 Jakub Jelinek <jakub@redhat.com>
5289 PR translation/79018
5290 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
5293 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
5296 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
5297 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
5298 TARGET_LONG_JUMP_TABLE_OFFSETS.
5299 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
5300 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
5301 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
5302 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
5303 * config/m68k/m68k.md (tablejump expander): Likewise.
5304 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
5305 TARGET_LONG_JUMP_TABLE_OFFSETS.
5306 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
5307 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
5309 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
5310 David Holsgrove <david.holsgrove@xilinx.com>
5312 * common/config/microblaze/microblaze-common.c
5313 (TARGET_EXCEPT_UNWIND_INFO): Remove.
5314 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
5316 * config/microblaze/microblaze.c (microblaze_must_save_register)
5317 (microblaze_expand_epilogue, microblaze_return_addr): Handle
5319 (microblaze_eh_return): New function.
5320 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
5321 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
5322 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
5323 * config/microblaze/microblaze.md (eh_return): New pattern.
5325 2017-01-06 Jakub Jelinek <jakub@redhat.com>
5327 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
5328 GCC_DIAGNOSTIC_STRINGIFY): Define.
5330 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
5332 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5334 * config/arm/arm.md (<mcrr>): New.
5336 * config/arm/arm.c (arm_arch5te): New.
5337 (arm_option_override): Set arm_arch5te.
5338 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
5340 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
5341 (arm_mcrr_qualifiers): ... this. New.
5342 (MRRC_QUALIFIERS): Define to...
5343 (arm_mrrc_qualifiers): ... this. New.
5344 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
5346 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
5347 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
5348 (MRRCI, mrrc, MRRC): New.
5349 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
5350 VUNSPEC_MRRC2): New.
5352 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5354 * config/arm/arm.md (<mcr>): New.
5356 * config/arm/arm.c (arm_coproc_builtin_available): Add
5357 support for mcr, mrc, mcr2 and mrc2.
5358 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
5359 (arm_mcr_qualifiers): ... this. New.
5360 (MRC_QUALIFIERS): Define to ...
5361 (arm_mrc_qualifiers): ... this. New.
5362 (MCR_QUALIFIERS): Define to ...
5363 (arm_mcr_qualifiers): ... this. New.
5364 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
5366 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
5367 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
5368 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
5371 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5373 * config/arm/arm.md (*ldc): New.
5377 * config/arm/arm.c (arm_coproc_builtin_available): Add
5378 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
5379 (arm_coproc_ldc_stc_legitimate_address): New.
5380 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
5381 'qualifier_const_pointer'.
5382 (LDC_QUALIFIERS): Define to...
5383 (arm_ldc_qualifiers): ... this. New.
5384 (STC_QUALIFIERS): Define to...
5385 (arm_stc_qualifiers): ... this. New.
5386 * config/arm/arm-protos.h
5387 (arm_coproc_ldc_stc_legitimate_address): New.
5388 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
5389 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
5390 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
5391 stc2, stcl, stc2l): New.
5392 * config/arm/constraints.md (Uz): New.
5393 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
5394 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
5395 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
5396 VUNSPEC_STC2L): New.
5398 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5400 * config/arm/arm.md (<cdp>): New.
5401 * config/arm/arm.c (neon_const_bounds): Rename this ...
5402 (arm_const_bounds): ... this.
5403 (arm_coproc_builtin_available): New.
5404 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
5405 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
5406 (CDP_QUALIFIERS): Define to...
5407 (arm_cdp_qualifiers): ... this. New.
5409 (arm_expand_builtin_args): Add case for 6 arguments.
5410 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
5411 (arm_const_bounds): ... this.
5412 (arm_coproc_builtin_available): New.
5413 * config/arm/arm_acle.h (__arm_cdp): New.
5415 * config/arm/arm_acle_builtins.def (cdp): New.
5417 * config/arm/iterators.md (CDPI,CDP,cdp): New.
5418 * config/arm/neon.md: Rename all 'neon_const_bounds' to
5420 * config/arm/types.md (coproc): New.
5421 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
5422 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
5423 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
5424 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
5426 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5428 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
5429 (UBINOP_QUALIFIERS): New.
5431 (acle_builtin_data): New. Change comment.
5432 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
5433 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
5434 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
5435 arm_acle_builtins.def.
5436 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
5437 (arm_init_acle_builtins): New.
5438 (CRC32_BUILTIN): Remove.
5439 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
5440 crc32cb, crc32ch and crc32cw.
5441 (arm_init_crc32_builtins): Remove.
5442 (arm_init_builtins): Use arm_init_acle_builtins rather
5443 than arm_init_crc32_builtins.
5444 (arm_expand_acle_builtin): New.
5445 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
5446 * config/arm/arm_acle_builtins.def: New.
5448 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5450 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
5451 (arm_builtin_datum): ... this.
5452 (arm_init_neon_builtin): Rename to ...
5453 (arm_init_builtin): ... this. Add a new parameters PREFIX
5454 and USE_SIG_IN_NAME.
5455 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
5456 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
5457 'arm_builtin_datum'.
5458 (arm_init_vfp_builtins): Likewise.
5459 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
5460 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
5461 (arm_expand_neon_args): Rename to ...
5462 (arm_expand_builtin_args): ... this. Rename builtin_arg
5463 enum values and differentiate between ARG_BUILTIN_MEMORY
5464 and ARG_BUILTIN_NEON_MEMORY.
5465 (arm_expand_neon_builtin_1): Rename to ...
5466 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
5467 values, arm_expand_builtin_args and add bool parameter NEON.
5468 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
5469 (arm_expand_vfp_builtin): Likewise.
5470 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
5472 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5475 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
5476 * predict.c (tree_estimate_probability_bb): Reverse direction of
5477 polymorphic call predictor.
5479 2017-01-06 David Malcolm <dmalcolm@redhat.com>
5481 * passes.c (execute_one_pass): Split out pass-skipping logic into...
5482 (determine_pass_name_match): ...this new function and...
5483 (should_skip_pass_p): ...this new function.
5485 2017-01-06 Nathan Sidwell <nathan@acm.org>
5487 * ipa-visibility.c (function_and_variable_visibility): Reformat
5488 comments and long lines. Remove extrneous if.
5489 * symtab.c (symtab_node::make_decl_local): Fix code format.
5490 (symtab_node::set_section_for_node): Fix comment typo.
5492 2017-01-06 Martin Liska <mliska@suse.cz>
5495 * lra-constraints.c: Rename invariant to lra_invariant.
5496 * predict.c (set_even_probabilities): Initialize e to NULL.
5498 2017-01-05 Martin Sebor <msebor@redhat.com>
5500 PR tree-optimization/78910
5501 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
5502 (format_integer): Correct off-by-one error in the handling
5503 of precision with negative numbers in signed conversions..
5505 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
5507 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
5509 2017-01-05 Jakub Jelinek <jakub@redhat.com>
5511 PR tree-optimization/71016
5512 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
5513 factor_out_conditional_conversion. Formatting fix.
5514 (factor_out_conditional_conversion): Add cond_stmt argument.
5515 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
5516 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
5519 2017-01-05 David Malcolm <dmalcolm@redhat.com>
5521 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
5522 read-rtl-function.o, and selftest-rtl.o.
5523 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
5524 (selftest::aarch64_test_loading_full_dump): New function.
5525 (selftest::aarch64_run_selftests): New function.
5526 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
5527 selftest::aarch64_run_selftests.
5528 * config/i386/i386.c
5529 (selftest::ix86_test_loading_dump_fragment_1): New function.
5530 (selftest::ix86_test_loading_call_insn): New function.
5531 (selftest::ix86_test_loading_full_dump): New function.
5532 (selftest::ix86_test_loading_unspec): New function.
5533 (selftest::ix86_run_selftests): Call the new functions.
5534 * emit-rtl.c (maybe_set_max_label_num): New function.
5535 * emit-rtl.h (maybe_set_max_label_num): New decl.
5536 * function.c (instantiate_decls): Guard call to
5537 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
5538 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
5540 * gensupport.c (gen_reader::gen_reader): Pass "false"
5541 for new "compact" param of rtx_reader.
5542 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
5543 rather than an empty string for NULL strings.
5544 * read-md.c: Potentially include config.h rather than bconfig.h.
5545 Wrap include of errors.h with #ifdef GENERATOR_FILE.
5546 (have_error): New global, copied from errors.c.
5547 (md_reader::read_name): Rename to...
5548 (md_reader::read_name_1): ...this, adding "out_loc" param,
5549 and converting "missing name or number" to returning false, rather
5551 (md_reader::read_name): Reimplement in terms of read_name_1.
5552 (md_reader::read_name_or_nil): New function.
5553 (md_reader::read_string): Handle "(nil)" by returning NULL.
5554 (md_reader::md_reader): Add new param "compact".
5555 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
5556 (md_reader::read_file): New method.
5557 * read-md.h (md_reader::md_reader): Add new param "compact".
5558 (md_reader::read_file): New method.
5559 (md_reader::is_compact): New accessor.
5560 (md_reader::read_name): Convert return type from void to file_location.
5561 (md_reader::read_name_or_nil): New decl.
5562 (md_reader::read_name_1): New decl.
5563 (md_reader::m_compact): New field.
5564 (noop_reader::noop_reader): Pass "false" for new "compact" param
5566 (rtx_reader::rtx_reader): Add new "compact" param.
5567 (rtx_reader::read_rtx_operand): Make virtual and convert return
5568 type from void to rtx.
5569 (rtx_reader::read_until): New decl.
5570 (rtx_reader::handle_any_trailing_information): New virtual function.
5571 (rtx_reader::postprocess): New virtual function.
5572 (rtx_reader::finalize_string): New virtual function.
5573 (rtx_reader::m_in_call_function_usage): New field.
5574 (rtx_reader::m_reuse_rtx_by_id): New field.
5575 * read-rtl-function.c: New file.
5576 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
5577 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
5578 (selftest::verify_three_block_rtl_cfg): New decl.
5579 * read-rtl-function.h: New file.
5580 * read-rtl.c: Potentially include config.h rather than bconfig.h.
5581 For host, include function.h, memmodel.h, and emit-rtl.h.
5582 (one_time_initialization): New function.
5583 (struct compact_insn_name): New struct.
5584 (compact_insn_names): New array.
5585 (find_code): Handle insn codes in compact dumps.
5586 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
5587 (bind_subst_iter_and_attr): Likewise.
5588 (add_condition_to_string): Likewise.
5589 (add_condition_to_rtx): Likewise.
5590 (apply_attribute_uses): Likewise.
5591 (add_current_iterators): Likewise.
5592 (apply_iterators): Likewise.
5593 (initialize_iterators): Guard usage of apply_subst_iterator with
5594 #ifdef GENERATOR_FILE.
5595 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
5596 (md_reader::read_mapping): Likewise.
5597 (add_define_attr_for_define_subst): Likewise.
5598 (add_define_subst_attr): Likewise.
5599 (read_subst_mapping): Likewise.
5600 (check_code_iterator): Likewise.
5601 (rtx_reader::read_rtx): Likewise. Move one-time initialization
5603 (one_time_initialization): New function.
5604 (rtx_reader::read_until): New method.
5605 (read_flags): New function.
5606 (parse_reg_note_name): New function.
5607 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
5608 Handle reuse_rtx ids.
5609 Wrap iterator lookup within #ifdef GENERATOR_FILE.
5610 Add parsing support for RTL dumps, mirroring the special-cases in
5611 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
5612 values, and calling handle_any_trailing_information.
5613 (rtx_reader::read_rtx_operand): Convert return type from void
5614 to rtx, returning return_rtx. Handle case 'e'. Call
5615 finalize_string on XSTR and XTMPL fields.
5616 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
5617 "(nil)" values were omitted. Call the postprocess vfunc on the
5619 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
5620 class ctor. Initialize m_in_call_function_usage. Call
5621 one_time_initialization.
5622 * rtl-tests.c (selftest::test_uncond_jump): Call
5623 set_new_first_and_last_insn.
5624 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
5625 * selftest-rtl.c: New file.
5626 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
5627 (selftest::get_insn_by_uid): New decl.
5628 * selftest-run-tests.c (selftest::run_tests): Call
5629 read_rtl_function_c_tests.
5630 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
5631 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
5634 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
5636 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
5637 operands in a special way. Assert that pos+len <= mode precision.
5639 2017-01-05 Jakub Jelinek <jakub@redhat.com>
5641 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
5642 3 argument Alias with unlimited for the negative form.
5643 (fno-vect-cost-model): Removed.
5645 2017-01-05 Martin Liska <mliska@suse.cz>
5647 * hsa-gen.c (gen_hsa_divmod): New function.
5648 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
5650 2017-01-05 Martin Liska <mliska@suse.cz>
5653 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
5656 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5658 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
5659 small constant length operands.
5661 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5663 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
5664 between loop iterations.
5666 2017-01-05 Martin Liska <mliska@suse.cz>
5669 * gimplify.c (gimplify_decl_expr): Compare to
5670 asan_poisoned_variables instread of checking flags.
5671 (gimplify_target_expr): Likewise.
5672 (gimplify_expr): Likewise.
5673 (gimplify_function_tree): Conditionally initialize
5674 asan_poisoned_variables.
5676 2017-01-04 Jeff Law <law@redhat.com>
5678 PR tree-optimizatin/78812
5679 * rtl.h (contains_mem_rtx_p): Prototype.
5680 * ifcvt.c (containts_mem_rtx_p): Move from here to...
5681 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
5682 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
5683 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
5684 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
5686 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5688 * input.c (assert_char_at_range): Default-initialize actual_range.
5690 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5692 * df-scan.c (df_ref_create_structure): Make regno unsigned,
5693 to match the caller.
5695 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5697 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
5698 insns after final jump in test to emit dummy move.
5700 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5702 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
5703 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
5705 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5707 * multiple_target.c (create_dispatcher_calls): Init e_next.
5708 * tree-ssa-loop-split.c (split_loop): Init border.
5709 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
5712 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
5717 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
5718 (altivec_register_operand): Do not return true if the operand
5719 contains a SUBREG mixing SImode and SFmode.
5720 (vsx_register_operand): Likewise.
5721 (vsx_reg_sfsubreg_ok): New predicate.
5722 (vfloat_operand): Do not return true if the operand contains a
5723 SUBREG mixing SImode and SFmode.
5724 (vint_operand): Likewise.
5725 (vlogical_operand): Likewise.
5726 (gpc_reg_operand): Likewise.
5727 (int_reg_operand): Likewise.
5728 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
5729 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
5730 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
5732 (rs6000_emit_move_si_sf_subreg): New helper function.
5733 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
5734 fixup SUBREGs involving SImode and SFmode.
5735 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
5736 numbers for the new peephole2 optimization.
5737 (peephole2 for SFmode unions): New peephole2 to optimize cases in
5738 the GLIBC math library that do AND/IOR/XOR operations on single
5739 precision floating point.
5740 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
5741 target macros to say whether we need to avoid SUBREGs mixing
5743 (TARGET_ALLOW_SF_SUBREG): Likewise.
5744 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
5745 (UNSPEC_SI_FROM_SF): Likewise.
5746 (iorxor): Change spacing.
5747 (and_ior_xor): New iterator for AND, IOR, and XOR.
5748 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
5749 (movdi_from_sf_zero_ext): Likewise.
5750 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
5751 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
5752 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
5753 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
5754 (fms<mode>4): Likewise.
5755 (fnma<mode>4): Likewise.
5756 (fnms<mode>4): Likewise.
5757 (nfma<mode>4): Likewise.
5758 (nfms<mode>4): Likewise.
5760 2017-01-04 Marek Polacek <polacek@redhat.com>
5763 * doc/invoke.texi: Document -Wpointer-compare.
5765 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5767 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
5770 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
5771 descriptions for -gdwarf-5 and emit them as uleb128 instead of
5774 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
5777 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
5778 documentation of the powerpc_popcntb_ok attribute.
5779 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
5780 code to issue warning messages if a requested CPU configuration is
5781 not supported by the binary (assembler and loader) toolchain.
5782 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
5783 made to define a built-in function that has been disabled.
5784 (paired_init_builtins): Add assertion to prevent ICE if attempt is
5785 made to define a built-in function that has been disabled.
5786 (altivec_init_builtins): Add comment explaining why definition
5787 of the DST built-in functions is not preceded by an assertion
5788 check. Add assertions to prevent ICE if attempts are made to
5789 define an altivec predicate or an abs* built-in function that has
5791 (htm_init_builtins): Add comment explaining why definition of the
5792 htm built-in functions is not preceded by an assertion check.
5794 2017-01-04 Jeff Law <law@redhat.com>
5796 PR tree-optimizatin/67955
5797 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
5798 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
5799 the points-to solution does not include pt_null. Use DECL_PT_UID
5802 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
5804 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
5805 Use gen_int_mode instead of gen_lopwart for const_int operands.
5807 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5809 PR tree-optimization/71563
5810 * match.pd: Simplify X << Y into X if Y is known to be 0 or
5811 out of range value - has low bits known to be zero.
5813 2017-01-04 Alan Modra <amodra@gmail.com>
5815 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
5816 * configure: Regenerate.
5817 * config.in: Regenerate.
5819 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5822 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
5823 a substring of the message, but strcmp with the whole message. Ifdef
5824 ENABLE_NLS, translate the message first using dgettext.
5826 2017-01-03 Jeff Law <law@redhat.com>
5828 PR tree-optimizatin/78856
5829 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
5830 (mark_threaded_blocks): Remove code to truncate thread paths that
5831 cross multiple loop headers. Instead invalidate the cached loop
5832 iteration information and handle case of a thread path walking
5833 into an irreducible region.
5835 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5838 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
5839 assertions. Add support for doing the signbit if the IEEE 128-bit
5840 floating point value is in a GPR.
5841 * config/rs6000/rs6000.md (Fsignbit): Delete.
5842 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
5843 Update the length attribute if the value is in a GPR.
5844 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
5845 the sign or zero extension instruction, since the value is always 0/1.
5846 (signbit<mode>2_dm2): Delete using <Fsignbit>.
5849 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
5850 extracting SImode to a GPR register so that we can generate a
5851 store, limit the vector to be in a traditional Altivec register
5852 for the vextuwrx instruction.
5854 2017-01-03 Ian Lance Taylor <iant@google.com>
5856 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
5858 2017-01-03 Martin Sebor <msebor@redhat.com>
5860 PR tree-optimization/78696
5861 * gimple-ssa-sprintf.c (format_floating): Correct handling of
5862 precision. Use MPFR for %f for greater fidelity. Correct handling
5864 (pass_sprintf_length::compute_format_length): Set width and precision
5865 specified by asrerisk to void_node for vararg functions.
5866 (try_substitute_return_value): Adjust dump output.
5868 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
5870 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
5872 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
5874 * doc/invoke.texi (SPARC options): Document -mlra as the default.
5875 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
5876 -mlra/-mno-lra was passed to the compiler.
5878 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
5880 PR rtl-optimization/65618
5881 * emit-rtl.c (try_split): Move initialization of "before" and
5882 "after" to just before the call to emit_insn_after_setloc.
5884 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
5886 * doc/md.texi (Standard Names): Remove reference to Java frontend.
5888 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
5890 * dwarf2out.c (gen_enumeration_type_die): When
5891 -gno-strict-dwarf, add a DW_AT_encoding attribute.
5893 2017-01-03 Jakub Jelinek <jakub@redhat.com>
5895 PR tree-optimization/78965
5896 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
5897 Change first argument from const call_info & to call_info &. For %n
5898 set info.nowrite to false.
5901 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
5902 possibly throwing calls.
5904 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
5905 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
5906 and fns handling, rather than in a separate case SSA_NAME.
5908 2017-01-02 Jeff Law <law@redhat.com>
5910 * config/darwin-driver.c (darwin_driver_init): Const-correctness
5911 fixes for first_period and second_period variables.
5913 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
5916 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
5917 (*insvqi_1): New insn pattern.
5918 (*insvqi_1_mem_rex64): Ditto.
5920 (*insvqi_3): Rename from *insvqi.
5922 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
5924 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
5926 * doc/cfg.texi (Edges): Remove reference to Java.
5927 (Maintaining the CFG): Ditto.
5929 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5932 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
5933 transparent aliases.
5935 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5938 * predict.def (PRED_CALL): Update hitrate.
5939 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
5940 * predict.c (tree_estimate_probability_bb): Split CALL predictor
5941 into direct/indirect/polymorphic variants.
5943 2017-01-01 Jakub Jelinek <jakub@redhat.com>
5945 Update copyright years.
5947 * gcc.c (process_command): Update copyright notice dates.
5948 * gcov-dump.c (print_version): Ditto.
5949 * gcov.c (print_version): Ditto.
5950 * gcov-tool.c (print_version): Ditto.
5951 * gengtype.c (create_file): Ditto.
5952 * doc/cpp.texi: Bump @copying's copyright year.
5953 * doc/cppinternals.texi: Ditto.
5954 * doc/gcc.texi: Ditto.
5955 * doc/gccint.texi: Ditto.
5956 * doc/gcov.texi: Ditto.
5957 * doc/install.texi: Ditto.
5958 * doc/invoke.texi: Ditto.
5960 Copyright (C) 2017 Free Software Foundation, Inc.
5962 Copying and distribution of this file, with or without modification,
5963 are permitted in any medium without royalty provided the copyright
5964 notice and this notice are preserved.